From majordom@nokomis.stat.umn.edu Tue Sep 12 12:52:00 1995 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA16395; Tue, 12 Sep 1995 12:52:00 -0400 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA16389; Tue, 12 Sep 1995 11:51:59 -0500 Date: Tue, 12 Sep 1995 11:51:59 -0500 From: Luke Tierney Message-Id: <9509121651.AA16389@nokomis.stat.umn.edu> Apparently-To: stat-lisp-news@nokomis.stat.umn.edu Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk a test From majordom@umnstat.stat.umn.edu Tue Sep 12 15:08:19 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA01677; Tue, 12 Sep 1995 15:08:19 -0400 Return-Path: Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA01671; Tue, 12 Sep 1995 14:08:16 -0500 From: Luke Tierney Message-Id: <9509121908.AA01671@umnstat.stat.umn.edu> Subject: change in list management To: stat-lisp-news@umnstat.stat.umn.edu Date: Tue, 12 Sep 1995 14:08:15 -0500 (CDT) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 2191 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I am in the process of switching this list over to use the majordomo list management software. Hopefully this will go smoothly and I won't need to send any more messages about it. It not, ... The change should not affect current list members. It does affect how you leave the list or join it. Here are the new instructions: --------------------------------------------------------------------------- All routine administrative requests (including subscriptions and unsubscriptions) concerning this mailing list are handled by an automated server. Please read this message carefully to find the information relevant to you. SUBSCRIBING =========== To subscribe to stat-lisp-news, send the following in the body (not the subject line) of an email message to "Majordomo@stat.umn.edu": subscribe stat-lisp-news This will subscribe the account from which you send the message to the stat-lisp-news list. If you wish to subscribe another address instead (such as a local redistribution list), you can use a command of the form: subscribe stat-lisp-news other-address@your_site.your_net UNSUBSCRIBING ============= To unsubscribe from stat-lisp-news, send the following in the body (not the subject line) of an email message to "Majordomo@stat.umn.edu": unsubscribe stat-lisp-news This will unsubscribe the account from which you send the message. If you are subscribed with some other address, you'll have to send a command of the following form instead: unsubscribe stat-lisp-news other-address@your_site.your_net If you don't know what address you are subscribed with, you can send the following command to see who else is on the list (assuming that information isn't designated "private" by the owner of the list):o who stat-lisp-news If you want to search non-privte lists at this server, you can do that by sending a command like: which string This will return a list of all entries on all lists that contain "string". HELP ==== To find out more about the automated server and the commands it understands, send the following command to "Majordomo@stat.umn.edu": help If you feel you need to reach a human, send email to stat-lisp-news-approval@stat.umn.edu From majordom@umnstat.stat.umn.edu Tue Sep 12 17:59:59 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02534; Tue, 12 Sep 1995 17:59:59 -0400 Return-Path: Received: from mathfs.math.montana.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02529; Tue, 12 Sep 1995 16:59:55 -0500 Received: by mathfs.math.montana.edu; (5.65/1.1.8.2/25Jan95-8.2MPM) id AA28742; Tue, 12 Sep 1995 15:59:07 -0600 Date: Tue, 12 Sep 1995 15:59:07 -0600 (MDT) From: Jim Robison-Cox To: lisp-stat-news@umnstat.stat.umn.edu Subject: instal problem on decstation Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'm having trouble getting xlispstat installed on a Decstation 5200 running Ultrix 4.4 and X11R5. Wondered if anyone has seen this one, it might be a problem in the X11 libraries? Here's what I get from make: > /psoft/Gnu-dev/bin/gcc -o X11WINDOWSxlisp xsdynload.o xssystem.o > unixprim.o xlarray.o xlbfun.o xlcont.o xldbug.o xldmem.o xleval.o xlfio.o > xlftab.o xlglob.o xlimage.o xlinit.o xlio.o xlisp.o xljump.o xllist.o > xlpp.o xlprin.o xlrand.o xlread.o xlseq.o xlstr.o xlsubr.o xlsym.o > xlsys.o xlstruct.o xlmath2.o xltvec.o xlbcode.o xlbcutil.o xlmodule.o > dummy.o basics.o common.o compound.o dists.o ddists.o hrdwrobs.o mats1.o > mats2.o math.o obinit.o objects.o optimize.o sortdata.o statinit.o > stats.o utils.o utils2.o gamln.o splines.o kernel.o lowess.o cfft.o > betab.o gammab.o nor.o ppnd.o studentb.o bivnor.o linalg.o cholesky.o > ludecomp.o qrdecomp.o rcondest.o svdecomp.o makerot.o eigen.o minimize.o > blas.o xsiview2.o xsivint.o xsnewplt.o xsgraph.o xsiview3.o xsivwin.o > xsiview.o xsivwin2.o menus.o dialogs.o iviewdat.o iviewscl.o iview.o > iviewint.o xshist.o xsnames.o xssctmat.o xsspin.o xssctplt.o windows.o > stmem.o graphics.o myplot.o term.o postscript.o X11BSDstuff.o X11graph.o > X11menus.o X11dialogs.o X11buttons.o X11text.o X11toggle.o X11choice.o > X11slider.o X11listitem.o X11resizebr.o X11scroll.o \ > -L/psoft/X11R5-mit/lib -lX11 -lm and then error message: > XConnDis.o(.text+0x684): undefined reference to `dnet_conn' > XConnDis.o(.text+0x6ac): undefined reference to `dnet_addr' > XConnDis.o(.text+0x740): undefined reference to `getnodebyname' Thanks, Jim Robison-Cox e-mail: jimrc@math.montana.edu Department of Mathematical Sciences 2-214 Wilson Hall Montana State University Bozeman, MT 59717 Home: 406 587-4977 Work: 406 994-5340 From majordom@umnstat.stat.umn.edu Wed Sep 13 08:10:16 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13072; Wed, 13 Sep 1995 08:10:16 -0400 Return-Path: Received: from lhc.nlm.nih.gov by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13067; Wed, 13 Sep 1995 07:10:12 -0500 Received: from play.nlm.nih.gov by lhc.nlm.nih.gov (4.1/SMI-4.0) id AA21213; Wed, 13 Sep 95 08:10:11 EDT Received: by play.nlm.nih.gov (940816.SGI.8.6.9/5.6) id IAA12782; Wed, 13 Sep 1995 08:16:18 -0400 Date: Wed, 13 Sep 1995 08:16:18 -0400 From: mabramson@nlm.nih.gov (Myriam Abramson) Message-Id: <199509131216.IAA12782@play.nlm.nih.gov> To: stat-lisp-news@umnstat.stat.umn.edu Subject: manova Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Thanks to all who responded to my inquiry about running xlispstat in batch. Now, to get this project off the ground, can anybody point me to some code to do MANOVA (I am not much of a statistician)? Thanks, myriam From majordom@umnstat.stat.umn.edu Thu Sep 14 04:04:21 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26733; Thu, 14 Sep 1995 04:04:21 -0400 Return-Path: Received: from pollux.edv.agrar.tu-muenchen.de by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26728; Thu, 14 Sep 1995 03:04:06 -0500 Received: by pollux.edv.agrar.tu-muenchen.de (AIX 3.2/UCB 5.64/4.03) id AA24546; Thu, 14 Sep 1995 10:03:13 +0100 From: walter@pollux.edv.agrar.tu-muenchen.de (Bernhard Walter (DVS)) Message-Id: <9509140903.AA24546@pollux.edv.agrar.tu-muenchen.de> Subject: Regression without intercept To: stat-lisp-news@umnstat.stat.umn.edu (XLispStat-List) Date: Thu, 14 Sep 1995 10:03:12 +0100 (GDT) X-Mailer: ELM [version 2.4 PL24alpha3] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 5547 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi all, Looks like XLispStat's regression-model-proto allows regression without the intercept, but doesn't redefine the R-squared to an uncentered one. Here is a little example: > (print-matrix (bind-columns x1 x3)) #2a( ( 9.43474 10.6733 ) ( 9.00556 12.4143 ) ( 10.4510 9.25088 ) ( 10.3146 9.35618 ) ( 9.36443 11.1377 ) ( 10.7270 8.76072 ) ( 11.5927 6.53162 ) ( 10.1615 9.46036 ) ( 8.86543 12.3553 ) ( 9.97447 10.3976 ) ( 8.67871 12.3337 ) ( 9.14190 11.3969 ) ( 8.86809 12.2169 ) ( 9.80247 9.80596 ) ( 10.8986 8.06527 ) ( 9.02447 12.2756 ) ( 10.3523 8.83951 ) ( 8.91739 12.5908 ) ( 8.02686 13.8066 ) ( 8.32836 13.4595 ) ) > (regression-model x3 x1 :predictor-names '("x3") :response-name "x1" :intercept nil) Least Squares Estimates: x3 0.850682 (5.386330E-2) R Squared: 0.000000 Sigma hat: 2.63131 Number of cases: 20 Degrees of freedom: 19 Computing the square of the pearson correlation coefficient gets: > (^ (pearson-correlation x1 x3) 2) 0.9763265551511946 So one would never expect an R-squared of 0.00000. Well, SAS knows about this: The SAS System 13 09:05 Thursday, September 14, 1995 OBS X1 X3 1 9.4347 10.6733 2 9.0056 12.4143 3 10.4510 9.2509 4 10.3146 9.3562 5 9.3644 11.1377 6 10.7270 8.7607 7 11.5927 6.5316 8 10.1615 9.4604 9 8.8654 12.3553 10 9.9745 10.3976 11 8.6787 12.3337 12 9.1419 11.3969 13 8.8681 12.2169 14 9.8025 9.8060 15 10.8986 8.0653 16 9.0245 12.2756 17 10.3523 8.8395 18 8.91739 12.5908 19 8.02686 13.8066 20 8.32836 13.4595 The SAS System 15 09:05 Thursday, September 14, 1995 Model: MODEL1 NOTE: No intercept in model. R-square is redefined. Dependent Variable: X1 Analysis of Variance Sum of Mean Source DF Squares Square F Value Prob>F Model 1 1726.99913 1726.99913 249.430 0.0001 Error 19 131.55177 6.92378 U Total 20 1858.55090 Root MSE 2.63131 R-square 0.9292 Dep Mean 9.59653 Adj R-sq 0.9255 C.V. 27.41936 Parameter Standard T for H0: Variable DF Estimate Error Parameter=0 Prob > |T| X3 1 0.850682 0.05386330 15.793 0.0001 So as can be seen from the output, SAS redefines the R-squared. It should be done like this: Let M be the usual centering-matrix I - 1/n \iota \iota' , \iota is the vector of ones. Let further be y^ the fits and e the residuals y-y^. The usual variance decompostion y'y = y^' M y^ + e^'e^ only holds, when there is an intercept in the equation. If not, you'll have to use y'y = y^' y^ + e^'e^, without the centering matrix. Then the uncentered R-squared is defined as 1 - e^'e / y'y. See for example in [1], page 29. Doing this formula in XlispStat by hand results in: > (def b (sum (* (/ x3 (sum (^ x3 2))) x1))) ; little strange, but it is LS-estimate B > b 0.8506823571205249 > (def x1^ (* b x3)) X1^ > (def e (- x1 x1^)) E > (def R2 (- 1 (/ (sum (^ e 2)) (sum (^ x1 2))))) R2 > r2 0.9292180955773006 So this is exactly the result of SAS and is conform to theory. In the XLispStat regression-model R-squared is wrong. Neither in the lisp documentation nor in [2] I found something about how to create the right regression-analysis for models without intercept. Did I miss something ? [1] D.A. Belsley "Conditioning diagnostics", 1991 Wiley & Sons. [2] XLISPSTAT: A Statistical Environment Based on the XLISP Language TR 528, July 1988. best regards -- Bernhard ---------------------------------------------------------------------------- Bernhard Walter walter@pollux.edv.agrar.tu-muenchen.de Techn. Univ. Muenchen ###### ##### +49-(0)8161-71-5055 (phone) DVS - Weihenstephan # # # # # +49-(0)8161-71-4409 (fax) Hohenbachernstr. 2 # # # # # (Germany) 85350 Freising # #### # # ---------------------------------------------------------------------------- From majordom@umnstat.stat.umn.edu Thu Sep 14 10:13:05 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27364; Thu, 14 Sep 1995 10:13:05 -0400 Return-Path: Received: from io.uwinnipeg.ca by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27359; Thu, 14 Sep 1995 09:12:57 -0500 Received: from io.UWinnipeg.ca (io) by io.uwinnipeg.ca (PMDF V5.0-2 #10778) id <0DEWFPAGD00KCO@io.uwinnipeg.ca>; Thu, 14 Sep 1995 09:17 -0500 (CDT) Date: Thu, 14 Sep 1995 09:17:34 -0500 (CDT) From: Bill Simpson Subject: Re: Regression without intercept In-Reply-To: <9509140903.AA24546@pollux.edv.agrar.tu-muenchen.de> To: Bernhard Walter Cc: XLispStat-List Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Content-Transfer-Encoding: 7BIT Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi Berhard et al., First, it seems like you have found a weakness in xlisp-stat's calculation of r in the case of a no-intercept model. (I am not sure about that; I am no expert). BUT may I point out that the no intercept model is completely silly for these data?! If you do a scatterplot of x1 (y axis) as a function of x3 (x axis) you will see that there is a strong NEGATIVE correlation and that there is a LARGE positive intercept. The ordinary linear fit is: y=14.696 -.4741 x; r=-.98809 If you plot your no-intercept fit line on these data you will see that there is a miserable fit. (In fact it seems that the "alternative definition" of r must be screwy--it seems to me that xlisp-stat's answer of 0 is more sensible). I confirm your fitted slope in the no intercept model of .85068. NOTE that this slope is the wrong sign compared to the actual slope of the data points!! To me this is a good example of what can go wrong if you don't follow the EDA philosophy which xlisp-stat supports with its various plots. Best wishes, Bill Simpson From majordom@umnstat.stat.umn.edu Thu Sep 14 14:09:24 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28629; Thu, 14 Sep 1995 14:09:24 -0400 Return-Path: Received: from pollux.edv.agrar.tu-muenchen.de by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28624; Thu, 14 Sep 1995 13:09:18 -0500 Received: by pollux.edv.agrar.tu-muenchen.de (AIX 3.2/UCB 5.64/4.03) id AA28786; Thu, 14 Sep 1995 20:08:38 +0100 From: walter@pollux.edv.agrar.tu-muenchen.de (Bernhard Walter (DVS)) Message-Id: <9509141908.AA28786@pollux.edv.agrar.tu-muenchen.de> Subject: Regression without intercept, more reasonable example To: stat-lisp-news@umnstat.stat.umn.edu (XLispStat-List) Date: Thu, 14 Sep 1995 20:08:37 +0100 (GDT) X-Mailer: ELM [version 2.4 PL24alpha3] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 3005 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi all, My last posting about this subject used, as Bill Simpson pointed out, a very silly example. I'm sorry about that, but I dealt the whole day with colllinearity diagnostics and after looking at very much regression outputs I lost a little bit the eye for the whole stuff. Here is a more reasonable example now: (Well, I hope so) -------------------------------------------------------- > x (0.02994 0.25656 0.11788 0.1842 0.43233 0.4886 0.19711 0.16788 0.33304 0.50809) > y (-0.2447 1.79348 0.47305 1.06725 1.79162 0.87857 -0.37403 -1.42984 -0.20444 2.28) > (regression-model x y :intercept nil) Least Squares Estimates: Variable 0 2.75081 (0.992655) R Squared: 0.302241 Sigma hat: 0.980817 Number of cases: 10 Degrees of freedom: 9 # And this is what SAS thinks about it: ------------------------------------- Model: MODEL1 NOTE: No intercept in model. R-square is redefined. Dependent Variable: Y Analysis of Variance Sum of Mean Source DF Squares Square F Value Prob>F Model 1 7.38755 7.38755 7.679 0.0217 Error 9 8.65802 0.96200 U Total 10 16.04557 Root MSE 0.98082 R-square 0.4604 Dep Mean 0.60310 Adj R-sq 0.4005 C.V. 162.63032 So the XLispStat formula seems to be another than the SAS formula. And the formula for uncentered R-square computed by hand in Xlispstat shows the same value for R-squared as in SAS: > (def y^ (* 2.75081 x)) Y^ > (def e (- y y^)) E > (- 1 (/ (sum (^ e 2)) (sum (^ yy 2)))) 0.4604130049448324 Even if the reasonability of R-squared may be questionable for itself when dealing with non-intercept models, it would be nice to have comparable results for reasonable datasets. Or, is still anything wrong ? Is there any literature regarding reasonability of R-squared when the intercept is omitted ? Any help is appreciated, with best regards -- Bernhard BTW, usually I still prefer regression with intercept if there isn't any good reason to leave it away. I just found this by chance and didn't understand the difference. Unfortunatly my first example was silly ;-) -- ---------------------------------------------------------------------------- Bernhard Walter walter@pollux.edv.agrar.tu-muenchen.de Techn. Univ. Muenchen ###### ##### +49-(0)8161-71-5055 (phone) DVS - Weihenstephan # # # # # +49-(0)8161-71-4409 (fax) Hohenbachernstr. 2 # # # # # (Germany) 85350 Freising # #### # # ---------------------------------------------------------------------------- From majordom@nokomis.stat.umn.edu Fri Sep 15 16:12:43 1995 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA26189; Fri, 15 Sep 1995 16:12:43 -0400 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA26182; Fri, 15 Sep 1995 15:12:40 -0500 From: Luke Tierney Message-Id: <9509152012.AA26182@nokomis.stat.umn.edu> Subject: future directions To: stat-lisp-news@nokomis.stat.umn.edu Date: Fri, 15 Sep 1995 15:12:39 -0500 (CDT) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 320 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk A technical report, "Recent Developments and Future Directions in Lisp-Stat," is available at http://stat.umn.edu/CurrentResearch/tr608.html (abstract) http://stat.umn.edu/CurrentResearch/tech-reports/tr608.ps (postscript) ftp://stat.umn.edu/pub/documents/tech-reports/tr608.ps (postscript) Comments welcome. luke From majordom@umnstat.stat.umn.edu Fri Sep 15 17:57:13 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13874; Fri, 15 Sep 1995 17:57:13 -0400 Return-Path: Received: from lhc.nlm.nih.gov by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13869; Fri, 15 Sep 1995 16:57:09 -0500 Received: from play.nlm.nih.gov by lhc.nlm.nih.gov (4.1/SMI-4.0) id AA10121; Fri, 15 Sep 95 17:57:08 EDT Received: by play.nlm.nih.gov (940816.SGI.8.6.9/5.6) id SAA17103; Fri, 15 Sep 1995 18:03:17 -0400 Date: Fri, 15 Sep 1995 18:03:17 -0400 From: mabramson@nlm.nih.gov (Myriam Abramson) Message-Id: <199509152203.SAA17103@play.nlm.nih.gov> To: stat-lisp-news@umnstat.stat.umn.edu Subject: prompt character Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Does anybody know how to change to prompt character ">"? I would really like to suppress it because I run xlispstat as a batch job and I would like to parse the output a little bit more easily. Thanks, myriam From majordom@umnstat.stat.umn.edu Fri Sep 15 18:48:29 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14055; Fri, 15 Sep 1995 18:48:29 -0400 Return-Path: Received: from cs.sfu.ca by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14050; Fri, 15 Sep 1995 17:48:25 -0500 Received: from skekoowahts (skekoowahts.math.sfu.ca) by cs.sfu.ca with SMTP id AA20803 (5.65c/IDA-1.4.4 for ); Fri, 15 Sep 1995 15:48:22 -0700 Message-Id: <199509152248.AA20803@cs.sfu.ca> Date: Fri, 15 Sep 95 15:48:23 -0700 From: Carl Schwarz Organization: Simon Fraser University X-Mailer: Mozilla 1.1N (X11; I; SunOS 5.4 sun4m) Mime-Version: 1.0 To: lisp-stat-news@umnstat.stat.umn.edu Subject: Problems with xlispstat on Sparc running Solaris X-Url: http://stat.umn.edu/~rcode/get-xlispstat.html Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=us-ascii Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I retreived and unzipped the xlispstat-3-44 files. After compiling, I get the following error message when trying to run xlispstat with one of the teaching modules from ucla. %xlispstat XLISP-PLUS version 2.1g .. XLISP-STAT 2.1 Release 3.44 (Beta) ; loading "ci.lsp" ; loading "OVL/overlay1.lsp" .. : loading "OVL/make-col.lsp" Error: can't allocate color Happened in: # (exit) I looked a the SUN Patch file, but it doesn't seem applicable. Any suggestions? Thanks. From majordom@umnstat.stat.umn.edu Fri Sep 29 11:29:09 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26897; Fri, 29 Sep 1995 11:29:09 -0400 Return-Path: Received: from triton.uqtr.uquebec.ca by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26891; Fri, 29 Sep 1995 10:29:04 -0500 Received: by triton.UQTR.UQuebec.ca (AIX 3.2/UCB 5.64/4.03) id AA07917; Fri, 29 Sep 1995 11:29:00 -0400 From: Belkacem_Abdous@uqtr.uquebec.ca Received: from bacchus.uqtr.uquebec.ca by triton.UQTR.UQuebec.ca (AIX 3.2/UCB 5.64/4.03) id AA34537; Fri, 29 Sep 1995 11:28:59 -0400 Received: by bacchus.UQTR.UQuebec.ca (5.x/SMI-SVR4) id AA00747; Fri, 29 Sep 1995 11:28:50 -0400 Date: Fri, 29 Sep 1995 11:28:50 -0400 Message-Id: <9509291528.AA00747@bacchus.UQTR.UQuebec.ca> To: stat-lisp-news@umnstat.stat.umn.edu X-Sun-Charset: US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, I am trying to use dynamic loading under Win32 (for windows 3.1). I ran into the following problem : 1) I used foo.c (Tierney (1990, p.362)). 2) I defined foo.def as suggested in the document "XLISP-STAT for Microsoft Windows 3.0 Version 2.1 Alpha 6" 3) The foo.dll was created with Borland C++ (Ver. 4.0). 4) Under Xlispstat, I obtained the following message > (setf foo-dll (load-dll "foo.dll")) 10254 > (call-cfun foo-dll "foo" 5 (float (iseq 1 5)) 0.0) Error: can't find function address - "foo" Happened in: # > Does somebody has a solution for this problem. Thank you. Belkacem Abdous Universite du Quebec a Trois-Rivieres. From majordom@umnstat.stat.umn.edu Sat Sep 30 14:44:36 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11768; Sat, 30 Sep 1995 14:44:36 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11763; Sat, 30 Sep 1995 13:44:32 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA20002; Sat, 30 Sep 95 11:44:31 PDT Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA05189; Sat, 30 Sep 95 11:44:29 PDT From: Jan Deleeuw Message-Id: <9509301844.AA05189@galton.math.ucla.edu> Subject: Style To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Sat, 30 Sep 1995 11:44:29 -0700 (PDT) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 461 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On ftp.stat.ucla.edu in pub/lisp/common/docs there is luv-slides.ps.gz, which has about 120 transparancies on Good Lisp Programming Style, written by Peter Norvig and Kent Pitman. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@nokomis.stat.umn.edu Wed Oct 11 12:00:35 1995 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA15412; Wed, 11 Oct 1995 12:00:35 -0400 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA15406; Wed, 11 Oct 1995 11:00:33 -0500 From: Luke Tierney Message-Id: <9510111600.AA15406@nokomis.stat.umn.edu> Subject: WWW page for Lisp-Stat To: stat-lisp-news@nokomis.stat.umn.edu Date: Wed, 11 Oct 1995 11:00:32 -0500 (CDT) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 371 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk I have set up a very minimal page for Lisp-Stat. It is available through the Projects & Research entry on the School of Statistics home page, http://stat.umn.edu, or directly as http://stat.umn.edu/~luke/xls/xlsinfo/xlsinfo.html (this URL may change as we reorganize things here). Let me know if you have suggestions, or if there is anything you would like to add. luke From majordom@umnstat.stat.umn.edu Tue Oct 17 07:13:47 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23814; Tue, 17 Oct 1995 07:13:47 -0400 Return-Path: Received: from mserv.rug.ac.be by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23809; Tue, 17 Oct 1995 06:12:48 -0500 Received: from hobbes.rug.ac.be by mserv.rug.ac.be with SMTP id AA27037 (5.67b/IDA-1.5 for <@mserv.rug.ac.be:stat-lisp-news@stat.umn.edu>); Tue, 17 Oct 1995 12:12:40 +0100 Received: by hobbes.rug.ac.be (940816.SGI.8.6.9/940406.SGI.950427.HV) for stat-lisp-news@stat.umn.edu id MAA27855; Tue, 17 Oct 1995 12:12:18 +0100 From: othas@hobbes.rug.ac.be (Olivier Thas) Posted-Date: Tue, 17 Oct 1995 12:12:18 +0100 Message-Id: <199510171112.MAA27855@hobbes.rug.ac.be> Subject: spatio-temporal data To: stat-lisp-news@umnstat.stat.umn.edu Date: Tue, 17 Oct 1995 12:12:18 +0100 (BDT) Reply-To: Olivier.Thas@hobbes.rug.ac.be Organization: University of Gent, BIOMATH Department X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1147 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, I just installed Xlisp-Stat and read some introductory course on it. Actually I was looking for some software package that can handle time series, spatial analysis techniques and related stuff. At this moment we only have SAS available at this site. I know I can solve a lot of my problems with SAS/IML but in the past I have worked with S-plus which has some features I am realy attracted to. Now I hope I can find some of these features back in Xlisp-Stat. But first I have some basic questions. 1. Are there allready programs available that can handle time series and spatial data (variograms, ARMA models, ...) and where can I find them? 2. What is the maximum size of data sets that can be handled with Xlisp-Stat? 3. Can I exchange data sets (raw) from SAS to Xlisp-Stat or visa versa? I appreciate any information, Regards, Olivier -- Olivier Thas University of Gent phone : +32 (9) 264 59 33 Department of Applied Mathematics fax : +32 (9) 223 49 41 Biometrics and Process Control e-mail: Olivier.Thas@rug.ac.be Coupure Links 653, 9000 Gent Belgium From majordom@umnstat.stat.umn.edu Wed Oct 18 16:52:46 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13728; Wed, 18 Oct 1995 16:52:46 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13718; Wed, 18 Oct 1995 15:52:42 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id QAA12664; Wed, 18 Oct 1995 16:52:34 -0400 Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 18 Oct 1995 16:51:00 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: uluru@gibbs.oit.unc.edu (Forrest Young) Subject: ViSta - The Visual Statistics System Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Announcing a new version of ViSta - The Visual Statistics System by Forrest W. Young FEATURES ViSta features state-of-the art visualization techniques for > visually guiding students and other novice data analysts > visualizing the overall structure of data analysis sessions > visualizing data structure and analysis results > visually re-estimated model parameters ViSta is useful for > teaching introductory statistics to undergraduates > teaching dynamic statistical graphics and computational statistics to graduate students > research and development in dynamic statistical graphics and in computational statistics ViSta runs under Microsoft Windows 3.01 and on Macs and PowerMacs. ViSta is open, extensible and is distributed at no cost. ABSTRACT ViSta is designed for an audience of users having a very wide range of data analysis sophistication, ranging from novice to expert. ViSta provides seamlessly integrated data analysis environments specifically tailored to the user's level of expertise. Guidance is available for students and novices, and tools are available for teachers and other experts to create guidance for these novices. A structured graphical user interface is available for competent users, and a command line interface is available for sophisticated users. The complete Lisp-Stat (Tierney, 1990) programming environment is included so that researchers and graduate students can extend ViSta's capabilities. NEW FEATURES in Versions 4.08 and 4.12 There have been two recent releases of ViSta. Version 4.08 (August, 1995) introduced several new features. Version 4.12 (October, 1995) fixed bugs, particularly in the Mac version. The new features include: > New datasheet editor is available for entering and editing data. > Data can be imported from text files. > Transformation menu has been completed. > Data manipulation features for matrix data have been expanded. > Double-clicking data icons now opens their datasheets. These new features are summarized below. DATASHEET EDITOR: A datasheet editor is now available. It allows you to edit existing data and to enter new data. You no longer have to prepare datafiles using lisp statements, although these previously described features are available. The datasheet is accessed through the SHOW DATASHEET item of the DATA menu, and the NEW DATA, OPEN DATA, and IMPORT DATA items of the FILE menu. The BROWSE DATA item of the DATA menu also displays data in datasheet form but does not allow editing. The datasheet supports standard editing functions. Click on a cell to edit it. Type the desired information. Use the delete key to delete what you have typed. Use the cursor, return or tab keys to move to another cell. The left column of cells contains observation labels. The top row of cells contains variable names. These can be edited like any other cells and may contain any information that you wish. The second row of cells contains variable types: Numeric, Ordinal and Category. The remaining cells of the datasheet contain the data. You can expand the datasheet (add rows, columns or matrices) and control the width of columns and the number of decimal places shown. You can redefine which variable is used to label observations. IMPORT DATA: A new FILE menu item is available for importing multivariate data. The data being imported must be in a text (ascii) file. The first line of the file specifies the data for the first observation. Succeeding lines specify the data for succeeding observations, and must contain the same number of data elements as are on the first line. Data elements must be separated by spaces. WORKMAP: Double-clicking a data icon opens the data to reveal the data's datasheet. Double-clicking a model icon opens the model to reveal the model's report. For MS-WINDOWS, you can now open a reduced (iconized) window from the icon. You can also open the window from the WINDOWS menu as well as the COMMAND window. TRANSFORMATION MENU: The transformation menu has been completed. There are 30 transformations now available, plus a user-defined menu item that gives access to additional user-defined transformations. MATRIX DATA: Data manipulation features for Matrix Data have been completed. You can now subset matrix data by matrix or by row/column, and you can merge separate matrix data objects into one. OPEN DATA, LOAD DATA and IMPORT DATA: These menu items now always open to the data folder in the Windows version. To support this, the data folder has been restructed. DATA and MODEL menus: These menus have been modified so that menu items for data and model object names are optionally appended to the end of the menus, rather than always being appended. The option is controlled by a new preference feature. This takes care of a MS-WINDOWS problem with menus being too long for the screen. BUG LIST: A BUG LIST menu item has been added to the COMMAND menu. It lists known bugs and tells you how to report unknown bugs that you discover. PREFERENCES: The initial startup PREFERENCES menu item has been expanded. ANALYSIS OF VARIANCE: SAVE MODEL and CREATE DATA menu items have been implemented. VISUALIZE MODEL is still not completed. CORRESPONDENCE ANALYSIS: Now analyzes all active numeric data rather than all numeric data. PRINCIPLE COMPONENTS: The visualization has been slightly improved. ___________________________________________________________________ Forrest Young e-mail: forrest@unc.edu UNC Psychometrics Lab WWW: http://forrest.psych.unc.edu CB-3270 Davie Hall Telephone: 919-962-5038 Chapel Hill NC 27599-3270 USA FAX: 919-962-2537 From majordom@umnstat.stat.umn.edu Wed Oct 18 16:53:35 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13736; Wed, 18 Oct 1995 16:53:35 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13731; Wed, 18 Oct 1995 15:53:31 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id QAA12734; Wed, 18 Oct 1995 16:53:29 -0400 Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 18 Oct 1995 16:51:55 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: uluru@gibbs.oit.unc.edu (Forrest Young) Subject: ViSta WWW and FTP Site Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk New World-Wide-Web site for ViSta - The Visual Statistics System You can now get ViSta, and more information about what it is and how to use it, from the new WWW site at forrest.psych.unc.edu. ViSta's home page forrest.psych.unc.edu/research/ViSta.html contains links to: 1) Descriptions of ViSta's features. 2) Freely downloadable code, testdata, examples, help, guidance. 3) Freely downloadable documentation. 4) The XLisp-Stat system needed to run ViSta (included and free). 5) Other related web sites. For those who do not have web access, you can ftp vista files directly: MACINTOSH: ViSta for Macintosh can be obtained directly from www.psych.unc.edu/pub/forrest/vista/mac. The appropriate file is one of the following: ViSta+XLSPPC.sea.hqx (for the PowerMac) ViSta+XLS881.sea.hqx (for Macs with math coprocessor) ViSta+XLS020.sea.hqx (for Macs without) Documentation for the Mac is in file vistadoc.sea.hqx from www.psych.unc.edu/pub/forrest/vista/mac WINDOWS: ViSta for Windows 3.1 with 32-bit extension can be obtained directly by ftp'ing the file vista32.zip from www.psych.unc.edu/pub/forrest/vista/mswindows. Currently, documentation for this version is in www.psych.unc.edu/pub/forrest/vista/ There are five files named cover.ps, contents.ps, chapter1.ps, chapter2.ps, and refernce.ps. The documentation will be in a zip file in the near future. Check for www.psych.unc.edu/pub/forrest/vista/mswindows/vistadoc.zip ___________________________________________________________________ Forrest Young e-mail: forrest@unc.edu UNC Psychometrics Lab WWW: http://forrest.psych.unc.edu CB-3270 Davie Hall Telephone: 919-962-5038 Chapel Hill NC 27599-3270 USA FAX: 919-962-2537 From majordom@umnstat.stat.umn.edu Thu Oct 19 04:42:21 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25901; Thu, 19 Oct 1995 04:42:21 -0400 Return-Path: Received: from hypatia-appli.ethz.ch by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25896; Thu, 19 Oct 1995 03:42:15 -0500 Received: from (maechler@sophie [129.132.58.2]) by hypatia.math.ethz.ch (8.6.12/Main-STAT-mailer) with SMTP id JAA07010; Thu, 19 Oct 1995 09:41:56 +0100 Received: by (4.1/D-MATH-stat-client) id AA02938 for stat-lisp-news@umnstat.stat.umn.edu; Thu, 19 Oct 95 09:42:36 +0100 Date: Thu, 19 Oct 95 09:42:36 +0100 Message-Id: <9510190842.AA02938@> From: Martin Maechler To: uluru@gibbs.oit.unc.edu Cc: Lisp-Stat Mailinglist In-Reply-To: (uluru@gibbs.oit.unc.edu) Subject: Re: ViSta - The Visual Statistics System : and Unix ? Reply-To: Martin Maechler Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I am still very interested; we almost only have Unix workstations (with lisp-stat (and MUCH more) installed) Why is ViSta not for Unix (X11)? I always thought that the underlying x-lisp-stat is completely portable between the 3 platforms -- including the dynamic graphics, mouse buttons,..? am I wrong? From majordom@umnstat.stat.umn.edu Thu Oct 19 04:56:46 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25938; Thu, 19 Oct 1995 04:56:46 -0400 Return-Path: Received: from uvapsy.psy.uva.nl by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25929; Thu, 19 Oct 1995 03:56:42 -0500 Received: from macmail.psy.uva.nl by uvapsy.psy.uva.nl with SMTP id AA13405 (5.67b/IDA-1.5 for ); Thu, 19 Oct 1995 09:56:29 +0100 Message-Id: Date: 19 Oct 1995 09:55:14 +0100 From: "pn Schooler" Subject: Common Lisp Version To: "stat-lisp" X-Mailer: Mail*Link SMTP-MS 3.0.1 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I remember a while back some mention of a common lisp version of xlispstat. Can anyone tell me where that stands? thanks, Lael From majordom@umnstat.stat.umn.edu Thu Oct 19 09:56:14 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26610; Thu, 19 Oct 1995 09:56:14 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26605; Thu, 19 Oct 1995 08:56:11 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id JAA29017; Thu, 19 Oct 1995 09:56:01 -0400 Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 19 Oct 1995 09:54:25 -0500 To: edstat-l@jse.stat.ncsu.edu, stat-lisp-news@umnstat.stat.umn.edu From: uluru@gibbs.oit.unc.edu (Forrest Young) Subject: ViSta Documentation zip file Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk ViSta - The Visual Statistics System There is now a zipped version of ViSta's documentation that can be downloaded from ViSta's home page at http://forrest.psych.unc.edu/research/ViSta.html or by ftp'ing file vistadoc.zip from www.psych.unc.edu/pub/forrest/vista/mswindows ___________________________________________________________________ Forrest Young e-mail: forrest@unc.edu UNC Psychometrics Lab WWW: http://forrest.psych.unc.edu CB-3270 Davie Hall Telephone: 919-962-5038 Chapel Hill NC 27599-3270 USA FAX: 919-962-2537 From majordom@umnstat.stat.umn.edu Fri Oct 20 13:35:02 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13344; Fri, 20 Oct 1995 13:35:02 -0400 Return-Path: Received: from mserv.rug.ac.be by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13338; Fri, 20 Oct 1995 12:34:57 -0500 Received: from hobbes.rug.ac.be by mserv.rug.ac.be with SMTP id AA12655 (5.67b/IDA-1.5 for <@mserv.rug.ac.be:stat-lisp-news@stat.umn.edu>); Fri, 20 Oct 1995 18:34:54 +0100 Received: by hobbes.rug.ac.be (940816.SGI.8.6.9/940406.SGI.950427.HV) for stat-lisp-news@stat.umn.edu id SAA16041; Fri, 20 Oct 1995 18:34:26 +0100 From: othas@hobbes.rug.ac.be (Olivier Thas) Posted-Date: Fri, 20 Oct 1995 18:34:26 +0100 Message-Id: <199510201734.SAA16041@hobbes.rug.ac.be> Subject: SAPA on Xlisp-Stat To: stat-lisp-news@umnstat.stat.umn.edu Date: Fri, 20 Oct 1995 18:34:26 +0100 (BDT) Reply-To: Olivier.Thas@hobbes.rug.ac.be Organization: University of Gent, BIOMATH Department X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 686 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, I installed Xlisp-Stat because there are some advanced tools available for time series analysis and spatial data analysis, and because of the flexibility of the system. I downloaded the SAPA-files, dt4ets and the times-files. Does anyone have any documentation on these programs? Regards, Olivier -- Olivier Thas University of Gent phone : +32 (9) 264 59 33 Department of Applied Mathematics fax : +32 (9) 223 49 41 Biometrics and Process Control e-mail: Olivier.Thas@rug.ac.be Coupure Links 653, 9000 Gent Belgium -- Olivier Thas University of Gent From majordom@umnstat.stat.umn.edu Sun Oct 22 20:21:28 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10040; Sun, 22 Oct 1995 20:21:28 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10035; Sun, 22 Oct 1995 19:21:23 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA00663; Sun, 22 Oct 95 17:21:21 PDT Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA00371; Sun, 22 Oct 95 17:21:19 PDT From: Jan Deleeuw Message-Id: <9510230021.AA00371@galton.math.ucla.edu> Subject: Xlisp-Stat code archive To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Sun, 22 Oct 1995 17:21:19 -0700 (PDT) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 820 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk In order to economize on disk space, and more importantly on time, we will make the Xlisp-Stat code archive a proper WWW hierarchy, where "other people's code" will come from "other people's machines" as much as possible. ViSta is the first example. It sits physically at UNC, which is where it should be, and which is where it is maintained and upgraded. There are some minor problems with this (such as the fact that my server tries to execute lsp files, instead of downloading or displaying them) but they can be solved rather easily. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon Oct 23 18:26:03 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25387; Mon, 23 Oct 1995 18:26:03 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25381; Mon, 23 Oct 1995 17:25:59 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA29646; Mon, 23 Oct 95 15:25:57 PDT Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA06317; Mon, 23 Oct 95 15:25:56 PDT From: Jan Deleeuw Message-Id: <9510232225.AA06317@galton.math.ucla.edu> Subject: added scheme to lisp-web To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Mon, 23 Oct 1995 15:25:56 -0700 (PDT) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 642 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Although this is not strictly relevant for the newsgroup here, I thought it was convenient to add (mostly because of the Grow/Guile/Java developments at GNU, Cygnus, and Sun) information on Scheme to my WWW pages. See http://www.stat.ucla.edu/develop/lisp/ Also see http://www.cygnus.com/~tiemann/grow/ on exciting guile-based developments. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Tue Oct 24 00:33:53 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27679; Tue, 24 Oct 1995 00:33:53 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27674; Mon, 23 Oct 1995 23:33:50 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA06200; Mon, 23 Oct 95 21:33:48 PDT Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA06934; Mon, 23 Oct 95 21:33:47 PDT From: Jan Deleeuw Message-Id: <9510240433.AA06934@galton.math.ucla.edu> Subject: hypertext for Xlisp2.1g manual To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Mon, 23 Oct 1995 21:33:47 -0700 (PDT) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 692 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I happened to come accross a version of Tom Almy/David Betz the Xlisp 2.1g manual and language definition in html (made with webmaker). The appropriate link is used in http://www.stat.ucla.edu/develop/lisp/ The orginal files (on CERN) consist of about 100 pieces of html, and they are not reachable by ftp, nor is the directory in which they sit browsable. So I didnt copy them to our server. --- Jan -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Wed Oct 25 11:19:38 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24503; Wed, 25 Oct 1995 11:19:38 -0400 Return-Path: Received: from mserv.rug.ac.be by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24493; Wed, 25 Oct 1995 10:19:21 -0500 Received: from hobbes.rug.ac.be by mserv.rug.ac.be with SMTP id AA27066 (5.67b/IDA-1.5 for <@mserv.rug.ac.be:stat-lisp-news@stat.umn.edu>); Wed, 25 Oct 1995 16:19:03 +0100 Received: by hobbes.rug.ac.be (940816.SGI.8.6.9/940406.SGI.950427.HV) for stat-lisp-news@stat.umn.edu id QAA08702; Wed, 25 Oct 1995 16:18:41 +0100 From: othas@hobbes.rug.ac.be (Olivier Thas) Posted-Date: Wed, 25 Oct 1995 16:18:41 +0100 Message-Id: <199510251518.QAA08702@hobbes.rug.ac.be> Subject: time series To: stat-lisp-news@umnstat.stat.umn.edu Date: Wed, 25 Oct 1995 16:18:40 +0100 (BDT) Reply-To: Olivier.Thas@hobbes.rug.ac.be Organization: University of Gent, BIOMATH Department X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 757 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, I am using XlispStat for analyzing time series. From the UCLA server I've downloaded some utilities: time and SAPA. However, the latter is developed for common-lisp and not for Xlisp-Stat which I am using. A lot of errors occur; some of them I already solved, but they keep comming. Does someone has any experience with the SAPA package, or - in general - does someone have some more information on time series analysis using XlispStat. Thanks, Olivier -- Olivier Thas University of Gent phone : +32 (9) 264 59 33 Department of Applied Mathematics fax : +32 (9) 223 49 41 Biometrics and Process Control e-mail: Olivier.Thas@rug.ac.be Coupure Links 653, 9000 Gent Belgium From majordom@umnstat.stat.umn.edu Fri Oct 27 15:25:27 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00111; Fri, 27 Oct 1995 15:25:27 -0400 Return-Path: Received: from comtch.iea.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00104; Fri, 27 Oct 1995 14:25:18 -0500 Received: by comtch.iea.com (8.6.10/SMI-4.1) id TAA00925; Fri, 27 Oct 1995 19:25:08 GMT Date: Fri, 27 Oct 1995 12:25:08 -0700 (PDT) From: Rod Anderson X-Sender: andersr@comtch To: Lisp Stat News Subject: ViSta Questions? Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Is there a ViSta mailing list or would it be OK to post ViSta type questions to this one? Rod From majordom@umnstat.stat.umn.edu Fri Nov 3 13:31:26 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13765; Fri, 3 Nov 1995 13:31:26 -0500 Return-Path: Received: from ribana.utia.cas.cz by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13739; Fri, 3 Nov 1995 12:28:55 -0600 Received: from gauss.utia.cas.cz by ribana.utia.cas.cz via SMTP (950215.SGI.8.6.10/940406.SGI) for id TAA01637; Fri, 3 Nov 1995 19:33:31 +0100 Date: Fri, 3 Nov 1995 19:29:54 +0100 (MET) From: Rudolf Kulhavy To: stat-lisp-news@umnstat.stat.umn.edu Subject: Bivariate histogram? Message-Id: Organization: Institute of Information Theory and Automation (UTIA AV CR) Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi everybody, Did someone write an XLISP-STAT function for plotting bivariate histograms (such as in Silverman's book on Density Estimation)? I tried Matlab 4.2c mesh/surf functions but they seem to be buggy (lines that should be hidden are visible in some views). Ruda Kulhavy --------------------------------------------------------- Mail: Institute of Information Theory and Automation Academy of Sciences of the Czech Republic P.O. Box 18, 182 08 Prague, Czech Republic E-mail: kulhavy@utia.cas.cz Tel: +42 2 6605 2313 Fax: +42 2 6641 4903 WWW: ftp://ftp.utia.cas.cz/pub/staff/kulhavy/home.html --------------------------------------------------------- From majordom@umnstat.stat.umn.edu Mon Nov 6 16:02:37 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23524; Mon, 6 Nov 1995 16:02:37 -0500 Return-Path: Received: from gold.tc.umn.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23519; Mon, 6 Nov 1995 15:02:34 -0600 Received: by gold.tc.umn.edu; Mon, 6 Nov 95 15:01:52 -0600 Date: Mon, 6 Nov 1995 15:01:51 -0600 (CST) From: Jan E Backlund Subject: xlispstat and tail-recursion To: stat-lisp-news@umnstat.stat.umn.edu Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I am trying to understand how xlispstat handles tail-recursion. In some cases, at least, it appears that the interpreter behaves in a properly tail-recusive manner (ie., occupying a constant memory space) with byte-compiled tail-recursive calls (an example is appended). What rules, if any, determine whether the byte-compiler will translate tail-recursive calls into iterative processes? Also, are there any functions available for monitoring the argument stack? Thanks Jan Erik Backlund back0003@gold.tc.umn.edu > (defun tail-count (n) (defun count-iter (current) (if (> current n) current (count-iter (+ current 1)))) (count-iter 1)) TAIL-COUNT > (tail-count 100) 101 > (tail-count 1000) error: argument stack overflow > (compile 'tail-count) TAIL-COUNT > (tail-count 1000) 1001 > (tail-count 1000000) 1000001 > (time (tail-count 1000000)) The evaluation took 36.30 seconds; 0.02 seconds in gc. 1000001 > From majordom@nokomis.stat.umn.edu Tue Nov 7 10:58:22 1995 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA18297; Tue, 7 Nov 1995 10:58:22 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA18291; Tue, 7 Nov 1995 09:57:51 -0600 From: Luke Tierney Message-Id: <9511071557.AA18291@nokomis.stat.umn.edu> Subject: Re: xlispstat and tail-recursion To: back0003@gold.tc.umn.edu (Jan E Backlund) Date: Tue, 7 Nov 1995 09:57:49 -0600 (CST) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Jan E Backlund" at Nov 6, 95 03:01:51 pm X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 758 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > > I am trying to understand how xlispstat handles tail-recursion. In some > cases, at least, it appears that the interpreter behaves in a properly > tail-recusive manner (ie., occupying a constant memory space) with > byte-compiled tail-recursive calls (an example is appended). What rules, > if any, determine whether the byte-compiler will translate tail-recursive > calls into iterative processes? > The interpreter is not tail-recusrive. Byte-compiled code is (at least supposed to be) tail-recursive. > Also, are there any functions available for monitoring the argument > stack? > Not really. Not likely to be useful inthe long run anyway since the may not be an argument stack as such in future releases. Hope that helps. luke From majordom@umnstat.stat.umn.edu Tue Nov 7 13:01:49 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA07648; Tue, 7 Nov 1995 13:01:49 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA07643; Tue, 7 Nov 1995 12:01:43 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA28775; Tue, 7 Nov 95 10:01:34 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA25850; Tue, 7 Nov 95 10:01:34 PST From: Jan Deleeuw Message-Id: <9511071801.AA25850@galton.math.ucla.edu> Subject: problems on the horizon To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Tue, 7 Nov 1995 10:01:33 -0800 (PST) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 4323 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk This rambles a bit, but it is well-meant. It is sort of reaction to Luke's "Future Developments" paper. I have been feeling somwwhat uneasy about the future of Xlisp-Stat, but maybe I am just getting old, and I merely think that things are moving so fast because I am moving so slow. O well, some random thoughts. The number of possible choices for interpreters (sometimes with byte-compilers, sometimes with compilers which produce C) which are integrated with portable graphics packages becomes larger and larger. This creates problems, not for individuals, but for larger units (departments, disciplines). Of course I am not talking about commercial products here. It is bad enough that these are around. First we had Xlisp-Stat. Common Lisp based systems such as Quail and Clasp were no competition, because of the bloatedness and price of Common Lisp, and because of the non-portability of the graphics. This is probably also why the gcl/mcl version of Lisp-Stat is still sitting there without graphics. There was also Winterp, which is non-portable because it only builds Motif. Then there was Tcl/Tk, which produced an enormous number of special purpose programs/scripts, but no statistics system (probably because of the limitations of Tcl, see the Tcl war papers). But there was also Stk, which integrated tk (generally considered to be a nice graphics lib) with Scheme -- non-portable, though, or at least non-ported. And then there were MrEd, Guile/Grow, scsh, Rscheme -- all based on syuitable enhanced scheme interpreters. MrEd exists for MSW, X11, almost for Mac. Guile/Grow is web aware, integrated with both Tcl/Tk and almost with Emacs, and it has a very nice interface to the Unix OS. scsh is a scheme based shell, i.e. a scheme interpreter with extensions that make it a convenient Unix shell as well. Rscheme will have interfaces with the portable Vibrant library. Guile will produce a version of TkWWW, the WWW browser, with Scheme support build in. Scsh comes with a http server in Scheme. In order to keep Xlisp-Stat current, top-of-the-line, the best parts of these other newer systems should be integrated as well. This means, I think, that we need an xlisp-stat lib, so that ordinary C programs can open Xlisp-Stat interpreters, and use Xlisp-Stat as an extension language (similar to Elk, Guile, Tcl). We need better interfaces to the OS, and in particular to the networking and the WWW. There are some things missing in the graphics at the C level (which is why the three main versions are not completely in sync -- a nuisance for programmers). I think this can be nicely introduced in a modular way at compile-time (do you want to compile a features into the executable) and at run time using packages. But it is clearly impossible for one person (who has many other things to do) to do all this, and it is also undesirable to have various incompatible versions in C floating around. I am not sure what the solution is, but there is a problem. So far, we only had to convince S users that Xlisp-Stat was better, at least for some purposes. Now, this may not be the most interesting battle any more. There are now ways of doing statistics on the Web, using forms. This is still rather primitive, but Java looms large, and creates the possibility to do dynamic graphics in WWW browsers without having to use cgi and clumsy helper applications. It seems to me that the new model that many people are working with is the "empty box", i.e. each computer is a terminal on the net. This is related to the hype which makes Marc Andreessen the new Bill Gates, but I do think it means that networking and communications features should be emphasized over traditional Lispish concerns such as prefix/infix (Dylan is in limbo) and the object system. Communications is meant here as with the OS, with other machines (if there is such a thing as other machines) and with data (i.e. I/O). Cygnus has 4 programmers working on Guile and comp.lang.java has more than 200 posts/day (comp.lang.lisp.x has maybe 10/week). -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Tue Nov 7 13:51:58 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA07991; Tue, 7 Nov 1995 13:51:58 -0500 Return-Path: Received: from io.uwinnipeg.ca by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA07980; Tue, 7 Nov 1995 12:50:09 -0600 Received: from io.UWinnipeg.ca (io) by io.uwinnipeg.ca (PMDF V5.0-5 #10778) id <0DHOSJMMY00DQU@io.uwinnipeg.ca>; Tue, 07 Nov 1995 12:54 -0600 (CST) Date: Tue, 07 Nov 1995 12:54:58 -0600 (CST) From: Bill Simpson Subject: Re: problems on the horizon In-Reply-To: <9511071801.AA25850@galton.math.ucla.edu> To: Jan Deleeuw Cc: Xlisp List Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Content-Transfer-Encoding: 7BIT Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On alternative freeware stats systems... I thought I would mention R, a freeware clone of S. Avail at: ftp://stat.auckland.ac.nz/pub/R/ Versions for unix, Mac, and windows (I think) exist. Bill Simpson From majordom@umnstat.stat.umn.edu Fri Nov 10 15:47:34 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27041; Fri, 10 Nov 1995 15:47:34 -0500 Return-Path: Received: from cda.mrs.umn.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27036; Fri, 10 Nov 1995 14:47:30 -0600 Received: by cda.mrs.umn.edu (5.57/Ultrix3.0-C) id AA29852; Fri, 10 Nov 95 14:47:29 -0600 From: dierkhra@cda.mrs.umn.edu (Ross A. Dierkhising) Message-Id: <9511102047.AA29852@cda.mrs.umn.edu> Subject: Boxplots To: stat-lisp-news@umnstat.stat.umn.edu Date: Fri, 10 Nov 1995 14:47:28 -0600 (CST) X-Mailer: ELM [version 2.4 PL22] Content-Type: text Content-Length: 349 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, Here is my question. I would like to display a boxplot, generate a new set of data, clear current boxplot, display new data's boxplot in the same boxplot window. I have tried sending the existing boxplot window variations of :clear and :add-points messages, but nothing works so far. Any suggestions? Regards, Ross Dierkhising UMM From majordom@umnstat.stat.umn.edu Sat Nov 11 13:48:43 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11284; Sat, 11 Nov 1995 13:48:43 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11279; Sat, 11 Nov 1995 12:48:41 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA04453; Sat, 11 Nov 95 10:48:38 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA00257; Sat, 11 Nov 95 10:48:38 PST From: Jan Deleeuw Message-Id: <9511111848.AA00257@galton.math.ucla.edu> Subject: appendix to rambling note To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Sat, 11 Nov 1995 10:48:37 -0800 (PST) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 608 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk See http://monty.cnri.reston.va.us/grail-0.2/ for a browser with Python applets and access to Tk. See http://www.stat.ucla.edu/consult/computing.html for some ways to compute with HTML forms (not as elegant as applets, but better than using helpers). I think this indicates the direction in which Xlisp-Stat should move. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Sat Nov 11 16:09:24 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11828; Sat, 11 Nov 1995 16:09:24 -0500 Return-Path: Received: from netcom21.netcom.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11823; Sat, 11 Nov 1995 15:09:22 -0600 Received: from localhost by netcom21.netcom.com (8.6.12/Netcom) id MAA25319; Sat, 11 Nov 1995 12:52:18 -0800 Message-Id: <199511112052.MAA25319@netcom21.netcom.com> To: Jan Deleeuw Cc: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Cc: winterp@netcom.com Subject: Re: appendix to rambling note In-Reply-To: Your message of "Sat, 11 Nov 95 10:48:37 PST." <9511111848.AA00257@galton.math.ucla.edu> Date: Sat, 11 Nov 95 12:52:17 -0800 From: "Niels P. Mayer" Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Jan -- I'm working on a WINTERP-based (aka extensible) WYSIWYG WWW browser/editor for Unix (WWWeasel II). To be released real soon now (???). See http://www.eit.com/software/wwweasel for info on the old prototype (which sucks hideously as a WYSIWYG editor, but does pretty well for other things, such as fulfulling the government contract that funded it's development, acting as a structure-oriented media composition tool, and giving us a platform to experiment with issues and needs involved with group document management and revision control. Potential, future work for WWWeasel II will be in "smartening" creation of tables and forms including back-ends for receiving/processing the form data. I have found it helpful to have a real, high-level, symbol-and-list manipulating language like XLISP/WINTERP to be very helpful for such work... PS: w/r/t your note's mention of WINTERP -- since when is Motif not portable -- after all, it's *THE* *STANDARD* Unix GUI toolkit. Oh, you mean it's not "multiplatform" -- yes that's true to the extent that Motif isn't multiplatform. However, I have heard reports that WINTERP 1.XX was ported to VMS (since Motif runs on VMS). I have also heard rumors of an NT port of Motif as well... Also, another aspect of portability is internationalization, and Motif (IMHO) wins in that regard.... not sure how Tk has progressed in that regard, last time I looked at it, it appeared sub-optimal as a language to interface to anything other than a string-based language like TCL; internationalizaiton looked sketchy and simplistic. !?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?! ! Niels Mayer ..... mayer@eit.com ! http://www.eit.com/~mayer ! ! Enterprise Integration Technologies ! http://www.eit.com/software/wwweasel! ! 800 El Camino Real, Menlo Park, CA. ! http://www.eit.com/software/winterp ! !?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?! From majordom@umnstat.stat.umn.edu Mon Nov 13 11:36:38 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA07626; Mon, 13 Nov 1995 11:36:38 -0500 Return-Path: Received: from landau.Mines.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA07621; Mon, 13 Nov 1995 10:36:34 -0600 Received: (from jscales@localhost) by landau.Mines.EDU (8.6.9/8.6.9) id JAA25245; Mon, 13 Nov 1995 09:36:31 -0700 Date: Mon, 13 Nov 1995 09:36:31 -0700 From: John Scales Message-Id: <199511131636.JAA25245@landau.Mines.EDU> To: stat-lisp-news@umnstat.stat.umn.edu Subject: xlispstat plotting question Reply-To: jscales@dix.mines.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'm using the graph-window-proto to make gray scale plots. Basically, within a nested set of loops I repeatedly send a paint-rect message with different levels of gray to draw the plot. This works fine except that the plot doesn't automatically redraw when uncovered. I gather I need to write my own redraw method. That's fine and is described in Tierney's book. The problem is that drawing the plot takes a while. 12 seconds or so on a pentium, much longer on my students' machines. So I'd like to avoid redrawing the thing from scratch every time. I made a quick attempt at double buffering but this seems to conflict with the other plots already on the screen. This issue is mentioned in the book, but I was wondering if it might now be possible to use more than one buffer. So is it true that I must accept the time penalty and redraw the thing from scratch every time it's uncovered or is there a better/faster way of doing this? John Scales email: jscales@dix.mines.edu phone: 303-273-3408 Center for Wave Phenomena, Colorado School of Mines Free books from Samizdat Press -- http://landau.mines.edu/~samizdat From majordom@umnstat.stat.umn.edu Wed Nov 15 08:40:16 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11647; Wed, 15 Nov 1995 08:40:16 -0500 Return-Path: Received: from post4.INRE.ASU.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11642; Wed, 15 Nov 1995 07:40:12 -0600 Received: from general1.asu.edu by asu.edu (PMDF V4.3-10 #7723) id <01HXNVE5NY4091YA2M@asu.edu>; Wed, 15 Nov 1995 06:39:59 -0700 (MST) Received: from general1.asu.edu (localhost [127.0.0.1]) by general1.asu.edu (8.6.12/8.6.12) with SMTP id GAA17816; Wed, 15 Nov 1995 06:40:00 -0700 Date: Wed, 15 Nov 1995 06:40:00 -0700 (MST) From: behrens@asu.edu Subject: xlispstat via internet In-Reply-To: <9511111848.AA00257@galton.math.ucla.edu> X-Sender: jbehrens@general1.asu.edu To: Jan Deleeuw Cc: Xlisp List Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Content-Transfer-Encoding: 7BIT Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Jan: Again I applaud and share your vision for improving the world with xlispstat. We have begun using form-based xlispstat computing over the net in our classes. A good example is located at: http://seamonkey.ed.asu.edu/~behrens/rci.html The page that xlispstat writes is Netscape 2.0 enhanced (sub and superscripts). We have a couple other pages but they need a little cleaning. We hope to complete a full set for an intro class and basic effect size meta-analysis work over Christmas holidays. Glad to know you will be doing power analysis. I think this will dramatically reduce the amount of hand computation I do on the white-board for demonstration. In addition the students are loving it not only because it does the computation, but we spell out the steps. This will allow them to check their own work. I anticipate it will therefore promote practice and reduce grading. I can tell them to keep doing practice problems until they can match the computer. As usual, it seems our biggest limitation (besides time) is our imagination. Cheers, John John T. Behrens Division of Psychology in Education Measurement, Statistics, and Methodological Studies Program Arizona State University Tempe, AZ 85287-0611 (602) 965-3993 - Voice From majordom@umnstat.stat.umn.edu Wed Nov 15 10:55:33 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12315; Wed, 15 Nov 1995 10:55:33 -0500 Return-Path: Received: from euler.bd.psu.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12310; Wed, 15 Nov 1995 09:55:27 -0600 Received: from localhost by euler.bd.psu.edu via SMTP (940816.SGI.8.6.9/940406.SGI) for id KAA27176; Wed, 15 Nov 1995 10:55:50 -0500 Message-Id: <199511151555.KAA27176@euler.bd.psu.edu> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: xlispstat via internet In-Reply-To: Your message of "Wed, 15 Nov 1995 06:40:00 -0700 (MST)" References: X-Mailer: Mew beta version 0.98 on Emacs 19.29.1 Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Date: Wed, 15 Nov 1995 10:55:41 -0500 From: Balasubramanian Narasimhan Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk John> Jan: Again I applaud and share your vision for improving the John> world with xlispstat. We have begun using form-based John> xlispstat computing over the net in our classes. A good John> example is located at: John> http://seamonkey.ed.asu.edu/~behrens/rci.html ... We are doing something similar---I'll be post details later. The main limitation here is that students don't have access to the Web. However, we are setting up a new Linux lab with xlisp-stat as the main statistical computing environment and things will be better then. ________________________________________________________________ B. Narasimhan naras@euler.bd.psu.edu From majordom@umnstat.stat.umn.edu Wed Nov 15 14:00:08 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14036; Wed, 15 Nov 1995 14:00:08 -0500 Return-Path: Received: from [129.252.16.112] by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14031; Wed, 15 Nov 1995 13:00:04 -0600 Received: (from rossini@localhost) by neuromancer.stat.sc.edu (8.6.12/8.6.12) id NAA07949; Wed, 15 Nov 1995 13:56:17 -0500 Date: Wed, 15 Nov 1995 13:56:17 -0500 (EST) From: Anthony Rossini To: Balasubramanian Narasimhan Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: xlispstat via internet In-Reply-To: <199511151555.KAA27176@euler.bd.psu.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Wed, 15 Nov 1995, Balasubramanian Narasimhan wrote: > > John> Jan: Again I applaud and share your vision for improving the > John> world with xlispstat. We have begun using form-based > John> xlispstat computing over the net in our classes. A good > John> example is located at: > > John> http://seamonkey.ed.asu.edu/~behrens/rci.html > ... > > We are doing something similar---I'll be post details later. > > The main limitation here is that students don't have access to the > Web. However, we are setting up a new Linux lab with xlisp-stat as the > main statistical computing environment and things will be better then. If you are using X11, you might consider the Hamlet approach -- it is an application which allows for dumping source code directly into a running program (xterm running xlispstat, xterm running splus, or xmaple; the latter is the usual platform, and it's nice for learning maple...). I've been working up some stuff for mupad(? forgot the name -- it's been a few weeks) and xlispstat. I can't recall the link to it, but it's on my homepage at South Carolina. The big advantage is that you can keep dumping source code into the _same_ running program. The big disadvantage is that it doesn't run without Motif or X (I've got access to a powermac lab, for my undergrads, so I'd prefer a solution for that other than just the long "download and load into xlispstat" paradigm). -tony ------ Anthony Rossini Asst. Prof. and General Nuisance rossini@stat.sc.edu Department of Statistics http://www.stat.sc.edu/~rossini/ University of South Carolina http://www.biostat.washington.edu/~arossini/ Columbia, SC 29208 http://www.stat.psu.edu/~rossini/ 803-777-3578(O) ..7-4048 (fax) From majordom@umnstat.stat.umn.edu Wed Nov 15 15:06:19 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14437; Wed, 15 Nov 1995 15:06:19 -0500 Return-Path: Received: from [129.252.16.112] by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14432; Wed, 15 Nov 1995 14:06:15 -0600 Received: (from rossini@localhost) by neuromancer.stat.sc.edu (8.6.12/8.6.12) id PAA08004; Wed, 15 Nov 1995 15:02:35 -0500 Date: Wed, 15 Nov 1995 15:02:34 -0500 (EST) From: Anthony Rossini To: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: xlispstat via internet In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Wed, 15 Nov 1995, Anthony Rossini wrote: > If you are using X11, you might consider the Hamlet approach -- it is an > application which allows for dumping source code directly into a running > program (xterm running xlispstat, xterm running splus, or xmaple; the > latter is the usual platform, and it's nice for learning maple...). > > I've been working up some stuff for mupad(? forgot the name -- it's been > a few weeks) and xlispstat. I can't recall the link to it, but it's on > my homepage at South Carolina. That's the link to the Hamlet homepage, not the stuff I'm developing. It'll be available soon enough, though... -tony ------ Anthony Rossini Asst. Prof. and General Nuisance rossini@stat.sc.edu Department of Statistics http://www.stat.sc.edu/~rossini/ University of South Carolina http://www.biostat.washington.edu/~arossini/ Columbia, SC 29208 http://www.stat.psu.edu/~rossini/ 803-777-3578(O) ..7-4048 (fax) From majordom@nokomis.stat.umn.edu Thu Nov 16 07:58:20 1995 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA16080; Thu, 16 Nov 1995 07:58:20 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA16074; Thu, 16 Nov 1995 06:58:15 -0600 From: Luke Tierney Message-Id: <9511161258.AA16074@nokomis.stat.umn.edu> Subject: Re: appendix to rambling note To: deleeuw@math.ucla.edu (Jan Deleeuw) Date: Thu, 16 Nov 1995 06:58:15 -0600 (CST) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: <9511111848.AA00257@galton.math.ucla.edu> from "Jan Deleeuw" at Nov 11, 95 10:48:37 am X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 2716 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > for some ways to compute with HTML forms (not as elegant as applets, > but better than using helpers). > You need to be a little careful here. Forms-based approaches and applet/helper approaches are very different -- forms-based runs on the server, applets or helpers run on the client. They have different advantages and problems (in particular different security problems). I'm not sure why you make a big deal of the difference between applets and helpers -- aside from some minor one-time setup issues for running helpers, the only real difference a user sees is that applets run in the browser window while helpers run in thier own window. Even this distinction will go away soon. Netscape 2.0 is supporting PlugIn's (http://home.netscape.com/comprod/products/navigator/version_2.0/plugins/index.html) which provide a protocol for having applications embed themselves in the browser. At this point this is just another wonderful/annoying (depending on your attitude to netscape) nonstandard netscape feature, but given the OLE/OpenDoc wave on the desktop it is probably ging to be standardized in some form. It is also a better way to go in the long run: instead of needing a browser that understands python for python applets, one that understands java for java applets, ..., you only need one that understands a basic embedding protocol. I would like xlispstat to work well in this kind of framework. There are certainly too many variations of things one might do for any one person to do them, and even if they did, new possibilites would come up that were not anticipated. So the challenge is to figure out a system structure that makes it easier for others to adapt the system to these kinds of needs. The current system is reasonably extensible within itself but is not suited very well as a tool for extending other systems, for adding functionality to other systems. My long term objective is to redesign the system to make it more flexible in this respect. In the process it should become much more modular, and this will make it easier to allow maintenance of parts of the system to be taken over by others. This is all very wooly at this point -- hopefully it will become more concrete in the next few months. In the meantime, hamlet (http://www.cs.utah.edu/~hamlet/) looks pretty useful on the UNIX end. Writing something similar for Mac or Windows using AppleEvents or DDE/OLE should be farily easy (getting commands into xlispstat can be done now; simulating user typing can be done on the mac but might need a little work on Windows). For what it is worth, another interesting article on using the net to help with instruction is http://thejournal.com/october/10andriole.html. luke From majordom@umnstat.stat.umn.edu Thu Nov 16 09:36:43 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28232; Thu, 16 Nov 1995 09:36:43 -0500 Return-Path: Received: from post3.INRE.ASU.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28227; Thu, 16 Nov 1995 08:36:39 -0600 Received: from general1.asu.edu by asu.edu (PMDF V4.3-10 #7723) id <01HXPBN60BMO937OYF@asu.edu>; Thu, 16 Nov 1995 07:36:11 -0700 (MST) Received: from general1.asu.edu (localhost [127.0.0.1]) by general1.asu.edu (8.6.12/8.6.12) with SMTP id HAA29417; Thu, 16 Nov 1995 07:36:08 -0700 Date: Thu, 16 Nov 1995 07:36:08 -0700 (MST) From: behrens@asu.edu Subject: Re: xlispstat via internet In-Reply-To: X-Sender: jbehrens@general1.asu.edu To: Anthony Rossini Cc: stat-lisp-news@umnstat.stat.umn.edu Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Content-Transfer-Encoding: 7BIT Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Tony: We dont use x-window with the students which is why we use html based computing on the net. However for interactive graphics Ive tried to define .fsl files as a new mime type in the srm.conf. Have you tried this? I was hoping clicking on the URL would download the file and launch the demo but its not working for us. John John T. Behrens Division of Psychology in Education Measurement, Statistics, and Methodological Studies Program Arizona State University Tempe, AZ 85287-0611 (602) 965-3993 - Voice From majordom@nokomis.stat.umn.edu Thu Nov 16 09:58:18 1995 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA16150; Thu, 16 Nov 1995 09:58:18 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA16143; Thu, 16 Nov 1995 08:58:10 -0600 From: Luke Tierney Message-Id: <9511161458.AA16143@nokomis.stat.umn.edu> Subject: Re: xlispstat via internet To: behrens@asu.edu Date: Thu, 16 Nov 1995 08:58:10 -0600 (CST) Cc: rossini@stat.sc.edu, stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "behrens@asu.edu" at Nov 16, 95 07:36:08 am X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1405 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > Tony: > We dont use x-window with the students which is why we use html based > computing on the net. However for interactive graphics Ive tried to > define .fsl files as a new mime type in the srm.conf. Have you tried > this? I was hoping clicking on the URL would download the file and > launch the demo but its not working for us. > John > > John T. Behrens > Division of Psychology in Education > Measurement, Statistics, and > Methodological Studies Program > Arizona State University > Tempe, AZ 85287-0611 > (602) 965-3993 - Voice > > If you are using mac's, you currently have to use .lsp as the extension. Also, you can only download each file once per session, at least with netscape. The reason is that xlisp requires a .lsp extension for files loaded in response to an ODoc Apple Event, which is how Netscape sends files to its helpers. When you download a file fred.lsp the first time as a helper data file, netscape brings it down as fred.lsp. The second time it comes down as fred.lsp-1234 or some such, which fails to work. There mey be some way to tell Netscape to overwrite the previous fred.lsp, but I don't know how. I hope to release a new mac version that fixes this (and several other bugs) in a month or so. I don;t know how things work on Windows. If anyone has tried using xlispstat as a windows web helper application, i would like to hear how it went. luke From majordom@nokomis.stat.umn.edu Thu Nov 16 10:12:42 1995 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA16183; Thu, 16 Nov 1995 10:12:42 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA16143; Thu, 16 Nov 1995 08:58:10 -0600 From: Luke Tierney Message-Id: <9511161458.AA16143@nokomis.stat.umn.edu> Subject: Re: xlispstat via internet To: behrens@asu.edu Date: Thu, 16 Nov 1995 08:58:10 -0600 (CST) Cc: rossini@stat.sc.edu, stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "behrens@asu.edu" at Nov 16, 95 07:36:08 am X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1405 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > Tony: > We dont use x-window with the students which is why we use html based > computing on the net. However for interactive graphics Ive tried to > define .fsl files as a new mime type in the srm.conf. Have you tried > this? I was hoping clicking on the URL would download the file and > launch the demo but its not working for us. > John > > John T. Behrens > Division of Psychology in Education > Measurement, Statistics, and > Methodological Studies Program > Arizona State University > Tempe, AZ 85287-0611 > (602) 965-3993 - Voice > > If you are using mac's, you currently have to use .lsp as the extension. Also, you can only download each file once per session, at least with netscape. The reason is that xlisp requires a .lsp extension for files loaded in response to an ODoc Apple Event, which is how Netscape sends files to its helpers. When you download a file fred.lsp the first time as a helper data file, netscape brings it down as fred.lsp. The second time it comes down as fred.lsp-1234 or some such, which fails to work. There mey be some way to tell Netscape to overwrite the previous fred.lsp, but I don't know how. I hope to release a new mac version that fixes this (and several other bugs) in a month or so. I don;t know how things work on Windows. If anyone has tried using xlispstat as a windows web helper application, i would like to hear how it went. luke From majordom@umnstat.stat.umn.edu Thu Nov 16 18:03:47 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02852; Thu, 16 Nov 1995 18:03:47 -0500 Return-Path: Received: from neopath.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02847; Thu, 16 Nov 1995 17:03:43 -0600 Received: from denali.neopath.com by neopath.com (5.x/SMI-SVR4) id AA09793; Thu, 16 Nov 1995 15:02:02 -0800 Received: from pilchuck.neopath.com by denali.neopath.com (5.x/SMI-SVR4) id AA23419; Thu, 16 Nov 1995 15:02:03 -0800 Received: by pilchuck.neopath.com (5.x/SMI-SVR4) id AA05723; Thu, 16 Nov 1995 15:02:00 -0800 Date: Thu, 16 Nov 1995 15:02:00 -0800 From: keith@neopath.com (Keith Frost) Message-Id: <9511162302.AA05723@pilchuck.neopath.com> To: luke@umnstat.stat.umn.edu Cc: deleeuw@math.ucla.edu, stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <9511161258.AA16074@nokomis.stat.umn.edu> (message from Luke Tierney on Thu, 16 Nov 1995 06:58:15 -0600 (CST)) Subject: Re: appendix to rambling note Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk In regards to: making xlispstat more easily extensible, you might take a look at http://betz.biostr.washington.edu/slisp/ to see what they have done with xlisp, in order to accomplish this very goal. Regards for xlispstat: a user-friendly and useful package. Keith Frost From majordom@umnstat.stat.umn.edu Fri Nov 17 21:45:58 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA20174; Fri, 17 Nov 1995 21:45:58 -0500 Return-Path: Received: from comoro.yorku.ca by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA20169; Fri, 17 Nov 1995 20:45:55 -0600 Received: from (roseanne02.slip.yorku.ca [130.63.122.110]) by comoro.yorku.ca (8.6.12/8.6.11) with SMTP id VAA24484 for ; Fri, 17 Nov 1995 21:45:53 -0500 From: "John Fox" Message-Id: <9511172145.ZM18623@unknown.zmail.host> Date: Fri, 17 Nov 1995 21:45:23 -0500 X-Mailer: ZM-Win (3.2.1 11Sep94) To: stat-lisp-news@umnstat.stat.umn.edu Subject: xlispstat binaries for Linux? Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Does anyone know if (and where) I can find xlispstat binaries for Linux? I've looked in what I thought were the obvious places for lisp-stat and Linux software without any luck. Thanks, John From majordom@umnstat.stat.umn.edu Sun Nov 19 13:41:54 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13196; Sun, 19 Nov 1995 13:41:54 -0500 Return-Path: Received: from euler.bd.psu.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13191; Sun, 19 Nov 1995 12:41:49 -0600 Received: from localhost by euler.bd.psu.edu via SMTP (940816.SGI.8.6.9/940406.SGI) id NAA16386; Sun, 19 Nov 1995 13:43:54 -0500 Message-Id: <199511191843.NAA16386@euler.bd.psu.edu> To: fox@YorkU.CA Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: xlispstat binaries for Linux? In-Reply-To: Your message of "Fri, 17 Nov 1995 21:45:23 -0500" References: <9511172145.ZM18623@unknown.zmail.host> X-Mailer: Mew beta version 0.98 on Emacs 19.29.1 Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Date: Sun, 19 Nov 1995 13:43:50 -0500 From: Balasubramanian Narasimhan Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk John> Does anyone know if (and where) I can find xlispstat John> binaries for Linux? I've looked in what I thought were the John> obvious places for lisp-stat and Linux software without any John> luck. In March, I wrote an article on Xlisp-Stat for the August 1995 issue of Linux Journal. (If you want a copy, I can make it available). At that time, I made a binary, which is still available from ftp://euler.bd.psu.edu/pub/lj. Note that with new versions of stuff being released for Linux so frequently, it is possible that your versions of shared libraries might be different from mine. So if you get a warning, just plonk the correct version in from sunsite. ________________________________________________________________ B. Narasimhan naras@euler.bd.psu.edu From majordom@umnstat.stat.umn.edu Sun Nov 19 17:14:40 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14352; Sun, 19 Nov 1995 17:14:40 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14347; Sun, 19 Nov 1995 16:14:37 -0600 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.6.4.3/10.1) id RAA25920; Sun, 19 Nov 1995 17:14:11 -0500 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sun, 19 Nov 1995 17:14:23 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: memory usage Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Does anyone know how to tell what a reasonable "expand" value is for the (expand value) function? And, how big a memory partition one should use? I would like to be able to do this automatically when ViSta first starts up, because it seems that ViSta doesn't run too well without using expand. I've also determined that if expand is too big, relative to memory size, then ViSta totally dies at some point. So far, what I've done, on the Mac, is set the maximum memory size to some value (I've used 6, 7, 8, and 10Meg) and tried arbitrary "expand" values, including (expand 50), (expand 100) and (expand 150). Then I look at "About this Macintosh" and make a judgment about how the memory bar-graph looks. If the dark part of it is about 3/4's across, I find everything seems to work. If it takes up the whole bar, then ViSta certainly crashes soon. But I don't understand what's happening. And I don't have a clue about what happens under Windows and Unix. Will an expand value good for the Mac be good for other environments? Can anyone help? ____________________________________________________________________________ Forrest & Patty Young | Forrest Young | email: forrest@unc.edu 7323 Apple Mill Road | CB3270 Davie Hall | www: forrest.psych.unc.edu Efland, NC 27243 | UNC Psychometric Lab | work: 1-919-962-5038 USA | Chapel Hill, NC 27599 | home: 1-919-563-4464 From majordom@umnstat.stat.umn.edu Sun Nov 19 22:53:32 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15865; Sun, 19 Nov 1995 22:53:32 -0500 Return-Path: Received: from garam.kreonet.re.kr by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15860; Sun, 19 Nov 1995 21:52:26 -0600 Received: from yurim.skku.ac.kr (yurim.skku.ac.kr [134.75.162.2]) by garam.kreonet.re.kr (8.6.9H1/8.6.12) with SMTP id MAA02555 for ; Mon, 20 Nov 1995 12:51:50 +0900 Received: by yurim.skku.ac.kr (4.1/yurim-MX-4.1) id AA13713; Mon, 20 Nov 95 12:52:50 KST Date: Mon, 20 Nov 1995 12:38:08 +0900 (KST) From: Huh moonyul Subject: Q:link c functions on Win31 To: lisp stat Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Dear Colleagues, I have trouble in calling c functions on Windows 3.1 I have the following message: > (setf foo-dll (load-dll"c:\\borlandc\\bin\\foo.dll")) >4678 > (call-cfun foo-dll "_FOO" 5 (float (iseq 1 5)) 0.0) > Error: Can't find function address - "_FOO" I tried various combinations of the second parameter of call-cfun such as "_foo", "FOO_", etc. I will appreciate to have the solution on the above problem. Thanks. Huh, Moon Yul >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> Deapartment of Statistics Sung Kyun Kwan University 53, 3-ka, Myungryun-Dong Chongro-Ku, Seoul 110, Korea Tel: 82 2 760 0463 Fax: 82 2 764 0622 e-mail: myhuh@yurim.skku.ac.kr >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> From majordom@umnstat.stat.umn.edu Mon Nov 20 05:59:09 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26021; Mon, 20 Nov 1995 05:59:09 -0500 Return-Path: Received: from inet.d48.lilly.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26016; Mon, 20 Nov 1995 04:59:06 -0600 Received: from mr.lilly.com by INET.D48.LILLY.COM (PMDF V5.0-4 #6224) id <01HXUTF1OLBK00IY0Q@INET.D48.LILLY.COM> for stat-lisp-news@stat.umn.edu; Mon, 20 Nov 1995 05:59:01 -0500 (EST) Received: with PMDF-MR; Mon, 20 Nov 1995 10:58:55 +0000 (GMT) Mr-Received: by mta CORP00; Relayed; Mon, 20 Nov 1995 10:58:55 +0000 Mr-Received: by mta MCDEV1; Relayed; Mon, 20 Nov 1995 10:58:56 +0000 Alternate-Recipient: prohibited Disclose-Recipients: prohibited Date: Mon, 20 Nov 1995 11:02:01 +0000 (GMT) From: KINLEY ROBERT Subject: Non-central distributions To: stat-lisp-news@umnstat.stat.umn.edu Message-Id: <01HXUTF2WSRQ00IY0Q@mr.lilly.com> Mime-Version: 1.0 Content-Type: TEXT/PLAIN; CHARSET=US-ASCII Content-Transfer-Encoding: 7BIT Importance: normal Priority: normal X400-Mts-Identifier: [;55850102115991/4826821@CRPVAX] Hop-Count: 2 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Non-central distributions I am a new (and very impressed) user of XLISPSTAT (MS Windows version). A facility which would be useful to me would be the ability to obtain the quantile (given the probability) and the probability (given the quantile) for the non-central chi-squared (and t and F) distributions. Has anyone developed these ? ...... thanks ... Bob Kinley From majordom@umnstat.stat.umn.edu Mon Nov 20 06:35:00 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26080; Mon, 20 Nov 1995 06:35:00 -0500 Return-Path: Received: from garam.kreonet.re.kr by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26075; Mon, 20 Nov 1995 05:33:41 -0600 Received: from yurim.skku.ac.kr (yurim.skku.ac.kr [134.75.162.2]) by garam.kreonet.re.kr (8.6.9H1/8.6.12) with SMTP id UAA07615 for ; Mon, 20 Nov 1995 20:33:20 +0900 Received: by yurim.skku.ac.kr (4.1/yurim-MX-4.1) id AA24187; Mon, 20 Nov 95 20:34:20 KST Date: Mon, 20 Nov 1995 20:27:44 +0900 (KST) From: Huh moonyul Subject: Q2: loading c functions on Win31 To: lisp stat In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Dear Colleagues, I could successfully load C functions on WIN31. But only with 16 bit versions. Anyone has been successful with 32 bit WXLS versions? > (setf foo-dll (load-dll"c:\\borlandc\\bin\\foo.dll")) >4678 > (call-cfun foo-dll "foo" 5 (float (iseq 1 5)) 0.0) > Error: Can't find function address - "_FOO" Thanks, Huh, Moon Yul >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> Deapartment of Statistics Sung Kyun Kwan University 53, 3-ka, Myungryun-Dong Chongro-Ku, Seoul 110, Korea Tel: 82 2 760 0463 Fax: 82 2 764 0622 e-mail: myhuh@yurim.skku.ac.kr >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> From majordom@umnstat.stat.umn.edu Mon Nov 20 12:16:22 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28104; Mon, 20 Nov 1995 12:16:22 -0500 Return-Path: Received: from inet.d48.lilly.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28097; Mon, 20 Nov 1995 11:16:18 -0600 Received: from mr.lilly.com by INET.D48.LILLY.COM (PMDF V5.0-4 #6224) id <01HXV6KDF9O000I4WI@INET.D48.LILLY.COM> for stat-lisp-news@stat.umn.edu; Mon, 20 Nov 1995 12:15:47 -0500 (EST) Received: with PMDF-MR; Mon, 20 Nov 1995 17:15:27 +0000 (GMT) Mr-Received: by mta CORP00; Relayed; Mon, 20 Nov 1995 17:15:27 +0000 Mr-Received: by mta MCDEV1; Relayed; Mon, 20 Nov 1995 17:15:28 +0000 Alternate-Recipient: prohibited Disclose-Recipients: prohibited Date: Mon, 20 Nov 1995 15:47:22 +0000 (GMT) From: KINLEY ROBERT Subject: XLISPSTAT on SunOS 5.5 (Solaris 2.4) To: stat-lisp-news@umnstat.stat.umn.edu Message-Id: <01HXV6L6HILU00I4WI@mr.lilly.com> Mime-Version: 1.0 Content-Type: TEXT/PLAIN; CHARSET=US-ASCII Content-Transfer-Encoding: 7BIT Importance: normal Priority: normal X400-Mts-Identifier: [;72517102115991/4833398@CRPVAX] Hop-Count: 2 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk XLISPSTAT on SunOS 5.5 (Solaris 2.4) Flushed with the success of the Windows version, I downloaded the Unix version, only to find that I can't compile it (my fault, not Xlispstat's). Could it be because the version on the net is for Solaris 4, while my machine is on Solaris 5 ? If so, has anyone else experienced (and solved) this problem ?? Thanks ..... Bob Kinley From majordom@umnstat.stat.umn.edu Thu Nov 23 08:44:48 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15843; Thu, 23 Nov 1995 08:44:48 -0500 Return-Path: Received: from euler.bd.psu.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15838; Thu, 23 Nov 1995 07:44:45 -0600 Received: from localhost by euler.bd.psu.edu via SMTP (940816.SGI.8.6.9/940406.SGI) id IAA05647; Thu, 23 Nov 1995 08:46:52 -0500 Message-Id: <199511231346.IAA05647@euler.bd.psu.edu> To: fox@YorkU.CA Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: xlispstat binaries for Linux? In-Reply-To: Your message of "Sun, 19 Nov 1995 13:43:50 -0500" References: <199511191843.NAA16386@euler.bd.psu.edu> X-Mailer: Mew beta version 0.98 on Emacs 19.29.1 Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Date: Thu, 23 Nov 1995 08:46:49 -0500 From: Balasubramanian Narasimhan Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The Linux Journal article can now be obtained from ftp://euler.bd.psu.edu/pub/lj/article.tar.gz ________________________________________________________________ B. Narasimhan naras@euler.bd.psu.edu From majordom@umnstat.stat.umn.edu Thu Nov 23 09:14:41 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15882; Thu, 23 Nov 1995 09:14:41 -0500 Return-Path: Received: from fw-ismw3.wiwi.uni-karlsruhe.de by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15869; Thu, 23 Nov 1995 08:13:44 -0600 Received: (from gerd@localhost) by fw-ismw3.wiwi.uni-karlsruhe.de (8.6.9/8.6.9) id PAA00259; Thu, 23 Nov 1995 15:25:49 +0100 Date: Thu, 23 Nov 1995 15:25:49 +0100 From: Gerd Martin Klesse Message-Id: <199511231425.PAA00259@fw-ismw3.wiwi.uni-karlsruhe.de> To: fox@YorkU.CA Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <9511172145.ZM18623@unknown.zmail.host> (message from John Fox on Fri, 17 Nov 1995 21:45:23 -0500) Subject: Re: xlispstat binaries for Linux? Comments: Hyperbole mail buttons accepted, v3.16. Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Well, it is easy to compile the sources under LinuX. XlispStat is very well usable under LinuX in combination with Emacs. The only thing you have to do is to change the sources to make dynamic loading possible. If you are interested, I can send you the necessary modifications. I have send them to Luke Tierney, but he refused to include tthem in the official sources for XLispStat. It Compiles very well under LinuX. Gerd -- (o o) -----------oOO--(_)--OOo------------------------------------- Institut für Statistik und Mathematische Wirtschaftstheorie Kollegium am Schloß, Bau II Universität Karlsruhe Postfach 6980 PLZ 76128 Karlsruhe --------------------------------------------------------------------- tel: +49-721-608-3793 fax: +49-721-608-3793 email : Gerd.Klesse@wiwi.uni-karlsruhe.de WWW: http://wi70.rz.uni-karlsruhe.de/~lh04 ===================================================================== From majordom@umnstat.stat.umn.edu Mon Nov 27 13:53:51 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02617; Mon, 27 Nov 1995 13:53:51 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02607; Mon, 27 Nov 1995 12:53:46 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA03342; Mon, 27 Nov 95 10:53:43 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA05214; Mon, 27 Nov 95 10:53:42 PST From: Jan Deleeuw Message-Id: <9511271853.AA05214@galton.math.ucla.edu> Subject: Approximating the Binomial To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Mon, 27 Nov 1995 10:53:41 -0800 (PST) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 683 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On ftp://ftp.stat.ucla.edu/pub/xlispstat/code/ ftp.stat.ucla.edu:/pub/xlispstat/code http://www.stat.ucla.edu/archive/xlispstat/code/ there is new or revised code in: statistics/introstat/modules/approx.lsp By Jason Bond. Binomial parameters N and p on sliders -- two windows show approximation by Poisson and Normal -- with overlay to superimpose a normal density. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Wed Nov 29 12:40:18 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04884; Wed, 29 Nov 1995 12:40:18 -0500 Return-Path: Received: from sunset.ma.huji.ac.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04876; Wed, 29 Nov 1995 11:40:04 -0600 Received: by sunset.ma.huji.ac.il id AA02259 (5.67b/HUJI 4.152 for stat-lisp-news@umnstat.stat.umn.edu); Wed, 29 Nov 1995 19:41:13 +0200 Date: Wed, 29 Nov 1995 19:41:13 +0200 From: "Harvey J. Stein" Message-Id: <199511291741.AA02259@sunset.ma.huji.ac.il> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Compiler bug. Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Xlispstat bug: The compile function seems to lose environments. For example, consider the following code: (defun make-add-n-fcn (n) (lambda (x) (+ x n))) (setf (symbol-function 'add-10) (make-add-n-fcn 10)) (add-10 3) (add-10 7) (compile 'add-10) (add-10 3) This code makes a function add-10 which adds 10 to its argument. After compiling add-10, however, it no longer works. Here's the xlispstat session: > (defun make-add-n-fcn (n) (lambda (x) (+ x n))) MAKE-ADD-N-FCN > (setf (symbol-function 'add-10) (make-add-n-fcn 10)) # > (add-10 3) 13 > (add-10 7) 17 > (compile 'add-10) ADD-10 > (add-10 3) Error: The variable N is unbound. Happened in: # > It looks like the compile lost the environment of the function. It seems to have treated the function like a lambda expression at top level instead of like a closure. Version info: XLISP-PLUS version 2.1g Portions Copyright (c) 1988, by David Betz. Modified by Thomas Almy and others. XLISP-STAT 2.1 Release 3.47 (Beta). Copyright (c) 1989-1994, by Luke Tierney. Running under Linux, kernel version 1.2.13. Compiled with gcc version 2.5.8 (I think - it was a long time ago, and I'm not sure if I upgraded the compiler since then...). Thanks, Dr. Harvey J. Stein Berger Financial Research hjstein@math.huji.ac.il From majordom@nokomis.stat.umn.edu Wed Nov 29 14:18:04 1995 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA06163; Wed, 29 Nov 1995 14:18:04 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA06155; Wed, 29 Nov 1995 13:17:09 -0600 From: Luke Tierney Message-Id: <9511291917.AA06155@nokomis.stat.umn.edu> Subject: Re: Compiler bug. To: hjstein@math.huji.ac.il (Harvey J. Stein) Date: Wed, 29 Nov 1995 13:17:09 -0600 (CST) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: <199511291741.AA02259@sunset.ma.huji.ac.il> from "Harvey J. Stein" at Nov 29, 95 07:41:13 pm X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 2195 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > Xlispstat bug: > > The compile function seems to lose environments. For example, > consider the following code: > > (defun make-add-n-fcn (n) > (lambda (x) (+ x n))) > (setf (symbol-function 'add-10) (make-add-n-fcn 10)) > (add-10 3) > (add-10 7) > (compile 'add-10) > (add-10 3) > > This code makes a function add-10 which adds 10 to its argument. > After compiling add-10, however, it no longer works. Here's the > xlispstat session: > > > (defun make-add-n-fcn (n) > (lambda (x) (+ x n))) > MAKE-ADD-N-FCN > > (setf (symbol-function 'add-10) (make-add-n-fcn 10)) > # > > (add-10 3) > 13 > > (add-10 7) > 17 > > (compile 'add-10) > ADD-10 > > (add-10 3) > Error: The variable N is unbound. > Happened in: # > > > > It looks like the compile lost the environment of the function. It > seems to have treated the function like a lambda expression at top > level instead of like a closure. > > Version info: > > XLISP-PLUS version 2.1g > Portions Copyright (c) 1988, by David Betz. > Modified by Thomas Almy and others. > XLISP-STAT 2.1 Release 3.47 (Beta). > Copyright (c) 1989-1994, by Luke Tierney. > > Running under Linux, kernel version 1.2.13. Compiled with gcc version > 2.5.8 (I think - it was a long time ago, and I'm not sure if I > upgraded the compiler since then...). > > Thanks, > > Dr. Harvey J. Stein > Berger Financial Research > hjstein@math.huji.ac.il > According to the Common Lisp standard the behavior of compile is "undefined" if the surrouning environment of a function being compiled contains lexical variable or function bindings. Currently compile just ignores such bindings. It would be better if it at least generated an error message -- I will make sure it does in the next release. Eventually I may arrange for closures with non-null environments to be compiled, but it isn't often useful. In this case you can achieve what you want by compiling make-add-n-fcn: > (defun make-add-n-fcn (n) #'(lambda (x) (+ x n))) MAKE-ADD-N-FCN > (make-add-n-fcn 3) # > (compile 'make-add-n-fcn) MAKE-ADD-N-FCN > (make-add-n-fcn 3) # > (funcall * 2) 5 From majordom@umnstat.stat.umn.edu Thu Nov 30 04:02:46 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA20632; Thu, 30 Nov 1995 04:02:46 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA20627; Thu, 30 Nov 1995 03:02:42 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA07252; Thu, 30 Nov 95 01:02:39 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA23055; Thu, 30 Nov 95 01:02:37 PST Date: Thu, 30 Nov 95 01:02:37 PST Message-Id: <9511300902.AA23055@galton.math.ucla.edu> From: Jan Deleeuw To: stat-lisp-news@umnstat.stat.umn.edu, ucla-stat@math.ucla.edu Subject: Normal Distribution Homepage Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk As far as I know, the Normal Distribution is the first probability distribution with its own homepage http://www.stat.ucla.edu/textbook/form-based/cdf/normal/ The computations on the page are done by forms and CGI programs in C on our server. The dynamic demo is done on the client machine. For this you need to have Xlisp-Stat installed. But you needed that anyway. From majordom@umnstat.stat.umn.edu Fri Dec 1 15:59:18 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11309; Fri, 1 Dec 1995 15:59:18 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11302; Fri, 1 Dec 1995 14:59:03 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA13725; Fri, 1 Dec 95 12:58:54 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA03699; Fri, 1 Dec 95 12:58:49 PST From: Jan Deleeuw Message-Id: <9512012058.AA03699@galton.math.ucla.edu> Subject: running Xlisp-Stat from Web browsers To: ndy@stat.ucla.edu (Don Ylvisaker), kcli@stat.ucla.edu (Ker-Chau Li), berk@stat.ucla.edu (Dick Berk), rij@stat.ucla.edu (Bob Jennrich), tom@math.ucla.edu (Tom Ferguson), iyx4ben@mvs.oac.ucla.edu (Bengt Muthen), bentler@psych.sscnet.ucla.edu (Peter Bentler), masonwm@ucla.edu (Bill Mason), eleamer@agsm.ucla.edu (Ed Leamer), cameron@econ.sscnet.ucla.edu (Trudy Cameron), woodward@psych.sscnet.ucla.edu (Art Woodward), judea@cs.ucla.edu (Judea Pearl), jmacquee@agsm.ucla.edu (Jim MacQueen), scp@math.ucla.edu (Sid Port), emigh@stat.ucla.edu (Rebecca Emigh), reiff@nicco.sscnet.ucla.edu (Jan Reiff), tlin@stat.ucla.edu (Tony Lin), rgould@stat.ucla.edu (Rob Gould), vledesma@stat.ucla.edu (Victoria Ledesma) Date: Fri, 1 Dec 1995 12:58:49 -0800 (PST) Cc: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 337 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Try http://www.stat.ucla.edu/textbook/xlispstat.html -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Fri Dec 1 20:59:49 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12875; Fri, 1 Dec 1995 20:59:49 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12870; Fri, 1 Dec 1995 19:59:46 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA22254; Fri, 1 Dec 95 15:35:44 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA04323; Fri, 1 Dec 95 15:35:43 PST From: Jan Deleeuw Message-Id: <9512012335.AA04323@galton.math.ucla.edu> Subject: moving-average.lsp To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Fri, 1 Dec 1995 15:35:43 -0800 (PST) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 531 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On ftp://ftp.stat.ucla.edu/pub/xlispstat/code/ ftp.stat.ucla.edu:/pub/xlispstat/code there is new or revised code in: statistics/smoothers/moving-average.lsp Time series plot, with slider for MA bandwidth. By Frederic Udina. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Fri Dec 1 21:58:00 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13091; Fri, 1 Dec 1995 21:58:00 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13086; Fri, 1 Dec 1995 20:57:58 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA29640; Fri, 1 Dec 95 18:57:55 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA04620; Fri, 1 Dec 95 18:57:54 PST From: Jan Deleeuw Message-Id: <9512020257.AA04620@galton.math.ucla.edu> Subject: Xlisp-Stat from WWW browsers To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Fri, 1 Dec 1995 18:57:54 -0800 (PST) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1422 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk In case somebody is interested: here is the html and the c (the c uses Tom Bouteil's cgic library). ======================================================================== Submit Xlisp-Stat logo line-eyes

Submit Xlisp-Stat

Type in a number of lines with legitimate Xlisp-Stat code (no graphics please).

blueline
Jan de Leeuw
UCLA Statistics
deleeuw@stat.ucla.edu
========================================================================= #include #include "cgic.h" int cgiMain() { FILE *fp; char **responses; int i = 0; fp = popen("/misc2/galton/bin/APPS/xlispstat","w"); cgiFormStringMultiple("xlisp",&responses); while (responses[i]){ fprintf(fp, responses[i]); i++; } pclose(fp); } -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Sat Dec 2 17:54:43 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26484; Sat, 2 Dec 1995 17:54:43 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26479; Sat, 2 Dec 1995 16:54:40 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA23590; Sat, 2 Dec 95 14:54:37 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA16286; Sat, 2 Dec 95 14:54:36 PST From: Jan Deleeuw Message-Id: <9512022254.AA16286@galton.math.ucla.edu> Subject: Xlisp-Stat on WWW browsers To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Sat, 2 Dec 1995 14:54:36 -0800 (PST) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 412 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk At Naras suggestion I installed a version of Xlisp-Stat which has both system and popen disabled. I feel much better now. --- Jan -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Sat Dec 2 23:31:49 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27778; Sat, 2 Dec 1995 23:31:49 -0500 Return-Path: Received: from peacock.tcinc.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27773; Sat, 2 Dec 1995 22:31:45 -0600 Received: from animas.tcinc.com (animas [165.137.131.8]) by peacock.tcinc.com (8.7.1/8.7.1) with SMTP id VAA07362 for ; Sat, 2 Dec 1995 21:31:39 -0700 (MST) Received: from den-mon1. by animas.tcinc.com (SMI-8.6/SMI-SVR4) id VAA09824; Sat, 2 Dec 1995 21:31:37 -0700 Received: from den-mon1 (localhost) by den-mon1. (5.x/SMI-SVR4) id AA22322; Sat, 2 Dec 1995 21:29:57 -0700 Message-Id: <9512030429.AA22322@den-mon1.> Date: Sat, 02 Dec 95 21:29:57 -0700 From: Julian Colmenero X-Mailer: Mozilla 1.1 (X11; U; SunOS 5.4 sun4m) Mime-Version: 1.0 To: lisp-stat-news@umnstat.stat.umn.edu Subject: xlisp-stat X-Url: http://stat.umn.edu/~rcode/get-xlispstat.html Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=us-ascii Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk -- Hi Is xlisp-stat an object-program. If it is, how to i make a graph. I retrieved the documents, but i do not have a mc to print them. ************************************************* * * * Julian V. Colmenero * * e-mail colmenero.julian@tcinc.com * * phone# (303) 267-6324 * * * ************************************************* From majordom@umnstat.stat.umn.edu Sun Dec 3 06:29:12 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA07853; Sun, 3 Dec 1995 06:29:12 -0500 Return-Path: Received: from sunset.ma.huji.ac.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA07848; Sun, 3 Dec 1995 05:27:13 -0600 Received: by sunset.ma.huji.ac.il id AA00117 (5.67b/HUJI 4.152 for stat-lisp-news@umnstat.stat.umn.edu); Sun, 3 Dec 1995 13:26:00 +0200 Date: Sun, 3 Dec 1995 13:26:00 +0200 From: "Harvey J. Stein" Message-Id: <199512031126.AA00117@sunset.ma.huji.ac.il> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Future direction desires... Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The discussion of future directions for xlispstat has spurred me into thinking about what I'd like to see from it. So, here's my wish list... 1. Documentation. The biggest problem I have with using xlispstat is incomplete documentation. I have the lisp-stat book, the xlisp reference manual, and CLTL2. But, these three documents are far from enough. One problem is that the lisp-stat book is not up to date - some things are done differently than the book explains. Even the online documentation is out of date. The other problem is that the basic objects and their usage aren't completely documented. I often end up grepping the C & lisp code in the distribution when trying to figure out how to do something, or even just when trying to figure out what's available. For me, having good documentation for xlispstat is of primary importance. For example, > (help 'make-array) MAKE-ARRAY [function-doc] Args: (dimensions &key initial-element (initial-contents nil) (displaced-to nil)) Creates an array of the specified DIMENSIONS. The default for INITIAL- ELEMENT depends on ELEMENT-TYPE. NIL It mentions element-type, but doesn't say what it may be, or how to specify it. I only found out by digging through cltl2 & reading the xlisp source code (the two don't completely agree). Tom Almy's xlisp-plus manual (version 2.1g) doesn't mention element-type. As for the blas routines, I couldn't find any documentation whatsoever. Furthermore, many object messages are undocumented. So, I guess a function & object reference manual would be a great help. Additionally, I'd like to see documentation combined with either references and discussion. Something like a reference manual which goes down to the tutorial level. Something which doesn't just say: COOKS-DISTANCES Message args: () Computes Cook's distances. Being a beginner in statistical analysis, I'd like to at least see a reference to some discussion of what this is and what it's good for. Ideally, the documentation would include definitions and usage. This is a bit much to ask for, since it would make the documentation a fairly complete statistics textbook. Maybe just a reference manual with afew statistics textbooks listed in the introduction as required background reading would be sufficient. On the other hand, this aspect can go pretty far - consider for example an html statistics textbook with xlispstat as a helper application. 2. Efficiency. I'd like to see xlispstat become more memory efficient and faster. For example, I had to upgrade my PC (a 486dx66 running Linux) from 16 megabytes to 32 megabytes to be able to do some of the computations I was interested in. Now, I can just barely do 15 variable regressions with 22,000 data points. One would expect this to require at least 8*16*20,000 = 2,560,000 bytes of memory, but my xlispstat binary grew to 29 meg. Just reading in 22,000 data points (8 variables) causes it to swell from 2.5 meg to 12.6 meg. I'm sure that some of this is my fault for reading in lists instead of arrays. But to read in an unboxed array, I'd still have to use (make-array 22000 :element-type 'c-double :initial-contents '(...)), so I'd still create a large amount of garbage. I'd like xlispstat to be industrial-strength in its data handling & numerical computations. For this, it should get all of blas (and maybe lapack), have a compiler (not just a byte compiler), and have a foreign function interface (which hopefully wouldn't be needed often). In other words, I'd like to see enhancements along the lines of the article "Fast Floating-Point Processing in Common Lisp" by Fatemean, Broughan, Willcock and Rettig. I think it's close, that having a fast free lisp is extremely important, and that being fast and memory efficient is also important for statistics. On the other hand, to what extent it *should* be fast and able to handle large datasets depends on what audience xlispstat is intended for. Is it to be a teaching tool, a statistics research tool, or a full fledged statistics package? I'd like it to also function as a numerical algorithms testbed. As a teaching tool, the speed and ability to handle large datasets isn't so necessary. On the other hand, it'd be nice for students to be able to animate distributions - for example, to draw a bivariate normal pdf and watch it change as the correlation coefficient changes. This is currently doable, but would look alot nicer if xlispstat was faster. Furthermore, it'd be nice if students could learn to use a stat package that they can continue to use as professionals. As a full fledged statistics package, it should be able to crunch data in a big way. I'm not a statistician, and I'm not familiar with other statistics packages, so I don't know how many data points one typically needs to handle. But I have about 800,000 10 dimensional data points. I'm currently using various techniques to reduce the amount of data before loading it into xlispstat, but it'd be nice to be able to at least handle 100,000 data points. Also, alot of the functions (regression, kernel smoothing, etc) seem to create alot of garbage. A re-write based on blas et al might significantly speed up these functions as well as reduce the amount of garbage produced (thus increasing the size of problems addressable). At the very least, it'd be nice to have some documentation on how to run xlispstat efficiently. Under what conditions should I use arrays instead of lists, what iteration functions are fastest, etc. A profiling package would also help. 4. Objects or Functions? The flexibility of the object system is hampered by the use of functions. For example, there's a function plot-function which is an easy interface to the scatterplot-proto. If I like a couple of extra things in my scatterplots, I can only add methods to scatterplot-proto. I can't define a new object which inherets from scatterplot-proto, because then I lose the use of these helper functions. Furthermore, any other code which uses the helper functions would then not get my new functionality because the helper functions instantiate copies of scatterplot-proto. This means that if I want to enhance the system, I can only enhance the standard objects, I can't inherit from them. If I inherit from them I also have to modify or replace all the helper functions. If I'm only adding new functionality, then this isn't such a problem. I just define new methods. The problematic situations are the ones where one would like to add a feature to a given message - i.e. - where one would override a message and use call-next-method to invoke the parent's behavior. This is maybe a minor point, but I do bump into it regularly. 3. Package documentation. Analogous to point 1, but somewhat orthogonal is the sparse documentation for the packages in the xlispstat ftp repository. For example, in the statistics/regression directory, there are 17 subdirectories. Many have much code and no documentation whatsoever. For example, .../regression/sir contains 2 subdirectories full of lisp code. There's not even a README file saying what to load first, how to use the package, etc. I don't understand how anyone can be expected to get any use at all out of the code. It's a large body of lisp code, and one would have to read it all to figure out how to use it. Furthermore, one would have to read it without any idea of what it does or what the top level interface is supposed to be. Not that I want to single out this package in particular, or that I want to complain that the author who's giving away this package for free didn't do more work. But having a source repository which is full of such packages makes it very difficult to find useful packages that are easy to use. It would take weeks to go through the regression directory trying to figure out what each package is supposed to do & how to use it. Having all this undocumented code there makes the directory *less* valuable. One possibility is to only accept packages which include a README file containing a description of what they do and how to start them up. Another possibility is to split the repository into documented & undocumented. In any case, I think that a well maintained source repository is very important. Looking forward to fast numerical processing in xlispstat, Dr. Harvey J. Stein Berger Financial Research hjstein@math.huji.ac.il From majordom@umnstat.stat.umn.edu Sun Dec 3 22:30:27 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10532; Sun, 3 Dec 1995 22:30:27 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10527; Sun, 3 Dec 1995 21:30:25 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA18886; Sun, 3 Dec 95 19:30:22 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA29727; Sun, 3 Dec 95 19:30:21 PST From: Jan Deleeuw Message-Id: <9512040330.AA29727@galton.math.ucla.edu> Subject: Xlisp-Stat from WWW To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Sun, 3 Dec 1995 19:30:20 -0800 (PST) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 688 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Jose Garcia and Jason Bond have improved the WWW interface, which now read and writes on the same page (using Tom Christensen's open2.pl). http://www.stat.ucla.edu/cgi-bin/Xlisp-Stat.cgi Jose is working on uploading files from a library (server-side) and from the client. Graphics and Help are next. At some point we may have to restrict access, but of course the perl or C is always available. --- Jan -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Tue Dec 5 14:26:36 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10808; Tue, 5 Dec 1995 14:26:36 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10803; Tue, 5 Dec 1995 13:26:32 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA20239; Tue, 5 Dec 95 11:26:28 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA12321; Tue, 5 Dec 95 11:26:27 PST From: Jan Deleeuw Message-Id: <9512051926.AA12321@galton.math.ucla.edu> Subject: Correlation Homepage To: ucla-stat@math.ucla.edu (ucla-stat mailing list), stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Tue, 5 Dec 1995 11:26:26 -0800 (PST) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1301 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk FYI After the normal distribution http://www.stat.ucla.edu/textbook/form-based/cdf/normal/ the correlation coefficient now also has its homepage (I guess my psychometric origins are showing). This is http://www.stat.ucla.edu/textbook/correlation/ This shows the working format for the textbook. Discussion, if possible by using quotations from the history of statistics. Then some traditional material (formulas, computing). This will include actual forms for doing the computations. Then interactive demos, for which you need Xlisp-Stat, which you needed anyway. I will include material written by others (with acknowledgements and permissions, of course). Because there are no limitations on space and creativity, and because anything on the WWW is "under construction" anyway, I think the textbook can in principle serve both undergraduate, graduate, and postdoc purposes. Thus I hereby invite contributions which fit in the framework established so far. For the moment, I am the only reviewer. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Wed Dec 6 01:26:38 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14165; Wed, 6 Dec 1995 01:26:38 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14160; Wed, 6 Dec 1995 00:26:34 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA10750; Tue, 5 Dec 95 22:26:32 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA14083; Tue, 5 Dec 95 22:26:30 PST Date: Tue, 5 Dec 95 22:26:30 PST Message-Id: <9512060626.AA14083@galton.math.ucla.edu> From: Jan Deleeuw To: ucla-stat@math.ucla.edu, stat-lisp-news@umnstat.stat.umn.edu Subject: bivariate normal calculator Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk In http://www.stat.ucla.edu/textbook/gendarme/ you will find a nice dynamic demo of the bivariate normal density (by B. Narasimhan), plus a form-based calculator for the bivariate normal cdf. For the demo you need Xlisp-Stat on your WWW client machine. The calculator is http://www.stat.ucla.edu/textbook/gendarme/bivariate-normal-cdf.html The CGI is written in C. You get the same results from Lisp by using http://www.stat.ucla.edu/cgi-bin/Xlisp-Stat.cgi and then asking for (bivnorm-cdf x-value y-value r-value) --- Jan From majordom@umnstat.stat.umn.edu Wed Dec 6 12:29:06 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25356; Wed, 6 Dec 1995 12:29:06 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25351; Wed, 6 Dec 1995 11:29:02 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA02639; Wed, 6 Dec 95 09:28:46 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA18817; Wed, 6 Dec 95 09:28:41 PST From: Jan Deleeuw Message-Id: <9512061728.AA18817@galton.math.ucla.edu> Subject: Power Calculator on WWW To: ucla-stat@math.ucla.edu (ucla-stat mailing list), stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Wed, 6 Dec 1995 09:28:41 -0800 (PST) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 931 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk This is still in early alpha. Inspired by stplan by Barry Brown et al at the MD Anderson Cancer Center, University of Texas, Houston (consists of FORTRAN programs with a command line interface), reprogrammed in Xlisp-Stat by Jason Bond, HTML and CGI in perl by Jason Bondand Jose Garcia. You dont need Xlisp-Stat or perl for this, everything is run on our server. http://www.stat.ucla.edu/~jbond/HTMLPOWER/newindex.html Problems: the interface is barebones, the output from Xlisp-Stat is copious (the answer you asked for is on the last line). Nevertheless, it would be nice if some of you pounded on it, and reported flaws to jbond@stat.ucla.edu -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Fri Dec 8 03:20:41 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27558; Fri, 8 Dec 1995 03:20:41 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27552; Fri, 8 Dec 1995 02:20:35 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA20664; Fri, 8 Dec 95 00:20:33 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA27420; Fri, 8 Dec 95 00:20:30 PST From: Jan Deleeuw Message-Id: <9512080820.AA27420@galton.math.ucla.edu> Subject: Binomial Calculator To: ucla-stat@math.ucla.edu (ucla-stat mailing list), stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Fri, 8 Dec 1995 00:20:30 -0800 (PST) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 779 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk http://www.stat.ucla.edu/textbook/probmodels/bionomial/ now has a calculator for the binomial cdf (based on the relation with the incomplete beta), plus a dynamic demo illustrating the normal and Poisson approximation to the binomial. The normal approximation actually illustrates the De Moivre version of the central limit theorem. We will put in, in the next couple of days, the binomial pdf and the functional inverse of the cdf, plus a calculator illustrating the weak law of large numbers. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Fri Dec 8 12:05:31 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29891; Fri, 8 Dec 1995 12:05:31 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29886; Fri, 8 Dec 1995 11:05:27 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA07388; Fri, 8 Dec 95 09:05:24 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA02017; Fri, 8 Dec 95 09:05:24 PST Date: Fri, 8 Dec 95 09:05:24 PST Message-Id: <9512081705.AA02017@galton.math.ucla.edu> From: Jan Deleeuw To: stat-lisp-news@umnstat.stat.umn.edu Subject: [sci.stat.consult] Re: Bivariate Normal Calculator and Demo Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk ------- Start of forwarded message ------- Path: nnrp.info.ucla.edu!usenet From: deleeuw@galton.math.ucla.edu (Jan Deleeuw) Newsgroups: sci.stat.edu,sci.stat.consult,sci.stat.math Subject: Re: Bivariate Normal Calculator and Demo Date: 08 Dec 1995 00:48:42 -0800 Organization: UCLA Center for Statistics Lines: 39 Sender: deleeuw@galton.math.ucla.edu Message-ID: References: NNTP-Posting-Host: galton.math.ucla.edu In-reply-to: partchev@mip.paed.uni-muenchen.de's message of Wed, 6 Dec 1995 10:15:36 X-Newsreader: Gnus v5.1 Xref: nnrp.info.ucla.edu sci.stat.edu:4329 sci.stat.consult:9257 sci.stat.math:5327 In article partchev@mip.paed.uni-muenchen.de (PARTCHEV IVAILO) writes: I think times are ripe now for a new group -- something like, um, alt.anti-xlisp-stat This is fine. If you look at the list of alt newsgroups you will find even weirder groups. Remember, however, that for the bivariate normal, binomial, normal, logistic, regression, and correlation calculators in the UCLA Statistics Textbook project you do not need Xlisp-Stat. They are done in C on our server. For the power calculator, you also dont need Xlisp-Stat. It is run transparantly on our machine. For the dynamic demos you need to have Xlisp-Stat installed on your Mac, Windows, or Unix web server, but only as a helper. There is no need to learn a single line of Lisp. In short, you get everything for free, even CPU cycles, and you dont have to learn a single line of Lisp. All you need to be able to do is click your mouse and type some numbers in a form. It seems to me that it make sense IN THIS CONTEXT to subscribe to alt.anti-xlisp-stat, if you object to the fact that software is free, that programs run unmodified on all kinds of machines, and that source code is available without limitations by ftp or on request. I agree these are serious disadvantages. I personally would much prefer a black box, which does not run on Macs or FreeBSD or Linux, costs an arm and a leg, and looks like a mainframe program with a facelift. By the way, the UCLA Statistics WWW server now gets 8000 hits, more than 3000 accesses, each day. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw ------- End of forwarded message ------- From majordom@umnstat.stat.umn.edu Sun Dec 10 19:00:57 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28432; Sun, 10 Dec 1995 19:00:57 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28427; Sun, 10 Dec 1995 18:00:52 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA21438; Sun, 10 Dec 95 16:00:50 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA16083; Sun, 10 Dec 95 16:00:45 PST From: Jan Deleeuw Message-Id: <9512110000.AA16083@galton.math.ucla.edu> Subject: Xlisp-Stat on the WWW To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Sun, 10 Dec 1995 16:00:45 -0800 (PST) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 730 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The CGI script has been improved -- we are now getting rid of Xlisp-Stat orphans -- we impose a max of 15 secs user time -- we write the output to a scrollable window on the form -- there is a button for a hypertext version of Tim Mikkelsen Xlisp Reference -- there is a button for a hypertext version of Tom Almy/David Betz User Manual -- the CGI is still in perl, but we are working on a C version Comments and suggestions to jose@stat.ucla.edu -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon Dec 11 12:00:09 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10787; Mon, 11 Dec 1995 12:00:09 -0500 Return-Path: Received: from cs.sfu.ca by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10782; Mon, 11 Dec 1995 11:00:04 -0600 Received: from [142.58.140.2] (schwarz.math.sfu.ca) by cs.sfu.ca with SMTP id AA20243 (5.65c/IDA-1.4.4 for ); Mon, 11 Dec 1995 08:59:59 -0800 Message-Id: <199512111659.AA20243@cs.sfu.ca> X-Sender: cschwarz@cs.sfu.ca (Unverified) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 11 Dec 1995 09:02:42 -0800 To: stat-lisp-news@umnstat.stat.umn.edu From: cschwarz@cs.sfu.ca (Carl Schwarz) Subject: Tools for Modal dialogs Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Are there additional tools for modal dialogues that will prompt the user for a number, edit the responses, and then only return when valid? It would be similar to the edit-text-item-proto except you would pass it a number type (integer, real, etc) and a valid range and it would ensure that the users input is valid and within range. At the moment, I'm using the edit-text-item-proto and then converting the string to numbers using a read, but is there an easier way? Thanks. Carl James Schwarz Department of Mathematics and Statistics Simon Fraser University Burnaby, BC V5A 1S6 Phone: (604) 291-3376 Fax (604) 291-4947 Email cschwarz@cs.sfu.ca WWW http://www.math.sfu.ca/stats/ From majordom@umnstat.stat.umn.edu Mon Dec 11 13:10:52 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11430; Mon, 11 Dec 1995 13:10:52 -0500 Return-Path: Received: from euler.bd.psu.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11425; Mon, 11 Dec 1995 12:10:48 -0600 Received: from localhost by euler.bd.psu.edu via SMTP (940816.SGI.8.6.9/940406.SGI) for id NAA04212; Mon, 11 Dec 1995 13:12:53 -0500 Message-Id: <199512111812.NAA04212@euler.bd.psu.edu> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Tools for Modal dialogs In-Reply-To: Your message of "Mon, 11 Dec 1995 09:02:42 -0800" References: <199512111659.AA20243@cs.sfu.ca> X-Mailer: Mew beta version 0.98 on Emacs 19.30.1 Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Date: Mon, 11 Dec 1995 13:12:48 -0500 From: Balasubramanian Narasimhan Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Carl> Are there additional tools for modal dialogues that will Carl> prompt the user for a number, edit the responses, and then Carl> only return when valid? It would be similar to the Carl> edit-text-item-proto except you would pass it a number type Carl> (integer, real, etc) and a valid range and it would ensure Carl> that the users input is valid and within range. Carl> At the moment, I'm using the edit-text-item-proto and then Carl> converting the string to numbers using a read, but is there Carl> an easier way? Code that would help can be found in my utilities package which is part of an (as yet) incomplete literate program. I will distribute the utilities toolbox at some point. The code is completely visible at http://euler.bd.psu.edu/~naras/bpois.ps. ________________________________________________________________ B. Narasimhan naras@euler.bd.psu.edu http://euler.bd.psu.edu/~naras From majordom@umnstat.stat.umn.edu Tue Dec 12 10:02:26 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25149; Tue, 12 Dec 1995 10:02:26 -0500 Return-Path: Received: from mimsy.cs.umd.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25144; Tue, 12 Dec 1995 09:02:22 -0600 Received: by mimsy.cs.UMD.EDU (8.6.11/UMIACS-0.9/04-05-88) id KAA13518; Tue, 12 Dec 1995 10:01:08 -0500 Received: (from pcolsen@localhost) by n2ell.columbia.md.us (8.6.12/8.6.9) id JAA05567; Tue, 12 Dec 1995 09:43:56 -0500 Date: Tue, 12 Dec 1995 09:43:56 -0500 From: "Peter C. Olsen" Message-Id: <199512121443.JAA05567@n2ell.columbia.md.us> To: stat-lisp-news@umnstat.stat.umn.edu, gcl@cli.com Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk genetic-programming@cs.stanford.edu, informatik.uni-stuttgart.de!snns Subject: Call For Entries: Mathematical Contest in Modelling Reply-To: pcolsen@acm.org The the annual national undergraduate Mathematical Contest in Modelling is soliciting entries. The is the LaTeX2e for a brief announcement. %%%%%%%%%%%%%%%%%%%%%%%%%% cut here %%%%%%%%%%%%%%%%%%%%%%%%% \documentclass[12pt]{article} \begin{document} \title{Mathematical Contest in Modelling} \author{} \date{November 14, 1995} \maketitle{} The Mathematical Contest in Modelling (MCM), a national contest for college undergraduates, will hold its twelfth competition in February 1996. The MCM is designed to stimulate and improve problem-solving and writing skills in a team setting. Last year, more than 320 teams submitted solution papers, representing 199 schools in 9 countries. Outstanding solution papers will be published in October 1996 in a special issue of \emph{The UMAP Journal}. You can be part of the MCM effort by serving as a faculty advisor. The advisor helps to organize the team, distribute contest materials, and return solution papers to COMAP. Further information is available from Clarice Callahan at The Consortium for Mathematics and its Applications (COMAP), via \begin{itemize} \item email: \texttt{mcm@comap.com} \item U.~.S.~Mail: COMAP, Suite 210, 57 Bedford Street, Lexington, MA 02173. \item Telephone: 617-862-7878 (ext 37) \item FAX: 617-863-1202 \end{itemize} \end{document} \centerline{\textbf{Applications must be received by 6 Jan 1996}} %%% Local Variables: %%% mode: latex %%% TeX-master: t %%% End: From majordom@umnstat.stat.umn.edu Tue Dec 12 13:05:42 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26282; Tue, 12 Dec 1995 13:05:42 -0500 Return-Path: Received: from super.super.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26277; Tue, 12 Dec 1995 12:05:39 -0600 Received: from gotham.super.org (gotham [192.239.79.2]) by super.super.org (8.7.1/8.6.12.1) with ESMTP id NAA08657; Tue, 12 Dec 1995 13:00:35 -0500 (EST) Received: from hume.super.org (hume [192.239.79.11]) by gotham.super.org (8.6.12/8.6.12.1) with ESMTP id NAA04520; Tue, 12 Dec 1995 13:03:34 -0500 From: pcolsen@super.org (Peter C Olsen) Received: (pcolsen@localhost) by hume.super.org (8.6.12/8.6.12.client) id NAA07568; Tue, 12 Dec 1995 13:03:34 -0500 Date: Tue, 12 Dec 1995 13:03:34 -0500 Message-Id: <199512121803.NAA07568@hume.super.org> To: stat-lisp-news@umnstat.stat.umn.edu, gcl@cli.com, genetic-programming@cs.stanford.edu, informatik.uni-stuttgart.de!snns@super.org Subject: Call For Entries: Mathematical Contest in Modelling Cc: fgiordan@carroll.edu, mcm@comap.org X-Disclaimer: These opinions are mine, alone. Reply-To: pcolsen@acm.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The the annual national undergraduate Mathematical Contest in Modelling is soliciting entries. The is the LaTeX for a brief announcement. %%%%%%%%%%%%%%%%%%%%%%%%%% cut here %%%%%%%%%%%%%%%%%%%%%%%%% \documentstyle[12pt]{article} \title{Mathematical Contest in Modelling} \author{} \date{November 14, 1995} \begin{document} \maketitle{} \pagestyle{empty} The Mathematical Contest in Modelling (MCM), a national contest for college undergraduates, will hold its twelfth competition in February 1996. The MCM is designed to stimulate and improve problem-solving and writing skills in a team setting. Last year, more than 320 teams submitted solution papers, representing 199 schools in 9 countries. Outstanding solution papers will be published in October 1996 in a special issue of {\em The UMAP Journal}. You can be part of the MCM effort by serving as a faculty advisor. The advisor helps to organize the team, distribute contest materials, and return solution papers to COMAP. Further information is available from Clarice Callahan at The Consortium for Mathematics and its Applications (COMAP), via \begin{itemize} \item email: {\tt mcm@comap.com} \item U.~.S.~Mail: COMAP, Suite 210, 57 Bedford Street, Lexington, MA 02173. \item Telephone: 617-862-7878 (ext 37) \item FAX: 617-863-1202 \end{itemize} \centerline{\bf Applications must be received by 6 Jan 1996} \end{document} %%% Local Variables: %%% mode: latex %%% TeX-master: t %%% End: From majordom@umnstat.stat.umn.edu Mon Dec 18 13:26:28 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17525; Mon, 18 Dec 1995 13:26:28 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17520; Mon, 18 Dec 1995 12:26:23 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id NAA14258; Mon, 18 Dec 1995 13:26:14 -0500 X-Sender: uluru@gibbs.oit.unc.edu (Unverified) Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 18 Dec 1995 13:25:18 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: uluru@gibbs.oit.unc.edu (Forrest Young) Subject: New Release of ViSta Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Announcing a New Macintosh Version of ViSta - The Visual Statistics System Version 4.21, December 17th, 1995 Copyright (c) 1991-95 by Forrest W. Young This release of ViSta adds Univariate Analysis and fixes bugs in the Macintosh and Power Macintosh versions. If you are using earlier releases of ViSta on an Apple Macintosh or PowerMac, I strongly recommend you obtain this new release, which is much more robust. This release is for Macintoshes only. The Windows version, which does not suffer from the problems in preceeding Macintosh versions, will be updated soon. ====================================================================== Changes since the previous publically announced release (4.12) ====================================================================== 1) A new Univariate Analysis module is included, which does t- or Z-tests (and confidence intervals) for one sample, two paired samples or two independent samples, and does Mann-Whitney and Wilcoxen tests in appropriate cases. The visualization is not yet available. 2) Changes have been made so that opened data icons reveal an uneditable datasheet for browsing. This datasheet may be edited with the new "Edit Datasheet" menu item (which may also be used to open data icons to an editable datasheet). As part of this change, the "Show-Datasheet" item has been changed to "Edit Datasheet"; "Open Data" now does a "Browse Data" rather than a "Show Datasheet" after the data are loaded; and double-clicking a data icon does a "Browse Data" rather than a "Show Datasheet". 3) The File and Analysis menus and the WorkMap's toolbar have been revised in preparation for Univariate Analysis and for other modules under development. ====================================================================== Bugs addressed since the previous publically announced release (4.12) ====================================================================== 1) A bug was fixed which, on the Macintosh, sometimes caused a fatal error when quitting ViSta with many open windows. This bug did not effect ViSta's computations, since it occured after the decision to quit, but it could completely freeze up the hardware, requiring a restart. 2) A bug was fixed which, on the Macintosh, sometimes prevented the workmap from being redrawn after closing dialog boxes. This was a fatal, non-reproducible bug which prevented data analysis from proceeding. This bug was fixed in release 4.20, which has been available for a few weeks but was not announced. 3) Errors due to insufficient memory are difficult to deal with, particularly on the Macintosh, where they may cause the system to die. If insufficient memory is allocated ViSta simply "disappears". A memory warning dialog has been added that suggests the ammount of memory that may be needed. The warning appears after ViSta exits inappropriately. 4) Bugs have been removed from "Create Data". 3) A bug has been removed which caused errors when converting table data to multivariate class data. ====================================================================== Availability ====================================================================== The complete code for ViSta, including documentation, test data, examples, help, guidance, and the XLisp-Stat base, is available from our WWW and FTP servers. WWW: http://forrest.psych.unc.edu FTP from www.psych.unc.edu, directory pub/forrest/vista. ====================================================================== The remainder of this message summarizes earlier messages about ViSta. ====================================================================== ======================================= What is ViSta? ======================================= ViSta, is a Visual Statistics system that is > open > extensible > freely redistributable PLATFORMS: ViSta is available for > Microsoft Windows 3.1 > Macintosh and PowerMac microcomputers > A Unix version is in preparation AUDIENCE: ViSta is useful for > UNDERGRADUATES: Teaching introductory statistics to undergraduates > GRADUATES: Teaching dynamic statistical graphics and computational >statistics to graduate students > RESEARCHERS: Research and development in dynamic statistical graphics and >in computational statistics ViSta is designed for an audience of users having a very wide range of data analysis sophistication, ranging from novice to expert. ViSta provides seamlessly integrated data analysis environments specifically tailored to the user's level of expertise. Guidance is available for students and novices, and tools are available for teachers and other experts to create guidance for these novices. A structured graphical user interface is available for competent users, and a command line interface is available for sophisticated users. The complete Lisp-Stat (Tierney, 1990) programming environment is available to researchers, programmers and graduate students who wish to extend ViSta's capabilities. ======================================= ViSta's main features ======================================= > WORKMAPS Maps that visualize the overall structure of an on-going data analysis session. These maps are created by ViSta as the data analysis session progresses. They aid novice and competent analysts in remembering their analysis session, and permit the analyst to return to earlier steps of the analysis. > GUIDEMAPS Maps that visually guide students and other novice data analysts through their data analysis sessions. These maps are created by experts using ViSta's Authoring tools. > SPREADPLOTS Multi-window dynamic statistical graphics that visualize data structure and analysis results > MODEL RE-VISION Dynamic interactive graphical tools that permit the user to visually re-estimate model parameters > ViDAL ViSta's Data Analysis Language. ViDAL is an extension of the underlying Lisp-Stat language that can be used by sophisticated data analysts for data analysis and by programmers to enhance ViSta's capabilities. > DATASHEET A graphical datasheet for displaying and editing data. > AUTHORING TOOLS Visual tools for authoring new GuideMaps. ======================================= Data Analysis Capabilities ======================================= ViSta's data analysis capabilities include: > DYNAMIC GRAPHICS Histograms, Boxplots, Diamond Plots, Dotplots, Scatterplots, Biplots, Spinplots, Scatterplot Matrices. Plots support brushing and labeling, and are dynamically linked. > DESCRIPTIVE STATISTICS Means, Standard Deviations, Variances, Ranges, Quartiles, Medians, Correlations, Covariances, Distances >UNIVARIATE ANALYSIS T- and Z-tests and confidence intervals for single sample, paired samples and two independent samples data, with Wilcoxon Signed-Rank and Mann-Whitney tests in appropriate situations. > ANOVA Univariate analysis of variance of orthogonal one- or multi-way data. Model may or may not include two-way interactions. The model visualization is a spreadplot comprised of boxplot, diamond plot, quantile plot, quantile-quantile plot and effects plot. > REGRESSION Simple and multiple OLS regression analysis (multivariate, robust and monotonic regression in preparation). The model visualization is a spreadplot comprised of a biplot, spinplot, histogram and scatterplot-matrix. > PRINCIPAL COMPONENTS Analysis of correlations or covariances. The model visualization is a spreadplot comprised of a biplot, spin-plot, scree-plot and scatterplot-matrix. > MULTIDIMENSIONAL SCALING Analysis of one or more symmetric or asymmetric matrices. The model visualization is a spreadplot comprised of a scatterplot, spin-plot, scree-plot and scatterplot-matrix. The spreadplot supports graphical re-estimation of model parameters. > CORRESPONDENCE ANALYSIS Analysis of two-way contingency tables. The model visualization is a spreadplot comprised of a biplot, spinplot, residuals plot and scree-plot. The spreadplot supports graphical re-estimation of model parameters. ======================================= Hardware Requirements ======================================= MACINTOSH: There are two versions of ViSta for the Macintosh, one for the PowerMac and one for 680xx Macintoshes. For 680xx Macs, we recommend a 68040 processor (at the minimum, a 68030 + 68881 math coprocessor). Both versions require System 7 or later, 3 MB of free hard disk space and at least 6 MB of RAM (8 for the PowerMac). We recommend at least a 17", 8-bit (256 color) screen (14" minimum). For 12-bit or 24-bit screens, a larger memory partition is required (add 1/2 MB for 12-bit, 1 MB for 24-bit). WINDOWS: The Windows version of ViSta is for MS-Windows 3.1 with the 32-bit extension. Users report that this version of ViSta also works under Windows 95, though no testing has been done. This version requires 3 MB of free hard disk space and 8 MB of RAM. We recommend at least a 17", 8-bit (256 color) screen (14" minimum). For 12-bit or 24-bit screens, more memory is required (about 1/2 MB for 12-bit, 1 MB for 24-bit). ======================================= ViSta is a work-in-progress ======================================= ViSta is a work-in-progress. Some features are incomplete. While quality assurance testing has been done, this is a beta-release system that is not fully mature. At the current stage of development, ViSta is viable as a supplemental statistical software system used for extra credit exercises in introductory statistics, and as the focus of graduate courses on research in computational and graphical methods in statistics. It may be premature to use ViSta in undergraduate statistics classes that need a complete and robust statistic analysis system. Currently, the Delete Data, Delete Model and Simulate Data menu items are not implemented; the ANOVA visualization is incomplete; the univariate analysis visualization is not available; MULREG does not do multivariate tests (it does many separate multiple regressions); transformations are not available with matrix data; model interpretation is unimplemented; and the Help and GuideMap systems are incomplete. A univariate multiple regression module (which complements the current multivariate multiple regression module with more extensive visualizations and with the choice of robust or monotonic regression) is being written but is not yet available. ======================================= Documentation ======================================= The following two monographs describe how to use ViSta: "ViSta, The Visual Statistics System", is available from the server mentioned above. A nicely bound and printed version is also available for $5.00 from the address below. "ViSta Corresp", a description of the Correspondence Analysis procedure in ViSta, is available for $5.00 from Forrest Young at the address below. The following paper has been recently published: Young, F.W. & Lubinsky, D.J. "Guiding Data Analysis with Visual Statistical Strategies". J. Computational and Graphical Statistics, Dec., 1995. In addition, the following paper is available for $5.00 from Forrest Young at the address below. Young, F.W., et al. "Visual Statistical Analysis - Papers presented at the Special Contributed Session on Visual Statistical Analyais at Interface '94. Psychometric Research Memorandum 94-2, July, 1994. ___________________________________________________________________ Forrest Young e-mail: forrest@unc.edu UNC Psychometrics Lab WWW: http://forrest.psych.unc.edu CB-3270 Davie Hall Telephone: 919-962-5038 Chapel Hill NC 27599-3270 USA FAX: 919-962-2537 ___________________________________________________________________ From majordom@umnstat.stat.umn.edu Mon Dec 18 17:28:47 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19393; Mon, 18 Dec 1995 17:28:47 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19388; Mon, 18 Dec 1995 16:28:42 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA28082; Mon, 18 Dec 95 14:28:36 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA10777; Mon, 18 Dec 95 14:28:35 PST From: Jan Deleeuw Message-Id: <9512182228.AA10777@galton.math.ucla.edu> Subject: Convergence of transition matrices To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Mon, 18 Dec 1995 14:28:34 -0800 (PST) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 676 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On ftp://ftp.stat.ucla.edu/pub/xlispstat/code/ ftp.stat.ucla.edu:/pub/xlispstat/code http://www.stat.ucla.edu/code/ there is new or revised code in: statistics/temporal/markov.lsp Graphical demo to illustrate iteration and convergence of the transition matrix of a stationary chain -- and convergence of the stationary distribution. By Jason Bond (jbond$stat.ucla.edu). -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon Dec 25 13:13:48 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16937; Mon, 25 Dec 1995 13:13:48 -0500 Return-Path: Received: from angel.heaven.net by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16932; Mon, 25 Dec 1995 12:13:42 -0600 Received: from [198.69.29.97] (josh.netheaven.com [198.69.29.97]) by angel.heaven.net (8.6.12/8.6.12) with SMTP id NAA20701; Mon, 25 Dec 1995 13:12:48 -0500 Message-Id: <199512251812.NAA20701@angel.heaven.net> X-Sender: josh@netheaven.com Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 25 Dec 1995 13:14:30 -0500 To: clips-list@cosmic.uga.edu, stat-lisp-news@umnstat.stat.umn.edu From: josh@angel.heaven.net (Joshua Kuperman) Subject: Interprogram communication on Macs with AppleScript and Alpha Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I just wanted to throw this out for discussion. I am working on a project. CLIPS is ideal for its rule based production systems and Xlispstat is a truly wonderful statistical and mathematical workhorse. The text editor Alpha, Xlispstat, and a variant of CLIPS (see the third conference proceedings) all support AppleScript to some degree. None support AppleScript as intently as they might, but that doesn't matter. CLIPS is similar, enough in sytax to LISP so that one editor environment could support both adequately. Since Alpha supports AppleScript and is programable (in tcl shell) it can send expressions to the programs to be evaluated and store the results as tcl variable. I would just like to know that ideas of others on how to do this. It seems that a way of defining objects independent of the systems that generate them, like open-doc or ole, might be better. Yet AppleScript seems to be a very viable way of doing things - though not really standardized in any sense for CLIPS. AppleScipt's DoScript command, as long as the value is returned and as long as at least on application (for my purposes Alpha) is capable of sending the messages and receiving the return value should be sufficient for have a truly multiprogram architecture for invoking and using external programs as tools for an expert system. I'm just going to keep plugging away and see what I wind up with, but I'd love to hear any of the ideas of others. BTW, to see what an application that actually supports AppleScript is like, take a look at MacPerl, Netscape, or Excel. Josh Kuperman josh@racing.saratoga.ny.us http://racing.saratoga.ny.us/~josh (barely under construction) From majordom@umnstat.stat.umn.edu Wed Dec 27 13:01:37 1995 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10046; Wed, 27 Dec 1995 13:01:37 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10041; Wed, 27 Dec 1995 12:01:33 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA21112; Wed, 27 Dec 95 10:01:27 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA04466; Wed, 27 Dec 95 10:01:27 PST From: Jan Deleeuw Message-Id: <9512271801.AA04466@galton.math.ucla.edu> Subject: Wishlist To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Wed, 27 Dec 1995 10:01:26 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1289 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The following things seem important to have in Xlisp-Stat, mostly because of integration with the WWW. (a) Xlisp-Stat as an extension language, i.e. a library that can be linked into arbitrary applications (as in elk, guile, tcl), so that one can open an Xlisp-Stat interpreter etc. This would be great, because then we can put Xlisp-Stat interpreters in CGI applications. (b) A safe version of Xlisp-Stat, in which no files and pipes can be opened, and in which (system "") is disabled. This would be good for our WWW interface to Xlisp-Stat, which now uses a kludge to disable system and popen. See http://www.stat.ucla.edu/cgi-bin/Xlisp-Stat.cgi (c) A version of Xlisp-Stat which supports pipe and fork and exec and low-level IO. Good on general principles. (d) A version or extension of Xlisp-Stat which writes gif files on the fly to stdout (similar to the gnuplot stuff now build in for character terminals) using Tom Boutell's libgd. This makes it possible to write directly into browser windows. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Tue Jan 2 18:07:46 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22553; Tue, 2 Jan 1996 18:07:46 -0500 Return-Path: Received: from martin.luther.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22548; Tue, 2 Jan 1996 17:07:42 -0600 Message-Id: <9601022307.AA22548@umnstat.stat.umn.edu> Received: from [198.247.63.70] by martin.luther.edu with SMTP (1.37.109.4/16.2) id AA24920; Tue, 2 Jan 96 17:15:29 -0600 Date: Tue, 2 Jan 96 17:15:29 -0600 X-Sender: brilgo@martin.luther.edu (Unverified) X-Mailer: Windows Eudora Version 1.4.4 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: stat-lisp-news@umnstat.stat.umn.edu From: brilgo@martin.luther.edu (Gordon Bril) Subject: time series Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I am interested in finding out if anyone has written any XLISP-STAT code for Time Series, particularly Box-Jenkins ARIMA models. I have glanced at the contributed code labelled "Times" and "Sapaclisp" on Statlib, but this looks like its more for spectral type analysis. Thanks. Gordon Bril Dr. Gordon Bril Department of Mathematics Luther College Decorah, IA 52101 phone: (319) 387-1121 e-mail: brilgo@martin.luther.edu fax: 319 387 1088 From majordom@umnstat.stat.umn.edu Wed Jan 3 03:35:37 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03181; Wed, 3 Jan 1996 03:35:37 -0500 Return-Path: Received: from hypatia-appli.ethz.ch by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03176; Wed, 3 Jan 1996 02:35:30 -0600 Received: from (maechler@sophie [129.132.58.2]) by hypatia.math.ethz.ch (8.6.12/Main-STAT-mailer) with SMTP id JAA27867; Wed, 3 Jan 1996 09:35:22 +0100 Received: by (4.1/D-MATH-stat-client) id AA00467 for maechler@stat.math.ethz.ch; Wed, 3 Jan 96 09:37:04 +0100 Date: Wed, 3 Jan 96 09:37:04 +0100 Message-Id: <9601030837.AA00467@> From: Martin Maechler To: brilgo@martin.luther.edu Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <9601022307.AA22548@umnstat.stat.umn.edu> (brilgo@martin.luther.edu) Subject: Re: time series -- AR(I)MA code for Lisp-Stat Reply-To: Martin Maechler Cc: maechler@stat.math.ethz.ch Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I doubt a bit that code is available ((and I'll be too glad if I'm proven wrong!)). This is my experience: Time-domain time-series analysis ``aka Box-Jenkins ..'' has been a domain where dozens of "different" packages have been sold to the economics and econometrics community. And I've not seen any free source code for it at all (lest some AR code which is the trivial part of it). The one thing you'll find in the ApStat directory (of statlib) are subroutines to compute the ARMA-likelihood, but you have to stuff it in your own optimizer, and then this exact likelihood method is not the only one you'd want. The above situation is probably also the reason why S was week in this, till S-plus added that part. If anybody has someone (skilled) to program this (hopefully even in C) and make it freely available, this would be a real good deed for the public.. BTW: A Happy New Year to you all ! ----------------------------- Martin Maechler <>< Seminar fuer Statistik, SOL F5 ETH (Federal Inst. Technology) 8092 Zurich SWITZERLAND phone: x-41-1-632-3408 fax: ...-1086 http://www.stat.math.ethz.ch/~maechler/ From majordom@umnstat.stat.umn.edu Wed Jan 3 03:51:21 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03198; Wed, 3 Jan 1996 03:51:21 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03193; Wed, 3 Jan 1996 02:51:18 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA02692; Wed, 3 Jan 96 00:51:15 PST Received: by galton.math.ucla.edu (Sendmail 4.1/1.13) id AA20683; Wed, 3 Jan 96 00:51:13 PST From: Jan Deleeuw Message-Id: <9601030851.AA20683@galton.math.ucla.edu> Subject: WWW interfaces to Xlisp To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Wed, 3 Jan 1996 00:51:13 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 859 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Jose Garcia has translated his Perl CGI script to run Xlisp-Stat from Web browsers to C. See http://www.stat.ucla.edu/~jose/Xlisp-Stat.cgi and try to fancy help buttons ! I have made an interface to the Xlisp-Stat regression routines, also in C, which is in http://www.stat.ucla.edu/cgi-bin/textbook/regres-comp.cgi This will get a bunch of buttons to put in the Xlisp-Stat optional arguments. I still have to integrate the graphics in some way, and then GLM will be next. Of course we'll put in file upload as well. I'll try to keep the netscapisms to a minimum. --- Jan -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@nokomis.stat.umn.edu Thu Jan 4 16:03:39 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA14606; Thu, 4 Jan 1996 16:03:39 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA14600; Thu, 4 Jan 1996 15:03:36 -0600 Date: Thu, 4 Jan 1996 15:03:36 -0600 From: Luke Tierney Message-Id: <9601042103.AA14600@nokomis.stat.umn.edu> To: stat-lisp-news@nokomis.stat.umn.edu Subject: lisp-stat documents Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk Some xlispstat documents are now available on the WWW through links in the References section of the Lisp-Stat home page, http://stat.umn.edu/~luke/xls/xlsinfo/xlsinfo.html The documents are tutorial tech report glim documentation Windows version notes release notes for current version PostScript versions can also be obtained through links in the References section. luke From majordom@umnstat.stat.umn.edu Mon Jan 15 12:14:36 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA08552; Mon, 15 Jan 1996 12:14:36 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA08547; Mon, 15 Jan 1996 11:14:28 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id MAA07700; Mon, 15 Jan 1996 12:14:19 -0500 Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 15 Jan 1996 12:11:59 -0500 To: edstat-l@jse.stat.ncsu.edu, stat-lisp-news@umnstat.stat.umn.edu From: uluru@gibbs.oit.unc.edu (Forrest Young) Subject: New Release of ViSta Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Announcing a New Release of ViSta - The Visual Statistics System Version 4.22, January 8th, 1996 Copyright (c) 1991-96 by Forrest W. Young ====================================================================== Summary of this release ====================================================================== All Versions: > ViSta now includes a new Univariate Analysis module > Data Icons are now opened in an improved manner > Bugs have been removed from "Create Data" MS-WINDOWS: > ViSta is now available for the 16-bit version of MS Windows 3.1 > ViSta now incorporates Luke Tierney's Configuration code > This release fixes some minor window management bugs MACINTOSH: If you are using releases prior to 4.21 on a Macintosh or PowerMac, I strongly recommend you obtain this new release, which is much more robust due to fixes of two major bugs. WEB SITE: The ViSta Web Site (http://forrest.psych.unc.edu) has been expanded to provide much more complete information about ViSta's capabilities. ====================================================================== Changes in releases 4.21 and 4.22 ====================================================================== 1) A new Univariate Analysis module is included, which does t- or Z-tests (and confidence intervals) for one sample, two paired samples or two independent samples, and does Mann-Whitney and Wilcoxen tests in appropriate cases. The visualization is not yet available. 2) Changes have been made so that opened data icons reveal a datasheet which may be browsed. It may be edited with the new "Edit Datasheet" menu item (which may also be used to open data icons directly to an editable datasheet). As part of this change, the "Show-Datasheet" item has been changed to "Edit Datasheet"; "Open Data" now does a "Browse Data" rather than a "Show Datasheet" after the data are loaded; and double-clicking a data icon does a "Browse Data" rather than a "Show Datasheet". 3) The File and Analysis menus and the WorkMap's toolbar have been revised in preparation for Univariate Analysis and for other modules under development. 4) Luke Tierney's MS-Windows configuration code has been incorporated. The first time that ViSta is run a ViSta/XLS program group is created for the program manager. In addition screen font type and size are set, removing the problem that some users had with screen fonts that were inappropriate for ViSta's windows. 5) Import Data has been enhanced to import data with variables that have string values as well as symbol and number values. ====================================================================== Bugs fixed in releases 4.21 and 4.22 ====================================================================== 1) A bug was fixed which, on the Macintosh, sometimes caused a fatal error when quitting ViSta with many open windows. This bug did not effect ViSta's computations, since it occured after the decision to quit, but it could completely freeze up the hardware, requiring a restart. (Fixed in 4.21). 2) A bug was fixed which, on the Macintosh, sometimes prevented the workmap from being redrawn after closing dialog boxes. This was a fatal, non-reproducible bug which prevented data analysis from proceeding. (Fixed in 4.20). 3) Errors due to insufficient memory are difficult to deal with, particularly on the Macintosh, where they may cause the system to die. If insufficient memory is allocated ViSta simply "disappears". A memory warning dialog has been added that suggests the ammount of memory that may be needed. The warning appears after ViSta exits inappropriately. (Fixed in 4.21). 4) Bugs have been removed from "Create Data". (Fixed in 4.21). 5) A bug has been removed which caused errors when converting table data to multivariate class data. (Fixed in 4.21). 6) Several minor bugs in the size, location and closing of spreadplot windows in the Windows version have been eliminated. (Fixed in 4.22). ====================================================================== Availability ====================================================================== Complete information about ViSta, including code, documentation, test data, examples, help, guidance, and the XLisp-Stat base, is available from these WWW and FTP servers: WWW: http://forrest.psych.unc.edu FTP from www.psych.unc.edu, directory pub/forrest/vista. ====================================================================== The remainder of this message summarizes earlier messages about ViSta. ====================================================================== ======================================= What is ViSta? ======================================= ViSta, is a Visual Statistics system that > Performs univariate and multivariate statistical data analysis and >visualization > Provides visual maps for summarizing and guiding analysis sessions > Includes a data analysis language and a programming language > Is open, extensible, and freely redistributable PLATFORMS: ViSta is available for > Microsoft Windows 3.1 (Windows 95 in preperation) > Macintosh and PowerMac microcomputers > A Unix version is in preparation AUDIENCE: ViSta is useful for > UNDERGRADUATES: Teaching introductory statistics to undergraduates > GRADUATES: Teaching dynamic statistical graphics and computational >statistics to graduate students > RESEARCHERS: Research and development in dynamic statistical graphics and >in computational statistics ViSta is designed for an audience of users having a very wide range of data analysis sophistication, ranging from novice to expert. ViSta provides seamlessly integrated data analysis environments specifically tailored to the user's level of expertise. Visual guidance is available for students and novices, and graphical tools are available for teachers and other experts to create guidance for these novices. A structured graphical user interface is available for competent users, and a command line interface is available for sophisticated users. The complete Lisp-Stat (Tierney, 1990) programming environment is available to researchers, programmers and graduate students who wish to extend ViSta's capabilities. ViSta is "freely redistributable". This means that it is not in the public domain, but that it may be freely copied and redistributed, with certain restrictions. See the COPYING and COPYVSTA files included in the distribution for details. ======================================= ViSta's main features ======================================= > WORKMAPS Maps that visualize the overall structure of an on-going data analysis session. These maps are created by ViSta as the data analysis session progresses. They aid novice and competent analysts in remembering their analysis session, and permit the analyst to return to earlier steps of the analysis. > GUIDEMAPS Maps that visually guide students and other novice data analysts through their data analysis sessions. These maps are created by experts using ViSta's Authoring tools. > SPREADPLOTS Multi-window dynamic statistical graphics that visualize data structure and analysis results > MODEL RE-VISION Dynamic interactive graphical tools that permit the user to visually re-estimate model parameters > ViDAL ViSta's Data Analysis Language. ViDAL is an extension of the underlying Lisp-Stat language that can be used by sophisticated data analysts for data analysis and by programmers to enhance ViSta's capabilities. > DATASHEET A graphical datasheet for displaying and editing data. > AUTHORING TOOLS Visual tools for authoring new GuideMaps. ======================================= Data Analysis Capabilities ======================================= ViSta's data analysis capabilities include: > DYNAMIC GRAPHICS Histograms, Boxplots, Diamond Plots, Dotplots, Scatterplots, Biplots, Spinplots, Scatterplot Matrices. Plots support brushing and labeling, and are dynamically linked. > DESCRIPTIVE STATISTICS Means, Standard Deviations, Variances, Ranges, Quartiles, Medians, Correlations, Covariances, Distances >UNIVARIATE ANALYSIS T- and Z-tests and confidence intervals for single sample, paired samples and two independent samples data, with Wilcoxon Signed-Rank and Mann-Whitney tests in appropriate situations. > ANOVA Univariate analysis of variance of orthogonal one- or multi-way data. Model may or may not include two-way interactions. The model visualization is a spreadplot comprised of boxplot, diamond plot, quantile plot, quantile-quantile plot and effects plot. > REGRESSION Simple and multiple OLS regression analysis (multivariate, robust and monotonic regression in preparation). The model visualization is a spreadplot comprised of a biplot, spinplot, histogram and scatterplot-matrix. > PRINCIPAL COMPONENTS Analysis of correlations or covariances. The model visualization is a spreadplot comprised of a biplot, spin-plot, scree-plot and scatterplot-matrix. > MULTIDIMENSIONAL SCALING Analysis of one or more symmetric or asymmetric matrices. The model visualization is a spreadplot comprised of a scatterplot, spin-plot, scree-plot and scatterplot-matrix. The spreadplot supports graphical re-estimation of model parameters. > CORRESPONDENCE ANALYSIS Analysis of two-way contingency tables. The model visualization is a spreadplot comprised of a biplot, spinplot, residuals plot and scree-plot. The spreadplot supports graphical re-estimation of model parameters. ======================================= Hardware Requirements ======================================= MACINTOSH: There are two versions of ViSta for the Macintosh, one for the PowerMac and one for 680xx Macintoshes. For 680xx Macs, we recommend a 68040 processor (at the minimum, a 68030 + 68881 math coprocessor). Both versions require System 7 or later, 3 MB of free hard disk space and at least 6 MB of RAM (8 for the PowerMac). WINDOWS: There are two versions of ViSta available for Microsoft Windows 3.1, one for WIndows 3.1 with the WIN32S 32-bit extension, and one without. Both require 3 MB of free hard disk space and 8 MB of RAM. The Windows 95 version is in preparation. Versions for Windows NT and OS/2 are not contemplated. If you wish, you can try the 32-bit version of ViSta with 95 or NT or the 16-bit version with OS/2. While I have not tried ViSta under these operating systems, some people report positive results. Let me know your experience. MONITORS: We recommend at least a 14", 8-bit (256 color) screen, but 17" is preferred. For 12-, 24- or 32-bit screens, a larger memory partition is required (add 1/2 MB for 12-bit, 1 MB for 24- or 32-bit). ======================================= ViSta is a work-in-progress ======================================= ViSta is a work-in-progress. Some features are incomplete. While quality assurance testing has been done, this is a beta-release system that is not fully mature. At the current stage of development, ViSta is viable as a supplemental statistical software system used for extra credit exercises in introductory statistics, and as the focus of graduate courses on research in computational and graphical methods in statistics. It may be premature to use ViSta in undergraduate statistics classes that need a complete and robust statistic analysis system. Currently, the Delete Data, Delete Model and Simulate Data menu items are not implemented; the ANOVA visualization is incomplete; the univariate analysis visualization is not available; MULREG does not do multivariate tests (it does many separate multiple regressions); transformations are not available with matrix data; model interpretation is unimplemented; and the Help and GuideMap systems are incomplete. A univariate multiple regression module (which complements the current multivariate multiple regression module with more extensive visualizations and with the choice of robust or monotonic regression) is being written but is not yet available. ======================================= Documentation ======================================= The following two monographs describe how to use ViSta: "ViSta, The Visual Statistics System", is available from the server mentioned above. A nicely bound and printed version is also available for $5.00 from the address below. "ViSta Corresp", a description of the Correspondence Analysis procedure in ViSta, is available for $5.00 from Forrest Young at the address below. The following paper has been recently published: Young, F.W. & Lubinsky, D.J. "Guiding Data Analysts with Visual Statistical Strategies". J. Computational and Graphical Statistics, Vol 4(4), December, 1995, pp. 229-260. In addition, the following paper is available for $5.00 from Forrest Young at the address below. Young, F.W., et al. "Visual Statistical Analysis - Papers presented at the Special Contributed Session on Visual Statistical Analyais at Interface '94. Psychometric Research Memorandum 94-2, July, 1994. Forrest Young UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 From majordom@umnstat.stat.umn.edu Wed Feb 7 15:37:34 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10894; Wed, 7 Feb 1996 15:37:34 -0500 Return-Path: Received: from uvaarpa.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10889; Wed, 7 Feb 1996 14:37:24 -0600 Received: from elvis.med.virginia.edu by uvaarpa.virginia.edu id aa23583; 7 Feb 96 15:29 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA05409; Wed, 7 Feb 1996 15:28:59 -0500 Date: Wed, 7 Feb 1996 15:28:58 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Lisp Statistics Subject: XLISP external functions / directory list Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk There is a xlsx.h header file with the Mac xlisp-stat distribution. Are there any instructions on how to add an external function ? What I need in particular is a function to return a list of filenames in a directory. I have not been able to find a way of doing this from lisp - the common lisp function 'directory' is missing. Is there some other way to do this ? ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Thu Feb 8 22:02:42 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29370; Thu, 8 Feb 1996 22:02:42 -0500 Return-Path: Received: from angel.heaven.net by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29365; Thu, 8 Feb 1996 21:02:36 -0600 Received: from [198.69.29.97] (josh.netheaven.com [198.69.29.97]) by angel.heaven.net (8.6.12/8.6.12) with SMTP id WAA05269 for ; Thu, 8 Feb 1996 22:01:12 -0500 Message-Id: <199602090301.WAA05269@angel.heaven.net> X-Sender: josh@netheaven.com (Unverified) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 8 Feb 1996 22:03:40 -0500 To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) From: josh@racing.saratoga.ny.us (Joshua Kuperman) Subject: getting at the clipboard on the Macintosh Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I have been experimenting with using the AppleEvents support in Xlispstat for the Macintosh as a way to embed SQL in lisp code and using FoxPro Mac as a quasi SQL-server. FoxPro only supports the "dosc" (do script) AppleEvent and does not return values through the AppleEvent mechanism. So at the moment, whenever I need a value from FoxPro I have to have FoxPro write it to a file and read the file in Xlispstat. FoxPro has a built in variable _CLIPTEXT which is always the value of the clipboard. Xlispstat (on the Macintosh). [ In theory I could also write an AppleScript, have Xlispstat send a do script message to FoxPro to run the Script and have the Script send data to Xlispstat - but as I fear I'm very near the abyss of total programming weirdness I thought it best to try to avoid invoking another language (AppleScript) ] Xlispstat, supports methods to access the clipboard in the window-proto objects. So, far I haven't figured out how to access them directly. As far as I can tell they are simply supported by menu-items in objects created by the edit-window-proto prototype. Yet, it seems that I should be able to access the data once I have a window-proto object using the :paste-from-clip method. So far I haven't been able to do it. Perhaps, I'm barking up the wrong tree and wasting my time. Is there thorough documentation on the methods associated with the various windowing object? If I were to use an edit-window or display-window could I get at the clipboard through those objects? Should I perhaps actually be using menu objects instead? Josh Kuperman josh@racing.saratoga.ny.us http://racing.saratoga.ny.us/~josh (barely under construction) From majordom@umnstat.stat.umn.edu Fri Feb 9 09:52:18 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10248; Fri, 9 Feb 1996 09:52:18 -0500 Return-Path: Received: from versicolor.uwaterloo.ca by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10243; Fri, 9 Feb 1996 08:52:11 -0600 Received: by versicolor.uwaterloo.ca id <7039>; Fri, 9 Feb 1996 09:51:56 -0500 Subject: Stat computing positions available at Waterloo From: "R.W. Oldford" To: stat-lisp-news@umnstat.stat.umn.edu Date: Fri, 9 Feb 1996 09:51:54 -0500 X-Mailer: ELM [version 2.3 PL11] Message-Id: <96Feb9.095156edt.7039@versicolor.uwaterloo.ca> Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Subject:Stat Computing Positions Available at U of Waterloo This may be an abuse of this mailing list. If so, I apologize in advance. The University of Waterloo has positions available in the Department of Statistics and Actuarial Science (tenure-track, definite term, and post-doc). We are looking for someone interested in research in any aspect of Statistical Computing. As you may know, one of the major projects under development here is the Quail project (http://setosa.uwaterloo.ca/~ftp/Quail/Quail.html) but research interests are not restricted to programming environments. There is substantial opportunity for collaborative research within the department and with other departments on campus. For more on the university see http://www.uwaterloo.ca/home.html . I enclose the text of the advertisement that was submitted to Amstat News and other places. Two things need adding. First: Please note that we are *not* required to hire Canadians (or landed immigrants) only, so non-Canadian applicants are strongly encouraged to apply. Law requires us to consider Canadian applicants first, not exclusively. Second: For some of these, the "official" deadline has come and gone. Do not worry. If you are interested in such a position, contact me immediately via e-mail (rwoldford@uwaterloo.ca) or fax (519 746 1875). R.W. Oldford Department of Statistics and Actuarial Science University of Waterloo Waterloo, Ontario N2L 3G1 CANADA ----------------------------------------------------------- POST-DOCTORAL POSITIONS IN STATISTICS University of Waterloo The Department of Statistics and Actuarial Science invites applications for one or two Post-Doctoral positions in Statistics. Successful applicants will work with members of the Department on research problems in statistics. Candidates with an interest in any area of statistics, biostatistics, industrial statistics or statistical computing are encouraged to apply. Candidates must have completed all requirements for a PhD in Statistics or closely related area at the time the appointment begins. Appointment may begin January 1, 1996 or later. Applications will be accepted until February 28, 1996. Please submit a curriculum vitae as well as information on teaching and consulting experience, and arrange for three letters of reference to be sent to: Professor K.S. Brown, Chair, Department of Statistics and Actuarial Science, University of Waterloo, Waterloo, Ontario N2L 3G1 CANADA These appointments are subject to the availability of funds. The University of Waterloo encourages applications from all qualified individuals including women, members of visible minorities, native peoples, and persons with disabilities. POSITION IN STATISTICS University of Waterloo Applications are being accepted for a tenure track or definite term position in Statistics at the Assistant Professor level. Applicants must have a Ph.D. in statistics, statistical computing or probability with proven ability in, or potential for research, and good teaching and communication skills. Duties include undergraduate and graduate teaching, and development of an independent research program. Salary commensurate with qualifications and experience. This appointment is subject to the availability of funds. Appointment effective July 1, 1996 or later. The closing date for applications is January 31, 1996. Please submit a curriculum vitae and arrange for three letters of reference to be sent to: Professor K.S. Brown, Chair, Department of Statistics and Actuarial Science, University of Waterloo, Waterloo, Ontario N2L 3G1 CANADA In accordance with Canadian immigration requirements, this advertisement is directed to Canadian citizens and permanent residents. The University of Waterloo encourages applications from all qualified individuals including women, members of visible minorities, native peoples, and persons with disabilities. ------------------------------------------ From majordom@umnstat.stat.umn.edu Fri Feb 9 13:51:31 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11927; Fri, 9 Feb 1996 13:51:31 -0500 Return-Path: Received: from uvaarpa.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11922; Fri, 9 Feb 1996 12:51:23 -0600 Received: from elvis.med.virginia.edu by uvaarpa.virginia.edu id aa09136; 9 Feb 96 13:38 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA16975; Fri, 9 Feb 1996 13:38:32 -0500 Date: Fri, 9 Feb 1996 13:38:31 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Lisp Statistics Subject: namelist methods and documentation needed. Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Although there are a number of entries for namelist in the Lisp-Stat book, these all point to about 3 inches of not very helpful text. The documentation for the FUNCTION 'namelist' is sufficient, but I can't find much on name-list-proto OBJECTS, their methods, and how to subclass them. Searching the the lisp sources I find: A couple of lines in graph2.lsp that set defaults in the name-list-proto template. The namelist methods must be implemented in C -- searching thru those sources, it looks like they are part of the idraw package. ( I've had luck deciphering some of the other C sources, but the stuff in idraw remains pretty opaque to me, so I stopped there. ) An example of their use in the function 'labels' in dotplot.lsp, which was enough to help me figure out how to write a subclass of name-list-proto. ( at the end of this message -- not very clean perhaps -- I'm just trying to figure it out. ) This does pretty close to what I want: it highlights the name in the list with the same color that it uses to redraw that line in the graph. Some problems: Sometimes it takes two clicks on an item to get the color right - it will initially be highlighted in the previous color. It initially draws in black & white. Everthing sometimes disappears when the window is obscured and the uncovered -- i.e. contents are not redrawn. ( I tried adding a redraw method, but didn't quite work the way I had hoped, and it seemed to slow things down too much. ) Is there any other documentation on name-list-proto methods ? Anyone have any tips or examples on how to do this sort of thing properly ? Thanks. ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] ( defproto ref-name-list-proto '( fit-model names elems colors) () name-list-proto ) ( defmeth ref-name-list-proto :isnew ( fitmod ) ( let (( u ( send fitmod :unk )) ( r ( send fitmod :refs ))) ( call-next-method 0 :title ( send u :name )) ( setf ( slot-value 'fit-model ) fitmod ) ( setf ( slot-value 'elems ) ( mapcar #'(lambda (x) (send x :elem)) r )) ( setf ( slot-value 'names ) ( mapcar #'(lambda (x) (send x :name)) r )) ( setf ( slot-value 'colors ) ( send ( send color-list :new ) :slot-value 'colors )) ( send self :add-points (length r)) ( send self :point-label (iseq (length r)) ( slot-value 'names )) ( send self :showing-labels T ) ( send ( send u :slot-value 'plot ) :add-subordinate self ) ( send self :redraw ))) ( defmeth ref-name-list-proto :do-click ( &rest args ) ( apply #'call-next-method args ) ( let* (( n ( car ( send self :selection ))) ( c ( elt ( slot-value 'colors ) n )) ( fit ( slot-value 'fit-model ))) ( let (( r ( elt ( send fit :refs ) n )) ( a ( elt ( send fit :coef-estimates ) n )) ( p ( send ( send fit :unk ) :slot-value 'plot ))) ( if ( < n ( send self :num-points )) ( progn ( send self :draw-color c ) ( send ( * r a ) :add-lines p :color c :width 2 ) ( elem-label p ( elt ( slot-value 'elems ) n ) :color c )) ( send self :redraw ))))) From majordom@umnstat.stat.umn.edu Fri Feb 9 23:34:58 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15564; Fri, 9 Feb 1996 23:34:58 -0500 Return-Path: Received: from angel.heaven.net by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15559; Fri, 9 Feb 1996 22:34:54 -0600 Received: from [198.69.29.97] (josh.netheaven.com [198.69.29.97]) by angel.heaven.net (8.6.12/8.6.12) with SMTP id XAA02064 for ; Fri, 9 Feb 1996 23:33:32 -0500 Message-Id: <199602100433.XAA02064@angel.heaven.net> X-Sender: josh@netheaven.com (Unverified) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 9 Feb 1996 23:36:02 -0500 To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) From: josh@racing.saratoga.ny.us (Joshua Kuperman) Subject: Retrieving the clipboard. Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Well I figured out how to get the data from the clipboard to show up. I can send the window in which the user (me) is interacting with XLISP-STAT the :paste-from-clip message and use the method that way. So if I use the cut/paste option, as below, I wind up sticking "(send *listener* :paste-from-clip)", and when it's pasted it retrieves itself and is displayed in the *listener* window. > (send *listener* :paste-from-clip) (send *listener* :paste-from-clip)NIL Oddly, this appears to be a mere side-effect, and the value of the clipboard escapes. Am I still missing something. Josh Kuperman josh@racing.saratoga.ny.us http://racing.saratoga.ny.us/~josh (barely under construction) From majordom@umnstat.stat.umn.edu Tue Feb 13 21:13:13 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13461; Tue, 13 Feb 1996 21:13:13 -0500 Return-Path: Received: from angel.heaven.net by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13454; Tue, 13 Feb 1996 20:13:07 -0600 Received: from [198.69.29.97] (josh.netheaven.com [198.69.29.97]) by angel.heaven.net (8.6.12/8.6.12) with SMTP id VAA02396; Tue, 13 Feb 1996 21:11:39 -0500 Message-Id: <199602140211.VAA02396@angel.heaven.net> X-Sender: josh@netheaven.com (Unverified) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Tue, 13 Feb 1996 21:14:16 -0500 To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) From: josh@racing.saratoga.ny.us (Joshua Kuperman) Subject: Re: Retrieving the clipboard. Cc: luke@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I did come up with a round about way to do this. (defun read-clip () "Args none: will return the value of the expression on the clipboard provided it contains a single expression acceptable to the read function followed by a carriage return. The second argument to read is simply a flag(t the default value or nil) to indicate that an error messages should be returned on eof. The function will hang (at best) if the expression on the clipboard is not terminated by a carriage return" (read t (send *listener* :paste-from-clip))) >> >> Well I figured out how to get the data from the clipboard to show up. >> I can send the window in which the user (me) is interacting with XLISP-STAT >> the :paste-from-clip message and use the method that way. >> >> So if I use the cut/paste option, as below, I wind up sticking "(send >> *listener* :paste-from-clip)", and when it's pasted it retrieves itself and >> is displayed in the *listener* window. >> >> > (send *listener* :paste-from-clip) >> (send *listener* :paste-from-clip)NIL >> >> Oddly, this appears to be a mere side-effect, and the value of the >> clipboard escapes. Am I still missing something. >> >> Josh Kuperman >> josh@racing.saratoga.ny.us >> http://racing.saratoga.ny.us/~josh (barely under construction) >> > >The durrent code clls mac toolbox routines that transfer the text >directy without alocating a lisp string. So there is currently no way >to get at the content directly. > >luke Josh Kuperman josh@racing.saratoga.ny.us http://www.racing.saratoga.ny.us/~josh (barely under construction) From majordom@umnstat.stat.umn.edu Wed Feb 14 16:07:21 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28020; Wed, 14 Feb 1996 16:07:21 -0500 Return-Path: Received: from ACS3.BU.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28015; Wed, 14 Feb 1996 15:07:17 -0600 Received: by acs3.bu.edu (8.6.11/BU_SmartClient-1.0) id QAA121623; Wed, 14 Feb 1996 16:04:52 -0500 Date: Wed, 14 Feb 1996 16:04:52 -0500 From: rdmurphy@acs.bu.edu Message-Id: <199602142104.QAA121623@acs3.bu.edu> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Data constraints? Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'm trying to find out what kind of constraints XLisp-Stat has on the size of data sets it can work with. Can anyone provide any guidance? In particular, I'd like to use it on some panel data sets with 5,000+ observations. I would be running a Unix version. Thanks for any assistance. Russell D. Murphy Department of Economics Boston University 270 Bay State Road Boston, Massachusetts (617) 353-4124 rdmurphy@bu.edu From majordom@umnstat.stat.umn.edu Thu Feb 15 08:32:43 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11330; Thu, 15 Feb 1996 08:32:43 -0500 Return-Path: Received: from ts3tp8.NetVision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11325; Thu, 15 Feb 1996 07:32:33 -0600 Received: (from hjstein@localhost) by blinky.cpaf.com (8.6.11/8.6.9) id OAA17660; Thu, 15 Feb 1996 14:56:37 +0200 Date: Thu, 15 Feb 1996 14:56:37 +0200 From: "Harvey J. Stein" Message-Id: <199602151256.OAA17660@blinky.cpaf.com> To: rdmurphy@acs.bu.edu Subject: Data constraints? Cc: hjstein@blinky.cpaf.com, stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk rdmurphy@acs.bu.edu writes: > I'm trying to find out what kind of constraints XLisp-Stat has > on the size of data sets it can work with. Can anyone provide any > guidance? > > In particular, I'd like to use it on some panel data sets with 5,000+ > observations. I would be running a Unix version. > It all depends on how much memory your machine has. The more memory you have, the larger the problem you can work on. I don't know of any inherent built in limits. As for how much memory different tasks require, I can easily do nonlinear regressions with 2,000 2dim data points on a 16meg Linux machine. On a 32 meg machine I can easily (albet slowly because of the complexity of the mean-function) do the same with 8,500 data points - the xlispstat process gets up to a size of about 12 meg. For 12,000 data points, the process grows to a size of about 17 meg, occasionally hitting 18 meg, so this would already be past the limits for a 16 meg machine. On a Linux machine with 16 meg of ram, the limit on process size before things become too slow due to swapping is probably around 14 meg, so for my particuar problem, this would probably be around 10,000 datapoints. Also, this is without any effort made to be especially memory efficient - I'm using lists, not unboxed arrays, lots of garbage is being generated, etc. Good luck, Dr. Harvey J. Stein Berger Financial Research abel@netvision.net.il (temporary address) From majordom@umnstat.stat.umn.edu Fri Feb 16 11:19:44 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28946; Fri, 16 Feb 1996 11:19:44 -0500 Return-Path: Received: from halley.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28933; Fri, 16 Feb 1996 10:18:53 -0600 Received: from libiya.upf.es by halley.upf.es with SMTP (1.37.109.15/16.2) id AA235861236; Fri, 16 Feb 1996 17:20:36 GMT Received: by libiya.upf.es (1.38.193.4/15.6) id AA20493; Fri, 16 Feb 1996 17:16:28 +0100 Date: Fri, 16 Feb 1996 17:16:27 +0100 (MET) From: Frederic Udina To: lisp stat list Subject: compstat 96 in Barcelona Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=ISO-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I apologize in advance if this is a cross-posting for someone... Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es =20 Balmes 132 | appleLink: spa0172 =20 08008 Barcelona | Compuserve: 100555.603 SPAIN | W W Web: http://libiya.upf.es/ ___________________________________/_\_________________________________ -------------- ABOUT COMPSTAT -------------- COMPSTAT is a forum for the interplay of Statistics and Computer Science. This important international event will take place under the auspices of the European Regional Section (ERS) of the International Association for Statistical Computing (IASC), a section of the International Statistical Institute (ISI). It is organized by the Universitat Politecnica de Catalunya (UPC) and will be held in the new campus of UPC where all facilities required by a conference of this size are available. Scientific Programme Committee Local Organizing Committee ------------------------------ -------------------------- Albert Prat (Chairman) Albert Prat (Chairman) Henri Caussinus (France) Pia Margarit (Secretary) David Edwards (Denmark) Tomas Aluja (UPC) Wilfried Grossmann (Austria) Josep M. Catot (UPC) Carlo Lauro (Italy) Jesus Lores (UdLL) Matthias Nagel (Germany) Jordi Oca~na (UB) Roger Payne (United Kingdom) Enric Ripoll (IEC) =20 Keith Worsley (Canada) Xavier Tort-Martorell (UPC) Frederic Udina (UPF) Frederic Utzet (UAB)=20 Consultative members -------------------- Rudolf Dutter (Austria) Agustin Maravall (Spain) Manuel Marti (Spain) Mateo Valero (Spain) EARS IASC - Chooichiro Asano (Japan) EUROSTAT - Daniel Defays (France) INTERFACE - Mike Meyer (USA) =20 ABOUT BARCELONA Barcelona is a metropolis which is fully equipped for hosting conferences and conventions. A city with a pleasant climate and a prime location. It has a history which goes back 2.000 years, and is culturally rich and varied. It has its own culture which has been constantly enriched by the different peoples who have passed through, and has open, enterprising citizens. It is built on a human scale. Its avenues are wide, its buildings not too tall and its climate, almost all the year round, is springlike. The city's attractions are basically cultural, commercial and recreational. Barcelona's attractions extend to its immediate surroundings. Within a radius of little more than one hundred kilometres the visitor may choose between a wide variety of landscapes and activities.=20 ----------- ABOUT UPC ----------- Though the Technical University of Catalunya is a relatively young university, it has more than a century of educational traditions to draw upon. UPC origins are found in the Instituto Politecnico Superior, which was founded in 1968, by the grouping together of the Technical Schools of Barcelona, e.g the Superior Technical School of Architecture and the Superior Technical School of Industrial Engineering both of which date back to the 19th century. In the year 1971, the Instituto Superior Politecnico was formed into a University, with the given name of Universitat Politecnica de Barcelona; this name was changed in 1984 to the Universitat Politecnica de Catalunya given its enlarged sphere of influence, taking in not just Barcelona but the whole of Catalunya. In the last years Catalunya and its capital Barcelona have themselves been driven by strong economic and social changes, which has manifested itself not only within the business and technical world of Catalunya but also the culture itself. UPC has contributed actively in this process of change. Today UPC offers a wide range of degrees, 24 of a three year duration, 11 of a five year duration and 12 doctorates, there are many courses available for students who wish to continue/specialise in their field of study together with an extensive programme of collaboration with both private and public organisations. =20 ------------------------ THE SCIENTIFIC PROGRAMME ------------------------ The scientific programme of COMPSTAT'96 covers a wide range of aspects, from user-experiences and evaluation of software through to the development and implementation of new statistical ideas. The programme is intended to appeal to anyone interested in statistics and its computational issues, whether in universities, industrial companies,=20 research institutes or as software developers. Therefore COMPSTAT'96 has accepted papers in all fields of statistics but belonging to one of the three following categories: C1- Statistical methods (preferable new ones) that require a substantial use of computing. C2- Computer environments, algorithms, tools and software useful in statist= ics. C3- Applications of computational statistics in areas of substantial intere= st (environment, industry, health, education, etc). ---------------- KEYNOTE SPEAKERS ---------------- Opening Session: George E.P Box (U.S.A) "Scientific Statistics Teaching, Learning and the Computer" Closing Session: Angel Jordan (U.S.A) =20 "Trends in the Information Technologies Markets-The Future" ---------------- INVITED SPEAKERS ---------------- - Persi Diaconis ( U.S.A ) "A Practical Look at Random Number Generation" - William Eddy ( U.S.A ) "The Statistical and Computational Challenges of Functional Magnetic Resona= nce Imaging"=20 - Antoni Espasa ( SPAIN ) "Automatic Modelling of Daily Series of Economic Activity"=20 - David Hand (UNITED KINGDOM)=20 "Classification and Computers: Shifting the Focus"=20 - Wolfgang H=84rdle ( GERMANY ) "An Interactive Software For Smoothing in High and Low Dimensions" - Martin Janzura (CZECH REPUBLIC) "Image Processing, Markov Chain Approach" =20 - Martin Maechler (SWITZERLAND) "Density Estimation: New Spline Approaches and Partial Review" - Agustin Maravall & V Gomez (SPAIN) "New methods for Quantative Analysis of short-term Economic Activity" =20 - Pedro A. Morettin ( BRASIL ) "From Fourier to Wavelet Analysis of Time Series" - Dr. Christian Ritter (BELGIUM)=20 "Likehood and Bayesian Settings" - Peter J. Rousseeuw (BELGIUM) "New Applications of Robust Methods" =20 - Gilbert Saporta ( FRANCE ) =20 "Interactive Software for Learning Statistics"=20 - Gunther Sawitzki (GERMANY) =20 "Evaluation and Testing Statistical Software" =20 - Phillipe Vieu ( FRANCE ) "Choice of Nonparametric Models for Multivariate Regression" =20 Victor J. Yohai ( ARGENTINA ) =20 "Robust prcedures for regression models with ARIMA errors" ------------------------------------------------------------ CONTRIBUTED PAPERS, SHORT COMMUNICATIONS, AND POSTERS TOPICS ------------------------------------------------------------ About 60 contributed papers and over 120 short communications and posters are going to be presented at the symposium, on topics such as: =20 1 . Experimental Design 2 . Multivariate Analysis 3 . Non-Linear Regression 4 . Robust Methods 5 . Cluster Analysis Classification 6 . Spatial Data Analysis 7 . Graphical Models/Catergorical Data Analysis 8 . Statistical Software 9 . Bayesian Methods 10. Non- Parametric Statistics 11. Survival Analysis 12. Biostatistics 13. Simulation 14. Statistical Education 15. Survey Methods and Databases 16. Time Series 17. Mixed Linear Models 18. Regression 19. Neural Networks 20. Structural Equations 21. Expert Systems 22. Generalized Linear Models 23. Applicatons to Physics 24. Computational Inference/Resampling =20 -------------------------------- PROFESSIONAL SOFTWARE EXHIBITION -------------------------------- A permanent software exhibition of professional software vendors will take place in the conference building. The demonstration stands of=20 these software producers will be adjacent to the conference rooms and the new products will be presented. --------------- BOOK EXHIBITION --------------- A book exhibition on computational statistics will be set up in the conference area. ------------ PUBLICATIONS ------------ The results of COMPSTAT 96 will be published in several volumes: * Invited and contributed papers (Physica-Verlag) * Short communications and poster presentations(UPC) =20 =20 ------------------- GENERAL INFORMATION ------------------- Conference Venue ---------------- The conference is to be held in the North Campus of UPC, to be found just off the main thoroughfare of Barcelona: the Diagonal, and directly behind the Palau Reial (The Royal Palace). In recent years UPC has made important investments that have led to the creation of 90,000 sq.m of new space for teaching, research, cultural and general purpose facilities for the University community. One of the main achievements is the North Campus, one of the most important technology campuses located in an urban context and an ideal site for COMPSTAT'96. Symposium Language ------------------ The offical language for meetings and correspondance is English. There will be no simultaneous translations available. Communication and Presentation Media ------------------------------------ Each lecture room is equiped with blackboards and overhead projectors. Authors wishing to use slide projection, PCs, or PC overhead projection panels should contact the organising commitee before July 1, 1996. Lunches and Coffee Breaks ------------------------- The University offers a good restaurant/coffee bar service where you can choose from a variety of hot/cold meals and beverages. The meals are offered each day, three courses for 700Pts, the restaurant also cater for vegetarians, with a vegetarian option always available. For those persons that plan to have lunch at the University restaurant, we strongly recommend that you pay for the lunch tickets using the registration form. During the Symposium, a limited number of lunch tickets will be sold at the registration desk. Its inexpensive and friendly atmosphere makes it an ideal place to eat and to absorb the University/Conference atmosphere. ---------------- REGISTRATION FEE ---------------- The conference fees for full conference participation are: =20 =09=09=09BEFORE Mar 31 BEFORE June 30 AFTER July 1 =20 ISI/IASC MEMBERS 35.000Pts 40.000Pts 45.000Pts OTHERS 40.000Pts 45.000Pts 50.000Pts ACCOMPANYING PERSONS=0910.000Pts 12.500Pts 15.000Pts The fee includes=20 * the proceedings * attendance at all scientific sessions, poster and software exhibitions * participation in social programs * beverages during coffee breaks Meals, conference dinner and accommodation are not included. REGISTRATION ------------ For final registration please contact us at the CORRESPONCENCE ADDRESS and we will send you the required Registration Form. CANCELLATION ------------ For cancellation prior to June 30 1996, a cancellation fee of 10.000Pts,- will be retained. For cancellation after that date no refund can be made. REGISTRATION AND INFORMATION ---------------------------- The information/registration desk will be located on the ground floor of the conference building. Services will be provided for registration, hotel accommodation, programme for accompanying persons, social events and tourist information. The Information/registration desk will be open on: =20 Sunday, August 25 3pm-8pm. Monday, August 26 8am-6pm Other Conference days 8.15am-6pm SOCIAL PROGRAMME(see registration fees) --------------------------------------- August 25 Informal get together in the conference building August 26 Concert=20 August 27 Mayor's reception August 28 General excursion August 29 Conference dinner (optional Pts.6.500'-) IBERIA has been appointed Offical Carrier to the 1996 symposium and we kindly ask you to contact your local Iberia agency for=20 assistance with your travel arrangements. TRAVEL AGENCY ------------- During the congress, the Offical Travel Agency will have a desk in the Congress venue offering their services to all those attending the congress. They will give information and book a wide range of optional trips, excursions and tours. ACCOMODATION ------------ There are numerous hotels ranging from moderate to deluxe within close distance to the conference site. Rooms have been reserved in seversl hotels (see page Registration Form). Due to the limited number of spaces available within each price bracket, hotels will be allocated on a first come first serve basis, i.e a prompt return of the registration form including payment of the hotel deposit will probably secure the price and hotel you require.=20 Full prepayment is required in students residence. ------------ SPONSORSHIPS ------------ CIRIT INSTITUT CATALA D'ESTADISTICA MINISTERIO DE EDUCACION Y CIENCIA MINITAB UNIVERSITAT AUTONOMA DE BARCELONA UNIVERSITAT DE BARCELONA UNIVERSITAT POLITECNICA DE CATALUNYA =20 -------------------------- ADDRESS FOR CORRESPONDENCE -------------------------- =20 COMPSTAT'96 Att. Prof . Albert Prat Dpt. d'Estadistica - ETSEIB UNIVERSITAT POLITECNICA DE CATALUNYA=20 Av. Diagonal, 647 08028 BARCELONA SPAIN Tel +34 3 401 65 69 - Fax +34 3 401 65 75 E-mail: compstat@eio.upc.es ********* www: http://sunrai.upc.es/compstat ********* $$END_OF_MESSAGE From majordom@umnstat.stat.umn.edu Sat Feb 17 19:09:33 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15219; Sat, 17 Feb 1996 19:09:33 -0500 Return-Path: Received: from angel.heaven.net by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15214; Sat, 17 Feb 1996 18:09:28 -0600 Received: from [198.69.29.97] (josh.netheaven.com [198.69.29.97]) by angel.heaven.net (8.6.12/8.6.12) with SMTP id TAA05253 for ; Sat, 17 Feb 1996 19:08:00 -0500 Message-Id: <199602180008.TAA05253@angel.heaven.net> X-Sender: josh@netheaven.com Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sat, 17 Feb 1996 19:10:39 -0500 To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) From: josh@racing.saratoga.ny.us (Joshua Kuperman) Subject: [Q] Slatex or noweb useable on a Mac? Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I downloaded a paper by B. Narasimhan that made extensive use of noweb. I was very impressed with the layout and presentation of the code. (As to the quality of the content I'm clueless.) Enough so, that I thought I might try to use noweb or at least Slatex with the idea of generating readable documentation using LaTex (Textures). Unfortunately, noweb, at least from my casual examination seems very much a Unix tool - or at least not a Mac tool. Then I thought Slatex might work and I was unable to get it to run successfully in Xlisp-stat on my Mac. I figure the problem with Slatex might simply be the options I selected. Has anyone managed to get either Slatex or noweb to work on a Macintosh. If you have could you give me a few pointers: what changes I need to make in the distribution(s) and what documentation beyond the standard distributions would be most profitable for me to read, etc. etc. ... Josh Kuperman josh@racing.saratoga.ny.us http://www.racing.saratoga.ny.us/~josh (barely under construction) From majordom@umnstat.stat.umn.edu Tue Feb 20 05:19:07 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25228; Tue, 20 Feb 1996 05:19:07 -0500 Return-Path: Received: from ts3ap9.NetVision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25223; Tue, 20 Feb 1996 04:18:57 -0600 Received: (from hjstein@localhost) by blinky.cpaf.com (8.6.11/8.6.9) id MAA25006; Tue, 20 Feb 1996 12:26:42 +0200 Date: Tue, 20 Feb 1996 12:26:42 +0200 From: "Harvey J. Stein" Message-Id: <199602201026.MAA25006@blinky.cpaf.com> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Small bug in xlispstat Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk There seems to be a small bug in xlispstat 2.1 Release 3.47 (Beta), X11 version. In X11graph.c, RGB color values gotten from X via the call to XParseColor are converted to a real range by dividing by MULTIPLIER. MULTIPLIER is defined to be 62535. I believe this should be 65535 instead. The problem can be demonstrated by calling parse-color from the interpreter: > (parse-color "white") (1.047973135044375 1.047973135044375 1.047973135044375) > (* 1.047973135044375 62535) 65534.99999999999 The expression (parse-color "white") should return (1.0 1.0 1.0). Multiplying the returned value by MULTIPLIER shows that the call to XParseColor was returning rgb values of 65535, thus demonstrating that MULTIPLIER should be 65535 instead of 62535. This is confirmed by this section of the XCreateColormap man page: STRUCTURES The XColor structure contains: typedef struct { unsigned long pixel;/* pixel value */ unsigned short red, green, blue;/* rgb values */ char flags; /* DoRed, DoGreen, DoBlue */ char pad; } XColor; The red, green, and blue values are always in the range 0 to 65535 inclusive, independent of the number of bits actually used in the display hardware. The server scales these values down to the range used by the hardware. Black is represented by (0,0,0), and white is represented by (65535,65535,65535). In some functions, the flags mem- ber controls which of the red, green, and blue members is used and can be the inclusive OR of zero or more of DoRed, DoGreen, and DoBlue. Thanks, Dr. Harvey J. Stein Berger Financial Research abel@netvision.net.il (temporary address) From majordom@umnstat.stat.umn.edu Thu Feb 22 13:48:22 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03616; Thu, 22 Feb 1996 13:48:22 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03610; Thu, 22 Feb 1996 12:48:13 -0600 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.6.4.3/10.1) id NAA26862; Thu, 22 Feb 1996 13:48:06 -0500 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 22 Feb 1996 13:51:51 -0500 To: edstat-l@jse.stat.ncsu.edu, stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: Updated ViSta Documentation Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Announcing updated documentation for ViSta, the Visual Statistics System. The documentation has been updated to correspond with the current release (4.22). The documentation is the same for all operating systems. It includes cover and contents, an introductory chapter, a tutorial chapter, and references. The documentation exists as postscript files. So far, the updated documentation has been saved only as bin-hexed, self-extracting stuffit archives, a file format that is most appropriate for downloading onto Macintoshes. The documentation will soon be saved in a file format more appropriate for MS-Windows. You can obtain the documentation in three ways: 1) World-Wide-Web The documentation may be downloaded from the ViSta server at http://forrest.psych.unc.edu/research/vista.html 2) FTP You may ftp the same self-extracting-archives directly from www.psych.unc.edu/pub/forrest/vista/mac. These archives are in files: * front.ps.sea.hqx * chapter1.ps.sea.hqx * chapter2.ps.sea.hqx * refernce.ps.sea.hqx 3) Mail You may mail a US$5.00 check made payable to "UNC Psychometric Lab" to Forrest Young at the address given in the signature of this message. For more information about vista, visit the ViSta web site, or request more information at the address given below. Forrest Young Professor, Psychometrics CB# 3270 Davie Hall email: forrest@unc.edu UNC Psychometric Lab www: forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 From majordom@umnstat.stat.umn.edu Sun Feb 25 18:23:58 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04600; Sun, 25 Feb 1996 18:23:58 -0500 Return-Path: Received: from dmssyd.syd.dms.CSIRO.AU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04595; Sun, 25 Feb 1996 17:23:47 -0600 Received: from dmsim.syd.dms.CSIRO.AU by dmssyd.syd.dms.CSIRO.AU (4.1/5.17) id AA23068; Mon, 26 Feb 96 09:46:39 EST (from Prabhakar.Neboori@dms.CSIRO.AU (Prabhakar Neboori)) To: stat-lisp-news@umnstat.stat.umn.edu Date: Mon, 26 Feb 1996 09:46:38 +1100 Message-Id: <27045.825288398@dmsim.syd.dms.CSIRO.AU> From: Prabhakar Neboori Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk SIGNOFF _Prabhakar Neboori, CSIRO Div Maths & Stats, / \ | Locked Bag 17, North Ryde NSW 2113 | | Ph: 61 2 325 3173 fax (02) 325 3200 | | (Building E6B, Macquarie University) | | E-mail: neboori@syd.dms.csiro.au | +--------------------------------------------+ From majordom@umnstat.stat.umn.edu Mon Feb 26 22:14:28 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA09972; Mon, 26 Feb 1996 22:14:28 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA09967; Mon, 26 Feb 1996 21:14:22 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa15503; 26 Feb 96 20:20 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA27247; Mon, 26 Feb 1996 20:20:14 -0500 Date: Mon, 26 Feb 1996 20:20:12 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: xlispstat color plot problems Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On XlispStat 2.1 R3.47 for Mac ( I've seen the same behaviour with 3.43 / 3.44 ): Setting a plot's :back-color to any color not in the built-in set assigned to *colors* does not work properly ( at least on Mac xlisp-stat ) Redefining *colors* to include the new color symbols has no effect on this. Does anyone know why ? I ran into this limitation some time ago, but was able to avoid bumping into it until now. ( I can probably still find a kludge around it, but the bump on my head reminded me to ask about it! ;-) Also, setting draw-mode to XOR for line-drawing doesn't seem to work right - I get strange artefacts around the fringe of the lines. I was going to use XOR mode drawing to erase and redraw a line using a different thickness when selected. Drawing a thicker line when selected works fine, but I'ld like to replace the thick line with a thinner one when its unselected. Any tips ? BTW: When moving development from Unix/X11 to Mac, I missed having parse-color on the Mac, so I wrote something to parse an X11 "rgb.txt" file into a "RGB.LSP" to give the Mac a parse-color and make the same color names available. Source is appended at bottom for anyone interested. ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] ;;; Mac doesn't have parse-color, but we can emulate it ;;; if you download a rgb.txt file from unix/X11 host. ( require "general-utils" ) ( provide "colors" ) ( setf rgb-table ( make-hash-table :test 'equal )) ( if ( probe-file "RGB.LSP" ) ; pre-parsed file ( with-open-file ( input "RGB.LSP" ) ( format T "Reading pre-parsed file RGB.LSP... ~%" ) ( do (( entry ( read input nil ) (read input nil ))) (( null entry )) ( setf ( gethash ( car entry ) rgb-table ) (cdr entry)))) ( when ;; else: read in raw rgb.txt and parse & write output ( probe-file "rgb.txt" ) ( format T "Parsing RGB.TXT file -- this will take a minute... ~%" ) ( let (( rgb ( read-rows-from-file "rgb.txt" ))) ( map-into rgb #'( lambda (x) (let (( hd ( subseq x 0 3 )) ( tl ( subseq x 3 ))) (cons (string-upcase (apply #'concatenate 'string ( mapcar #'symbol-name tl ))) hd ))) rgb ) ( map 'nil #'(lambda (x) (setf (gethash ( car x ) rgb-table ) ( mapcar #'(lambda (x) (/ x 255.0)) (cdr x)))) rgb ) ( format T "Writing pre-parsed RGB.LSP file. ~%" ) ( with-open-file ( output "RGB.LSP" :direction :output ) ( maphash #'(lambda (key val) ( print ( cons key val ) output )) rgb-table ))))) ( defun parse-color (name) ( gethash ( if ( symbolp name ) ( symbol-name name ) ( string-upcase name )) rgb-table )) ( defun add-color ( cname ) ( let (( c ( parse-color cname ))) ( when c ( apply #'make-color cname c )))) ;; and try adding some new colors : ( mapcar #'add-color '( Pink HotPink Purple SkyBlue LightBlue tan YellowGreen LightGray SlateGray Gray Orange Sienna linen bisque honeydew )) ;;; color-list object is a way to rotate thru a list of colors ;;; with :next-color method, if you want, for example, to try ;;; to use a unique color for each line or region in a graph. ;;; note that 'BLACK and 'WHITE are always removed and reserved ;;; ( and currently, also GRAY, SLATEGRAY, LIGHTGRAY ) ( defproto color-list '( colors reserved )) ( defmeth color-list :reset ( &optional available-colors ) "( &optional available-colors ) (re)set the list of colors - default is (color-symbols)" ( slot-value 'colors ( if available-colors available-colors (color-symbols))) ( dolist ( res '( white black gray slategray lightgray )) ( send self :remove res ) ( slot-value 'reserved ( cons res ( slot-value 'reserved ))))) ( defmeth color-list :remove (color-symbol) ( slot-value 'colors ( remove color-symbol ( slot-value 'colors ))) ( slot-value 'colors )) ( defmeth color-list :next () ( let (( cl ( slot-value 'colors ))) ( let (( nxt ( car cl )) ( rem ( cdr cl ))) ( slot-value 'colors ( append rem (list nxt))) nxt ))) ( defmeth color-list :isnew () ( send self :reset )) ;;; and here is the function from general-utils that is used in the ;;; above code: ( provide "general-utils" ) ( defun read-rows-from-file (file) "( read-rows-from-file file-name ) reads in a file turning each line into a lisp list, and returns the list of lists. " ( with-open-file ( ffile file ) ( do (( line ( read-line ffile ) ( read-line ffile Nil)) ( data ())) ( ( null line ) ( reverse data ) ) ;; ( print line ) ( if ( not ( zerop ( length ( string-trim " \t" line )))) ( setf data ( cons ( with-input-from-string ( str line ) ( let (( error-token ( gensym "eof" ))) ( do (( x (read str Nil error-token) (read str Nil error-token)) ( l () )) ( ( eq x error-token ) (reverse l) ) ( setf l ( cons x l ))))) data )))))) From majordom@umnstat.stat.umn.edu Wed Feb 28 11:10:13 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05552; Wed, 28 Feb 1996 11:10:13 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05547; Wed, 28 Feb 1996 10:10:02 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id LAA15914; Wed, 28 Feb 1996 11:09:22 -0500 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 28 Feb 1996 11:07:49 -0500 To: edstat-l@jse.stat.ncsu.edu, stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: MS-Windows ViSta Documentation Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Announcing updated MS-Windows documentation for ViSta, the Visual Statistics System. The MS-Windows documentation has been updated to correspond with the current release of ViSta (4.22) and with the earlier announced Macintosh documentation. The documentation includes cover and contents, an introductory chapter, a tutorial chapter, and references. The Windows documentation exists as zipped postscript files. The Macintosh documentation exists as self-extracting-archives of the same postscript files. These postscript files must be printed on a postscript printer. Adobe Acrobat pdf files will be available soon. These files can be read and displayed by Acrobat Reader and then printed on a very wide variety of printers, including non-postscript printers. Acrobat Reader is available from Adobe at no cost from their site at http://w1000.mv.us.adobe.com/Acrobat/Acrobat0.html. In addition, you can download Amber, a free beta version of the next release of Acrobat Reader from the same site. In addition to the capabilities of Acrobat Reader, Amber can display the documentation within a NetScape Navigator 2 window. You can obtain the documentation in three ways: 1) World-Wide-Web The MS-Windows and Macintosh versions of the documentation may be downloaded from the ViSta server at http://forrest.psych.unc.edu/research/ViSta/doco.html 2) FTP You may ftp the zipped postscript files for downloading and printing from MS-Windows. The ftp site is: ftp://www.psych.unc.edu/pub/forrest/vista/mswindows. The relevant files are: * frontps.zip * chap01ps.zip * chap02.ps.zip * refrncps.zip You may also ftp the self-extracting-archive postscript files for downloading and printing from a Macintosh. The ftp site is: ftp://www.psych.unc.edu/pub/forrest/vista/mac. The relevant files are: * front.ps.sea.hqx * chapter1.ps.sea.hqx * chapter2.ps.sea.hqx * refernce.ps.sea.hqx 3) Mail You may mail a US $5.00 check made payable to "UNC Psychometric Lab" to Forrest Young at the address given in the signature of this message. For more information about vista, visit the ViSta web site at the www address given below, or request more information at the US-Mail address given below. Forrest Young UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta software site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Thu Feb 29 15:08:16 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17299; Thu, 29 Feb 1996 15:08:16 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17294; Thu, 29 Feb 1996 14:08:02 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa23446; 29 Feb 96 13:50 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA30540; Thu, 29 Feb 1996 13:50:12 -0500 Date: Thu, 29 Feb 1996 13:50:11 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: :DO-CLICK / :POINT-SELECTED / :SELECTION / :POINT-SELECTED ? Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Well - First I wrote a subclass of name-list-proto: It prints element names in the color that the elemental peaks are plotted in on another graph. When you click on an element, the appropriate plot is drawn with a thicker line ( and, if you've used the menu to change the selection color, a new color ) and a marker. I rewrote the :isnew and :do-click methods to implement this class. PROBLEM: If a selection is make by some method other than clicking, for example using the "Selection..." menu item, the do-click method isn't called. Actually, my real need is to link the two windows more symmetrically, so that a click in either window updates the state of the other. ( This differs from the standard Lisp-Stat linked plots in that it is not respective points that are linked, but points or items in a list are linked to overlays in the scatterplot. ) OK - so it seemed to make more sense to use the parents :do-click, and instead to customize the selection method. There are three that seem appropriate: POINTS-SELECTED Method args: (&optional list) Sets or Return indices of current selection. SELECTION Method args: (&optional list) Sets or Return indices of current selection. POINT-SELECTED Method args: (point &optional selected) Sets or returns selection status (true or NIL) of POINT. Sends :ADJUST-SCREEN message if states are set. Vectorized. Of those, :POINT-SELECTED looked like the most basic, as it is able to select or unselect points. So - I re-wrote :point-selected. Only problem is -- it doesn't appear to get called when I make a selection. The item in the name list is highlited as selected, but my method is never called. What is the actual flow of control of selections in graphics methods ? I can't figure this out. ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Fri Mar 1 17:07:23 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11473; Fri, 1 Mar 1996 17:07:23 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11464; Fri, 1 Mar 1996 16:07:03 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa13776; 1 Mar 96 16:35 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA19660; Fri, 1 Mar 1996 16:35:42 -0500 Date: Fri, 1 Mar 1996 16:35:42 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: Re: :DO-CLICK / :POINT-SELECTED / :SELECTION / :POINT-SELECTED ? In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk To followup my own question: I finally found the answer in the section on Linking in the LISP-STAT book. :adjust-points-in-rect and :adjust-screen-point appear to be the methods to catch changes in the point's state. ( I'm still not sure of the exact flow of control -- i.e. whether these methods are called whenever a points state changes via any method. ) ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Fri Mar 1 18:13:52 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12431; Fri, 1 Mar 1996 18:13:52 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12426; Fri, 1 Mar 1996 17:13:47 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id ac27291; 1 Mar 96 17:07 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA19619; Fri, 1 Mar 1996 17:07:17 -0500 Date: Fri, 1 Mar 1996 17:07:16 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: inconsistency in Xlisp-Stat type hierarchy Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk In Common Lisp, "A string is a specialized vector (one-dimensional array) whose types are characters." ( CLTL1 ), and in Xlisp: > ( arrayp "a string" ) T And arrays are sequences: > ( sequencep ( vector 1 2 3 )) T > ( sequencep "a string" ) T However: > ( compound-data-p '( 1 2 3 )) T > ( compound-data-p ( vector 1 2 3 )) T > ( compound-data-p "a string" ) NIL I can understand overlooking strings as not very useful as a form of COMPOUND-DATA, but omitting them introduces an undesirable type inconsistency. In a pure OO language, a type like COMPOUND-DATA would naturally be usable anywhere another type of sequence would be used. I've been successfully using the package at the bottom to rectify this problem in Xlisp-Stat. It redefines #'LENGTH and #'ELT to work on both regular Lisp sequences and COMPOUND-DATA sequences, however, I finally got bitten by the above type inconsistency when I tried: > ( length "a string" ) Error: not a compound data item - "a string" Happened in: # and found I had to use: > ( xlisp::length "a string" ) 8 because strings ARE sequences, but they AREN'T compound-data-sequences. ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] ( unless ( find-package "GENSEQ" ) ( make-package "GENSEQ" :nicknames '( "Generic-Sequences" ) :use '( LISP ))) ;; Note: even though the docs state that ":use '(LISP)" is the default, ;; the default case doesn't seem to work, so you have to explicitly list it. ;; Xlisp:IN-PACKAGE also does not work as documented - it does not create a ;; package if it doesn't exist ( thus the UNLESS clause above ). ; The memnonic is: ; Put IN Seven EXtremely Random USEr Interface COmmands ( Provide "GENSEQ" ) ( IN-package "GENSEQ" ) ( Shadow '( length elt sequencep seq )) ; dis-inherit these from package LISP ( EXport '( length elt seq make-xy-seq sequencep )) ; Require () ; Use-package '( LISP ) [ set in make-package ] ; Import () ; COmmands ... ;; alias for compound-data-seq ( setf ( symbol-function 'seq ) ( function compound-data-seq )) ( setf ( documentation 'seq 'function ) ( documentation 'compound-data-seq 'function )) ;; To make compound-data-sequences "plug-replacible" with built-in Lisp sequences, ;; we have to redefine several built-in functions to work on either. ;; LENGTH :: shadowed alias for compound-data-length ( setf ( symbol-function 'length ) ( function compound-data-length )) ( setf ( documentation 'length 'function ) ( documentation 'compound-data-length 'function )) ;; ELT :: redefine shadowed ELT to work on built-in Lisp or compound-data sequences. ( defun elt ( sequence n ) "Args: (a i) Returns element I of Lisp sequence or compound-data-sequence A" ( LISP::ELT ( seq sequence ) n )) ;; SEQUENCEP ( setf ( symbol-function 'sequencep ) ( function compound-data-p )) ( setf ( documentation 'sequencep 'function ) ( documentation 'compound-data-p 'function )) ;; ... we also need a setter macro so ELT can continue to be used with SETF forms ( defsetf elt ( a i ) ( val ) `( setf ( lisp:elt ( seq ,a ) ,i ) ,val )) From majordom@umnstat.stat.umn.edu Sat Mar 2 00:23:47 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23200; Sat, 2 Mar 1996 00:23:47 -0500 Return-Path: Received: from mimsy.cs.umd.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23194; Fri, 1 Mar 1996 23:23:41 -0600 Received: by mimsy.cs.UMD.EDU (8.7.3/UMIACS-0.9/04-05-88) id AAA07123; Sat, 2 Mar 1996 00:23:33 -0500 (EST) Received: (from pcolsen@localhost) by n2ell.columbia.md.us (8.6.12/8.6.9) id MAA00476; Fri, 1 Mar 1996 12:20:55 -0500 Date: Fri, 1 Mar 1996 12:20:55 -0500 Message-Id: <199603011720.MAA00476@n2ell.columbia.md.us> From: "Peter C. Olsen" To: stat-lisp-news@umnstat.stat.umn.edu Subject: How to use color ... Reply-To: Peter.C.Olsen@n2ell.columbia.md.us, "P.E." Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Can anyone point me to a good tutorial on how to use color in Xlispstat graphics? I've been able to make a few simple colored plots, but I'm finding that I need to do much more complex programming in order to simplify the resulting visuals. I'd appreciate anything --- from a well-written, comprehensive, and indexed tutorial some kludgy code "with only a few misleading comments" :-) Thanks! Peter From majordom@umnstat.stat.umn.edu Mon Mar 4 00:40:17 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04021; Mon, 4 Mar 1996 00:40:17 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04016; Sun, 3 Mar 1996 23:40:12 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa09772; 4 Mar 96 0:39 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA11168; Mon, 4 Mar 1996 00:39:20 -0500 Date: Mon, 4 Mar 1996 00:39:18 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Peter.C.Olsen@n2ell.columbia.md.us, "P.E." Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: How to use color ... In-Reply-To: <199603011720.MAA00476@n2ell.columbia.md.us> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Fri, 1 Mar 1996, Peter C. Olsen wrote: > Can anyone point me to a good tutorial on how to use color in > Xlispstat graphics? > > I've been able to make a few simple colored plots, but I'm > finding that I need to do much more complex programming in order > to simplify the resulting visuals. I'd appreciate anything --- > from a well-written, comprehensive, and indexed tutorial some > kludgy code "with only a few misleading comments" :-) > I've found color, and particularly colored area plots quite useful when the plots get "busy" - however, when the get so busy that you really NEED color, you also probably need finer control of the drawing sequence. Drawing objects in overlays gives you control of the redraw sequence -- however, overlays, as designed in Xlisp-Stat, are not intended to overlay the content area of the plot - the were actually intended for controls rather than graphics overlays. A Plot's :REDRAW method calls :REDRAW-BACKGROUND, :REDRAW-OVERLAYS and then :REDRAW-CONTENT, in that order. However, the problem is that :REDRAW-CONTENT clears the content area, so any overlays there will be erased. I have some overlay classes, which along with a modification of the redraw order, will work. However, I'm currently "between versions" : some of my recent queries to this list have been about how to link a name-list to the overlays in the plot. ( I've almost got this figured out - things ain't quite how they are documented in the book! ) What I'm doing with it is: Plot an Xray Energy Spectrum with a colored area plot; Fit the spectrum against a set of elemental reference peaks and draw the peaks scaled by the coeffecients in different color lines. Draw the residual background using a gray line. Color the areas under the residual in two different continuum measurement regions in another color. Clicking on a peak, or on the element name in the linked namelist, causes that peak to be selected, which causes it to be redrawn with a thicker line width, and some peak markers to be displayed. That last paragraph is the part currently under construction: I've tested several unsatisfactory hacks that semi-worked, and I think I've finally figure out how to do it "right". I hope to have that complete sometime this week -- I'll be happy to post it when finished. ( Unoptimized, and doing a few more screen :redraw's than absolutely necessary. ) ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Mon Mar 4 15:44:55 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA07527; Mon, 4 Mar 1996 15:44:55 -0500 Return-Path: Received: from mac.ripon.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA07521; Mon, 4 Mar 1996 14:44:47 -0600 Message-Id: <9603042044.AA07521@umnstat.stat.umn.edu> Date: 4 Mar 1996 14:46:17 U From: "LongA" Subject: hidden line surface plots To: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Does anyone already have a hidden line surface plotter (before I spend oodles of time writing my own?) Andy Long From majordom@umnstat.stat.umn.edu Tue Mar 5 12:39:25 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13340; Tue, 5 Mar 1996 12:39:25 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13332; Tue, 5 Mar 1996 11:39:16 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa26926; 5 Mar 96 12:37 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA20125; Tue, 5 Mar 1996 12:37:03 -0500 Date: Tue, 5 Mar 1996 12:36:55 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Lisp Statistics Subject: RGB values of XLISP-STAT standard built-in Mac *colors* ? Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Is there any way to get the RGB colors of the standard built-in *colors* on Mac Xlisp-Stat ? I'm using the code at bottom along with the mac version of parse-color I previosly posted. Parse color returns the "pure" values for the standard colors - e.g. ( parse-color 'red ) => ( 1.0 0.0 0.0 ), ( parse-color 'green ) => ( 0.0 1.0 0.0 ) However, when I add: ( lighter-color 'green ) ; LIGHT-GREEN = ( 2.0 1.0 2.0 ) ] ( darker-color 'green ) ; DARK-GREEN = ( 0.0 0.8 0.0 ) ] ( make-color 'real-green 0.0 1.0 0.0 ) And plot them along with the standard built-in color 'GREEN, the latter is by far the darkest of them all. I haven't been able to find a function that returns the actual RGB colors used ( and I can't just free and replace a standard color.) ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] ( defun guarded-parse-color (name) ( let (( result ( parse-color name ))) ( if result result ( error "Not in color table: ~S" ( string name ))))) ( defun color-complement ( color-symbol &optional new-color ) "Args: (COLOR-SYMBOL &optional NEW-COLOR) makes color complement of COLOR-SYMBOL. Returns new-color-symbol. If a new-color symbol is not given, it interns the new color as '*-complement' (*=old-color) and returns that symbol." ( let (( new-color ( if new-color new-color ( intern ( concatenate 'string ( string color-symbol ) "-complement" ))))) ( apply #'make-color new-color ( - 1 ( guarded-parse-color color-symbol ))) new-color )) ( defun darker-color ( color-symbol &optional new-color (dec 0.2)) "Args: ( color-symbol &optional new-color (dec 0.2)) Makes and returns symbol for a new color darker that the old one." ( let (( new-color ( if new-color new-color ( intern ( concatenate 'string "DARK-" ( string color-symbol )))))) ( apply #'make-color new-color ( mapcar #'(lambda (x) (max x 0.0)) (- ( guarded-parse-color color-symbol ) dec))) new-color )) ( defun lighter-color ( color-symbol &optional new-color (inc 0.2)) "Args: ( color-symbol &optional new-color (inc 0.2)) Makes and returns symbol for a new color lighter that the old one." ( let (( new-color ( if new-color new-color ( intern ( concatenate 'string "LIGHT-" ( string color-symbol )))))) ( apply #'make-color new-color ( mapcar #'(lambda (x) (min x 1.0)) (+ (guarded-parse-color color-symbol) inc))) new-color )) ( defun free-extra-colors () "Args: () Frees all (color-symbols) not in *colors*" ( dolist ( c (color-symbols)) ( if ( not ( find c *colors* )) ( free-color c ))) (color-symbols)) From majordom@umnstat.stat.umn.edu Tue Mar 5 13:44:13 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14019; Tue, 5 Mar 1996 13:44:13 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14014; Tue, 5 Mar 1996 12:44:06 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa29966; 5 Mar 96 13:42 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA22484; Tue, 5 Mar 1996 13:42:38 -0500 Date: Tue, 5 Mar 1996 13:42:28 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Lisp Statistics Subject: NOTE: Bug in previous code, but problem description accurate [was: , RGB values... ] In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk There may have been a bug in the code I posted -- I think I dropped a mapcar that forced the range to ( 0.0 - 1.0 ) which was in an earlier edit. However, the problem was and still is as I described -- at least with GREEN -- some other colors I have checked look correct. Better version of the code is below: Fixed the max,min problem to keep RGB values between 0 and 1. functions return: ( Values color-symbol rgb-value-list ) [ Since generated symbol names like |BLUE-complement| will also not return a value from parse-color ] Sorry to be so quick to post before I've double checked everything, but I'm seeing a severe delay in mail getting out to the list, which has made me quick on the trigger with these questions. The problem may be on our end rather than with the list -- UVA has a *severly* overloaded central mail server. ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] ( defun guarded-parse-color (name) ( let (( result ( parse-color name ))) ( if result result ( error "Not in color table: ~S" ( string name ))))) ( defun color-complement ( color-symbol &optional new-color ) "Args: (COLOR-SYMBOL &optional NEW-COLOR) makes color complement of COLOR-SYMBOL. Returns new-color-symbol. If a new-color symbol is not given, it interns the new color as '*-complement' (*=old-color) and returns that symbol." ( let* (( new-color ( if new-color new-color ( intern ( concatenate 'string ( string color-symbol ) "-complement" )))) ( rgb (- 1 (guarded-parse-color color-symbol)))) ( apply #'make-color new-color rgb ) ( values new-color rgb ))) ( defun darker-color ( color-symbol &optional new-color (dec 0.25)) "Args: ( color-symbol &optional new-color (dec 0.25)) Makes and returns symbol for a new color darker that the old one." ( let* (( new-color ( if new-color new-color ( intern ( concatenate 'string "DARK-" ( string color-symbol ))))) ( rgb ( mapcar #'(lambda (x) (max x 0)) ( - ( guarded-parse-color color-symbol ) dec )))) ( apply #'make-color new-color rgb ) ( values new-color rgb ))) ( defun lighter-color ( color-symbol &optional new-color (inc 0.25)) "Args: ( color-symbol &optional new-color (inc 0.25)) Makes and returns symbol for a new color lighter that the old one." ( let* (( new-color ( if new-color new-color ( intern ( concatenate 'string "LIGHT-" ( string color-symbol ))))) ( rgb (mapcar #'(lambda (x) (min x 1.0)) (+ (guarded-parse-color color-symbol) inc)))) ( apply #'make-color new-color rgb ) (values new-color rgb))) From majordom@umnstat.stat.umn.edu Tue Mar 5 15:05:27 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14790; Tue, 5 Mar 1996 15:05:27 -0500 Return-Path: Received: from ts024p15.pop3a.netvision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14785; Tue, 5 Mar 1996 14:05:15 -0600 Received: (from hjstein@localhost) by bogart.nnt.com (8.6.11/8.6.9) id WAA00752; Tue, 5 Mar 1996 22:02:04 +0300 Date: Tue, 5 Mar 1996 22:02:04 +0300 From: "Harvey J. Stein" Message-Id: <199603051902.WAA00752@bogart.nnt.com> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Combo nreg-model & newtonmax available? Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'd like to do some large nonlinear regressions. Unfortunately, doing this in a straight forward manner causes xlispstat to suck up memory until it sends my machine into a swapping frenzy. This is under Linux on a Pentium 120 with 32mb of memory. I have about N = 20,000 data points, each data point consisting of 3 parameters X and a response Y, and a P = 9 parameter model. In other words, I have a 9 parameter family of functions of R^3 to R, and I'd like to choose the one which best fits my 20,000 data points. I can just barely manage this on the above machine using nreg-model. However, I'd like to add some annual seasonality, allowing for a shift and a scaling that depends on the month of the year. This adds 24 parameters to the function. I'd also like to use larger data sets. I tried just using nreg-model, but as I said, it sends my machine into fits of swapping. I killed xlispstat when it grew to about 36meg. Since I'm only trying to minimize something like (Y-f(X))(Y-f(X))^t as a function of the parameters, one should only need to work with a px1 jacobian matrix, where p is the number of parameters. However, nreg-model computes the derivative of Y-f(X) as a function of the parameters, thus yielding a Nxp matrix which it manipulates. This obviously takes much more space, in fact, too much to be handled on a 32 meg machine. Currently, I'm planning on explicitly minimizing the above equation using newtonmax, but then I lose all the nice statistics that nreg-model gives me. So, does anyone have any ways around this? Perhaps an alternative nreg model proto which relies on newtonmax instead of make-jacobian and make-sweep-matrix? Thanks, Dr. Harvey J. Stein Berger Financial Research abel@netvision.net.il (temporary address) From majordom@umnstat.stat.umn.edu Wed Mar 6 14:51:39 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA21890; Wed, 6 Mar 1996 14:51:39 -0500 Return-Path: Received: from euler.bd.psu.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA21884; Wed, 6 Mar 1996 13:51:32 -0600 Received: from localhost by euler.bd.psu.edu via SMTP (940816.SGI.8.6.9/940406.SGI) for id OAA19122; Wed, 6 Mar 1996 14:52:58 -0500 Message-Id: <199603061952.OAA19122@euler.bd.psu.edu> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Web version of code X-Mailer: Mew version 1.02 on Emacs 19.30.1 Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Date: Wed, 06 Mar 1996 14:52:53 -0500 From: Balasubramanian Narasimhan Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I finally got around to converting my older program (stbl.lsp) into HTML using Noweb in anticipation of the release of my GUI version of Diehard. (Diehard is a battery of tests for Random Number Generators developed by George Marsaglia). You can view the program at http://euler.bd.psu.edu/lispstat/stbl-prog.html In case you are curious about Diehard, snapshots of Diehard can be viewed at http://euler.bd.psu.edu/~naras/diehard/snaphots.html ________________________________________________________________ B. Narasimhan naras@euler.bd.psu.edu http://euler.bd.psu.edu/~naras From majordom@umnstat.stat.umn.edu Wed Mar 6 15:29:42 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22285; Wed, 6 Mar 1996 15:29:42 -0500 Return-Path: Received: from euler.bd.psu.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22272; Wed, 6 Mar 1996 14:29:35 -0600 Received: from localhost by euler.bd.psu.edu via SMTP (940816.SGI.8.6.9/940406.SGI) for id PAA19632; Wed, 6 Mar 1996 15:31:00 -0500 Message-Id: <199603062031.PAA19632@euler.bd.psu.edu> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Correction to URL X-Mailer: Mew version 1.02 on Emacs 19.30.1 Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Date: Wed, 06 Mar 1996 15:30:56 -0500 From: Balasubramanian Narasimhan Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I mispelt the URL in my prev note. The URL for diehard snapshots is http://euler.bd.psu.edu/~naras/diehard/snapshots.html The home page is actually is http://stat.fsu.edu/~geo/diehard.html ________________________________________________________________ B. Narasimhan naras@euler.bd.psu.edu http://euler.bd.psu.edu/~naras From majordom@umnstat.stat.umn.edu Mon Mar 11 14:37:55 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA01222; Mon, 11 Mar 1996 14:37:55 -0500 Return-Path: Received: from ts023p4.pop3a.netvision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA01203; Mon, 11 Mar 1996 13:34:10 -0600 Received: (from hjstein@localhost) by blinky.cpaf.com (8.6.11/8.6.9) id VAA01033; Mon, 11 Mar 1996 21:41:47 +0200 Date: Mon, 11 Mar 1996 21:41:47 +0200 From: "Harvey J. Stein" Message-Id: <199603111941.VAA01033@blinky.cpaf.com> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Getting closer... Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk In my last message, I asked about aternative nonlinear regression code. My problem was that nreg-model computes the jacobian of the mean-function. My mean-function takes 33 parameters, and returns a vector of length 20,000. This causes the function returned by make-jacobian to suck up all available memory and go into a swapping frenzy on my Linux box with 32meg of memory. To avoid this, I tried using newtonmax as a replacement for nreg-model. I worked out how to get all the statistics I desired without computing the above matrix directly. However, it turns out that this only trades my space problem for a time problem. Newtonmax computes the Hessian of the function that it is maximizing. In my case, this takes over 1000 function calls. Each function call takes about 5.5 seconds, which means this will take about 1.5 hours per iteration! I tried using nelmeadmax instead of newtonmax. This has the advantage of not computing the Hessian. However, it doesn't seem to do a very good job of finding the minimum even in the 9 parameter case. It keeps stopping with an "iteration limit exceeded" message. So, my question is, does anyone have any other maximization or nonlinear regression code for xlisp-stat? Something that's both conservative in space and number of function evaluations? Perhaps a gradient based maximizer which doesn't appeal to the Hessian? Or, does anyone have any ideas on other methods for doing this computation? Thanks, Dr. Harvey J. Stein Berger Financial Research abel@netvision.net.il From majordom@umnstat.stat.umn.edu Tue Mar 12 10:55:46 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16546; Tue, 12 Mar 1996 10:55:46 -0500 Return-Path: Received: from gateway.ets.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16541; Tue, 12 Mar 1996 09:55:41 -0600 Received: from research.ets.org by gateway.ets.org with SMTP id AA12189 (5.65c/IDA-1.4.4 for ); Tue, 12 Mar 1996 10:52:17 -0500 Received: from comet.research.ets.org by research.ets.org with SMTP id AA19380 (5.67b/IDA-1.4.3 for abel@netvision.net.il); Tue, 12 Mar 1996 10:52:06 -0500 Received: by comet.research.ets.org (SMI-8.6/SMI-SVR4) id KAA04200; Tue, 12 Mar 1996 10:52:00 -0500 Date: Tue, 12 Mar 1996 10:52:00 -0500 Message-Id: <199603121552.KAA04200@comet.research.ets.org> From: Russell Almond To: abel@netvision.net.il Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <199603111941.VAA01033@blinky.cpaf.com> (abel@netvision.net.il) Subject: Efficiency tricks (was Re: Getting closer...) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk In my last message, I asked about aternative nonlinear regression code. My problem was that nreg-model computes the jacobian of the mean-function. My mean-function takes 33 parameters, and returns a vector of length 20,000. This causes the function returned by make-jacobian to suck up all available memory and go into a swapping frenzy on my Linux box with 32meg of memory. Have you tried applying all the usual efficiency tricks to your code? If not, the problem may lie in your mean-function. For the benefit of those readers on the list without a whole lot of lisp experience, here are some things to try: 1) Make sure you are using arrays not lists (for small data sets the flexibility of lists outweighs their overhead). 2) Try to avoid "consing" functions which allocate new storage automatically. Things like append, push, mapcar, etc all generate new storage. Especially try and avoid things which generate and free storage in the middle of a loop. If you know the size of the list/vector you need to generate, you can pre-allocate storage for it outside the loop (using let) and then pass the scratch space as an argument (ala BLAS code) or as a lexical closure, e.g., (let ((BIG-SCRATCH (array 20000))) (defun my-mean-fun (params) (dotimes (i 20000) (setf (elt BIG-SCRATCH i) (...)) ...) ...) ) This insures that 2a) I'm not sure if the vectorized functions of Lisp-Stat fall into this category. I seem to recall that early versions of ISP had problems with the storage for intermediate results seemed to cause a problem. I'm not sure how the vectorization works in Lisp-stat, but I recall it is the naive implementation (i.e., it just applies a mapping function and generates an intermediate event). This works fine for problems which don't demand to much in the way of resources but is really inefficent in more complex circumstances (i.e., operations on vectors of length 20,000). You may need to rewrite them as you would in Fortran or C, i.e., specifically allocate intermediate storage and write explicit loops which do the operations. You might also need to do that inside the nreg-model functions. (One big advantage of XLISP-STAT is that it is an open implementation. If you need to re-implement part of the system because it doesn't work for your project you can do it. It more than makes up for the fact that not all of the implementations are "clean." Good luck. If you need more tricks, I would refer you to any number of elementary texts on Lisp. I actually am quite fond of Peter Norvig's book on AI programming in Lisp. --Russell Almond Educational Testing Services Research Statistics Group, 15-T Princeton, NJ 08541 Phone: 609-734-1241 FAX: 609-734-5420 Email: almond@acm.org, ralmond@ets.org http://bayes.stat.washington.edu/almond/almond.html From majordom@umnstat.stat.umn.edu Wed Mar 13 07:42:15 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03781; Wed, 13 Mar 1996 07:42:15 -0500 Return-Path: Received: from ts008p2.pop3a.netvision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03776; Wed, 13 Mar 1996 06:41:59 -0600 Received: (from hjstein@localhost) by blinky.cpaf.com (8.6.11/8.6.9) id OAA04750; Wed, 13 Mar 1996 14:11:54 +0200 Date: Wed, 13 Mar 1996 14:11:54 +0200 From: "Harvey J. Stein" Message-Id: <199603131211.OAA04750@blinky.cpaf.com> To: Russell Almond Subject: Efficiency tricks (was Re: Getting closer...) In-Reply-To: <199603121552.KAA04200@comet.research.ets.org> References: <199603111941.VAA01033@blinky.cpaf.com> <199603121552.KAA04200@comet.research.ets.org> Cc: hjstein@blinky.cpaf.com, stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Russell Almond writes: > > In my last message, I asked about aternative nonlinear regression > code. My problem was that nreg-model computes the jacobian of the > mean-function. My mean-function takes 33 parameters, and returns a > vector of length 20,000. This causes the function returned by > make-jacobian to suck up all available memory and go into a swapping > frenzy on my Linux box with 32meg of memory. > > Have you tried applying all the usual efficiency tricks to your code? > If not, the problem may lie in your mean-function. For the benefit of > those readers on the list without a whole lot of lisp experience, here > are some things to try: > > 1) Make sure you are using arrays not lists (for small data sets the > flexibility of lists outweighs their overhead). > > 2) Try to avoid "consing" functions which allocate new storage > automatically. > ... > > 2a) I'm not sure if the vectorized functions of Lisp-Stat fall into > this category. > ... > You might also need to do that inside the nreg-model functions. (One > big advantage of XLISP-STAT is that it is an open implementation. If > you need to re-implement part of the system because it doesn't work > for your project you can do it. It more than makes up for the fact > that not all of the implementations are "clean." I input all the data as typed-vectors (type c-double). My mean-function does produce garbage, but not 30meg of it. Although the excessive garbage creation could be slowing things down, I verified that it's not filling things up - I forced a gc in my mean-function, and xlispstat still slowly grew to 40meg. Regarding the use of vectorized functions. I had always believed that although they produce alot of garbage (by virtue of the fact that they're returning the results), they're quite fast. I had thought that avoiding them would probably make things too slow. However, I just did afew preliminary tests, and I'm quite surprised to see that this might not be the case. Consider: > (type-of (setq m (make-array 25000 :initial-contents (normal-rand 25000) :element-type 'c-double))) TYPED-VECTOR > (time (type-of (* m m))) The evaluation took 0.28 seconds; 0.09 seconds in gc. VECTOR > (time (type-of (* m m))) The evaluation took 0.36 seconds; 0.16 seconds in gc. VECTOR > (time (type-of (* m m))) The evaluation took 0.35 seconds; 0.15 seconds in gc. VECTOR > (time (type-of (* m m))) The evaluation took 0.34 seconds; 0.15 seconds in gc. VECTOR > (time (type-of (* m m))) The evaluation took 0.34 seconds; 0.14 seconds in gc. VECTOR > (time (type-of (* m m))) The evaluation took 0.37 seconds; 0.14 seconds in gc. VECTOR > (time (type-of (* m m))) The evaluation took 0.35 seconds; 0.14 seconds in gc. VECTOR > (time (type-of (* m m))) The evaluation took 0.34 seconds; 0.14 seconds in gc. VECTOR > (time (type-of (* m m))) The evaluation took 0.34 seconds; 0.15 seconds in gc. VECTOR > (time (type-of (* m m))) The evaluation took 0.36 seconds; 0.13 seconds in gc. VECTOR > (defun vect-prod (x y result) (dotimes (i (length x)) (setf (aref result i) (* (aref x i) (aref y i))))) VECT-PROD > (compile 'vect-prod) VECT-PROD > (type-of (setq acc (make-array 25000 :initial-element 0.0 :element-type 'c-double))) TYPED-VECTOR > (time (type-of (vect-prod m m acc))) The evaluation took 0.30 seconds; 0.02 seconds in gc. SYMBOL > (time (type-of (vect-prod m m acc))) The evaluation took 0.31 seconds; 0.03 seconds in gc. SYMBOL > (time (type-of (vect-prod m m acc))) The evaluation took 0.31 seconds; 0.03 seconds in gc. SYMBOL > (time (type-of (vect-prod m m acc))) The evaluation took 0.31 seconds; 0.03 seconds in gc. SYMBOL > (time (type-of (vect-prod m m acc))) The evaluation took 0.31 seconds; 0.03 seconds in gc. SYMBOL > (time (type-of (vect-prod m m acc))) The evaluation took 0.31 seconds; 0.03 seconds in gc. SYMBOL > (time (type-of (vect-prod m m acc))) The evaluation took 0.32 seconds; 0.03 seconds in gc. SYMBOL (I use type-of to avoid accumulating garbage in the history variables *, **, etc.) Note that (* m m) seems to stabilize at about .36 seconds, and spends over 1/3 of its time in gc, whereas the corresponding non-garbage producing vect-prod function ends up taking .32 seconds. So, it seems that byte-compiled vector operations can actually end up being faster than the built in vectorized operations. This is quite a surprise for me. Presumably, if there were a way to use the unvectorized scalar multiplication function in vect-prod, then vect-prod could be further speeded up. In any case, in my setting, the problem with nreg-model seems to stem primarily from make-jacobian. One call to the function it returns causes a 20meg increase in the size of the xlispstat process. And this is with my modified make-jacobian which uses one-sided derivatives so as to half the number of function calls. Now that I see that explicit array accesses can be as fast as vectorized operations, I'll try rewriting my mean-function (as well as make-jacobian, and possibly parts of nreg-model) and see what happens... Thanks, Dr. Harvey J. Stein Berger Financial Research abel@netvision.net.il From majordom@umnstat.stat.umn.edu Wed Mar 13 11:02:55 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05024; Wed, 13 Mar 1996 11:02:55 -0500 Return-Path: Received: from STOUT.BBN.COM by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05019; Wed, 13 Mar 1996 10:02:50 -0600 Received: from stout (localhost [127.0.0.1]) by stout.BBN.COM (8.6.9/8.6.5) with ESMTP id KAA24111; Wed, 13 Mar 1996 10:57:44 -0500 Message-Id: <199603131557.KAA24111@stout.BBN.COM> To: "Harvey J. Stein" Cc: Russell Almond , hjstein@blinky.cpaf.com, stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Efficiency tricks (was Re: Getting closer...) In-Reply-To: Your message of "Wed, 13 Mar 1996 14:11:54 +0200." <199603131211.OAA04750@blinky.cpaf.com> Date: Wed, 13 Mar 1996 10:57:42 +0000 From: Ken Anderson Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk (I use type-of to avoid accumulating garbage in the history variables *, **, etc.) Note that (* m m) seems to stabilize at about .36 seconds, and spends over 1/3 of its time in gc, whereas the corresponding non-garbage producing vect-prod function ends up taking .32 seconds. So, it seems that byte-compiled vector operations can actually end up being faster than the built in vectorized operations. This is quite a surprise for me. Presumably, if there were a way to use the unvectorized scalar multiplication function in vect-prod, then vect-prod could be further speeded up. It may be that * is written very generically and is basically doing what your version is doing. So there is probably a optimization opportunity here. From majordom@nokomis.stat.umn.edu Wed Mar 13 15:06:15 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA25106; Wed, 13 Mar 1996 15:06:15 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA25088; Wed, 13 Mar 1996 14:04:28 -0600 From: Luke Tierney Message-Id: <9603132004.AA25088@nokomis.stat.umn.edu> Subject: Re: Efficiency tricks (was Re: Getting closer...) To: abel@netvision.net.il (Harvey J. Stein) Date: Wed, 13 Mar 1996 14:04:27 -0600 (CST) Cc: rga6711@comet.research.ets.org, hjstein@blinky.cpaf.com, stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: <199603131211.OAA04750@blinky.cpaf.com> from "Harvey J. Stein" at Mar 13, 96 02:11:54 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 3070 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > > (type-of (setq m (make-array 25000 :initial-contents (normal-rand 25000) :element-type 'c-double))) > TYPED-VECTOR > > (time (type-of (* m m))) > The evaluation took 0.28 seconds; 0.09 seconds in gc. ... > > (time (type-of (* m m))) > The evaluation took 0.36 seconds; 0.13 seconds in gc. > VECTOR Looks like the GC could use a bit more tuning for these situations -- I'll think about that. > > (defun vect-prod (x y result) > (dotimes (i (length x)) > (setf (aref result i) (* (aref x i) (aref y i))))) > > VECT-PROD > > (compile 'vect-prod) > VECT-PROD > > (type-of (setq acc (make-array 25000 :initial-element 0.0 :element-type 'c-double))) > TYPED-VECTOR > > (time (type-of (vect-prod m m acc))) > The evaluation took 0.30 seconds; 0.02 seconds in gc. ... > Note that (* m m) seems to stabilize at about .36 seconds, and spends > over 1/3 of its time in gc, whereas the corresponding non-garbage > producing vect-prod function ends up taking .32 seconds. So, it seems > that byte-compiled vector operations can actually end up being faster > than the built in vectorized operations. This is quite a surprise for > me. Presumably, if there were a way to use the unvectorized scalar > multiplication function in vect-prod, then vect-prod could be further > speeded up. I'd be a bit careful about jumping to conclusions here. The byte code compiler does reduce the penalty of doing lisp-level looping, but vectorized arithmetic in principle has far greater potential if it properly takes into account the type of the data. Typed vectors have been added only recently, and while they work with the vectorized arithmetic, the type information is ignored -- that is why the result comes back as type 'VECTOR. Eventually I will change this so that * applied to two typed floating point arrays will return a typed floating point array and be implemented internally as a tight C loop. (There are still a few issues I'm not sure about yet, like what should happen when log is applied to a typed floating point vector containing 1.0 and -1.0 -- should the result be a generic vector containing a float and a complex, or a typed complex vector? Similar issues need to be thought through for integer overflow). Another point to keep in mind is that speed isn't the only reason to use vectorized arithmetic -- vectorized code is usually clearer and hence easier to get right. > > In any case, in my setting, the problem with nreg-model seems to stem > primarily from make-jacobian. One call to the function it returns > causes a 20meg increase in the size of the xlispstat process. And > this is with my modified make-jacobian which uses one-sided > derivatives so as to half the number of function calls. > One of the things I am thinking about is modifying the compiler to rewrite functions using vectorized arithmetic so they use as little memory as possible and also do as much of the computation in native data types (unboxed values or typed vectors) as possible. THis is longer term sutff, so who knows when it will come to fruition. From majordom@umnstat.stat.umn.edu Thu Mar 14 05:55:01 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA20558; Thu, 14 Mar 1996 05:55:01 -0500 Return-Path: Received: from ts017p11.pop3a.netvision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA20553; Thu, 14 Mar 1996 04:54:45 -0600 Received: (from hjstein@localhost) by blinky.cpaf.com (8.6.11/8.6.9) id NAA05522; Thu, 14 Mar 1996 13:02:41 +0200 Date: Thu, 14 Mar 1996 13:02:41 +0200 From: "Harvey J. Stein" Message-Id: <199603141102.NAA05522@blinky.cpaf.com> To: Luke Tierney Subject: Re: Efficiency tricks (was Re: Getting closer...) In-Reply-To: <9603132004.AA25088@nokomis.stat.umn.edu> References: <199603131211.OAA04750@blinky.cpaf.com> <9603132004.AA25088@nokomis.stat.umn.edu> Cc: rga6711@comet.research.ets.org, hjstein@blinky.cpaf.com, stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Luke Tierney writes: > > Note that (* m m) seems to stabilize at about .36 seconds, and spends > > over 1/3 of its time in gc, whereas the corresponding non-garbage > > producing vect-prod function ends up taking .32 seconds. So, it seems > > that byte-compiled vector operations can actually end up being faster > > than the built in vectorized operations. This is quite a surprise for > > me. Presumably, if there were a way to use the unvectorized scalar > > multiplication function in vect-prod, then vect-prod could be further > > speeded up. > > I'd be a bit careful about jumping to conclusions here. The byte code > compiler does reduce the penalty of doing lisp-level looping, but > vectorized arithmetic in principle has far greater potential if it > properly takes into account the type of the data. I rewrote my vectorized mean-function to be iterative (and to return the result in a supplied array so as to avoid gc). The routine is now over twice as fast as it used to be. It used to take about 5.5 seconds to run, with about 1.5 seconds in gc. Now it takes about 2.5 seconds to run, with no time in gc. Also, timing (* m m) when m is a list instead of an array seems to take somewhat longer. Incidentally, I also rewrote make-jacobian to use an internal typed 2 dimensional work matrix instead of building up the jacobian each time. Whereas calling the function returned by the stock make-jacobian would cause xlispstat to swell by about 20meg, the new version initially allocates 5meg and never grows any larger. It seems like I'll be able to use nreg-model again after afew more changes of a similar nature. I agree that in principal the vectorized code should be faster. But, in the current implementation, it seems that there's alot of overhead to the vectorized functions when the args are compound data types. Even more so than explicitly using map! As the following shows, it seems that using mapcar to multiply to lists of floats is about 10% faster than using the native multiplication! (The times are much longer here because they're being run on a 486dx66 instead of a pentium 120 - I ran these tests on a different machine with a different xlispstat binary to make sure that the surprising timing results were *not* due to my having messed up the xlispstat binary. None the less, I'd appreciate it if others could verify my results.) XLISP-PLUS version 2.1g Portions Copyright (c) 1988, by David Betz. Modified by Thomas Almy and others. XLISP-STAT 2.1 Release 3.47 (Beta). Copyright (c) 1989-1994, by Luke Tierney. > (setq *gc-flag* t) T > (type-of (setq x (normal-rand 20000))) [ gc: total 58826, 3034 free ] [ gc: total 58826, 4488 free ] CONS > (time (type-of (* x x))) [ gc: total 66850, 218 free ] [ gc: total 66850, 218 free ] [ gc: total 80892, 252 free ] [ gc: total 80892, 252 free ] [ gc: total 96940, 300 free ] [ gc: total 96940, 300 free ] The evaluation took 2.04 seconds; 1.15 seconds in gc. CONS > (time (type-of (* x x))) [ gc: total 117000, 368 free ] [ gc: total 117000, 40371 free ] The evaluation took 1.10 seconds; 0.39 seconds in gc. CONS > (time (type-of (* x x))) [ gc: total 117000, 14436 free ] [ gc: total 117000, 40374 free ] The evaluation took 1.14 seconds; 0.45 seconds in gc. CONS > (time (type-of (* x x))) [ gc: total 117000, 14450 free ] [ gc: total 117000, 40379 free ] The evaluation took 1.17 seconds; 0.46 seconds in gc. CONS > (time (type-of (* x x))) [ gc: total 117000, 14458 free ] [ gc: total 117000, 40378 free ] The evaluation took 1.18 seconds; 0.47 seconds in gc. CONS > (time (type-of (mapcar #'* x x))) [ gc: total 117000, 14457 free ] [ gc: total 117000, 40379 free ] The evaluation took 1.02 seconds; 0.44 seconds in gc. CONS > (time (type-of (mapcar #'* x x))) [ gc: total 117000, 369 free ] [ gc: total 117000, 40378 free ] The evaluation took 1.01 seconds; 0.44 seconds in gc. CONS > (time (type-of (mapcar #'* x x))) [ gc: total 117000, 369 free ] [ gc: total 117000, 40378 free ] The evaluation took 1.01 seconds; 0.44 seconds in gc. CONS > (time (type-of (mapcar #'* x x))) [ gc: total 117000, 369 free ] [ gc: total 117000, 40378 free ] The evaluation took 1.03 seconds; 0.44 seconds in gc. CONS > (time (type-of (mapcar #'* x x))) [ gc: total 117000, 369 free ] [ gc: total 117000, 40381 free ] The evaluation took 1.03 seconds; 0.45 seconds in gc. CONS So, although using the vectorized functions does typically lead to much simpler and straight forward implementations, I'm afraid that the current implementation extracts an efficiency cost for this. Also, it's difficult to recover the time in the current implementation, because the unvectorized versions are unavailable at the interpreter level. As I mentioned above, I still find all this rather startling. So, can someone please verify some of these results? Even better would be a more complete study of what is fast and what isn't fast in xlispstat. Thanks, Dr. Harvey J. Stein Berger Financial Research abel@netvision.net.il From majordom@umnstat.stat.umn.edu Thu Mar 14 11:04:19 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA21886; Thu, 14 Mar 1996 11:04:19 -0500 Return-Path: Received: from ts019p9.pop3a.netvision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA21881; Thu, 14 Mar 1996 10:03:56 -0600 Received: (from hjstein@localhost) by blinky.cpaf.com (8.6.11/8.6.9) id QAA05708; Thu, 14 Mar 1996 16:20:53 +0200 Date: Thu, 14 Mar 1996 16:20:53 +0200 From: "Harvey J. Stein" Message-Id: <199603141420.QAA05708@blinky.cpaf.com> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Partial success... Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I managed to hack nreg-model to handle the problem I've been working on (namely 20,000 data points, 33 parameter mean-function). The key savings was rewriting make-jacobian to build an output matrix when it's called, and have the jacobian function always build up the jacobian within this matrix (and return this matrix). It's probably clearer to look at the code, so here it is: (defun make-jacobian-2 (f n m) "Args: (f n m) F is a function of an N-vector, and returns an M-vector. Returns a function that approximates the jacobian function iof F by a one-sided difference." (let* ((h .0000001) (res (make-array (list m n) :element-type 'c-double)) (del (* h (column-list (identity-matrix n)))) (fx0 (make-array m :element-type 'c-double)) (fx0+h ())) (lambda (b) (blas-dcopy n (funcall f b) 0 1 fx0 0 1) (dotimes (i n) (setq fx0+h (funcall f (+ b (nth i del)))) (dotimes (j m) (setf (aref res j i) (/ (- (aref fx0+h j) (aref fx0 j)) h)))) res))) So, basically, after: (setq j (make-jacobian-2 #'f 33 20000)) then (funcall j param-vector) builds up the jacobian within the typed matrix res, and returns res. There's still additional garbage being created, but it's quite minor compared to the make-jacobian in nonlin.lsp. I also subclassed nreg-model-proto to use the above make-jacobian-2, and wrote a version of nreg-model to use the above modifications, and for completeness, to take a mean-function which works in a supplied work array instead of one which always returns a new mean vector. In any case, after these two modifications, the xlispstat process only grows to about 16meg when doing the nonlinear regression, thus comfortably running on a 32meg machine. There are other places which could be modified for less garbage creation, but it's only of order number-of-parameters^2 as opposed to number-of-parameters * number-of-data-points, so it's not really needed. OOPS! I spoke too soon. I had it running while I was typing this. During the minimization it was doing fine at 16meg. But when it got to the output phase, it started eating memory until it got to about 40meg (on my 32meg machine), and died with a segmentation fault! So, I guess I still have a little more work to do to keep the memory usage down... Incidentally, recoding the mean-function to use a work space and to be iterative (as opposed to the original which used the vectorized functions) speeded up the mean-function by a factor of 2! I had always assumed that the fastest code is produced by using the vectorized functions, so I'm quite surprised to see that I've been paying a significant speed penalty for using them. I still find this surprising, but it really indicates a need to investigate just how fast the vectorized functions really are... I guess it's not so surprising, given that it's built to handle things like: (* '((1 2 3) 4 (5 6)) '(7 (8 9 10) (11 12))) Maybe the vectorization needs some fine tuning. Perhaps some of the tests could be moved up and out of the iteration. Or, maybe common subcases could be handled at the top... Thanks to everyone for their help, Dr. Harvey J. Stein Berger Financial Research abel@netvision.net.il From majordom@umnstat.stat.umn.edu Thu Mar 14 12:36:53 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22630; Thu, 14 Mar 1996 12:36:53 -0500 Return-Path: Received: from ts019p9.pop3a.netvision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22625; Thu, 14 Mar 1996 11:36:41 -0600 Received: (from hjstein@localhost) by blinky.cpaf.com (8.6.11/8.6.9) id TAA06324; Thu, 14 Mar 1996 19:39:32 +0200 Date: Thu, 14 Mar 1996 19:39:32 +0200 From: "Harvey J. Stein" Message-Id: <199603141739.TAA06324@blinky.cpaf.com> To: Ken Anderson Subject: Re: Efficiency tricks (was Re: Getting closer...) In-Reply-To: <199603141407.JAA29299@stout.BBN.COM> References: <199603141102.NAA05522@blinky.cpaf.com> <199603141407.JAA29299@stout.BBN.COM> Cc: hjstein@blinky.cpaf.com, stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Ken Anderson writes: > I think your timing tests are to short. You are only getting about a > 0.1 sec accuracy, which is about the size of the difference between * and > mapcar. You're right. When I upped the list size to 100000, * and mapcar seem to be about the same... However... I just built a version of xlispstat with all the scalar operations available at the interpreter (called scalar*, scalar+, ...). I find that (mapcar #'scalar* m m) does beat (* m m) by about 10% (when m is initialized to be (normal-rand 100000)). Here's the timings of the last 10 runs of each: (* m m) (mapcar #'scalar* m m) time gc time-gc time gc time-gc Trial 1 1.57 0.67 0.90 1.22 0.49 0.73 2 1.35 0.45 0.90 1.23 0.49 0.74 3 1.58 0.67 0.91 1.23 0.50 0.73 4 1.35 0.45 0.90 1.22 0.49 0.73 5 1.57 0.67 0.90 1.22 0.49 0.73 6 1.35 0.45 0.90 1.22 0.49 0.73 7 1.57 0.67 0.90 1.23 0.51 0.72 8 1.36 0.45 0.91 1.22 0.49 0.73 9 1.57 0.67 0.90 1.22 0.49 0.73 10 1.38 0.45 0.93 1.22 0.49 0.73 Note that time - gc is very stable, and indicates that mapping scalar* gives a sizable win (about a 20% savings in time) computationally over just using *. The alternating of the gc column for (* m m) indicates that (.67+.45)/2 is a pretty accurate estimate of the average gc time for (* m m), which comes out to .56, which is also significantly higher than the average gc time for the mapcar version (which just happens to be stable here). Thanks, Dr. Harvey J. Stein Berger Financial Research abel@netvision.net.il From majordom@umnstat.stat.umn.edu Thu Mar 14 13:32:26 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23040; Thu, 14 Mar 1996 13:32:26 -0500 Return-Path: Received: from ts019p9.pop3a.netvision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23023; Thu, 14 Mar 1996 12:31:54 -0600 Received: (from hjstein@localhost) by blinky.cpaf.com (8.6.11/8.6.9) id UAA06374; Thu, 14 Mar 1996 20:39:53 +0200 Date: Thu, 14 Mar 1996 20:39:53 +0200 From: "Harvey J. Stein" Message-Id: <199603141839.UAA06374@blinky.cpaf.com> To: "Harvey J. Stein" Subject: More timing tests (and errors noted) In-Reply-To: <199603141739.TAA06324@blinky.cpaf.com> References: <199603141102.NAA05522@blinky.cpaf.com> <199603141407.JAA29299@stout.BBN.COM> <199603141739.TAA06324@blinky.cpaf.com> Cc: hjstein@blinky.cpaf.com, kanderso@bbn.com, stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I just wrote that (mapcar #'scalar* m m) is significantly faster than (* m m). But, before you all go out and recompile your xlispstat binary to take advantage of this, note that scalar* will *not* necessarily be faster than * for byte compiled code. I just tested scalar* vs * using: (defun vect-prod (x y result) (dotimes (i (length x)) (setf (aref result i) (* (aref x i) (aref y i))))) vs (defun vect-prod2 (x y result) (dotimes (i (length x)) (setf (aref result i) (scalar* (aref x i) (aref y i))))) It turns out that vect-prod is marginally slower than vect-prod2 on a c-double array of size 100000 (presumably because interpreter overhead swamps the difference between * and scalar*). However, after byte compiling, vect-prod is *faster* than vect-prod2. Out of 10 runs, vect-prod takes 1.184 sec (standard dev = 0.0143), and vect-prod2 takes 1.629 sec (standard dev = 0.0100)! I'd guess that the byte compiler has a byte code operand for *, and that dispatching on a byte code operand is much faster than dispatching to an arbitrary function. If I get ambitious, I might look into the byte compiler to see if this is the case, and if it is, maybe I'll try adding scalar* and see what happens... Also, in the same run, I tried out (* m m) again. The last 10 runs give an avg time of 1.12 sec (standard dev = 0.0115). This time, vect-prod is *not* beating (* m m), in fact, it's marginally slower. As others have pointed out, my previous claim that a byte compiled vect-prod can be faster than * is unjustified, in fact, the converse seems to be supported (by a small percentage). Note that timing (* v v), where v is set by (coerce m 'list), gives the same results as in my previous message (avg time 1.465 sec). So, it should be noted that vectorized * is significantly faster on c-double typed arrays than it is on lists. Thanks, Dr. Harvey J. Stein Berger Financial Research abel@netvision.net.il From majordom@nokomis.stat.umn.edu Thu Mar 14 17:18:50 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA11227; Thu, 14 Mar 1996 17:18:50 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA11220; Thu, 14 Mar 1996 16:18:47 -0600 From: Luke Tierney Message-Id: <9603142218.AA11220@nokomis.stat.umn.edu> Subject: Re: Efficiency tricks (was Re: Getting closer...) To: stat-lisp-news@nokomis.stat.umn.edu Date: Thu, 14 Mar 1996 16:18:46 -0600 (CST) In-Reply-To: <199603141739.TAA06324@blinky.cpaf.com> from "Harvey J. Stein" at Mar 14, 96 07:39:32 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 3985 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > Here's the timings of the last 10 runs of each: > > (* m m) (mapcar #'scalar* m m) > time gc time-gc time gc time-gc > Trial > 1 1.57 0.67 0.90 1.22 0.49 0.73 > 2 1.35 0.45 0.90 1.23 0.49 0.74 > 3 1.58 0.67 0.91 1.23 0.50 0.73 > 4 1.35 0.45 0.90 1.22 0.49 0.73 > 5 1.57 0.67 0.90 1.22 0.49 0.73 > 6 1.35 0.45 0.90 1.22 0.49 0.73 > 7 1.57 0.67 0.90 1.23 0.51 0.72 > 8 1.36 0.45 0.91 1.22 0.49 0.73 > 9 1.57 0.67 0.90 1.22 0.49 0.73 > 10 1.38 0.45 0.93 1.22 0.49 0.73 > If you insert (mapcar #'* m m) then this gives a result in between. On my system: > (time (progn (* m m) nil)) The evaluation took 1.83 seconds; 0.62 seconds in gc. NIL > (time (progn (mapcar #'* m m) nil)) The evaluation took 1.44 seconds; 0.42 seconds in gc. NIL > (time (progn (mapcar #'scalar* m m) nil)) The evaluation took 1.32 seconds; 0.47 seconds in gc. NIL The reason for the difference is most likely extra function calls -- things done with C functions, rather that macros or switches, to make the code simpler. When * is called with scalar arguments, there are two more C function calls than with scalar*. Similarly, when the mapping is done in *, there are two more function calls per element (accessing sequence elements) than in mapcar. The difference wetween the vectorized arithmetic mapping and mapcar is easy to eliminate, since it all comes down to one function, map in compound.c. I'll rewrite that to eliminate the function calls. The difference between * ans scalar* is a bit harder to deal with, but I might be able to handle it with a macro. I'll loox into that. It will result in increased code size (still a worry on Mac's and PC's) but may be worth it. > I just wrote that (mapcar #'scalar* m m) is significantly faster than > (* m m). But, before you all go out and recompile your xlispstat > binary to take advantage of this, note that scalar* will *not* > necessarily be faster than * for byte compiled code. > > I just tested scalar* vs * using: > > (defun vect-prod (x y result) > (dotimes (i (length x)) > (setf (aref result i) (* (aref x i) (aref y i))))) > > vs > > (defun vect-prod2 (x y result) > (dotimes (i (length x)) > (setf (aref result i) (scalar* (aref x i) (aref y i))))) > > > It turns out that vect-prod is marginally slower than vect-prod2 on a > c-double array of size 100000 (presumably because interpreter overhead > swamps the difference between * and scalar*). However, after byte > compiling, vect-prod is *faster* than vect-prod2. Out of 10 runs, > vect-prod takes 1.184 sec (standard dev = 0.0143), and vect-prod2 > takes 1.629 sec (standard dev = 0.0100)! > > I'd guess that the byte compiler has a byte code operand for *, and > that dispatching on a byte code operand is much faster than > dispatching to an arbitrary function. > That is exactly what happens. The scalar versions of these sunctions will eventually be availale once I get around to taking proper advantage of the package mechanism by putting all the lisp-stat related stuff in its own package. Then the scalar functions will be in package COMMON-LISP or CL (eg cl:*) and the vectorized ones in package LISP-STAT or LS (so ls:*). There are a lot of details I still need to work out about how to handle putting things in packages, so this won't happen overnight, but will happen eventually. (For what it is worth, the main reason I added the package facility was to be able to isolate the internals of the compiler -- this it does quite nicely. Working through the rest of the code to take advantage of the package system has lower priority.) luke From majordom@umnstat.stat.umn.edu Thu Mar 14 20:09:23 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26453; Thu, 14 Mar 1996 20:09:23 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26446; Thu, 14 Mar 1996 19:09:06 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa25494; 14 Mar 96 19:28 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA22039; Thu, 14 Mar 1996 19:28:50 -0500 Date: Thu, 14 Mar 1996 19:28:49 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Cc: Luke Tierney Subject: IVIEW, HARDWARE-OBJECTS and linking plots. Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk If I make an object with a 'hardware-address slot equal to Nil, the: "Error: no slot by this name - HARDWARE-ADDRESS" error when linking to it and selecting points, changes to: "Error: No IView installed in this window" If I set the hardware-address equal to the hardware-address of a third plot, then the code generates no errors, but does not propagate state to the linked objects. If I modify :adjust-screen-point to also check and reset the state, then everything works as it should, until I close the dummy plot whose hardware-address slot I have borrowed. The "No IView" error then reappears. My guess is that state setting is the function that violates encapsulation here: point state is not set or read via the :point-state or :last-point-state methods, but objects diddle with each other's implementation specific internal-state. Further evidence is that no :POINT-STATE messages were traced until I added the explicit call in :ADJUST-SCREEN-POINT, and when I "borrow" the HARDWARE-ADDRESS slot of the Green plot, it does not, because it is not linked, get a call to :adjust-screen, but if I send it either a :REDRAW, or a :NEEDS-ADJUSTING T and a :ADJUST-SCREEN, the selection is updated to be identical to the RED and BLUE plots. The test code at the bottom demonstrates the problem if you load it, select some points, and the close the GREEN plot and select or deselect points. Is there any way I can "fake out" IView's need for a HARDWARE-ADDRESS slot without actually having a dummy window on the screen ? ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] ;;; LINKABLE.LSP ;;; linkable object Mixin ( defproto linkable-proto '( links linked ) () ) ( defmeth linkable-proto :linked ( &optional ( value nil set )) ( if set ( slot-value 'linked value ) ( if ( slot-value 'linked ) ( send self :links )))) ( defmeth linkable-proto :links ( &optional ( value nil set )) ( if set ( slot-value 'links value ) ( slot-value 'links ))) ( defmeth linkable-proto :add-link (obj) ( let (( links ( send self :links ))) ( if ( not ( find obj links :test #'eq )) ( send self :links ( cons obj links ))))) ( defmeth linkable-proto :remove-link (obj) ( send self :links ( remove obj ( send self :links ) :test #'eq ))) ( defmeth linkable-proto :broadcast ( &rest msg ) ( dolist ( other ( send self :links )) ( if ( send other :has-method ( car msg )) ( apply #'send other msg )))) ( provide "LINKABLE" ) ;;; TEST-MAPPER.LSP ( require "linkable" ) ( defproto test-mapper-proto '( a b hardware-address ) () linkable-proto ) ( defmeth test-mapper-proto :isnew ( a b ) ( slot-value 'a a ) ( slot-value 'b b ) ( send a :links ( cons self ( send a :links ))) ( send b :links ( cons self ( send b :links ))) ( send self :links ( list a b )) ( send self :linked T )) ( defmeth test-mapper-proto :point-state ( n &optional ( val nil set )) ( print ( list ( slot-value 'proto-name ) :point-state n val )) ( when set ( send ( slot-value 'a ) :point-state n val ) ( send ( slot-value 'b ) :point-state n val ))) ( defmeth test-mapper-proto :last-point-state ( n &optional ( val nil set )) ( print ( list ( slot-value 'proto-name ) :last-point-state n val )) ( when set ( send ( slot-value 'a ) :last-point-state n val ) ( send ( slot-value 'b ) :last-point-state n val ))) ( defmeth test-mapper-proto :adjust-screen-point ( n ) ( print ( list ( slot-value 'proto-name ) :adjust-screen-point n )) ( send ( slot-value 'a ) :adjust-screen-point n ) ( send ( slot-value 'b ) :adjust-screen-point n )) ( defmeth test-mapper-proto :adjust-screen ( ) ( print ( list ( slot-value 'proto-name ) :adjust-screen )) ( send ( slot-value 'a ) :adjust-screen ) ( send ( slot-value 'b ) :adjust-screen )) ( defmeth scatterplot-proto :links ( &optional ( value nil set )) ( unless ( send self :has-slot 'links ) ( send self :add-slot 'links )) ( if set ( setf ( slot-value 'links ) value ) ( slot-value 'links ))) ( defmeth scatterplot-proto :linked ( &optional arg ) ( if ( send self :links ) T Nil )) ( defmeth scatterplot-proto :adjust-screen () ( print ( list ( slot-value 'proto-name ) ( if ( send self :has-method :title ) ( send self :title )) :adjust-screen )) ( call-next-method )) ( defmeth scatterplot-proto :adjust-screen-point ( n ) ( print ( list ( slot-value 'proto-name ) ( if ( send self :has-method :title ) ( send self :title )) :adjust-screen-point n )) ;;;;;;;;;;;;;;;;;; insert link code -- ( when ( send self :linked ) ( dolist ( link ( send self :links )) ( send link :point-state n ( send self :point-state n )))) ;;;;;;;;;;;;;;;;;; ( call-next-method n )) ( defmeth scatterplot-proto :point-state ( n &optional ( value nil set )) ( print ( list ( slot-value 'proto-name ) ( if ( send self :has-method :title ) ( send self :title )) :point-state n value )) ( if set ( call-next-method n value ) ( call-next-method n ))) ( defmeth scatterplot-proto :last-point-state ( n &optional ( value nil set )) ( print ( list ( slot-value 'proto-name ) ( if ( send self :has-method :title ) ( send self :title )) :last-point-state n value )) ( if set ( call-next-method n value ) ( call-next-method n ))) ( setf nnn ( iseq 20 )) ( setf red ( plot-points nnn nnn :color 'red :title "RED" )) ( setf dx (+ 8 (car ( send red :size )))) ( setf x0 ( car ( send red :location ))) ( setf y0 ( cadr ( send red :location ))) ( setf green ( plot-points nnn nnn :color 'green :title "GREEN" :location (list ( + x0 dx ) (+ 20 y0) ))) ( setf blue ( plot-points nnn nnn :color 'blue :title "BLUE" :location (list ( + x0 dx dx ) (+ 24 y0) ))) ( setf mm ( send test-mapper-proto :new red blue )) ( send mm :slot-value 'hardware-address ( send green :slot-value 'hardware-address )) From majordom@umnstat.stat.umn.edu Tue Mar 19 10:20:53 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10182; Tue, 19 Mar 1996 10:20:53 -0500 Return-Path: Received: from gateway.kellogg.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10169; Tue, 19 Mar 1996 09:19:48 -0600 Received: by gateway.kellogg.com id AA06503 (InterLock SMTP Gateway 3.0 for stat-lisp-news@umnstat.stat.umn.edu); Tue, 19 Mar 1996 10:16:04 -0500 Received: by gateway.kellogg.com (Protected-side Proxy Mail Agent-1); Tue, 19 Mar 1996 10:16:04 -0500 Mime-Version: 1.0 Date: Tue, 19 Mar 1996 10:17:24 -0500 Message-Id: <14ed0700@cornelius.scp.com> From: Charles.Kincaid@kellogg.com (Charles Kincaid) Subject: compiling 3.39 To: stat-lisp-news@umnstat.stat.umn.edu Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Description: cc:Mail note part Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, I'm just now getting around to compiling 3.39 (I'm pretty sure that's the version. I don't have it here) on my NeXTstation running NS3.2. When I compiled the previous version 2.1R2(?) I had mouseX running and had all of the X11 libraries. However, now I have Cub'X user version, i.e. no libraries. Is there a way to compile this and draw X11 windows rather than gnuplot? Can I download the mouseX stuff and load those (X11R4) libraries or does it have to be X11R5 which is what Cub'X uses (I'm pretty sure)? Thank you for any info. Chuck Kincaid From majordom@umnstat.stat.umn.edu Wed Mar 20 19:23:29 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00572; Wed, 20 Mar 1996 19:23:29 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00563; Wed, 20 Mar 1996 18:23:20 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa05720; 20 Mar 96 16:04 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA24588; Wed, 20 Mar 1996 16:04:03 -0500 Date: Wed, 20 Mar 1996 16:03:59 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List , Luke Tierney Subject: Re: IVIEW, HARDWARE-OBJECTS and linking plots. In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Sorry if this message was a little short on context. This was a followup on two previous messages, both with different subject lines, about my problems trying to link points in a namelist plot with graphic overlay objects. Unfortunately, the current linking protocol seems to violate object encapsulation: you can only link to an object with an allocated HARDWARE-ADDRESS representing an allocated IDRAW graphics window. [ Now that I've finally managed to rebuild 3.44 on the mac, if I don't hear any better suggestions, I may try to modify the idraw sources to provide a "dummy" graphics object.] On Thu, 14 Mar 1996, Steven D. Majewski wrote: > > If I make an object with a 'hardware-address slot equal to Nil, the: > "Error: no slot by this name - HARDWARE-ADDRESS" > error when linking to it and selecting points, changes to: > "Error: No IView installed in this window" > > If I set the hardware-address equal to the hardware-address of a > third plot, then the code generates no errors, but does not propagate > state to the linked objects. > > If I modify :adjust-screen-point to also check and reset the state, > then everything works as it should, until I close the dummy plot > whose hardware-address slot I have borrowed. The "No IView" error > then reappears. > > My guess is that state setting is the function that violates > encapsulation here: point state is not set or read via the > :point-state or :last-point-state methods, but objects diddle > with each other's implementation specific internal-state. > > Further evidence is that no :POINT-STATE messages were traced > until I added the explicit call in :ADJUST-SCREEN-POINT, and > when I "borrow" the HARDWARE-ADDRESS slot of the Green plot, > it does not, because it is not linked, get a call to :adjust-screen, > but if I send it either a :REDRAW, or a :NEEDS-ADJUSTING T and > a :ADJUST-SCREEN, the selection is updated to be identical to > the RED and BLUE plots. > > > The test code at the bottom demonstrates the problem if you load > it, select some points, and the close the GREEN plot and select > or deselect points. > > > Is there any way I can "fake out" IView's need for a > HARDWARE-ADDRESS slot without actually having a dummy window on > the screen ? > > > ---| Steven D. Majewski (804-982-0831) |--- > ---| Computer Systems Engineer University of Virginia |--- > ---| Department of Molecular Physiology and Biological Physics |--- > ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- > [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] > > > ;;; LINKABLE.LSP > > ;;; linkable object Mixin > > > ( defproto linkable-proto '( links linked ) () ) > > > ( defmeth linkable-proto :linked ( &optional ( value nil set )) > ( if set ( slot-value 'linked value ) > ( if ( slot-value 'linked ) ( send self :links )))) > > ( defmeth linkable-proto :links ( &optional ( value nil set )) > ( if set ( slot-value 'links value ) > ( slot-value 'links ))) > > > ( defmeth linkable-proto :add-link (obj) > ( let (( links ( send self :links ))) > ( if ( not ( find obj links :test #'eq )) > ( send self :links ( cons obj links ))))) > > ( defmeth linkable-proto :remove-link (obj) > ( send self :links > ( remove obj ( send self :links ) :test #'eq ))) > > ( defmeth linkable-proto :broadcast ( &rest msg ) > ( dolist ( other ( send self :links )) > ( if ( send other :has-method ( car msg )) > ( apply #'send other msg )))) > > ( provide "LINKABLE" ) > > > > ;;; TEST-MAPPER.LSP > > ( require "linkable" ) > > ( defproto test-mapper-proto '( a b hardware-address ) () linkable-proto ) > > ( defmeth test-mapper-proto :isnew ( a b ) > ( slot-value 'a a ) > ( slot-value 'b b ) > ( send a :links ( cons self ( send a :links ))) > ( send b :links ( cons self ( send b :links ))) > ( send self :links ( list a b )) > ( send self :linked T )) > > > > ( defmeth test-mapper-proto :point-state ( n &optional ( val nil set )) > ( print ( list ( slot-value 'proto-name ) :point-state n val )) > ( when set > ( send ( slot-value 'a ) :point-state n val ) > ( send ( slot-value 'b ) :point-state n val ))) > > ( defmeth test-mapper-proto :last-point-state ( n &optional ( val nil set )) > ( print ( list ( slot-value 'proto-name ) :last-point-state n val )) > ( when set > ( send ( slot-value 'a ) :last-point-state n val ) > ( send ( slot-value 'b ) :last-point-state n val ))) > > ( defmeth test-mapper-proto :adjust-screen-point ( n ) > ( print ( list ( slot-value 'proto-name ) :adjust-screen-point n )) > ( send ( slot-value 'a ) :adjust-screen-point n ) > ( send ( slot-value 'b ) :adjust-screen-point n )) > > ( defmeth test-mapper-proto :adjust-screen ( ) > ( print ( list ( slot-value 'proto-name ) :adjust-screen )) > ( send ( slot-value 'a ) :adjust-screen ) > ( send ( slot-value 'b ) :adjust-screen )) > > > > ( defmeth scatterplot-proto :links ( &optional ( value nil set )) > ( unless ( send self :has-slot 'links ) ( send self :add-slot 'links )) > ( if set > ( setf ( slot-value 'links ) value ) > ( slot-value 'links ))) > > ( defmeth scatterplot-proto :linked ( &optional arg ) > ( if ( send self :links ) T Nil )) > > ( defmeth scatterplot-proto :adjust-screen () > ( print ( list > ( slot-value 'proto-name ) > ( if ( send self :has-method :title ) ( send self :title )) > :adjust-screen )) > ( call-next-method )) > > ( defmeth scatterplot-proto :adjust-screen-point ( n ) > ( print ( list > ( slot-value 'proto-name ) > ( if ( send self :has-method :title ) ( send self :title )) > :adjust-screen-point n )) > ;;;;;;;;;;;;;;;;;; insert link code -- > ( when ( send self :linked ) > ( dolist ( link ( send self :links )) > ( send link :point-state n ( send self :point-state n )))) > ;;;;;;;;;;;;;;;;;; > ( call-next-method n )) > > > ( defmeth scatterplot-proto :point-state ( n &optional ( value nil set )) > ( print ( list ( slot-value 'proto-name ) > ( if ( send self :has-method :title ) ( send self :title )) > :point-state > n value )) > ( if set > ( call-next-method n value ) > ( call-next-method n ))) > > > ( defmeth scatterplot-proto :last-point-state ( n &optional ( value nil > set )) > ( print ( list ( slot-value 'proto-name ) > ( if ( send self :has-method :title ) ( send self :title )) > :last-point-state > n value )) > ( if set > ( call-next-method n value ) > ( call-next-method n ))) > > > > > ( setf nnn ( iseq 20 )) > ( setf red ( plot-points nnn nnn :color 'red :title "RED" )) > ( setf dx (+ 8 (car ( send red :size )))) > ( setf x0 ( car ( send red :location ))) > ( setf y0 ( cadr ( send red :location ))) > > ( setf green ( plot-points nnn nnn :color 'green :title "GREEN" > :location (list ( + x0 dx ) (+ 20 y0) ))) > ( setf blue ( plot-points nnn nnn :color 'blue :title "BLUE" > :location (list ( + x0 dx dx ) (+ 24 y0) ))) > > > ( setf mm ( send test-mapper-proto :new red blue )) > > > ( send mm :slot-value 'hardware-address > ( send green :slot-value 'hardware-address )) > > > > ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Wed Mar 20 19:46:37 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00761; Wed, 20 Mar 1996 19:46:37 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00756; Wed, 20 Mar 1996 18:46:09 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa26202; 20 Mar 96 15:43 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA23662; Wed, 20 Mar 1996 15:42:24 -0500 Date: Wed, 20 Mar 1996 15:42:24 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Luke Tierney Cc: Xlisp List Subject: Building Xlisp-Stat on Mac with CodeWarrior 8 Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Compiling Xlisp-Stat 3-44 on PPC Macintosh using CodeWarrior 8 [ Note: Success with 3.44, but I still haven't been able to build 3.47. Question: Are there any instructions, examples or documentation on how to build and use external function on the Mac ? ] Source code changes: [1] linalg.h:: wrap #define abs with conditional #ifndef #ifnedf abs #define abs [...] #endif [2] macstuff.c:: The definition of struct PPCPortRec has changed slightly -- Change the several occurances of: name.u.port.type => name.u.port.portType name.u.port.creator => name.u.port.portCreator [3] TransSkel1.c:: Changed "(long)" cast to "(long *)": # ifdef __CONDITIONALMACROS__ InitDialogs ((long *) initParams->skelResumeProc); # else InitDialogs (initParams->skelResumeProc); # endif (I have not investigated why this change - this was a guess guided by the compiler complaints. ) Open and convert the old project file. Project Preferences Changes: add {Project f}:: to access paths. Change Prefix File from "CWHeader" to "CWHeader.h" in C/C++ Language prefs and Turn on MPW Pointer Type Rules option. ( This got rid of the first type complaint when compiling TransSkel1.c -- the source mode above got rid of the other. ) add console.stub.c & SIOUX.PPC.LIB to the project. Make sure that you have checked the convert newlines option when untaring on the Mac, or else convert the .lsp source code. If you don't, load of file will return T, but you won't get the side effects of loading the file. CW and ThinkC libraries seem to handle this differently. ( load "cmpsys" ) generates a number of stack overflow errors compiling some files. Increasing the stack size from 32k to 96k and relinking did not fix this problem. Everything else I've tested so far seems to work fine. 3.47 does not build - the distribution is missing a file: "rcondest.c". Copying the rcondest.c file from the 3.44 distribution and compiling, generates a page full of errors that aren't generated in the 3.44 system compile. Now that I have 3.44 successfully compiling I'll take another look at the problem. ( Why is this file missing from 3.47 ? ) ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Wed Mar 20 23:36:31 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA01732; Wed, 20 Mar 1996 23:36:31 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA01727; Wed, 20 Mar 1996 22:36:21 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa26202; 20 Mar 96 15:43 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA23662; Wed, 20 Mar 1996 15:42:24 -0500 Date: Wed, 20 Mar 1996 15:42:24 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Luke Tierney Cc: Xlisp List Subject: Building Xlisp-Stat on Mac with CodeWarrior 8 Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Compiling Xlisp-Stat 3-44 on PPC Macintosh using CodeWarrior 8 [ Note: Success with 3.44, but I still haven't been able to build 3.47. Question: Are there any instructions, examples or documentation on how to build and use external function on the Mac ? ] Source code changes: [1] linalg.h:: wrap #define abs with conditional #ifndef #ifnedf abs #define abs [...] #endif [2] macstuff.c:: The definition of struct PPCPortRec has changed slightly -- Change the several occurances of: name.u.port.type => name.u.port.portType name.u.port.creator => name.u.port.portCreator [3] TransSkel1.c:: Changed "(long)" cast to "(long *)": # ifdef __CONDITIONALMACROS__ InitDialogs ((long *) initParams->skelResumeProc); # else InitDialogs (initParams->skelResumeProc); # endif (I have not investigated why this change - this was a guess guided by the compiler complaints. ) Open and convert the old project file. Project Preferences Changes: add {Project f}:: to access paths. Change Prefix File from "CWHeader" to "CWHeader.h" in C/C++ Language prefs and Turn on MPW Pointer Type Rules option. ( This got rid of the first type complaint when compiling TransSkel1.c -- the source mode above got rid of the other. ) add console.stub.c & SIOUX.PPC.LIB to the project. Make sure that you have checked the convert newlines option when untaring on the Mac, or else convert the .lsp source code. If you don't, load of file will return T, but you won't get the side effects of loading the file. CW and ThinkC libraries seem to handle this differently. ( load "cmpsys" ) generates a number of stack overflow errors compiling some files. Increasing the stack size from 32k to 96k and relinking did not fix this problem. Everything else I've tested so far seems to work fine. 3.47 does not build - the distribution is missing a file: "rcondest.c". Copying the rcondest.c file from the 3.44 distribution and compiling, generates a page full of errors that aren't generated in the 3.44 system compile. Now that I have 3.44 successfully compiling I'll take another look at the problem. ( Why is this file missing from 3.47 ? ) ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Thu Mar 21 11:22:49 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13226; Thu, 21 Mar 1996 11:22:49 -0500 Return-Path: Received: from BALU.kfunigraz.ac.at by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13221; Thu, 21 Mar 1996 10:22:41 -0600 Received: from pcim82 (BIMIGW.kfunigraz.ac.at) by balu.kfunigraz.ac.at with SMTP (1.37.109.16/16.2) id AA281355315; Thu, 21 Mar 1996 17:21:55 +0100 Message-Id: <31519C29.4B2B@kfunigraz.ac.at> Date: Thu, 21 Mar 1996 17:12:57 -0100 From: Franz Quehenberger Organization: Karl-Franzens-Universitaet Graz X-Mailer: Mozilla 2.0 (WinNT; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: Q: Save XLISP-STAT Objects Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Q: I am looking for code to save XLISP 'objects' as a recursive function that writes an ASCII representation of a LISP object to a file. Unfortunately Mark Kantrowitz' SAVE_OBJ.CL requires Common-Lisp. At the moment I am adapting his code. For example objects are save through (derived from code from Luke Tierney's distribution): (defun object-dump-form (object) (labels ((save-slot (slot) `(send m :slot-value ',slot ',(get-dump-form (send object :slot-value slot))))) (let* ((proto (send object :slot-value 'proto-name)) (islots (send object :slot-value 'instance-slots))) (append (list 'let (list (list 'm (list 'send proto ':new)))) (mapcar #'save-slot islots) (list 'm))))) Conses are dumped through: (defun CONS-DUMP-FORM (item) `(CONS ,(get-dump-form (first item)) ,(get-dump-form (rest item)))) and so on. The function GET-DUMP-FORM calls the dump-form, that ist adequate to the type of the object. Does anyone know of related work? Thanks! -- franz.quehenberger@kfunigraz.ac.at (++43 316 385-3205, fax -3590) Dept. of Medical Informatics, Statistics, and Documentation Graz University, Auenbruggerplatz 9/III, A-8036 Graz, Austria From majordom@nokomis.stat.umn.edu Thu Mar 21 12:06:45 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA16535; Thu, 21 Mar 1996 12:06:45 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA16529; Thu, 21 Mar 1996 11:06:44 -0600 From: Luke Tierney Message-Id: <9603211706.AA16529@nokomis.stat.umn.edu> Subject: Message from forrest young asking for help To: stat-lisp-news@nokomis.stat.umn.edu Date: Thu, 21 Mar 1996 11:06:44 -0600 (CST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 2991 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk Majordomo bounced this message since its subject line contained the word 'help'. Forwarded message: > From majordom@umnstat.stat.umn.edu Thu Mar 21 11:04 EST 1996 > Date: Thu, 21 Mar 1996 11:04:50 -0500 > Message-Id: <9603211604.AA13155@umnstat.stat.umn.edu> > To: owner-stat-lisp-news@umnstat.stat.umn.edu > From: owner-stat-lisp-news@umnstat.stat.umn.edu > Subject: BOUNCE stat-lisp-news@stat.umn.edu: Admin request > > >From forrest@unc.edu Thu Mar 21 10:04:42 1996 > Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP > (1.38.193.4/16.2) id AA13142; Thu, 21 Mar 1996 10:04:42 -0600 > Return-Path: > Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.6.4.3/10.1) > id LAA25683; Thu, 21 Mar 1996 11:04:38 -0500 > X-Sender: uluru@gibbs.oit.unc.edu > Message-Id: > Mime-Version: 1.0 > Content-Type: text/plain; charset="us-ascii" > Date: Thu, 21 Mar 1996 11:06:33 -0500 > To: stat-lisp-news@umnstat.stat.umn.edu > From: forrest@unc.edu (Forrest Young) > Subject: HELP! > > 1) First Question: > Have I discovered a bug or is there something wrong with the following code: > > (dotimes (i n) > (send self :label-strings > (combine (send self :label-strings > (format nil "~a~d" "Obs" nobs+i))))) > > This code is supposed to add strings such as "Obs3" "Obs4" "Obs5" to the > strings that already exist in the label-strings slot. However, if, say, > n=14, nobs=1 and (send self :label-strings) returns ("Obs0") I get errors > such as > > Error: index out of bounds - 1 > Happened in: # > > or > > Error: index out of range - 1 > Happened in: # > > Furthermore, the number of times I can get through the loop before the > error occurs changes from time to time. Close examination shows that the > errors seem to be coming from the second (send self :label-strings). This > method is a simple slot-accessor method defined as follows: > > (defmeth datasheet-proto :label-strings (&optional (list nil set)) > "Message args: (&optional list) > Sets or retrieves the list of label strings." > (if set (setf (slot-value 'label-strings) list)) > (slot-value 'label-strings)) > > HELP! > > 2) Second Question: > If I replace the above code with > > (send self :label-strings > (combine (send self :label-strings) (repeat "Obs" n))) > > I have no problem. However, this code creates the strings "Obs" "Obs" "Obs" > etc., where what I really want is "Obs1" "Obs2" "Obs3" etc. Can someone > suggest a way of doing this that is different than the way I tried above? > > Forrest Young > Professor, Psychometrics > > CB# 3270 Davie Hall email: mailto://forrest@unc.edu > UNC Psychometric Lab www: http://forrest.psych.unc.edu > Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 > > ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html > > > From majordom@umnstat.stat.umn.edu Thu Mar 21 12:19:41 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13673; Thu, 21 Mar 1996 12:19:41 -0500 Return-Path: Received: from STOUT.BBN.COM by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13668; Thu, 21 Mar 1996 11:19:36 -0600 Received: from stout (localhost [127.0.0.1]) by stout.BBN.COM (8.6.9/8.6.5) with ESMTP id MAA22390; Thu, 21 Mar 1996 12:16:46 -0500 Message-Id: <199603211716.MAA22390@stout.BBN.COM> To: Franz Quehenberger Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Q: Save XLISP-STAT Objects In-Reply-To: Your message of "Thu, 21 Mar 1996 17:12:57 -0100." <31519C29.4B2B@kfunigraz.ac.at> Date: Thu, 21 Mar 1996 12:16:42 +0000 From: Ken Anderson Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Q: I am looking for code to save XLISP 'objects' as a recursive function that writes an ASCII representation of a LISP object to a file. Does anyone know of related work? I'll send you some code i have, also in Common Lisp. One issue when dumping to to avoid consing up something that is as big as the thing you are trying to dump. k From majordom@umnstat.stat.umn.edu Thu Mar 21 15:21:09 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15942; Thu, 21 Mar 1996 15:21:09 -0500 Return-Path: Received: from stat19.stat.rice.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15937; Thu, 21 Mar 1996 14:21:05 -0600 Received: (from scottdw@localhost) by stat19.stat.rice.edu (8.7.5/8.7.3) id OAA00803; Thu, 21 Mar 1996 14:20:48 -0600 (CST) Date: Thu, 21 Mar 1996 14:20:48 -0600 (CST) From: David Scott Message-Id: <199603212020.OAA00803@stat19.stat.rice.edu> To: s-news@utstat.toronto.edu, outnews+netnews.sci.stat.math@andrew.cmu.edu, outnews+netnews.sci.stat.edu@andrew.cmu.edu, outnews+netnews.sci.stat.consult@andrew.cmu.edu, bayes-news@stat.cmu.edu, stat-lisp-news@umnstat.stat.umn.edu Subject: tenure track job announcement Cc: thomp@stat19.stat.rice.edu, dbrown@stat19.stat.rice.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk RICE UNIVERSITY DEPARTMENT OF STATISTICS The Department of Statistics at Rice Univeristy announces an opening for a tenure track post in probability and statistics. Women and/or minorities are encouraged to apply. Although junior applicants are preferred, it is possible that exceptional candidates might be considered at the rank of associate or full professor. An interested applicant should send a vita and four names of reference to: James R. Thompson, Chairman Department of Statistics - MS 138 Rice University Houston, Texas 77251-1892 Deadline for application is May 1, 1996 or until position is filled. From majordom@umnstat.stat.umn.edu Thu Mar 21 16:35:01 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16615; Thu, 21 Mar 1996 16:35:01 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16610; Thu, 21 Mar 1996 15:34:56 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id QAA10714; Thu, 21 Mar 1996 16:34:53 -0500 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 21 Mar 1996 16:31:53 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: problem with LispStat Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I have two questions for the LispStat group: 1) First Question: Have I discovered a bug or is there something wrong with the following code? (dotimes (i n) (send self :label-strings (combine (send self :label-strings (format nil "~a~d" "Obs" nobs+i))))) This code is supposed to add strings such as "Obs3" "Obs4" "Obs5" to the strings that already exist in the label-strings slot. However, if, say, n=14, nobs=1 and (send self :label-strings) returns ("Obs0") I get errors such as Error: index out of bounds - 1 Happened in: # or Error: index out of range - 1 Happened in: # Furthermore, the number of times I can get through the loop before the error occurs changes from time to time. Close examination shows that the errors seem to be coming from the second (send self :label-strings). This method is a simple slot-accessor method defined as follows: (defmeth datasheet-proto :label-strings (&optional (list nil set)) "Message args: (&optional list) Sets or retrieves the list of label strings." (if set (setf (slot-value 'label-strings) list)) (slot-value 'label-strings)) HELP! 2) Second Question: If I replace the above code with (send self :label-strings (combine (send self :label-strings) (repeat "Obs" n))) I have no problem. However, this code creates the strings "Obs" "Obs" "Obs" etc., where what I really want is "Obs1" "Obs2" "Obs3" etc. Can someone suggest a way of doing this that is different than the way I tried above? Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Thu Mar 21 16:44:46 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16678; Thu, 21 Mar 1996 16:44:46 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16673; Thu, 21 Mar 1996 15:44:41 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id QAA11583; Thu, 21 Mar 1996 16:44:32 -0500 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 21 Mar 1996 16:41:33 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: problem with LispStat (correction) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Its been pointed out to me that there is a variable "nobs+i" in my code that doesn't make sense. True...it was left over from my obsessively changing code that was originally (+ nobs i). sorry 'bout that Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Fri Mar 22 08:48:00 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29509; Fri, 22 Mar 1996 08:48:00 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29504; Fri, 22 Mar 1996 07:47:49 -0600 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.6.4.3/10.1) id IAA27698; Fri, 22 Mar 1996 08:47:44 -0500 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 22 Mar 1996 08:49:38 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: Problem with LispStat (3rd time) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Let me restate my problem, perhaps this time without the mistakes I made last time. I've also incorporated a change suggested by Frederick Udina. My Question: Have I discovered a bug or is there something wrong with the my code? What I want to do is to generate a list of strings with increasing integers at their end and add them at the end of a list of strings that already exists in a slot. The slot is "label-strings". So if I do: (send self :label-strings) I get something like: ("a" "b" "c") Frederick noted that if I used: > (mapcar #'(lambda (n) (concatenate 'string "Obs" (num-to-string n))) (iseq 5)) I would get: ("Obs0" "Obs1" "Obs2" "Obs3" "Obs4") This is true. Thus I write (send self :label-strings (combine (send self :label-strings) (mapcar #'(lambda (i) (concatenate 'string "Obs" (num-to-string (+ nobs i)))) (iseq n)))) where nobs=3 (the length of the list already in the slot) and n is the number of new strings I want to add. If n were 5 I'd like to get ("a" "b" "c" "Obs0" "Obs1" "Obs2" "Obs3" "Obs4") Sometimes I do, sometimes I get errors. There errors are either Error: index out of bounds - 1 Happened in: # or Error: index out of range - 1 Happened in: # To repeat my question: Is this a bug or is there something wrong with the my code? Forrest Young Professor, Psychometrics CB# 3270 Davie Hall email: mailto://forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Sun Mar 24 09:49:18 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28819; Sun, 24 Mar 1996 09:49:18 -0500 Return-Path: Received: from angel.heaven.net by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28814; Sun, 24 Mar 1996 08:49:13 -0600 Received: from [198.69.29.97] (josh.netheaven.com [198.69.29.97]) by angel.heaven.net (8.6.12/8.6.12) with SMTP id JAA10877 for ; Sun, 24 Mar 1996 09:48:29 -0500 Message-Id: <199603241448.JAA10877@angel.heaven.net> X-Sender: josh@netheaven.com Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sun, 24 Mar 1996 09:50:38 -0500 To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) From: josh@racing.saratoga.ny.us (Josh Kuperman) Subject: [Q] byte-compiled code with loops on the Mac Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I compiled some code that used the loop macros. The interpreted version ran fine, though slowly. The compiled version would come up with an error - variable "until" not defined. I have two thoughts as to the cause. (1) I have an actual error in the code but the clause that is triggering it only occurs when the code is rearranged for optimization. (2) I missed some trick to make sure that the supporting code for the loop clause was known to the compiler. Josh Kuperman josh@racing.saratoga.ny.us http://www.racing.saratoga.ny.us/~josh (barely under construction) From majordom@umnstat.stat.umn.edu Sun Mar 24 14:34:51 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29659; Sun, 24 Mar 1996 14:34:51 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29654; Sun, 24 Mar 1996 13:34:46 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa18306; 24 Mar 96 14:34 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA24864; Sun, 24 Mar 1996 14:34:05 -0500 Newsgroups: comp.lang.lisp.x Date: Sun, 24 Mar 1996 14:33:56 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Cc: Luke Tierney Subject: Common Lisp directory function Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Now that I've managed to compile xlispstat-3.47 successfully, I've figured out enough of the internals to provide one of my own previous requests: A 'directory' function to return a list of a directory's contents. ( Note: I've still been searching for examples or documentation on how to write a dynamically loadable code extension for xlisp on Mac. There's little explaination in the xlsx.h header file except for a reference to "the New S book". ) This has only been tested on Mac ( using CW8 ), but the code ought to be portable to any system with dirent.h routines. See comments in the code for outstanding questions about Common Lisp compliance. I have paper copy of CLtL1 nearby, but I may have to wait until Im back iin the office to check what CLtL2 says. Comments on my comments appreciated! You need to add "DIRECTORY" to the function table. On my first attempt ( with the simple #'xyzzy ), I added some extra entries to the table and got some strange behaviour. ( And no - I didn't drop the end-of-table marker. ) When I instead just modified some of the extra not-implemented entries, it worked fine. Is the size of this table hardcoded somewhere ? Or is there anything else that would explain this. ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] #include "xlisp.h" #include "xlstat.h" #include "dirent.h" /* additional xlisp functions * * Steven D. Majewski / University of Virginia - Physiology / * */ /* just an initial attempt to add a trivial function */ LVAL xlxyzzy(void) { char *buf="(You are in a ((maze) of (((twisty) (little)) parentheses)), all the same!)" ; xllastarg(); return cvstring(buf); } /* * minimal common-lisp directory function * CLtL1 says: * ( directory pathname &key ) * But keys are not defined or required, they are allowable extensions. * ( I need to check CLtL2 to see if this still holds. ) * * Defined in CLtL1 to return a list of pathnames. I'm returning a list of filename * strings. xlisp 'pathname' is implemented as a .lsp closure that just coerces a * stream into it's truename, else returns a string. * * Returns NIL if not passwed a valid directory name ( or, more precisely, if opendir * fails for any reason ): Should it signal an error ? * * I also need to check CLtL2 for other file/directory management routines. * For example: is there any portable equivalent of fstat, particularly * to check is-a-directory ? Since '( directory regular-file )' returns NIL * it can be used as a boolean, but it isn't really is-a-directory, since empty * directories will also return NIL -- it's functionally a Not-an-end-node test. * * [ Note: the above use is the reason I'm not raising an error for that case. ] * */ LVAL directory( char *dirname ) { DIR *dirp; struct dirent *entryp; LVAL result; xlsave1(result); /* protect 'result' on stack */ result = NIL; /* initial list, or final return value if not a valid directory */ dirp = opendir( dirname ); /* failure here will return NL */ if ( dirp ) { /* build a list of file names */ while ( entryp = readdir( dirp )) { result = cons( cvstring(entryp->d_name), result ); } closedir( dirp ); } xlpop(); /* restore stack */ return result; } /* * This is split up into an internal and external function because that's what * Jeff Prothero's xlisp internals guide suggested ( and my initial attempt * left out the xlpop() and messed up the stack, so I retried it following * his model more exactly. ) I don't know that there's really any need for an * internal directory function, and it's simple to fold this back into one. */ LVAL xldirectory(void) { char *dirname; dirname = getstring(xlgetfname()); xllastarg(); return directory( dirname ); } From majordom@umnstat.stat.umn.edu Wed Mar 27 08:46:38 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19046; Wed, 27 Mar 1996 08:46:38 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19041; Wed, 27 Mar 1996 07:46:32 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id IAA06073; Wed, 27 Mar 1996 08:46:06 -0500 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 27 Mar 1996 08:43:11 -0500 To: Luke Tierney , "Steven D. Majewski" , Frederic Udina From: forrest@unc.edu (Forrest Young) Subject: Solution to my Problem with LispStat Cc: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Dear Luke, Steven and Frederic: Thanks to each of you for your replies to my messages about the problems I was having. All of you agreed that my code looked innocent enough. I've finally figured out what the problem was, and it's such a tricky situation that I thought I'd share it with you (and with the group in general). The problem wasn't the code, it was a timing problem. Note that I'm working on Macintoshes and under Windows. The code was a part of a method for the datasheet that I have written as part of ViSta. The datasheet is a graphical object that inherits from graph-proto. The method in question presented a dialog box which let the user add rows or columns to an already existing datasheet. When the dialog box was closed the slots of the object were updated to reflect the additional rows/columns and the window was redrawn using the updated information and showing the additional rows/columns. The problem was that there are TWO redraws involved: The first one to take care of the disappearance of the dialog box; the second one to take care of the changes in the datasheet. And SOMETIMES the first one would occur when the object's slots were still in the process of being updated, so that they contained inconsistent information. When I realized this it was simple to place the window in a "don't redraw" state until all slots were updated. This actually is the second time I've run into this situation, the first being for my workmap and it's dialog boxes. That first time it literally took me THREE MONTHS to catch this bug. After all, it was un-reproducible, occured rarely, didn't occur on all machines (due to differences in speed) nor with all displays (due to differences in relative locations of the dialog box and window). It was clarly the worst debugging situation I've had in my 5 years on this project. At least I've learned from my previous mistakes - This time it only took me a week to realize that I had the same type of bug all over again. Next time let's hope I recognize the situation immediately. Once again, thanks for your help. I thought I was crazy there for a while. Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@nokomis.stat.umn.edu Wed Mar 27 11:34:19 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA02545; Wed, 27 Mar 1996 11:34:19 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA02511; Wed, 27 Mar 1996 10:24:48 -0600 From: Luke Tierney Message-Id: <9603271624.AA02511@nokomis.stat.umn.edu> Subject: Re: Solution to my Problem with LispStat To: forrest@unc.edu (Forrest Young) Date: Wed, 27 Mar 1996 10:24:48 -0600 (CST) Cc: luke@nokomis.stat.umn.edu, sdm7g@Virginia.EDU, udina@upf.es, stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Forrest Young" at Mar 27, 96 08:43:11 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 3518 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > Dear Luke, Steven and Frederic: > > Thanks to each of you for your replies to my messages about the problems I > was having. All of you agreed that my code looked innocent enough. I've > finally figured out what the problem was, and it's such a tricky situation > that I thought I'd share it with you (and with the group in general). > > The problem wasn't the code, it was a timing problem. Note that I'm working > on Macintoshes and under Windows. > > The code was a part of a method for the datasheet that I have written as > part of ViSta. The datasheet is a graphical object that inherits from > graph-proto. The method in question presented a dialog box which let the > user add rows or columns to an already existing datasheet. When the dialog > box was closed the slots of the object were updated to reflect the > additional rows/columns and the window was redrawn using the updated > information and showing the additional rows/columns. > > The problem was that there are TWO redraws involved: The first one to take > care of the disappearance of the dialog box; the second one to take care of > the changes in the datasheet. And SOMETIMES the first one would occur when > the object's slots were still in the process of being updated, so that they > contained inconsistent information. When I realized this it was simple to > place the window in a "don't redraw" state until all slots were updated. > > This actually is the second time I've run into this situation, the first > being for my workmap and it's dialog boxes. That first time it literally > took me THREE MONTHS to catch this bug. After all, it was un-reproducible, > occured rarely, didn't occur on all machines (due to differences in speed) > nor with all displays (due to differences in relative locations of the > dialog box and window). It was clarly the worst debugging situation I've > had in my 5 years on this project. > > At least I've learned from my previous mistakes - This time it only took me > a week to realize that I had the same type of bug all over again. Next time > let's hope I recognize the situation immediately. > > Once again, thanks for your help. I thought I was crazy there for a while. > > Forrest Young > Professor, Psychometrics > > UNC Psychometrics Lab e-mail: forrest@unc.edu > CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu > Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 > > ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html > > > This is a very good point. There is a very limited (and very ad hoc) bit of multitasking that happens, in particular in the Mac and Windows versions. This is needed to process window events related to switching from one application to another, among others. As a result, a :redraw can be triggered almost any time a window is visible. If you need to prevent this, you need to arrange for your :redraw method to do nothing in these inopportune calls and to be triggered once the data is ready. Essentially, you have to invent a little semaphore mechanism like Forrest did. I tinker with this at the ends every so often to try and minimize the effect but it is still there. The right solution is probably to add support for preemptive threads, put the event handling process in its own thread, and provide a synchronization mechanism to deal with situations like this. This is in the works for a future release (rather long term future at tis point). luke From majordom@umnstat.stat.umn.edu Wed Mar 27 11:54:26 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA20936; Wed, 27 Mar 1996 11:54:26 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA20931; Wed, 27 Mar 1996 10:54:18 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa26703; 27 Mar 96 11:41 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA27867; Wed, 27 Mar 1996 11:41:25 -0500 Date: Wed, 27 Mar 1996 11:41:24 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: [xlisp-stat] Any way to catch all messages ? Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Is there any way for an object to catch all messages, even and especially those that are undefined for the object ? I've been writing my own linking mechanism that doesn't trigger the implementation dependent graphics linking, and I have added a :broadcast method: this allows the base class methods to be implemented in terms of :broadcast, an only ( or mostly ) :broadcast has to be changed to alter parts of the linking behaviour. In debugging this code and in trying to figure out how to get to my required behaviour, I first tried linking two plots using my alternate linking mehanism ( it gives the same behaviour as the built-in linking, as it should ) and secondly, I tried linking the plots thru an intermediate object, that just propagates the messages it gets to the others in the chain. It would be useful for such an object to respond to all possible messages, - perhaps just rebroadcasting them to others, perhaps modifying the messages or arguments, or for debugging, maybe just printing a trace message before rebroadcasting. In my particular case, I want to transform a certain set of messages into a different set -- the object becomes an adapter for between two different interfaces/protocols -- but I would like it to just pass on any other message ( perhaps first checking :has-method on the receiver. ) From what I *THINK* I know about how ( send obj :msg args ) is translated into a ( msg-closure obj args ), I don't see a good place to hook into this, and if the :msg is not part of the args ( It isn't, is it? ) then even dynamically adding methods ( all dispatched to the same re-send code ) wouldn't work. This is not vital for my current project, but if it was possible, it would allow me to write more general and reusable objects, and would be a great debugging tool. ( Well - I guess for debugging, you can globally just redefine #'send and get the same effect. ) ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Wed Mar 27 12:52:51 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA21623; Wed, 27 Mar 1996 12:52:51 -0500 Return-Path: Received: from netcom6.netcom.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA21615; Wed, 27 Mar 1996 11:52:45 -0600 Received: from localhost by netcom6.netcom.com (8.6.13/Netcom) id JAA21793; Wed, 27 Mar 1996 09:52:38 -0800 Message-Id: <199603271752.JAA21793@netcom6.netcom.com> To: Luke Tierney , sdm7g@Virginia.EDU, udina@upf.es, stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Solution to my Problem with LispStat In-Reply-To: Your message of "Wed, 27 Mar 96 10:24:48 CST." <9603271624.AA02511@nokomis.stat.umn.edu> Date: Wed, 27 Mar 96 09:52:29 -0800 From: "Niels P. Mayer" Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >bit of multitasking that happens, in particular in the Mac and Windows >versions. This is needed to process window events related to switching >from one application to another, among others. As a result, a :redraw >can be triggered almost any time a window is visible. If you need to >prevent this, you need to arrange for your :redraw method to do >nothing in these inopportune calls and to be triggered once the data >is ready. Essentially, you have to invent a little semaphore mechanism >like Forrest did. I tinker with this at the ends every so often to >try and minimize the effect but it is still there. The right solution >is probably to add support for preemptive threads, put the event >handling process in its own thread, and provide a synchronization >mechanism to deal with situations like this. This is in the works for >a future release (rather long term future at tis point). Yipes!! This stuff has been dealt with effectively way back in the late 80's in this archaic piece of technology known as the X window system and the X toolkit. Hasn't anybody built an abstraction to give a "nice" interface (perhaps w/ {X,Xt,}-like API) to events and windows that is multiplatform? What are people's experiences in deploying xlisp-stat for multi-platform applications -- do you have to do a lot of mucking about at the events level to make things work just right? And what about WxWindows and Mr. Ed? When I do custom graphics apps with WINTERP/Xtango (on Sun, Irix, Linux, & HP platforms), the ability for multiple things to be going on at the same time is typically the least of my worries. !?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?! ! Niels Mayer ..... mayer@eit.com ! http://www.eit.com/~mayer ! ! Enterprise Integration Technologies ! http://www.eit.com/software/wwweasel! ! 800 El Camino Real, Menlo Park, CA. ! http://www.eit.com/software/winterp ! !?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?!?! From majordom@umnstat.stat.umn.edu Wed Mar 27 15:03:46 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22870; Wed, 27 Mar 1996 15:03:46 -0500 Return-Path: Received: from gateway.ets.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22865; Wed, 27 Mar 1996 14:03:42 -0600 Received: from research.ets.org by gateway.ets.org with SMTP id AA04168 (5.65c/IDA-1.4.4 for ); Wed, 27 Mar 1996 14:58:21 -0500 Received: from comet.research.ets.org by research.ets.org with SMTP id AA06894 (5.67b/IDA-1.4.3 for sdm7g@Virginia.EDU); Wed, 27 Mar 1996 14:58:20 -0500 Received: by comet.research.ets.org (SMI-8.6/SMI-SVR4) id OAA07263; Wed, 27 Mar 1996 14:58:10 -0500 Date: Wed, 27 Mar 1996 14:58:10 -0500 Message-Id: <199603271958.OAA07263@comet.research.ets.org> From: Russell Almond To: sdm7g@Virginia.EDU Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: (sdm7g@Virginia.EDU) Subject: Re: [xlisp-stat] Any way to catch all messages ? Reply-To: almond@acm.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Is there any way for an object to catch all messages, even and especially those that are undefined for the object ? I did some similar stuff in ElToY. I defined an :update message which would pass a message along to all of its children objects. Thus I would send (send :update ). If had a method for it would execute it, then it would send the update message to every "child" object. The ubdate mechanism was attached to a base class object (el-object) which every ElToY object inherited from so it was more or less universal. Another alternative would be to define a :before or :after method which did the updating on the base class object for your system. (As xlisp doesn't support :before and :after methods, you will need to fake this with call-next-method). Then any message which must be propagated would get defined with that update message attached to your base object. You might want to take a look at the ElToY code. Unfortuantely, it seems to have disappeared from the UW archives. I'll post a new version soon. (Apparently ElToY exercises a bug in version 3.44 of Xlispstat, I'm not sure whether to release a patched version with a workaround or wait until Luke releases the next version of XLISPSTAT with the bug fix). Educational Testing Services Research Statistics Group, 15-T Princeton, NJ 08541 Phone: 609-734-1557 FAX: 609-734-5420 Email: almond@acm.org, ralmond@ets.org http://bayes.stat.washington.edu/almond/almond.html From majordom@umnstat.stat.umn.edu Wed Mar 27 15:41:08 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23354; Wed, 27 Mar 1996 15:41:08 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23348; Wed, 27 Mar 1996 14:40:55 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa28340; 27 Mar 96 15:10 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA11791; Wed, 27 Mar 1996 15:10:51 -0500 Date: Wed, 27 Mar 1996 15:10:51 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: "Niels P. Mayer" Cc: Luke Tierney , udina@upf.es, stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Solution to my Problem with LispStat In-Reply-To: <199603271752.JAA21793@netcom6.netcom.com> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Wed, 27 Mar 1996, Niels P. Mayer wrote: > Yipes!! This stuff has been dealt with effectively way back in the late > 80's in this archaic piece of technology known as the X window system > and the X toolkit. Hasn't anybody built an abstraction to give a > "nice" interface (perhaps w/ {X,Xt,}-like API) to events and > windows that is multiplatform? I don't think it's that trivially solved in X. The folks working on Python have been running into a few similar glitches trying to get ILU and Tk to work together in one image - they both expect to have their own main event loops, and that and the fact that they both make use of unix signals seem to cause some strange interference. One of the goals of a higher level toolkit like Tk or the graphics classes in xlisp-stat is to free the programmer from the event loop. But, it's making the event loop implicit that seems to cause some of these glicthes. ( In Forrest's case, I think that the fact that he is NOT explicitly dispatching from the event loop is just what makes the possibility that his actions may be interruptable by an externally generated :redraw. ) Multi-threading, would in fact, make most of these cases trivially solvable. ( I'm not sure that it wouldn't produce a different set of problems, though. ) > What are people's experiences in deploying xlisp-stat for > multi-platform applications -- do you have to do a lot of mucking about > at the events level to make things work just right? Typically, you don't much about at the event level. xlisp-stat, Tk, and some of the C++ graphics frameworks, like HyperCard, typically try to hide the event loop ( or at least minimize your concern about it ) and you define a bunch of ON-EVENT-DO type functions or classes ( In xlisp-stat's classes, they are things like :do-click, :redraw, :do-key ) and dispatching to those is what's done under the covers. In my experience, xlisp-stat works pretty well cross-platform, but [1] it presents higher level abstractions, so you don't run into many of the cross platform differences. [2] it just doesn't provide interfaces to some of the other desirable but less portable features ( drag-and-drop or list-boxes, for example. ) > And what about > WxWindows and Mr. Ed? When I do custom graphics apps with > WINTERP/Xtango (on Sun, Irix, Linux, & HP platforms), the ability for > multiple things to be going on at the same time is typically the least > of my worries. ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Wed Mar 27 18:22:37 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24851; Wed, 27 Mar 1996 18:22:37 -0500 Return-Path: Received: from solution.maths.unsw.EDU.AU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24846; Wed, 27 Mar 1996 17:22:31 -0600 Received: (frances@localhost) by solution.maths.unsw.EDU.AU (8.6.9/8.6.9.NET-BJB) id KAA04789 for stat-lisp-news@umnstat.stat.umn.edu; Thu, 28 Mar 1996 10:22:15 +1000 Date: Thu, 28 Mar 1996 10:22:15 +1000 From: frances@maths.unsw.EDU.AU Message-Id: <199603280022.KAA04789@solution.maths.unsw.EDU.AU> To: stat-lisp-news@umnstat.stat.umn.edu Content-Length: 676 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk To: stat-lisp-news@umnstat.stat.umn.edu Hi, Does anyone know if the UNIX version of Xlispstat works on DEC ALPHASERVER 8200 running DIGITAL UNIX (otherwise known as OFS/1) ? Thanks. Frances Fang ---------------------------------------------------------------- | Frances Fang email: frances@maths.unsw.edu.au | | Department of Statistics phone: +612 3852978 | | School of Mathematics | | University of New South Wales | | Australia | ----------------------------------------------------------------- From majordom@nokomis.stat.umn.edu Thu Mar 28 07:09:15 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA15243; Thu, 28 Mar 1996 07:09:15 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA15228; Thu, 28 Mar 1996 06:05:24 -0600 From: Luke Tierney Message-Id: <9603281205.AA15228@nokomis.stat.umn.edu> Subject: Re: Solution to my Problem with LispStat To: sdm7g@Virginia.EDU (Steven D. Majewski) Date: Thu, 28 Mar 1996 06:05:23 -0600 (CST) Cc: mayer@netcom.com, luke@nokomis.stat.umn.edu, udina@upf.es, stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Steven D. Majewski" at Mar 27, 96 03:10:51 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 3957 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > > On Wed, 27 Mar 1996, Niels P. Mayer wrote: > > > Yipes!! This stuff has been dealt with effectively way back in the late > > 80's in this archaic piece of technology known as the X window system > > and the X toolkit. Hasn't anybody built an abstraction to give a > > "nice" interface (perhaps w/ {X,Xt,}-like API) to events and > > windows that is multiplatform? > > I don't think it's that trivially solved in X. > > The folks working on Python have been running into a few similar > glitches trying to get ILU and Tk to work together in one image - > they both expect to have their own main event loops, and that > and the fact that they both make use of unix signals seem to cause > some strange interference. > > One of the goals of a higher level toolkit like Tk or the graphics > classes in xlisp-stat is to free the programmer from the event loop. > But, it's making the event loop implicit that seems to cause some of > these glicthes. ( In Forrest's case, I think that the fact that he > is NOT explicitly dispatching from the event loop is just what makes > the possibility that his actions may be interruptable by an externally > generated :redraw. ) > > Multi-threading, would in fact, make most of these cases trivially > solvable. ( I'm not sure that it wouldn't produce a different set > of problems, though. ) It does raise it's own problems, but it should provide a clear way of dealing with them and be ignorable if it isn't needed. The problem Forrest came across does not arise on UNIX but it does on the Mac and Windows because on these platforms multitasking with other application requires cooperation, which is accomplished by periodically checking for and processing certain events during long running calculations (in the oscheck routine). This checking is where user interrupts are detected on these platforms. On the Mac at least it appears to also be necessary to process window update events in order to get switching in and out of applications to work right; on Windows repaint events don't go directly through the event loop so any event checking ends up processing them (at least as i recall). This more or less forces :redraw events to get handled at these times. The net effect is that code executed from the listener window and event processing is to a certain extent preemptively multitasked on these platforms. All this isn't necessary on UNIX since UNIX does preemptive multitasking of processes on its own. Windows 95 and NT do also, so the Win32 version running on these platforms could dispense with the event processing. The need for multitasking of some sort arises in xlispstat because it it is common to run moderately long calculations, and when you start up such a calculation it is nice to be able to do something like go read your mail. On preemptively multitasking systems, like UNIX or Windows NT, you don't need to do a thing to your program to make this possible. On the Macintosh and Win16 you do. So one way or another the ability to switch from a long running xlispstat calculation to another program exists on all platforms. What doesn't exist, and would be nice, is the ability to switch within xlispstat from a long running calculation to some other activities, to background a long simulation while you look at some plots for example. Adding preemptive threads would make this possible. (You can do this in a heavy-weight fashion on UNIX and Win32 by running two copies of xlispstat, but that is usually not an optimal solution. Even that is very difficult on the Macintosh.) As Steve points out, it would raise problems (like making sure you don't clobber some of the data the long-running application is using). but it would also open up new opportunities, such as taking advantage of multiple processors. It's not clear if it will be worth it in the end, but finding out things like that is part of research is this area of statistical computing. luke From majordom@umnstat.stat.umn.edu Thu Mar 28 12:19:22 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA09855; Thu, 28 Mar 1996 12:19:22 -0500 Return-Path: Received: from uhura.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA09850; Thu, 28 Mar 1996 11:19:18 -0600 Received: from localhost by uhura.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id JAA20111; Thu, 28 Mar 1996 09:19:16 -0800 (PST) Date: Thu, 28 Mar 1996 09:19:16 -0800 (PST) From: Thomas Lumley X-Sender: thomas@uhura To: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Solution to my Problem with LispStat In-Reply-To: <9603281205.AA15228@nokomis.stat.umn.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Thu, 28 Mar 1996, Luke Tierney wrote: > Windows repaint events don't go directly through the event loop so any > event checking ends up processing them (at least as i recall). This > more or less forces :redraw events to get handled at these times. Yes, Windows requires paint messages to be processed. Most messages can be ignored or discarded by a program that would rather do something else but Windows will seize up if the .Paint method for a window is not called when it want the window redrawn. You *could* have a dummy .Paint method that didn't do any drawing but the synchronisation issues would still be there. thomas lumley PhD student, Biostatistics, Uni. Washington From majordom@umnstat.stat.umn.edu Thu Mar 28 13:43:02 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10458; Thu, 28 Mar 1996 13:43:02 -0500 Return-Path: Received: from uhura.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10453; Thu, 28 Mar 1996 12:42:59 -0600 Received: from localhost by uhura.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id KAA20406; Thu, 28 Mar 1996 10:42:47 -0800 (PST) Date: Thu, 28 Mar 1996 10:42:46 -0800 (PST) From: Thomas Lumley X-Sender: thomas@uhura To: stat-lisp-news@umnstat.stat.umn.edu Subject: ?Bug in 3.47(beta) Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Version 3.47 (running on Solaris 4.1.3) appears not to invert 1x1 matrices. For example, the command (inverse #2A((0.017))) produces the message Error: incompatible with access indices - #(0.0) Happened in: # I noticed this because a program that worked on a machine running 3.44 suddenly stopped working. lu-decomp and lu-solve don't have this problem. thomas lumley Biostatistics, Uni of Washington From majordom@umnstat.stat.umn.edu Fri Mar 29 01:43:47 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15713; Fri, 29 Mar 1996 01:43:47 -0500 Return-Path: Received: from mail.comnet.mt by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15708; Fri, 29 Mar 1996 00:43:40 -0600 Received: by michelle.magnet.mt id <21763>; Fri, 29 Mar 1996 07:55:15 +0100 X400-Received: by mta BLB49H43 in /PRMD=EGM/ADMD=MALTAGATE/C=MT/ ; Relayed ; Fri, 29 Mar 96 07:41:34 +0100 X400-Received: by mta MOG56H43 in /PRMD=EGM/ADMD=MALTAGATE/C=MT/ ; Relayed ; Fri, 29 Mar 96 07:43:15 +0100 X400-Received: by /PRMD=EGM/ADMD=MALTAGATE/C=MT/ ; Relayed ; Fri, 29 Mar 96 07:43:00 +0100 Date: Fri, 29 Mar 1996 07:43:00 +0100 X400-Mts-Identifier: [/PRMD=EGM/ADMD=MALTAGATE/C=MT/;G700216AE30000000101140D3CC3042E] X400-Originator: "Andrew Amato-Gauci" X400-Recipients: stat-lisp-news@umnstat.stat.umn.edu X400-Content-Type: P2-1984 (2) Original-Encoded-Information-Types: IA5-Text Priority: normal From: "Andrew Amato-Gauci" To: stat-lisp-news@umnstat.stat.umn.edu Importance: normal Subject: Message-Id: <96Mar29.075515gmt+0100.21763@michelle.magnet.mt> Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Can you possibly send me details on how I may obtain a version of the statistics software? Andrew From majordom@umnstat.stat.umn.edu Fri Mar 29 12:36:04 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27537; Fri, 29 Mar 1996 12:36:04 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27532; Fri, 29 Mar 1996 11:35:51 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa00698; 29 Mar 96 12:25 EST Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA16545; Fri, 29 Mar 1996 12:25:35 -0500 Date: Fri, 29 Mar 1996 12:25:34 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: Color symbols and packages Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk A while ago, I wrote a parse-color function for the Mac, so I could use the same color names on Mac as on Unix/X11 xlisp-stat. Recently, I added some more color management routines, and, because I had to redefine some procedures like ADD-COLOR, I reorganized the code to put it all into it's own package. It worked well that way, except that now I've been running into problems with the fact that the symbols returned by (color-symbols) are sometimes interned in different packages, depending (I suppose) on where and when they were defined/added. 'BLACK, 'BLUE, etc. all seem to be external symbols from package XLISP, which are inherited in package USER. However, some of the color symbols I added end up being interned in another package - for example 'COLORS::PURPLE. This side effect has broken some of my other code that tries to select or restrict various sets of color. I can probably fix those functions to take that possibility into account, but I wonder here about what is the "Right Thing" - i.e. if colors are going to be passed around as symbols, shouldn't they be keyword symbols ? Either that, or in their own package - the rationale being that they need to be unique. Anyone have any ideas on the best ( but backwardly compatible ) way to handle this ? ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Fri Apr 5 19:27:41 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00768; Fri, 5 Apr 1996 19:27:41 -0400 Return-Path: Received: from mathfs.math.montana.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00763; Fri, 5 Apr 1996 17:27:36 -0600 Received: (from jimrc@localhost) by mathfs.math.montana.edu (8.7.5/8.7.3) id QAA18707; Fri, 5 Apr 1996 16:27:35 -0700 Date: Fri, 5 Apr 1996 16:27:34 -0700 (MST) From: Jim Robison-Cox To: Xlisp List Subject: Installing on Sun Ultra-Sparc Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'm having trouble getting Xlisp-Stat to run on a SUN Ultra Sparc 170 and wonder if anyone has had similar troubles and can set me straight. I am getting a linking error when make tries to use gcc to link a long list of *.o files in: gcc -o X11WINDOWSxlisp xsdynload.o xssystem.o unixprim.o .... etc. -L/opt/X11R6.1/lib -lX11 -ldl -lm The error message is: Undefined first referenced symbol in file socket /opt/X11R6.1/lib/libX11.so getpeername /opt/X11R6.1/lib/libX11.so xdrmem_create /opt/X11R6.1/lib/libX11.so gethostbyname_r /opt/X11R6.1/lib/libX11.so setsockopt /opt/X11R6.1/lib/libX11.so authdes_create /opt/X11R6.1/lib/libX11.so inet_addr /opt/X11R6.1/lib/libX11.so shutdown /opt/X11R6.1/lib/libX11.so getsockname /opt/X11R6.1/lib/libX11.so getservbyname_r /opt/X11R6.1/lib/libX11.so connect /opt/X11R6.1/lib/libX11.so ld: fatal: Symbol referencing errors. No output written to X11WINDOWSxlisp So I guess I'm missing some include file. I don't know enough unix to find the file which defines these variables, so can someone 1) tell me what file defines these, or 2) tell me how to search libraries to find the definitions (Yuck) Thanks. Jim Robison-Cox e-mail: jimrc@math.montana.edu Department of Mathematical Sciences 2-214 Wilson Hall Montana State University Bozeman, MT 59717 Home: 406 587-4977 Work: 406 994-5340 From majordom@umnstat.stat.umn.edu Mon Apr 8 09:37:02 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11990; Mon, 8 Apr 1996 09:37:02 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11985; Mon, 8 Apr 1996 08:36:57 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id JAA02114; Mon, 8 Apr 1996 09:36:44 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 8 Apr 1996 09:36:38 -0500 To: edstat-l@jse.stat.ncsu.edu, stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: ViSta, The Visual Statistics System Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Documentation for ViSta, the Visual Statistics System, is now available at no cost in the form of Adobe Acrobat files, in addition to postscript files. The documentation includes cover, contents, introduction, tutorial and references. The documentation is the same for all operating systems. It can be downloaded from the ViSta web site at http://forrest.psych.unc.edu/research/ViSta. For those of you who have not seen previous announcements about ViSta, a previously posted announcement is attached to the end of this message. We strongly recommend getting the documentation in the form of Adobe Acrobat files because these files > are small > can be viewed on your computer screen > can be printed on any printer, not just postscript printers > are the same no matter what operating system you are using > are free You must have Adobe's Acrobat Reader to use the Acrobat files. It is available for free from Adobe at . Note: We do not recommend getting the beta version of the next release of Acrobat Reader (called "Adobe Acrobat Amber") because in our judgment it is not yet reliable enough. You can also get the documentation in the form of postscript files from the ViSta web site. However, we recommend Adobe Acrobat files rather than postscript files because postscript files are much larger, cannot be seen on screen and can only be printed on postscript printers. For those of you who have not seen previous announcements about ViSta, the remainder of this message repeats previous announcements. ====================================================================== The remainder of this message summarizes earlier messages about ViSta. ====================================================================== Announcing a New Release of ViSta - The Visual Statistics System Version 4.22, January 8th, 1996 Copyright (c) 1991-96 by Forrest W. Young If you are using releases prior to 4.21 on a Macintosh or PowerMac, I strongly recommend you obtain this new release, which is much more robust due to fixes of two major bugs. ======================================= What is ViSta? ======================================= ViSta, is a Visual Statistics system that > Performs univariate and multivariate data analysis and visualization > Provides visual maps for summarizing and guiding analysis sessions > Includes a data analysis language and a programming language > Is open, extensible, and freely redistributable PLATFORMS: ViSta is available for > Microsoft Windows 3.1 (Windows 95 in preperation) > Macintosh and PowerMac microcomputers > A Unix version is in preparation AUDIENCE: ViSta is useful for > UNDERGRADUATES: Teaching introductory statistics to undergraduates > GRADUATES: Teaching multivariate statistics, dynamic statistical graphics >and computational statistics to graduate students > RESEARCHERS: Research and development in dynamic statistical graphics and >in computational statistics ViSta is designed for an audience of users having a very wide range of data analysis sophistication, ranging from novice to expert. ViSta provides seamlessly integrated data analysis environments specifically tailored to the user's level of expertise. Visual guidance is available for students and novices, and graphical tools are available for teachers and other experts to create guidance for these novices. A structured graphical user interface is available for competent users, and a command line interface is available for sophisticated users. The complete Lisp-Stat (Tierney, 1990) programming environment is available to researchers, programmers and graduate students who wish to extend ViSta's capabilities. ViSta is "freely redistributable". This means that it is not in the public domain, but that it may be freely copied and redistributed, with certain restrictions. See the COPYING and COPYVSTA files included in the distribution for details. ======================================= ViSta's main features ======================================= > WORKMAPS Maps that visualize the overall structure of an on-going data analysis session. These maps are created by ViSta as the data analysis session progresses. They aid novice and competent analysts in remembering their analysis session, and permit the analyst to return to earlier steps of the analysis. > GUIDEMAPS Maps that visually guide students and other novice data analysts through their data analysis sessions. These maps are created by experts using ViSta's Authoring tools. > SPREADPLOTS Multi-window dynamic statistical graphics that visualize data structure and analysis results > MODEL RE-VISION Dynamic interactive graphical tools that permit the user to visually re-estimate model parameters > ViDAL ViSta's Data Analysis Language. ViDAL is an extension of the underlying Lisp-Stat language that can be used by sophisticated data analysts for data analysis and by programmers to enhance ViSta's capabilities. > DATASHEET A graphical datasheet for displaying and editing data. > AUTHORING TOOLS Visual tools for authoring new GuideMaps. ======================================= Data Analysis Capabilities ======================================= ViSta's data analysis capabilities include: > DYNAMIC GRAPHICS Histograms, Boxplots, Diamond Plots, Dotplots, Scatterplots, Biplots, Spinplots, Scatterplot Matrices. Plots support brushing and labeling, and are dynamically linked. > DESCRIPTIVE STATISTICS Means, Standard Deviations, Variances, Ranges, Quartiles, Medians, Correlations, Covariances, Distances >UNIVARIATE ANALYSIS T- and Z-tests and confidence intervals for single sample, paired samples and two independent samples data, with Wilcoxon Signed-Rank and Mann-Whitney tests in appropriate situations. > ANOVA Univariate analysis of variance of orthogonal one- or multi-way data. Model may or may not include two-way interactions. The model visualization is a spreadplot comprised of boxplot, diamond plot, quantile plot, quantile-quantile plot and effects plot. > REGRESSION Simple and multiple OLS regression analysis (multivariate, robust and monotonic regression in preparation). The model visualization is a spreadplot comprised of a biplot, spinplot, histogram and scatterplot-matrix. > PRINCIPAL COMPONENTS Analysis of correlations or covariances. The model visualization is a spreadplot comprised of a biplot, spin-plot, scree-plot and scatterplot-matrix. > MULTIDIMENSIONAL SCALING Analysis of one or more symmetric or asymmetric matrices. The model visualization is a spreadplot comprised of a scatterplot, spin-plot, scree-plot and scatterplot-matrix. The spreadplot supports graphical re-estimation of model parameters. > CORRESPONDENCE ANALYSIS Analysis of two-way contingency tables. The model visualization is a spreadplot comprised of a biplot, spinplot, residuals plot and scree-plot. The spreadplot supports graphical re-estimation of model parameters. ======================================= Hardware Requirements ======================================= MACINTOSH: There are two versions of ViSta for the Macintosh, one for the PowerMac and one for 680xx Macintoshes. For 680xx Macs, we recommend a 68040 processor (at the minimum, a 68030 + 68881 math coprocessor). Both versions require System 7 or later, 3 MB of free hard disk space and at least 6 MB of RAM (8 for the PowerMac). WINDOWS: There are two versions of ViSta available for Microsoft Windows 3.1, one for WIndows 3.1 with the WIN32S 32-bit extension, and one without. Both require 3 MB of free hard disk space and 8 MB of RAM. The Windows 95 version is in preparation. Versions for Windows NT and OS/2 are not contemplated. If you wish, you can try the 32-bit version of ViSta with 95 or NT or the 16-bit version with OS/2. While I have not tried ViSta under these operating systems, some people report positive results. Let me know your experience. MONITORS: We recommend at least a 14", 8-bit (256 color) screen, but 17" is preferred. For 12-, 24- or 32-bit screens, a larger memory partition is required (add 1/2 MB for 12-bit, 1 MB for 24- or 32-bit). ======================================= ViSta is a work-in-progress ======================================= ViSta is a work-in-progress. Some features are incomplete. While quality assurance testing has been done, this is a beta-release system that is not fully mature. At the current stage of development, ViSta is viable as a supplemental statistical software system used for extra credit exercises in introductory statistics, and as the focus of graduate courses on research in computational and graphical methods in statistics. It may be premature to use ViSta in undergraduate statistics classes that need a complete and robust statistic analysis system. Currently, the Delete Data, Delete Model and Simulate Data menu items are not implemented; the ANOVA visualization is incomplete; MULREG does not do multivariate tests (it does many separate multiple regressions); transformations are not available with matrix data; model interpretation is unimplemented; and the Help and GuideMap systems are incomplete. A univariate multiple regression module (which complements the current multivariate multiple regression module with more extensive visualizations and with the choice of robust or monotonic regression) is being written but is not yet available. Multivariate tests, Redundancy Analysis and Canonical Analysis are being added to the multivariate regression module. ======================================= Documentation & References ======================================= The following two monographs describe how to use ViSta: "ViSta, The Visual Statistics System", is available from the server mentioned above. "ViSta Corresp", a description of the Correspondence Analysis procedure in ViSta, is available for $5.00 from Forrest Young at the address below. The following paper has been recently published: Young, F.W. & Lubinsky, D.J. "Guiding Data Analysts with Visual Statistical Strategies". J. Computational and Graphical Statistics, Vol 4(4), December, 1995, pp. 229-260. In addition, the following paper is available for $5.00 from Forrest Young at the address below. Young, F.W., et al. "Visual Statistical Analysis - Papers presented at the Special Contributed Session on Visual Statistical Analyais at Interface '94. Psychometric Research Memorandum 94-2, July, 1994. Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Thu Apr 11 07:43:27 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06343; Thu, 11 Apr 1996 07:43:27 -0400 Return-Path: Received: from maila.surrey.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06338; Thu, 11 Apr 1996 06:43:07 -0500 Received: from central.surrey.ac.uk by maila.surrey.ac.uk with SMTP (PP); Thu, 11 Apr 1996 12:35:48 +0100 Received: from annex1-1.surrey.ac.uk by central.surrey.ac.uk with SMTP (1.37.109.8/16.2) id AA18649; Tue, 9 Apr 1996 21:47:07 +0100 X-Sender: scs1ng@boris.soc.surrey.ac.uk Message-Id: Date: Tue, 9 Apr 1996 21:48:02 +0000 To: stat-lisp-news@umnstat.stat.umn.edu From: Nigel Gilbert Subject: Get-lambda-expression and equalp Cc: gng@soc.surrey.ac.uk Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk 1. The function get-lambda-expression seems to have disappeared from XLISP-STAT 2.1 Release 3.44 (Beta). It is used in function savefun, defined in common.lsp, but this of course fails: > (savefun xxx) Error: The function XLISP::GET-LAMBDA-EXPRESSION is unbound. Happened in: # Is there some work-around? 2. The function equalp should be capable of comparing two structures. However, the code for equalp in common.lsp does not do this. How does one compare two arbitrary structures? You would need first to compare their types with #'type-of. If these are eql, then you could compare each field of the structures in turn for equality. However, to do this, one needs a way of obtaining a list of the field names of a structure. I can't see a way of getting that. Thanks. ____________________________________________________________________ Prof G. Nigel Gilbert, Department of Sociology, University of Surrey Guildford GU2 5XH, UK. Tel: +44 1483 259173 Fax: +44 1483 306290 From majordom@umnstat.stat.umn.edu Thu Apr 11 09:10:20 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06583; Thu, 11 Apr 1996 09:10:20 -0400 Return-Path: Received: from halley.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06578; Thu, 11 Apr 1996 08:10:11 -0500 Received: from libiya.upf.es by halley.upf.es with SMTP (1.38.193.5/16.2) id AA26093; Thu, 11 Apr 1996 15:10:27 +0100 Received: by libiya.upf.es (1.38.193.4/15.6) id AA11463; Thu, 11 Apr 1996 15:09:57 +0200 Date: Thu, 11 Apr 1996 15:09:56 +0200 (METDST) From: Frederic Udina To: lisp stat list Subject: Re: Get-lambda-expression and equalp In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Tue, 9 Apr 1996, Nigel Gilbert wrote: > 1. The function get-lambda-expression seems to have disappeared from > XLISP-STAT 2.1 Release 3.44 (Beta). It is used in function savefun, > defined in common.lsp, but this of course fails: > > (savefun xxx) > Error: The function XLISP::GET-LAMBDA-EXPRESSION is unbound. > Happened in: # > It has been moved to function-lambda-expression, I think that following CL especification: > (defun ppp (x) (list x x x)) PPP > (function-lambda-expression #'ppp) (LAMBDA (X) (LIST X X X)) NIL PPP > Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Balmes 132 | appleLink: spa0172 08008 Barcelona | Compuserve: 100555.603 SPAIN | W W Web: http://libiya.upf.es/ ___________________________________/_\_________________________________ From majordom@umnstat.stat.umn.edu Mon Apr 15 05:41:03 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10455; Mon, 15 Apr 1996 05:41:03 -0400 Return-Path: Received: from mailb.surrey.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10449; Mon, 15 Apr 1996 04:40:47 -0500 Received: from soc.surrey.ac.uk (actually host boris-gate.soc.surrey.ac.uk) by mailb.surrey.ac.uk with SMTP (PP); Mon, 15 Apr 1996 10:39:17 +0100 Received: from socmac32 by boris.soc.surrey.ac.uk via Ethernet with SMTP id aa00102; 15 Apr 96 10:40 BST X-Sender: scs1ng@boris.soc.surrey.ac.uk Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 15 Apr 1996 10:40:27 +0100 To: stat-lisp-news@umnstat.stat.umn.edu From: Nigel.Gilbert@soc.surrey.ac.uk (Nigel Gilbert) Subject: Answers: Get-lambda-expression and equalp Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I asked the list two questions. Many thanks for the replies, which I summarise below. >1. The function get-lambda-expression seems to have disappeared from >XLISP-STAT 2.1 Release 3.44 (Beta). It is used in function savefun, >defined in common.lsp, but this of course fails: >> (savefun xxx) >Error: The function XLISP::GET-LAMBDA-EXPRESSION is unbound. >Happened in: # > >Is there some work-around? The function's name has changed to FUNCTION-LAMBDA-EXPRESSION, to be in accord with CLtL. [Note: the next release of common.lsp should be changed to suit]. > >2. The function equalp should be capable of comparing two structures. >However, the code for equalp in common.lsp does not do this. How does one >compare two arbitrary structures? You would need first to compare their >types with #'type-of. If these are eql, then you could compare each field >of the structures in turn for equality. However, to do this, one needs a >way of obtaining a list of the field names of a structure. I can't see a >way of getting that. The name of the constructor function and the field names and accessor functions of a structure are stored on the property list of the structure's type. Thanks to: Franz Quehenberger, Frederic Udina, and LongA for their help. _________________________________________________________________________ Prof G. Nigel Gilbert, Dept of Sociology, University of Surrey, Guildford, GU2 5XH, United Kingdom. Tel: +44 (0)1483 259173 Fax: +44 (0)1483 306290 From majordom@umnstat.stat.umn.edu Mon Apr 15 18:09:08 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14746; Mon, 15 Apr 1996 18:09:08 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14722; Mon, 15 Apr 1996 17:09:02 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA07169; Mon, 15 Apr 96 15:08:52 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA03701; Mon, 15 Apr 1996 15:08:49 -0700 From: Jan Deleeuw Message-Id: <9604152208.AA03701@galton.math.ucla.edu> Subject: XLisp-Stat code for shrinkage/ridge regression... (fwd) To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Mon, 15 Apr 1996 15:08:49 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I added this to the archive in the statistics/regression section. --- Jan ================================================================== XLisp-Stat code for shrinkage/ridge regression... Jan, I would very much appreciate it if you would consider adding the following link to your xlisp-stat/code/statistics/regression html file:
shrinkage
An introduction to maximum likelihood methods of shrinkage/ridge regression, plus numerous code/example archives for download ...by Bob Obenchain, softRX freeware. The corresponding XLisp-Stat code, rxridge.lsp, represents enhancements and changes (by me) to code originally developed by Bernhard Walter (see reference list) in 1994. Anyway, I hope you will agree with me that the code implements CONSIDERABLE functionality ...thanks, of course, to Luke Tierney! -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Thu Apr 18 16:47:22 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06227; Thu, 18 Apr 1996 16:47:22 -0400 Return-Path: Received: from bubba.NMSU.Edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06220; Thu, 18 Apr 1996 15:47:16 -0500 Received: from NMSU.Edu by bubba.NMSU.Edu (8.6.10/NMSU) id OAA05460; Thu, 18 Apr 1996 14:49:36 -0600 Received: from ccserver by NMSU.Edu (8.6.10/NMSU-1.18) id OAA12467; Thu, 18 Apr 1996 14:47:11 -0600 Date: Thu, 18 Apr 1996 14:47:11 -0600 Message-Id: <199604182047.OAA12467@NMSU.Edu> Received: from [128.123.30.78] by ccserver (4.1/NMSU) id AA17108; Thu, 18 Apr 96 14:47:09 MDT X-Sender: ddaniel@dante.nmsu.edu Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: stat-lisp-news@umnstat.stat.umn.edu From: ddaniel@nmsu.edu (David L. Daniel) Subject: log-scale plotting? Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'm wanting to plot data on a graph having one or more log-scale axes. Does anyone know of code that will do this? Thanks, -David David Daniel Dept. of Experimental Statistics New Mexico State University ddaniel@nmsu.edu ################################################################### Degenerate distributions - They're everywhere! They're everywhere! ################################################################### From majordom@umnstat.stat.umn.edu Mon Apr 22 18:12:41 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14307; Mon, 22 Apr 1996 18:12:41 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14298; Mon, 22 Apr 1996 17:12:35 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa11115; 22 Apr 96 18:12 EDT Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA24552; Mon, 22 Apr 1996 18:12:09 -0400 Date: Mon, 22 Apr 1996 18:12:09 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Luke Tierney Cc: Xlisp List Subject: [1] Bug or feature ? [2] generalized digital filters Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Up until now, I have been using some very unoptimized Lisp code for doing digital filtering of spectra. The original fortran code I'm translating and generalizing from uses a "top-hat" filter: +1 +---------------+ | | | | 0 --+ | | +-- | | | | | | | | -1 +-------+ +-------+ ( This is a bandpass filter that removes both hi-frequency and low frequency conponents. ) That code is currently the slowest non-graphics, non-io, bottleneck. I've been experimenting with some other ways to do it. [1] I've been poking around in the source for kernel-smooth to see how that is done in xlisp-stat. I discovered that the choice of kernels is hard coded in kernel.c. I added an additional case to C function kernel(): case 'H': /* sdm7g: TopHat filter */ /* w = 1.0 ; */ z = 2.0 * z ; if (( -0.5 < z) && (z < 0.5 )) k = 1.0 ; else if (( -1.0 < z) && (z < 1.0)) k = -1.0; else k = 0.0 ; break; [ A case also has to be added to the xlisp function wrapper in linalg.c. ] Which initially seemed to work, however I found that I had picked a lucky set of test parameters, because most of the time it didn't work as I expected, but produced all zero results. In C function kernel_smooth(), I discovered the line: if (y != nil) ys[i] = (wsum <= 0.0) ? ysum / wsum : 0.0; which I changed to: if (y != nil) ys[i] = (wsum != 0.0) ? ysum / wsum : 0.0; which seems to fix the problem. There are no comments in this code, so I'm not sure what is intended. Is this a 'bug' or a 'feature' ? [2] The above mods aren't really general enough for what I may need: kernel-smooth takes a single width parameter along with the type code, and I may need to control the widths of both the positive inner and the negative outer filter lobes independently. I might also want to try other filter kernel functions - sinc() for example. * Does anyone else have a need for more generalized digital filtering ? * Is there something else already in lisp-stat for this that I missed ? * Is there a real performance advantage to doing it on the C language level rather than in Lisp ? * Anyone have any ideas about the best interface for a more generalized routine ? It might be nice to be able to pass a lisp defined filter kernel function, but would this kill any speed advantage of the C code. ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Tue Apr 23 03:43:41 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27567; Tue, 23 Apr 1996 03:43:41 -0400 Return-Path: Received: from halley.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27561; Tue, 23 Apr 1996 02:42:49 -0500 Received: from libiya.upf.es by halley.upf.es with SMTP (1.38.193.5/16.2) id AA13696; Tue, 23 Apr 1996 09:43:25 +0100 Received: by libiya.upf.es (1.38.193.4/15.6) id AA04572; Tue, 23 Apr 1996 09:42:43 +0200 Date: Tue, 23 Apr 1996 09:42:43 +0200 (METDST) From: Frederic Udina To: Xlisp List Subject: Re: [1] Bug or feature ? [2] generalized digital filters In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Mon, 22 Apr 1996, Steven D. Majewski wrote: > > Up until now, I have been using some very unoptimized Lisp code > for doing digital filtering of spectra. The original fortran > code I'm translating and generalizing from uses a "top-hat" > filter: > > +1 +---------------+ > | | > | | > 0 --+ | | +-- > | | | | > | | | | > -1 +-------+ +-------+ > > ... > > In C function kernel_smooth(), I discovered the line: > > if (y != nil) ys[i] = (wsum <= 0.0) ? ysum / wsum : 0.0; > > which I changed to: > > if (y != nil) ys[i] = (wsum != 0.0) ? ysum / wsum : 0.0; > > which seems to fix the problem. > I guess that kernel-smooth is intended for use with non-negative weigth functions only. > [2] The above mods aren't really general enough for what I may need: > * Does anyone else have a need for more generalized digital filtering ? > * Is there something else already in lisp-stat for this that I missed ? > * Is there a real performance advantage to doing it on the C language > level rather than in Lisp ? > * Anyone have any ideas about the best interface for a more generalized > routine ? It might be nice to be able to pass a lisp defined filter > kernel function, but would this kill any speed advantage of the C > code. I would suggest you using fourier transforms if you want flexible and fast filtering. I do that in my kde code entirely in lisp (together with some rounding, of course) In my opinion, C code for xlisp-stat must be good enough and interesting enough to convince Luke to include it in the 'official' distribution. Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Balmes 132 | appleLink: spa0172 08008 Barcelona | Compuserve: 100555.603 SPAIN | W W Web: http://libiya.upf.es/ ___________________________________/_\_________________________________ From majordom@umnstat.stat.umn.edu Tue Apr 23 06:35:21 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28077; Tue, 23 Apr 1996 06:35:21 -0400 Return-Path: Received: from ts021p4.pop3a.netvision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28068; Tue, 23 Apr 1996 05:34:38 -0500 Received: (from hjstein@localhost) by blinky.cpaf.com (8.6.11/8.6.9) id NAA01672; Tue, 23 Apr 1996 13:38:03 +0200 Date: Tue, 23 Apr 1996 13:38:03 +0200 From: "Harvey J. Stein" Message-Id: <199604231138.NAA01672@blinky.cpaf.com> To: "Steven D. Majewski" Subject: Re: generalized filters... In-Reply-To: References: Cc: hjstein@blinky.cpaf.com, stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Here're two possibilities. 1. I wrote some smoothing code in lisp which I think is basically what you'd do if you if the lisp code is fast enough. It's simple code for computing the convolution of a set of data points against a kernel. I'm appending it to the end of this message. This code is *very* slow, but easy to use. It also can be modified to be much faster in special cases. For example, if your data is 1 dimensional, and you're using a box shaped kernel, you can sort the data first, and only work with the points within the box. 2. If you're running xlispstat on a platform which supports dynamic linking and loading of object code (running under Unix, say), you could modify the above to work with kernels coded in Fortran (for example). Then, you could code up the kernels in Fortran, link them on the fly into a running interpreter, and let the lisp code use the Fortran kernels. I can send you some modifications to xsdynload.c to make this a little faster and easier, if your interested in this route. It gave me a factor of 10 speedup when I used this approach with a nonlinear regression. Good luck, Dr. Harvey J. Stein Berger Financial Research abel@netvision.net.il ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;; Simple convolution code. ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;; Copyright (c) 1996, Harvey J. Stein ;;; Under GNU GPL ;;; ;;; Usage: ;;; A data point is a list of numbers of length N. Let DP be a list of ;;; data points, FV be a list of function values at the data points, and ;;; F be a function of N variables (to be used as a kernel function). ;;; Let SF be (MAKE-DATA-SMOOTHER (transpose DP) FV F). Then SF is also a ;;; function of N variables. Its value at a point X is (please excuse ;;; the mixed notation): ;;; ;;; (length DP)-1 ;;; SUM FV_i * F(DP_i - X) ;;; i=0 ;;; ------------------------------- ;;; (length DP)-1 ;;; SUM F(DP_i - X) ;;; i=0 ;;; ;;; The routines BOX-KERNEL and BVN-KERNEL generate commonly used ;;; kernel functions. (BOX-KERNEL da db) returns a function which is ;;; 1 -da < x < da and -db < y < db, and 1 outside. (BVN-KERNEL sa sb) ;;; returns a bivariate normal function with standard deviations sa & ;;; sb & rho = 0. ;;; ;;; ;;; So, for example, one could do the following: ;;; ;;; 1. Make a random array of x & y coordinates ranging from -3 to 3. ;;; ;;; (setq x (* 6 (- (uniform-rand 300) .5))) ;;; (setq y (* 6 (- (uniform-rand 300) .5))) ;;; ;;; 2. Let the function values at the points be the function ;;; x^2+y^2 + normally distributed noise. ;;; ;;; (setq FV (+ (mapcar (lambda (x y) (+ (* x x) (* y y))) ;;; x y) ;;; (normal-rand 300))) ;;; ;;; 3. Smooth the data by averaging points in 2x2 squares (i.e. - do a ;;; convolution with the fcn that's 1 for x & y btw -1 & 1, and 0 ;;; elsewhere. ;;; ;;; (setq F (make-data-smoother (list x y) FV (box-kernel 1 1))) ;;; ;;; 4. Plot the results. ;;; ;;; The data (including noise) ;;; (setq p (spin-plot (list x y fv))) ;;; ;;; The original function ;;; (send p :add-function (lambda (x y) (+ (* x x) (* y y))) -3 3 -3 3) ;;; ;;; The smoothed function ;;; (send p :add-function F -3 3 -3 3) ;;; The following 3 functions should be compiled before being used. (defun make-data-smoother (predict response kernel) "Args: (predict response kernel) Returns smoother fcn f of (length predict) args s.t. f(x,y,...) is kernel convoluted with the specified data points." (lambda (&rest l) (let* ((w (apply #'mapcar (lambda (&rest p) (apply kernel (- l p))) predict)) (tw (sum w))) (if (< (abs tw) 1d-30) ; Avoid division by zero... 0.0 (/ (sum (* response w)) tw))))) (defun box-kernel (da db) (let ((mda (- da)) (mdb (- db))) (lambda (x y) (if (and (< mda x da) (< mdb y db)) 1 0)))) (defun bvn-kernel (s1 s2) (let ((scale (/ 1 2 pi s1 s2))) (lambda (x y) (let ((xmu (/ x s1)) (ymu (/ y s2))) (* scale (exp (/ (+ (* xmu xmu) (* ymu ymu)) -2))))))) From majordom@umnstat.stat.umn.edu Tue Apr 23 09:05:54 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28471; Tue, 23 Apr 1996 09:05:54 -0400 Return-Path: Received: from halley.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28466; Tue, 23 Apr 1996 08:05:41 -0500 Received: from libiya.upf.es by halley.upf.es with SMTP (1.38.193.5/16.2) id AA16821; Tue, 23 Apr 1996 15:05:52 +0100 Received: by libiya.upf.es (1.38.193.4/15.6) id AA04784; Tue, 23 Apr 1996 15:05:09 +0200 Date: Tue, 23 Apr 1996 15:05:09 +0200 (METDST) From: Frederic Udina To: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: generalized filters... In-Reply-To: <199604231138.NAA01672@blinky.cpaf.com> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I would suggest Luke to include in xlisp-stat some basic routine (in compiled C, of course) for doing convolution. Most packages have it (Matlab, S+, ...) and this give them a clear advantage in simplicity for implementation of filetering, smoothing, etc... cheers, Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Balmes 132 | appleLink: spa0172 08008 Barcelona | Compuserve: 100555.603 SPAIN | W W Web: http://libiya.upf.es/ ___________________________________/_\_________________________________ From majordom@umnstat.stat.umn.edu Wed Apr 24 17:18:15 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19922; Wed, 24 Apr 1996 17:18:15 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19917; Wed, 24 Apr 1996 16:18:09 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa13632; 24 Apr 96 17:15 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA12852; Wed, 24 Apr 1996 17:15:46 -0400 Date: Wed, 24 Apr 1996 17:15:45 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Cc: Luke Tierney Subject: xlisp-stat: bug in #'sum Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk When I replaced the generic Lisp: #-xlisp-stat (defun sum-of-prod (&rest sss) "Sum-of-prod ( &rest sequences ) reduces to sum of the element wise products of sequences." (reduce #'+ (apply #'map 'vector #'* sss ))) with code that uses xlisp-stat's vectorizing functions: #+xlisp-stat ( defun sum-of-prod ( &rest args ) ( sum ( apply #'* args ))) I uncovered a bug in function sum. The documentation says: > ( help 'sum ) SUM [function-doc] Args: (&rest number-data) Returns the sum of all the elements of its arguments. Returns 0 if there are no arguments. Vector reducing. However: ( sum #() ) #() > ( sum () ) NIL > ( sum ) Error: too few arguments Happened in: # Zero not returned for any of the above cases. [ My particular code relied upon ( reduce #'+ #() ) ==> 0 : sum of product of empty sequences equals zero. ] ( Going back to the generic lisp code for now. ) ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Thu Apr 25 14:59:03 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05154; Thu, 25 Apr 1996 14:59:03 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05149; Thu, 25 Apr 1996 13:58:59 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa03420; 25 Apr 96 14:58 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA16281; Thu, 25 Apr 1996 14:58:19 -0400 Date: Thu, 25 Apr 1996 14:58:18 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: xlisp-stat: object method Gotcha! :-( Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk *** Object methods can not return multiple values! *** Not a particularly horrible limitation, unless it bites you unawares! I happen to have gotten into the habit of using multiple-value returns to return a value and it's associated standard deviation or standard error, but it doesn't work for methods. Obviously, the 'send' dispatching code wasn't designed to handle multiple values. ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@nokomis.stat.umn.edu Fri Apr 26 06:24:35 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA29056; Fri, 26 Apr 1996 06:24:35 -0400 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA29050; Fri, 26 Apr 1996 05:24:31 -0500 From: Luke Tierney Message-Id: <9604261024.AA29050@nokomis.stat.umn.edu> Subject: Re: xlisp-stat: object method Gotcha! :-( To: sdm7g@Virginia.EDU (Steven D. Majewski) Date: Fri, 26 Apr 1996 05:24:31 -0500 (CDT) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Steven D. Majewski" at Apr 25, 96 02:58:18 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 436 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > > > *** Object methods can not return multiple values! *** > I forgot to mark the internal send function as returning multiple values. This will be fixed in the next release. If you are compiling your own version, you can fix it by changing the line { "SEND", S, xmsend }, is osptrs.h with { "SEND", SM, xmsend }, luke From majordom@umnstat.stat.umn.edu Fri Apr 26 08:00:58 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19247; Fri, 26 Apr 1996 08:00:58 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19242; Fri, 26 Apr 1996 07:00:53 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id IAA28456; Fri, 26 Apr 1996 08:00:50 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 26 Apr 1996 07:59:47 -0500 To: Xlisp List From: forrest@unc.edu (Forrest Young) Subject: Re: xlisp-stat: object method Gotcha! :-( Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Which version/release of XLS are we talking about here? >> >> >> >> *** Object methods can not return multiple values! *** >> > >I forgot to mark the internal send function as returning multiple >values. This will be fixed in the next release. If you are compiling >your own version, you can fix it by changing the line > > >{ "SEND", S, xmsend }, > >is osptrs.h with > >{ "SEND", SM, xmsend }, > >luke Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@nokomis.stat.umn.edu Fri Apr 26 08:58:37 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA29239; Fri, 26 Apr 1996 08:58:37 -0400 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA29233; Fri, 26 Apr 1996 07:58:33 -0500 From: Luke Tierney Message-Id: <9604261258.AA29233@nokomis.stat.umn.edu> Subject: Re: xlisp-stat: object method Gotcha! :-( To: forrest@unc.edu (Forrest Young) Date: Fri, 26 Apr 1996 07:58:33 -0500 (CDT) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Forrest Young" at Apr 26, 96 07:59:47 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 269 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > Which version/release of XLS are we talking about here? > > >> > >> > >> > >> *** Object methods can not return multiple values! *** > >> All platforms, any version <= 3.47 -- it's in the common C code and has been ever since multiple values were added. luke From majordom@umnstat.stat.umn.edu Sun Apr 28 10:26:46 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA20678; Sun, 28 Apr 1996 10:26:46 -0400 Return-Path: Received: from mailb.surrey.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA20673; Sun, 28 Apr 1996 09:26:41 -0500 Received: from [131.227.133.17] (actually host annex1-7.surrey.ac.uk) by mailb.surrey.ac.uk with SMTP (PP) with ESMTP; Sun, 28 Apr 1996 15:25:16 +0100 X-Sender: scs1ng@boris.soc.surrey.ac.uk Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sun, 28 Apr 1996 12:08:23 +0000 To: stat-lisp-news@umnstat.stat.umn.edu From: Nigel Gilbert Subject: Files bound to edit windows Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Is it true that it is impossible to obtain the pathname of the file which the user is editing in an edit window (i.e. an instance of edit-window-proto)? It appears that the save file dialog in which the user specifies the file name is buried within the XLisp C code and does not emerge into the Lisp environment. This also means that while you can obtain the title of an ordinary window by sending the message :title to the window object, in the case of edit windows, NIL is always returned. If I am correct, could Luke consider adding some code to edit.c to allow user code to get at the window title and/or file pathname? Why would one want to know, anyway, you ask? Here's one reason: Assuming that you have an edit window open on the code you are developing, to go through the edit-load-test cycle with Xlisp-Stat involves: 1. Make changes to the edit window. 2. Save the edit window (command-S). 3. Find the listener window behind the edit window, bring it to the front 4. Load the file (command L; open-file dialog; locate the file) 5. Run the code ... and repeat indefinitely. Steps 2 - 4 could all be replaced by one menu command (say, "Save and Load") but to write this command would require determining the name of the file being edited. That's what I can't do. ____________________________________________________________________ Prof G. Nigel Gilbert, Department of Sociology, University of Surrey Guildford GU2 5XH, UK. Tel: +44 1483 259173 Fax: +44 1483 306290 From majordom@umnstat.stat.umn.edu Sun Apr 28 17:15:36 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22824; Sun, 28 Apr 1996 17:15:36 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22819; Sun, 28 Apr 1996 16:15:30 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA13313; Sun, 28 Apr 96 14:15:26 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA24030; Sun, 28 Apr 1996 14:15:25 -0700 From: Jan Deleeuw Message-Id: <9604282115.AA24030@galton.math.ucla.edu> Subject: regression page To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Sun, 28 Apr 1996 14:15:25 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The UCLA Xlisp-Stat Archive regression section has been updated somewhat. Its homepage at http://www.stat.ucla.edu/develop/lisp/xlisp/xlisp-stat/code/statistics/regression has been repainted and redecorated, and code is added by Thomas Lumley for Generalized Estimating Equations and by Bob Obenchain for ridge regression. It's a very nice collection now. And quite inexpensive. If you want to add material, feel free to send it to me. If you have a complete and well-documented product, think about submitting to http://www.stat.ucla.edu/journals/jss -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon Apr 29 04:18:22 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05802; Mon, 29 Apr 1996 04:18:22 -0400 Return-Path: Received: from halley.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05797; Mon, 29 Apr 1996 03:18:10 -0500 Received: from libiya.upf.es by halley.upf.es with SMTP (1.37.109.15/16.2) id AA298089527; Mon, 29 Apr 1996 10:18:47 +0100 Received: by libiya.upf.es (1.38.193.4/15.6) id AA07110; Mon, 29 Apr 1996 10:16:19 +0200 Date: Mon, 29 Apr 1996 10:16:19 +0200 (METDST) From: Frederic Udina To: Nigel Gilbert Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Files bound to edit windows In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Sun, 28 Apr 1996, Nigel Gilbert wrote: > Is it true that it is impossible to obtain the pathname of the file which > the user is editing in an edit window (i.e. an instance of > edit-window-proto)? I found this problem some months ago, and Luke confirmed me that the window name and the file name are not accessible. > Assuming that you have an edit window open on the code you are developing, > to go through the edit-load-test cycle with Xlisp-Stat involves: > > 1. Make changes to the edit window. > 2. Save the edit window (command-S). > 3. Find the listener window behind the edit window, bring it to the front > 4. Load the file (command L; open-file dialog; locate the file) > 5. Run the code > ... and repeat indefinitely. > > Steps 2 - 4 could all be replaced by one menu command (say, "Save and > Load") but to write this command would require determining the name of the > file being edited. That's what I can't do. > Actually, you can (send (front-window) :save) and also :eval-selection so the only part you can't do from xlisp-stat is to select all the text in the window! Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Balmes 132 | appleLink: spa0172 08008 Barcelona | Compuserve: 100555.603 SPAIN | W W Web: http://libiya.upf.es/ ___________________________________/_\_________________________________ From majordom@umnstat.stat.umn.edu Mon Apr 29 16:39:18 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11061; Mon, 29 Apr 1996 16:39:18 -0400 Return-Path: Received: from halley.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11056; Mon, 29 Apr 1996 15:38:59 -0500 Received: from libiya.upf.es by halley.upf.es with SMTP (1.37.109.15/16.2) id AA145452299; Mon, 29 Apr 1996 19:24:59 +0100 Received: by libiya.upf.es (1.38.193.4/15.6) id AA07350; Mon, 29 Apr 1996 19:23:44 +0200 Date: Mon, 29 Apr 1996 19:23:44 +0200 (METDST) From: Frederic Udina To: lisp stat list Subject: Mathematica to lisp translation Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk LispForm uses Mathematica pattern matching programming to translate some Mathematica expressions to lisp syntax. LispForm.mma can be found in ftp://ftp.upf.es/pub/stat/cosetes/LispForm/LispForm.mma together with this text file. The current version is really incomplete, I'm adding rules as I need them, but ringht now you can convert most Mathematica expressions to a string your lisp interpreter can read and store. For example, instead of the ugly expression In[27]:= ListCoefs[2,2] 2 2 2 2 2 2 4 Out[27]= {a (-1 + a ) , 2 a (1 - 3 a ) (1 - a ), 1 - 12 a + 15 a , 2 2 > 4 a (-2 + 5 a ), -2 + 15 a , 6 a, 1} using LispForm In[28]:= LispForm[%] you obtain the much more beatiful one Out[28]= (list (* (sqr a) (sqr (1- (sqr a))))(* 2 a (* (+ 1 (* -3 (sqr\ > a))) (+ 1 (- (sqr a)))))(+ 1 (* -12 (sqr a)) (* 15 (expt a 4\ > )))(* 4 a (+ -2 (* 5 (sqr a))))(+ -2 (* 15 (sqr a)))(* 6 a)\ > 1) If you type in Mathematica In[29]:= LispForm[some list of number pairs] >> somefile.lsp you will get "(list (list -1.2 0) (list -1.1 0) (list -1. 1.2326e-32) (list -0.9 0.01)\ (list -0.8 0.04) (list -0.7 0.09) (list -0.6 0.16) (list -0.5 0.25)\ (list -0.4 0.36) (list -0.3 0.49) (list -0.2 0.64) (list -0.1 0.81)\ (list -2.77556e-17 1.) (list 0.1 0.81) (list 0.2 0.64) (list 0.3 0.49)\ (list 0.4 0.36) (list 0.5 0.25) (list 0.6 0.16) (list 0.7 0.09) (list\ 0.8 0.04) (list 0.9 0.01) (list 1. 1.2326e-32) (list 1.1 0) (list 1.2 0\ )) " in your file somefile.lsp. You migth need to supress the quotes before using the file :<) Feel free to use it, and please send me any improvement you make on it. Frederic Udina udina@upf.es http://libiya.upf.es/ From majordom@umnstat.stat.umn.edu Tue Apr 30 13:19:17 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26680; Tue, 30 Apr 1996 13:19:17 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26675; Tue, 30 Apr 1996 12:19:11 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id NAA25747; Tue, 30 Apr 1996 13:19:04 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Tue, 30 Apr 1996 13:17:49 -0500 To: edstat-l@jse.stat.ncsu.edu, stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: ViSta Documentation Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk An earlier message about documentation for ViSta had the wrong web site address for the documentation. The web site is correct in this message. =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Documentation for ViSta, the Visual Statistics System, is now available at no cost in the form of Adobe Acrobat files, in addition to postscript files. The documentation includes cover, contents, introduction, tutorial and references. The documentation is the same for all operating systems. It can be downloaded from the ViSta web site at http://forrest.psych.unc.edu/research/ViSta.html. We strongly recommend getting the documentation in the form of Adobe Acrobat files because these files > are small > can be viewed on your computer screen > can be printed on any printer, not just postscript printers > are the same no matter what operating system you are using > are free You must have Adobe's Acrobat Reader to use the Acrobat files. It is available for free from Adobe at http://w1000.mv.us.adobe.com/Acrobat/Acrobat0.html. Note: We do not recommend getting the beta version of the next release of Acrobat Reader (called "Adobe Acrobat Amber") because in our judgment it is not yet reliable enough. You can also get the documentation in the form of postscript files from the ViSta web site. However, we recommend Adobe Acrobat files rather than postscript files because postscript files are much larger, cannot be seen on screen and can only be printed on postscript printers. For those of you who have not seen previous announcements about ViSta, please contact me at the address in the signature below for further information. Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Wed May 1 19:38:24 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17951; Wed, 1 May 1996 19:38:24 -0400 Return-Path: Received: from bubba.NMSU.Edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17946; Wed, 1 May 1996 18:38:19 -0500 Received: from NMSU.Edu by bubba.NMSU.Edu (8.6.10/NMSU) id RAA17166; Wed, 1 May 1996 17:40:14 -0600 Received: from ccserver by NMSU.Edu (8.6.10/NMSU-1.18) id RAA22031; Wed, 1 May 1996 17:37:41 -0600 Date: Wed, 1 May 1996 17:37:41 -0600 Message-Id: <199605012337.RAA22031@NMSU.Edu> Received: from [128.123.30.78] by ccserver (4.1/NMSU) id AA15519; Wed, 1 May 96 17:37:38 MDT X-Sender: ddaniel@dante.nmsu.edu Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: stat-lisp-news@umnstat.stat.umn.edu From: ddaniel@nmsu.edu (David L. Daniel) Subject: copy-list woes inside methods Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I want to make a *temporary* change to a slot-value from inside a method, and then change the slot-value back to its original value. To do this, I'm setting up a temporary variable to hold the original slot-value: (setf temp (copy-list (slot-value 'window-labels-loc))) But then when I change the slot-value, TEMP also changes. I've also tried the following: (setf temp2 (slot-value 'window-labels-loc)) (setf temp (copy-list temp2)) But TEMP still changes when I later change the slot-value. I thought COPY-LIST would overcome this problem. Any comments or suggestions? Thanks, -David David Daniel Dept. of Experimental Statistics New Mexico State University ddaniel@nmsu.edu From majordom@umnstat.stat.umn.edu Wed May 1 20:45:38 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA18316; Wed, 1 May 1996 20:45:38 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA18311; Wed, 1 May 1996 19:45:34 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa22302; 1 May 96 20:45 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA19702; Wed, 1 May 1996 20:45:28 -0400 Date: Wed, 1 May 1996 20:45:27 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: "David L. Daniel" Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: copy-list woes inside methods In-Reply-To: <199605012337.RAA22031@NMSU.Edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk A simple test prototype demonstrates that copy-seq works as expected inside a method for me ( using Mac Xlisp-Stat 3.47 ), so that should not be your problem. However, I don't know if copy-seq is recursive. What sort of values are in the lists you are copying ? ( defproto thing '( a b ) () ) ( defmeth thing :a (n v) ( setf tmp ( copy-seq (slot-value 'a))) ( setf ( elt tmp n ) v ) tmp ) ( defmeth thing :b (n v) ( setf tmp (copy-seq (send self :slot-value 'b ))) ( setf ( elt tmp n ) v ) tmp ) ( defmeth thing :isnew ( a b ) ( slot-value 'a a ) ( slot-value 'b b )) ( setf it ( send thing :new ( iseq 10 ) ( iseq 10 ))) > ( send it :slot-value 'a ) (0 1 2 3 4 5 6 7 8 9) > ( send it :a 4 -4 ) (0 1 2 3 -4 5 6 7 8 9) > ( send it :slot-value 'a ) (0 1 2 3 4 5 6 7 8 9) Method :a (send self :slot-value 'a) or :b (slot-value 'b) both seem to give the expected identical results. ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Mon May 6 12:21:13 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13231; Mon, 6 May 1996 12:21:13 -0400 Return-Path: Received: from super.super.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13226; Mon, 6 May 1996 11:21:08 -0500 Received: from gotham.super.org (gotham [192.239.79.2]) by super.super.org (8.7.1/8.6.12.1) with ESMTP id MAA06821 for ; Mon, 6 May 1996 12:16:05 -0400 (EDT) Received: from hume.super.org (hume [192.239.79.11]) by gotham.super.org (8.6.12/8.6.12.1) with ESMTP id MAA16354; Mon, 6 May 1996 12:21:01 -0400 From: pcolsen@super.org (Peter C Olsen) Received: (pcolsen@localhost) by hume.super.org (8.6.12/8.6.12.client) id MAA18792; Mon, 6 May 1996 12:21:00 -0400 Date: Mon, 6 May 1996 12:21:00 -0400 Message-Id: <199605061621.MAA18792@hume.super.org> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Limiting Printing Precision in the SAVEVAR Command X-Disclaimer: These opinions are mine, alone. Reply-To: pcolsen@acm.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I want to be able to limit the printing precision in the SAVEVAR command. As it stands now, I appear to be storing things with full precision --- apparently 18-digits of mantissa (plus exponent). Unfortunately, I don't need all that precision and I would much rather trade it for the savings in time and space that would result from saving only four or five digits of mantissa. I like the robustness and generality of using a standard function like SAVEVAR, so I would prefer not to write my own. Any suggestions? Peter -- Peter Olsen, n2ell, pcolsen@super.super.org ...!uunet!super!pcolsen P.O. Box 410, Simpsonville, MD 21150-0410; 410-997-8584 "Engineering is the art of applying a professional knowledge of mathematics and the physical sciences to improve the quality of life" From majordom@umnstat.stat.umn.edu Mon May 6 14:19:27 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14606; Mon, 6 May 1996 14:19:27 -0400 Return-Path: Received: from super.super.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14601; Mon, 6 May 1996 13:19:23 -0500 Received: from gotham.super.org (gotham [192.239.79.2]) by super.super.org (8.7.1/8.6.12.1) with ESMTP id OAA07236 for ; Mon, 6 May 1996 14:14:20 -0400 (EDT) Received: from hume.super.org (hume [192.239.79.11]) by gotham.super.org (8.6.12/8.6.12.1) with ESMTP id OAA17563 for ; Mon, 6 May 1996 14:19:21 -0400 Resent-From: pcolsen@super.org (Peter C Olsen) Received: (pcolsen@localhost) by hume.super.org (8.6.12/8.6.12.client) id OAA21041 for stat-lisp-news@umnstat.stat.umn.edu; Mon, 6 May 1996 14:19:21 -0400 Resent-Date: Mon, 6 May 1996 14:19:21 -0400 From: Resent-Message-Id: <199605061819.OAA21041@hume.super.org> Message-Id: <199605061819.OAA21041@hume.super.org> Resent-To: stat-lisp-news@umnstat.stat.umn.edu X-Disclaimer: These opinions are mine, alone. Reply-To: pcolsen@acm.org Subject: Suggestions for Large Matrices in Xlispstat? Date: Mon, 6 May 1996 14:13:44 -0400 Apparently-To: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'm looking for some suggestions to speed up code that must read and write large matrices. I would also like to speed up matrix multiplications and svd calculations. I'm doing some calculations with svds of large matrices that are "very tall and skinny" (on the order of 80000 rows and 30 columns). The matrices are not sparse and I need the singular values and both the left and right singular vectors. Peter -- Peter Olsen, n2ell, pcolsen@super.super.org ...!uunet!super!pcolsen P.O. Box 410, Simpsonville, MD 21150-0410; 410-997-8584 "Engineering is the art of applying a professional knowledge of mathematics and the physical sciences to improve the quality of life" From majordom@umnstat.stat.umn.edu Mon May 6 17:40:01 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16284; Mon, 6 May 1996 17:40:01 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16279; Mon, 6 May 1996 16:39:56 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa28277; 6 May 96 17:39 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA21563; Mon, 6 May 1996 17:39:41 -0400 Date: Mon, 6 May 1996 17:39:40 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: pcolsen@acm.org Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Limiting Printing Precision in the SAVEVAR Command In-Reply-To: <199605061621.MAA18792@hume.super.org> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Mon, 6 May 1996, Peter C Olsen wrote: > I want to be able to limit the printing precision in the SAVEVAR > command. > > As it stands now, I appear to be storing things with full precision > --- apparently 18-digits of mantissa (plus exponent). Unfortunately, > I don't need all that precision and I would much rather trade it for > the savings in time and space that would result from saving only four > or five digits of mantissa. > > Any suggestions? Set the variable *float-format* to a C printf format string. > *float-format* NIL > pi 3.141592653589793 > ( setf *float-format* "%g" ) "%g" > pi 3.14159 > ( setf *float-format* "%6.4g" ) "%6.4g" > pi 3.142 > ( / 10 3 ) 3.333 > ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Fri May 10 08:01:25 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22027; Fri, 10 May 1996 08:01:25 -0400 Return-Path: Received: from hilbert.edv.agrar.tu-muenchen.de by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22022; Fri, 10 May 1996 07:01:03 -0500 Received: by hilbert.edv.agrar.tu-muenchen.de; id AA01014; Fri, 10 May 1996 14:00:14 +0200 Date: Fri, 10 May 1996 14:00:11 +0200 (MET DST) From: Max Pilgram X-Sender: max@hilbert.edv.agrar.tu-muenchen.de To: Liste StatLisp Subject: xlispstat on AIX 3.2.5 Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, I have problems with the compilation of xlispstat-3-44 on an RS/6000 running AIX 3.2.5. The process stops after quite some time with the following error message: cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -D_BSD -qmaxmem=3000 -qlanglvl=ansi -Imachines/aix/dynload -DFOREIGNCALL -DIEEEFP -DANSI -c cholesky.c "./linalg.h", line 6.0: 1506-213 (S) Macro name abs cannot be redefined. make: 1254-004 The error code from the last command is 1. Maybe that anybody on the list was running in similar problems and has already found a solution. Thank's in advance to anyone who is responding, Max Max Pilgram Technische Universitaet Muenchen-Weihenstephan __ _ ___ |_ /_|| | | e-mail: max@pollux.edv.agrar.tu-muenchen.de | / |http://www.edv.agrar.tu-muenchen.de/dvs/persons/max/max.html From majordom@umnstat.stat.umn.edu Fri May 10 11:14:39 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23121; Fri, 10 May 1996 11:14:39 -0400 Return-Path: Received: from super.super.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23115; Fri, 10 May 1996 10:14:33 -0500 Received: from gotham.super.org (gotham [192.239.79.2]) by super.super.org (8.7.1/8.6.12.1) with ESMTP id LAA20673 for ; Fri, 10 May 1996 11:09:26 -0400 (EDT) Received: from hume.super.org (hume [192.239.79.11]) by gotham.super.org (8.6.12/8.6.12.1) with ESMTP id LAA07628; Fri, 10 May 1996 11:14:26 -0400 From: pcolsen@super.org (Peter C Olsen) Received: (pcolsen@localhost) by hume.super.org (8.6.12/8.6.12.client) id LAA03415; Fri, 10 May 1996 11:14:26 -0400 Date: Fri, 10 May 1996 11:14:26 -0400 Message-Id: <199605101514.LAA03415@hume.super.org> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Using PATH with popen ? X-Disclaimer: These opinions are mine, alone. Reply-To: pcolsen@acm.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'm writing to ask if anyone has been able to have popen search an explicit PATH. I'm using Xlispstat 3.47 to prototype a large application on several different Unix hosts (including Suns with SunOS 4.0+, SGIs with Irix 6.2, and IBM RISC machines running AIX). To maximize speed, I'm preprocessing some data using external programs (like "find", "wc", and some custom filters) called via popen. The problem arises because popen seems either to assume that the executable is in the current directory or, if not, that it will contain an explicit path. Unfortunately, the explicit paths do not seem to be constant across operating systems. (For example, IRIX uses /sbin/wc and SunOS uses /usr/ucb/wc.) There are several possible work-arounds (like writing lisp code to find an explicit path to each executable), but all of them seem to be horribly kludgy when all I really need is to have popen search the path itself. I will appreciate any suggestions. Peter -- Peter Olsen, n2ell, pcolsen@super.super.org ...!uunet!super!pcolsen P.O. Box 410, Simpsonville, MD 21150-0410; 410-997-8584 "Engineering is the art of applying a professional knowledge of mathematics and the physical sciences to improve the quality of life" From majordom@umnstat.stat.umn.edu Fri May 10 12:18:41 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23604; Fri, 10 May 1996 12:18:41 -0400 Return-Path: Received: from super.super.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23599; Fri, 10 May 1996 11:18:36 -0500 Received: from gotham.super.org (gotham [192.239.79.2]) by super.super.org (8.7.1/8.6.12.1) with ESMTP id MAA20970 for ; Fri, 10 May 1996 12:13:31 -0400 (EDT) Received: from hume.super.org (hume [192.239.79.11]) by gotham.super.org (8.6.12/8.6.12.1) with ESMTP id MAA08276; Fri, 10 May 1996 12:18:31 -0400 From: pcolsen@super.org (Peter C Olsen) Received: (pcolsen@localhost) by hume.super.org (8.6.12/8.6.12.client) id MAA03625; Fri, 10 May 1996 12:18:30 -0400 Date: Fri, 10 May 1996 12:18:30 -0400 Message-Id: <199605101618.MAA03625@hume.super.org> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Broken function Directory-files? X-Disclaimer: These opinions are mine, alone. Reply-To: pcolsen@acm.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I seem to have discovered a problem in the function "directory-files". I'm developing a large Xlispstat application on another machine (which, unfortunately, isn't connected to the internet). I say "unfortunately" because the "directory-files" function has taken to returning "nil" as a result even when (system "ls") shows that there are files in the directory --- and I can't easily include the dialog leading to the error (at least without typing several hundred lines of text). The one consistent precursor seems to be calling something like (popen "find directory -name '*.exten' -print") in a way that returns a Unix error (as when, for example, directory does not exist). As my users may do exactly this (for example, by calling the enclosing routine with a typo), and my application depends heavily on (directory-files), I'd be really appreciative of any ideas. Peter -- Peter Olsen, n2ell, pcolsen@super.super.org ...!uunet!super!pcolsen P.O. Box 410, Simpsonville, MD 21150-0410; 410-997-8584 "Engineering is the art of applying a professional knowledge of mathematics and the physical sciences to improve the quality of life" From majordom@umnstat.stat.umn.edu Fri May 10 13:53:14 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24223; Fri, 10 May 1996 13:53:14 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24218; Fri, 10 May 1996 12:53:07 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id ab02555; 10 May 96 13:52 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA14465; Fri, 10 May 1996 13:52:51 -0400 Date: Fri, 10 May 1996 13:52:50 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: pcolsen@acm.org Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Broken function Directory-files? In-Reply-To: <199605101618.MAA03625@hume.super.org> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Fri, 10 May 1996, Peter C Olsen wrote: > I seem to have discovered a problem in the function "directory-files". I didn't know there was a directory funcion in xlisp-stat! > (popen "find directory -name '*.exten' -print") Oh - it's a unix-only kludge... > in a way that returns a Unix error (as when, for example, directory > does not exist). As my users may do exactly this (for example, by > calling the enclosing routine with a typo), and my application depends > heavily on (directory-files), I'd be really appreciative of any ideas. I wrote and posted a common lispish directory function in C that I use on the Mac. It OUGHT to be portable to unix, as it's using posix-y systen calls, but I haven't actually tested it anywhere but on the Mac. I'll be happy to send you a copy if you want to recompile and give it a try. The common lisp manual was not specific about what the result of directory on an invalid pathname should be, but since there is no corresponding portable 'stat' function, or some equivalent to check whether a file is a directory or a plain file, I made it return Nil, to make it simpler to recursively descend directories. It only returns a list of files in the directory pathname arg - no filename globbing, although simple matching is easy in lisp: ( remove ".ext" (directory d) :test-not #'string-search ) or use the pathname functions. ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Mon May 13 14:12:51 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA09163; Mon, 13 May 1996 14:12:51 -0400 Return-Path: Received: from hilbert.edv.agrar.tu-muenchen.de by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA09153; Mon, 13 May 1996 13:12:06 -0500 Received: by hilbert.edv.agrar.tu-muenchen.de; id AA02060; Mon, 13 May 1996 20:11:10 +0200 Date: Mon, 13 May 1996 20:11:09 +0200 (MET DST) From: Max Pilgram X-Sender: max@hilbert.edv.agrar.tu-muenchen.de To: Liste StatLisp Subject: Piping output into command Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, I would like to redirect the output of a lispstat command into a shell command. Is this generally possible and if yes, how could I do it? Example: Piping the output of the command... > (select (send current-model :labels) (send plot :selection)) ("Volvo 240 GL" "Peugeot 694 SL") ...into a shell-script. Max Max Pilgram Technische Universitaet Muenchen-Weihenstephan __ _ ___ |_ /_|| | | e-mail: max@pollux.edv.agrar.tu-muenchen.de | / |http://www.edv.agrar.tu-muenchen.de/dvs/persons/max/max.html From majordom@umnstat.stat.umn.edu Thu May 16 16:56:03 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03139; Thu, 16 May 1996 16:56:03 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03134; Thu, 16 May 1996 15:55:54 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa17678; 16 May 96 16:55 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA05008; Thu, 16 May 1996 16:55:36 -0400 Date: Thu, 16 May 1996 16:55:35 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: Mac XlispStat background color bug found. Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk maciviewwindow2.c [Mac xlispstat 3.47] contains the following code, which is the source of some of the color bugs I had previously reported: VOID set_fore_color(StGWWinInfo *gwinfo) { if (gwinfo->drawColor < NumBasicColors) ForeColor(get_color(gwinfo->drawColor)); else RGBForeColor(get_rgb_color(gwinfo->drawColor)); } VOID set_back_color(StGWWinInfo *gwinfo) { if (gwinfo->backColor < NumBasicColors) BackColor(get_color(gwinfo->backColor)); else RGBForeColor(get_rgb_color(gwinfo->backColor)); } Both functions set the foreground color -- the second function should be calling RGBBackColor() instead. With this change, setting the background color to a non-built-in color works correctly. I am also now able to copy color graphics into the clipboard without colored regions being lost - something that was previously not working for me, but I have also made some changes in the Lisp code for scatterplot :redraw-content and some other methods, so I'm not sure which changes have fixed that problem. ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Thu May 16 17:31:30 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03343; Thu, 16 May 1996 17:31:30 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03338; Thu, 16 May 1996 16:31:24 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa00664; 16 May 96 17:30 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA21657; Thu, 16 May 1996 17:30:47 -0400 Date: Thu, 16 May 1996 17:30:46 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Forrest Young Cc: Luke Tierney , Xlisp List Subject: Vista data table objects & programming or architecture docs In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I've been rewriting some of my spectral data classes and considering writing some classes for handling tabular data, and I thought I'ld take a look at what you've done in Vista before re-inventing something similar. I'm impressed with some of the Vista user interface! ( I really like the workmaps and guidemaps. ) After a quick look at the code in dataobj*.lsp and tablobj*.lsp, it looks [1] too complex for what I need and [2] too monolithic to just steal a piece of it. ( But that's just a first impression based on a really quick look. ) Is there any programmer-oriented documentation, in addition to the PostScript User docs ? Or a guide to importing data into Vista, or writing a datafile that it will accept ? In Luke's "Recent Developments and Future Directions..." paper, he mentioned the desirability of a more complete data object model, to avoid having to incompatibly reinvent similar functionality. If I find I can't make use of your data objects, perhaps we need some discussion of what sort of base classes would be useful in the base implementation. ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Thu May 16 17:39:54 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03374; Thu, 16 May 1996 17:39:54 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03369; Thu, 16 May 1996 16:39:49 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA03068; Thu, 16 May 96 14:39:06 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA09494; Thu, 16 May 1996 14:39:05 -0700 From: Jan Deleeuw Message-Id: <9605162139.AA09494@galton.math.ucla.edu> Subject: Zoom Overlay To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Thu, 16 May 1996 14:39:04 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk In ftp://ftp.stat.ucla.edu/pub/xlispstat/code/statistics/plotting/zoom-overlay.lsp there is an overlay for zooming a scatterplot by Jason Bond (UCLA) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;; ;; Zoom Overlay prototype. When the "Zoom In" radio button is ;; pushed in the overlay, a box can be drawn around points (using the ;; mouse) and they will be "zoomed" in on. Lines that are drawn in the ;; plot whose linestarts lie completely inside of the box that is drawn ;; will also be drawn in the "zoom". The overlay is added by sending ;; the original plot the :add-overlay method. An example is: ;; ;; (def p1 (plot-points (list (sample (iseq 50) 60 t) ;; (sample (iseq 50) 60 t)))) ;; (send p1 :add-overlay (send zoom-overlay-proto :new)) ;; (send p1 :redraw) ;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Thu May 16 17:58:26 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03464; Thu, 16 May 1996 17:58:26 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03457; Thu, 16 May 1996 16:58:17 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA03679; Thu, 16 May 96 14:58:16 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA09609; Thu, 16 May 1996 14:58:14 -0700 From: Jan Deleeuw Message-Id: <9605162158.AA09609@galton.math.ucla.edu> Subject: table objects/spreadsheets/data To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Thu, 16 May 1996 14:58:12 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk For another approach, see ftp://ftp.stat.ucla.edu/pub/xlispstat/code/utilities/edit/cellsheet or the older one in ftp://ftp.stat.ucla.edu/pub/xlispstat/code/utilities/edit/spreadsheet or the even older one in ftp://ftp.stat.ucla.edu/pub/xlispstat/code/statistics/introstat/uclastat I think the cellsheet solution is fairly nice, the scrolling list solutions are pretty awful. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon May 20 12:17:37 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02335; Mon, 20 May 1996 12:17:37 -0400 Return-Path: Received: from dmsna.dms.unina.it by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02330; Mon, 20 May 1996 11:17:28 -0500 Received: (from icora@localhost) by dmsna.dms.unina.it (8.6.12/8.6.9) id RAA08202; Mon, 20 May 1996 17:20:34 +0200 Date: Mon, 20 May 1996 17:20:33 +0200 (MET DST) From: Isabella Corazzari To: stat-lisp-news@umnstat.stat.umn.edu Subject: is there someone who is working about multiway-multimode-data-analysis, specifically with , STATIS-model, with xlisp-stat? Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'm a Ph.D. student, and I'm working about data classified as unit x variables x times (occasions), and I'm looking for codes in xlisp-stat about the Statis and Tuckers' methods, for this kind of matrices' analysis. If someone has something about this argument, please send me to Isabella Corazziari icora@dmsna.dms.unina.it Many thanks to you From majordom@umnstat.stat.umn.edu Wed May 22 11:16:03 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA07622; Wed, 22 May 1996 11:16:03 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA07617; Wed, 22 May 1996 10:15:58 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id LAA23065; Wed, 22 May 1996 11:15:55 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 22 May 1996 11:13:31 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: table data object Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Who was it that sent the message to this group about wanting to write a table data object? I wanted to reply but got the message just before I left town for a week, and now I can't find it. If someone sends it to me again, I'll reply. Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Thu May 23 06:55:17 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23429; Thu, 23 May 1996 06:55:17 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23424; Thu, 23 May 1996 05:55:13 -0500 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.6.4.3/10.1) id GAA14077; Thu, 23 May 1996 06:55:06 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 23 May 1996 06:56:14 -0500 To: sdm7g@Virginia.EDU, stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: Re: Vista data table objects Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk "Steven D. Majewski" wrote: >I've been rewriting some of my spectral data classes and considering >writing some classes for handling tabular data, and I thought I'd >take a look at what you've done in Vista before re-inventing something >similar. I'm impressed with some of the Vista user interface! ( I >really like the workmaps and guidemaps. ) >After a quick look at the code in dataobj*.lsp and tablobj*.lsp, >it looks [1] too complex for what I need and [2] too monolithic to >just steal a piece of it. ( But that's just a first impression based >on a really quick look. ) Unfortunately, I think you came to the right conclusion on the basis of your quick look (it doesn't take a long look to see the problems in trying to re-use my messy code :-( ). I suspect that you'll find you can't use my code as you would wish. Furthermore, the table data object is the worst part of the data object code. I feel good about the basic approach I took in the multivariate data object, and a little less good about the matrix data object, but the table data object is a real kludge that I want to get rid of. >Is there any programmer-oriented documentation, in addition to >the PostScript User docs ? Or a guide to importing data into Vista, >or writing a datafile that it will accept ? Yes, there is more documentation than is on my server. Right now there is a completed chapter on getting data into ViSta, as well as three completed analysis modules chapters. There also is a somewhat out-of-date chapter (that I'm updating) that is oriented towards the programmer who wants to write new model objects. >In Luke's "Recent Developments and Future Directions..." paper, he >mentioned the desirability of a more complete data object model, to >avoid having to incompatibly reinvent similar functionality. If >I find I can't make use of your data objects, perhaps we need some >discussion of what sort of base classes would be useful in the >base implementation. I believe that without too much effort I could extract the multivariate data object from ViSta and make it a stand-alone XLS object. It could then be used as an object for other data types to inherit from. I also have several stand-alone XLS graphical objects and a stand-alone n-way anova object that I could make available. Perhaps Luke would want to include them in his distribution if I documented them? Is there any discussion of this? Would others find that useful? Forrest Young Professor, Psychometrics CB# 3270 Davie Hall email: mailto://forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Thu May 23 15:31:16 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27168; Thu, 23 May 1996 15:31:16 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27163; Thu, 23 May 1996 14:31:01 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa27648; 23 May 96 15:29 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA01713; Thu, 23 May 1996 15:29:12 -0400 Date: Thu, 23 May 1996 15:29:12 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: XLS Preferences: Can fonts be changed on the fly? Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Is there any way to change the values set in XLS preferences on the fly from withing Xlisp-Stat ? ( In particular, I would like to switch between a large and a small graphics font: large for viewing on a full screen, small when I'm clipping a smaller size version into the clipboard to paste into a report. ) ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Thu May 23 17:32:38 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28130; Thu, 23 May 1996 17:32:38 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28125; Thu, 23 May 1996 16:32:34 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa07452; 23 May 96 17:31 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA20366; Thu, 23 May 1996 17:31:52 -0400 Date: Thu, 23 May 1996 17:31:51 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Forrest Young Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: data and graphic objects/part 1 [was: Vista data table objects] In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Thu, 23 May 1996, Forrest Young wrote: > I believe that without too much effort I could extract the multivariate > data object from ViSta and make it a stand-alone XLS object. It could then > be used as an object for other data types to inherit from. I also have > several stand-alone XLS graphical objects and a stand-alone n-way anova > object that I could make available. Perhaps Luke would want to include > them in his distribution if I documented them? Is there any discussion of > this? Would others find that useful? > I could use the n-way anova -- I'm currently using oneway-model, but it really should be a two way analysis. I also have some graphics objects: I have completely rewritten my overlay code ( this is the third try, at least. I think I've finally got it "Right". ) The first module (realgraph.lsp) is pretty much in its final form. The second ( jitterplot.lsp ) is my first use of it -- which led to some changes in realgraph.lsp - it may go thru some changes, but is a good example of how to use the realgraph overlays. I have not yet gone back and changed my content-overlays objects to use and be compatible with these classes. ( The new classes are a better implementation, they appear to be faster and cleaner, and they fix some of the color problems I had run into, for example, clipping graphs into the mac clipboard preserves fill colors, which were getting mysteriously lost before -- I suspect by that erase of the content region. ) realgraph.lsp: redefines scatterplot-proto :REDRAW-CONTENT method to :REDRAW-DATA-LINES and :REDRAW-DATA-POINTS without doing an erase of the content region. redefines graph-overlay-proto :graph to be a setter as well as a getter, and redefines GRAPH-PROTO :ADD-OVERLAY & :DELETE-OVERLAY to use it, so that the overlay objects can be hierarchical and graph binding can be propagated. Adds :bring-to-front methods to graphs and overlays. Defines a REAL-GRAPH-OBJ-PROTO which defines common methods (:REAL-TO-CANVAS, :RESIZE, :REDRAW, etc. ) for overlay objects drawn with real coordinates. RECT-PROTO OVAL-PROTO ( also implements ARC with optional args ) LINE-PROTO POLY-PROTO POINTS-PROTO LABEL-PROTO COMPOSIT-GRAPH-OBJ-PROTO join primitives together with :ADD-PARTS, and access them with :PARTS propagates graph, line-color, line-width, fill-color changes to parts. Jitterplot.lsp: JITTERPLOT-PROTO: "jittered" ( randomized in the x direction so points don't fall on top of each other, and density of points is not obscured.) and groups data sets. example: (send self :group-points n :point-color 'blue ) (send self :group-points n '( :point-coordinate 1 )) BOXPLOT-PROTO (the standard boxplots, but as an overlay that has settable :line-width :line-color :fill-color ) and DISTRIB-CLOUD-PROTO ( ovals with major radius equal to multiples of the standard deviation, and centered around the mean or weighted mean line. ) are specific COMPOSIT-GRAPH-OBJ-PROTO descendants. A snapshot of the code and some gifs of example plots are at small and large images: A screen dump from the old content-overlay code is: I'll release a final version after I have reconciled the old code to the new graphic overlay objects. [ This *DOES* still have something to do with data-objects and data-tables, but I'm going to break this long message up into two parts. The 2nd will discuss data-objects and some of the advantages of having a consistent and standard composite data object interface. ] ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Thu May 23 18:41:34 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28555; Thu, 23 May 1996 18:41:34 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28550; Thu, 23 May 1996 17:41:29 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa06092; 23 May 96 18:41 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA18540; Thu, 23 May 1996 18:41:22 -0400 Date: Thu, 23 May 1996 18:41:21 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Forrest Young Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: data and graphic objects/Cont. [was: Vista data table objects] In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I want ahead and implemented my own table-proto: it takes an array with the first row containing the field/class names, and constructs named accessor methods for each column. An :indexes method returns the row indices of all of the rows that match a given test ( the default test is #'string-search ), which can be used to select the elements of the desired column. More complicated selections can combine or intersect the lists of indices, for example: ( select ( send table :CA_CONC ) ( intersection ( send table :indexes :GROUP "Control" ) ( send table :indexes :STRUCT "Cyto" ))) pulls out the CA_CONC column for all the observations/rows where GROUP column contains the string "Control" and STRUCT column contains the string "Cyto". This manages to do the thing which was puzzling me when I wrote the initial help message -- I need to extract data into one of my sequence data types which are subclasses of COMPOUND-DATA-SEQ: Typically, I am matching my row selectors as above, but I need several columns for each data sequence -- for example NAME, CA_CONC, CA_CONC_STD_DEV. I have found that using a consistent data-sequence object interface makes life a lot easier. Consistant access methods allow :PLOT methods to automatically label points and axes correctly, scale plot to regions of interest rather that max,min data points, handle data overlays and linking, consistent coloring of overlays, etc. ( An example that I haven't yet implemented: all of my points in my CA_CONC distribution plots have :NAME's, which are displayed when that point is selected. Typically those names are the same as the filename of the spectrum which was analyzed to give that concentration. [ The "R25-11" in the plot at for example. ] So, not only can I link Calcium concentration plots with Phosphorus concentration, but, if I add something to keep track of the root pathnames for my data sets, I can make double-clicking on a suspicious outlier point automatically bring up the spectrum plot of the data from which that point was derived. ) That is why I would like to see a similar data model and programming conventions standardized, both for sequences and 2 way tables. Actually, EVERY data element OUGHT to be able to be a composite object, with a NAME slot, a VALUE slot, and a slot for variance/std-dev. or some other associated statistic linked to the data value. In Python, there isn't such a big split between objects and built-in data types, so it would be easier to extend without diving into the implementation. In XlispStat, this might require the same sort of global modifications that were done to add vectorized arithmetic operations. If there were some standard slots and accessors, it would also make converting to or from other dimensional data objects or tables easier. ( like extracting a column of data from a table, and building a compound-data-seq, as described above. ) Following is a summary of the COMPOUND-DATA-SEQ objects I'm using in my current app. ( Some methods may migrate from one class to another, as I rationalize the design a bit more and try to figure out how to better integrate singletons, sequences and tables of data and conversion between them. [ I think the most important thing to standardize beyond COMPOUND-DATA-SEQ itself, is turning them into first class generic sequences in GENSEQ -- that is what makes the rest of the package work! ] ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] GENSEQ.LSP redefines length, elt, (setf (elt ... )) to work on COMPOUND-DATA-SEQ objects, which allows them to be used in most functions that expect a generic sequence of some type. Vectored functions like +, *, etc. already work for COMPOUND-DATA-SEQ, but PLOT-LINES and some other functions expect a LISP sequence and fail when passed a sequence object. This mod fixes all of the cases I've found so far. DATASEQ.LSP DATA-SEQ-PROTO defines a basic subclass of COMPOUND-DATA-SEQ with slots/accessors: NAME & COMMENT strings NAMES & COMMENTS sequence of strings [info on each observation/data-point] STD-ERRS ( or standard-deviations, but will probably change this to variances or inverse variances/weights - to optimize calculations and to be more consistent with weighted regression,etc. and methods: :MEAN ( &key weighted T ) :STANDARD-DEVIATION :PRINT ( prints summary info -- I will probably shorted this and add a separate :SUMMARY method ) :PLOT ( a distribution/boxplot ) XYSEQ.LSP: XY-SEQ-PROTO: inherits from DATA-SEQ-PROTO. IS-A Ordered data sequence adds slots/accessors: :OFFSET ( or initial X ) :dx :units and methods: :X :Y ( is an alias for the data-seq ) :X-RANGE :Y-RANGE :DIFFERENCE :MEDIAN-SMOOTH (actually called :MEDIAN -- should change name) ( :COLOR is here, but I'm not sure where it really belongs.) :PLOT is redefined to be PLOT-POINTS (send self :X) (send self :Y) SPEC.LSP: SPEC-PROTO: Adds methods to convert between CHANNEL-NUMBER or INDEX X coordinates and real valued X values. Shifts, subsequences, zero-padded extended sequences, etc. ( actually everything miscellaneous that didn't quite make it into the application specific XES ( X-ray Energy Spectra ) classes. XES-PROTO XES-REF-PROTO Add bandpass filtering functions, overlay plotting functions, and most of the application specific data functions. From majordom@umnstat.stat.umn.edu Tue Jun 4 13:23:55 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA01659; Tue, 4 Jun 1996 13:23:55 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA01653; Tue, 4 Jun 1996 12:23:49 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA17880; Tue, 4 Jun 96 10:23:46 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA11348; Tue, 4 Jun 1996 10:23:43 -0700 From: Jan Deleeuw Message-Id: <9606041723.AA11348@galton.math.ucla.edu> Subject: Polish-Plot To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Tue, 4 Jun 1996 10:23:43 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Added to archive: ftp://ftp.stat.ucla.edu/pub/xlispstat/code/statistics/plotting/polish-plot.lsp ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; polish-plot.lsp ; ;AUTHOR: David L. Daniel ; Dept. of Experimental Statistics ; New Mexico State University ; ddaniel@nmsu.edu ; ;UPDATED: 5/30/96 ; ;COMMENTS: ; If you have suggestions or comments about this code, or if you make ; useful changes to it, please let me know via email. Also, if you ; find this code useful, I'd appreciate your letting me know that as ; well. Thanks, -David ; ; ; ;OVERVIEW: ; The primary function in the code below is "polish-plot" which ; enhances the capabilities of the typical two dimensional plots ; available in Xlisp-Stat. The main objective of this code is ; to create plots which are better suited for final output to a ; printer. The function polish-plot adds the following ; attributes to two dimensional plots. ; ;ADVANTAGES: ;o Enables log-linear, linear-log, and log-log scale plotting. ; The logarithm base may be chosen by the user. ; ;o A title and multiple labels may be placed inside the plot ; window, and will be redrawn automatically when the plot ; is redrawn. ; ;o Function curves may be added which are represented by line ; segments, as with the :add-function method, but whose line ; segment end-points are equally spaced in the X direction ; even when the X axis is log scaled. ; ;o Formats for labeling the major tick marks can be set. ; ;o Minor tick marks can be included on linear or log scaled ; axes, and the number of minor ticks can be specified. ; ;o The minimum X or Y value for a plot may be offset by a ; specified amount from the axes' intersection so that points ; and curves are not plotted immediately at an axis line. ; ;o X and Y axis tick marks are labeled with horizontal strings ; so that chunky bitmapped images are not obtained when plots ; are sent to a printer. ; ;DISADVANTAGES (known): ;o Brushing does not work on log scaled plots. ; Note: This was a conscious decision between two alternatives - ; I chose this consequence in order to retain the ability to ; retrieve the original data values with standard xlisp-stat ; messages. Brushing could be implemented if the original ; data values were stored in a new slot, but the user would ; have to access these values with non-standard commands. One ; option would be to allow the user to switch between these ; two different modes of operation by sending the plot a ; message. Let me know if there is interest in this. ; -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Fri Jun 7 12:57:36 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19630; Fri, 7 Jun 1996 12:57:36 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19623; Fri, 7 Jun 1996 11:57:31 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa27504; 7 Jun 96 12:57 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA26088; Fri, 7 Jun 1996 12:57:12 -0400 Date: Fri, 7 Jun 1996 12:57:12 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: color grids Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Now that background colors are finally working on Mac XlispStat [ see previous posts for source fix ], I find you can draw a nice background grid with the contrast reduced sufficiently so that it doesn't distract too much from the data. I assume that bug was Mac specific, so this should work on other platforms that support color, but you also have to use the modified :redraw-content that I also posted sometime back, so that it doesn't erase the content area. Add the following methods ( :draw-grid, :grid-color and modified :redraw-background ) to scatterplot-proto and try something like: ;; assumes ( setf plot SOMEPLOT ... ) ( add-color 'ghostwhite ) ( add-color 'gray80 ) ( send plot :back-color 'ghostwhite ) ( send plot :grid-color 'gray80 ) ( send plot :redraw ) ;; or resize or uncover the plot ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] ( defmeth scatterplot-proto :draw-grid ( &key color ) ( let (( x-range ( send self :range 0 )) ( y-range ( send self :range 1 )) ( back-c ( send self :back-color )) ( draw-c ( send self :draw-color ))) ( if ( null color ) ( setf color draw-c )) ( send self :draw-color color ) ( dolist ( x ( apply #'rseq (append x-range ( last ( send self :x-axis ))))) ( apply #'send self :draw-line ( append ( send self :real-to-canvas x ( car y-range )) ( send self :real-to-canvas x ( cadr y-range ))))) ( dolist ( y ( apply #'rseq (append y-range ( last ( send self :y-axis ))))) ( apply #'send self :draw-line ( append ( send self :real-to-canvas ( car x-range ) y ) ( send self :real-to-canvas ( cadr x-range) y )))) ( send self :draw-color draw-c ))) ( defmeth scatterplot-proto :grid-color ( &optional (color nil set)) ( if ( not ( send self :has-slot 'grid-color )) ( send self :add-slot 'grid-color )) ( if set ( slot-value 'grid-color color ) ( if ( eq color T ) ( send self :draw-color ) ( slot-value 'grid-color )))) ( defmeth scatterplot-proto :redraw-background () (call-next-method) ( let (( grid ( send self :grid-color ))) ( if grid (send self :draw-grid :color grid)))) From majordom@umnstat.stat.umn.edu Mon Jun 10 08:01:33 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27716; Mon, 10 Jun 1996 08:01:33 -0400 Return-Path: Received: from lenzie.cent.gla.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27711; Mon, 10 Jun 1996 07:01:23 -0500 Message-Id: <593.199606101200@lenzie.cent.gla.ac.uk> From: James Currall Received: (from jepc1e@localhost sender jepc1e) by lenzie.cent.gla.ac.uk (8.7.5/UK-2.2a/cent-sparc) id NAA00593 for stat-lisp-news@umnstat.stat.umn.edu; Mon, 10 Jun 1996 13:00:51 +0100 (BST) Subject: XLISPSTAT query To: stat-lisp-news@umnstat.stat.umn.edu Date: Mon, 10 Jun 1996 13:00:50 +0100 (BST) X-Mailer: ELM [version 2.4 PL23] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi I have hit a baffling problem with XLISPSTAT 3.45 under Windows on a PC. The thing is, it should be quite simple. In order to decide how many tick marks to put on the y-axis of a graph, I need to calculate the y axis range, then add 1 (to get 1 tick per value). The following function returns the range (defun yRange (numList) (- (max numList) (min numList))) However, when I use this function in the following statement (anova 2 is the graph and the y axis range = 10): (send anova2 :y-axis t t (+ 1 (yRange (send anova2 :range 1)))) I get the following error: Error: bad argument type - 10 The problem does not appear to be with the function, but with any argument returned as a result of sending a message to the graph. Any other way of generating the result '10' seems fine. Any thoughts would be most welcome. James -- --------------------- Dr James Currall ----------------------- | E-mail: J.Currall@compserv.gla.ac.uk ~ ~ --- ~ | | Fax: (UK) 0141-330-4808 ~ / ^ / ~ ~ | | Phone: (UK) 0141-339-8855 Ext 4855 ~ /^ ^ --- ~ | | Telex: 777070 / Answerback UNIGLA ~ /^ ^ / ~ | | Post: Computing Service, University of ~ \ * -- ~ | | Glasgow, Glasgow G12 8QQ, Scotland. ~ / ^ ^ | ~ | -------------------------------------------------------------- From majordom@umnstat.stat.umn.edu Mon Jun 10 14:51:40 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04881; Mon, 10 Jun 1996 14:51:40 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04876; Mon, 10 Jun 1996 13:51:32 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id OAA06323; Mon, 10 Jun 1996 14:51:24 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 10 Jun 1996 14:48:02 -0500 To: edstat-l@jse.stat.ncsu.edu, stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: Additional Vista documentation Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Additional chapters of documentation for ViSta are now available. The new chapters are on defining data, and on regression analysis and correspondence analysis. The contents and references have been updated. The introduction and tutorial are still available but are unchanged. Additional chapters on exploring data, on each of the remaining analysis modules, and on adding analysis and visualization methods to ViSta are in preparation. =============================== The documentation is the same for all operating systems. However, it exists in three forms: 1) Adobe Acrobat pdf files (for any operating system) 2) Postscript in the form of MSWindows zipped files 3) Postscript in the form of self-extracting-archive files for Macintosh We recommend Adobe Acrobat files because they: > are small and download quickly (postscript files are much larger). > can be viewed on your screen (postscript cannot be viewed on screen). > can be printed on any printer (postscript only prints on postscript printers). > are the same for all operating systems (postscript is compressed >differently for different operating systems). Note that to display and print the Acrobat files you must have the Adobe Acrobat Reader. It is available from Adobe at no cost from http://w1000.mv.us.adobe.com/Acrobat/Acrobat0.html. No special software is needed for postscript files. =============================== You can obtain the documentation in three ways: 1) World-Wide-Web Any of the three forms of the documentation may be downloaded from the ViSta world-wide-web server at http://forrest.psych.unc.edu/research/ViSta/doco.html. 2) FTP You may ftp any of the three forms of the documentation from the ViSta ftp site at www.psych.unc.edu/pub/forrest/vista. 3) Mail You may obtain hardcopy of the documentation by sending a check or international money order. It should be in US dollars and drawn on a US bank. It should be made payable to "UNC Psychometric Lab" in one of the following amounts: >USA & Canada: $10.00US >Mexico: $15.00US >Elsewhere: $25.00US Mail the check or money order to the address in the signature of this message. =============================== For more information about vista, visit the ViSta web site at the www address in the signature, or request more information at the address given below. Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Fri Jun 14 08:52:47 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10221; Fri, 14 Jun 1996 08:52:47 -0400 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10216; Fri, 14 Jun 1996 07:52:42 -0500 Received: by kodakr.kodak.com id AA11408 (5.67b/IDA-1.5 for stat-lisp-news@stat.umn.edu); Fri, 14 Jun 1996 08:50:21 -0400 Date: Fri, 14 Jun 1996 08:50:21 -0400 From: morris@Kodak.COM Message-Id: <199606141250.AA11408@kodakr.kodak.com> Subject: GLIM problem Apparently-To: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Try as I might I can't seem to break away from XLS...To wit, I have encountered a small problem that I would like to solve/understand. When I try to fit a glim model of the binomial family, I have no problems with the default link, i.e. (binomialreg-model logdose r n) but when I ask for a probit link I get this error: Error: The variable PROBIT-LINK is unbound. Happened in: # Now...I only get this if I ask for the probit link as I first enter the session. If I first run the default log link then I can, for the rest of the session, ask for the probit. Why and how do I solve it. Thanks!!! Jeff Jeffrey D. Morris, Ph.D. Environmental Sciences Eastman Kodak Company Rochester, NY 14652 morris@kodak.com From majordom@umnstat.stat.umn.edu Fri Jun 14 12:55:59 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11871; Fri, 14 Jun 1996 12:55:59 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11866; Fri, 14 Jun 1996 11:55:52 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA04396; Fri, 14 Jun 96 09:55:44 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA07578; Fri, 14 Jun 1996 09:55:42 -0700 From: Jan Deleeuw Message-Id: <9606141655.AA07578@galton.math.ucla.edu> Subject: archive To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Fri, 14 Jun 1996 09:55:42 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk polish-plot.lsp by David Daniel has been upgraded a bit. It is now in code/statistics/plotting/strays/polish-plot.lsp Mirko Ravagnan contributed code/statistics/introstat/dist-fun/ yet another probability distribution package (documented). -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@nokomis.stat.umn.edu Mon Jun 17 14:28:47 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA00967; Mon, 17 Jun 1996 14:28:47 -0400 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA00959; Mon, 17 Jun 1996 13:28:38 -0500 From: Luke Tierney Message-Id: <9606171828.AA00959@nokomis.stat.umn.edu> Subject: Re: GLIM problem To: morris@Kodak.COM Date: Mon, 17 Jun 1996 13:28:37 -0500 (CDT) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: <199606141250.AA11408@kodakr.kodak.com> from "morris@Kodak.COM" at Jun 14, 96 08:50:21 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 913 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > Try as I might I can't seem to break away from XLS...To wit, I have > encountered a small problem that I would like to solve/understand. > When I try to fit a glim model of the binomial family, I have no > problems with the default link, i.e. > (binomialreg-model logdose r n) > but when I ask for a probit link I get this error: > Error: The variable PROBIT-LINK is unbound. > Happened in: # > Now...I only get this if I ask for the probit link as I first enter the > session. If I first run the default log link then I can, for the rest of > the session, ask for the probit. Why and how do I solve it. > Thanks!!! This is a bug in the simple-mindedn autoload facility -- only function or macro calls trigger an outoload, not variable references. For now you need to explicitly load the glim system if you want to use the probit link -- this will hopefuly be fixed soon. luke From majordom@umnstat.stat.umn.edu Tue Jun 18 10:36:50 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04913; Tue, 18 Jun 1996 10:36:50 -0400 Return-Path: Received: from gemini.oscs.montana.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04908; Tue, 18 Jun 1996 09:36:45 -0500 Received: by gemini.oscs.montana.edu (5.65/DEC-Ultrix/4.3) id AA19507; Tue, 18 Jun 1996 08:36:43 -0600 Date: Tue, 18 Jun 1996 08:36:43 -0600 From: uchih@gemini.oscs.montana.edu (Ina Hoeschele) Message-Id: <9606181436.AA19507@gemini.oscs.montana.edu> To: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk ***** Postdoctoral Position available ********************************* ***** Desired Start Date: September 1, 1996 (August 1 - October 1) **** ******************* Area of Research: ******************************* ***** Statistical Gene Mapping and Markov Chain Monte Carlo *********** ******************* Location: *************************************** ******** Department of Dairy Science ********************************** ** Virginia Polytechnic Institute and State University **************** I am offering a postdoctoral position with funding available for up to two years. Two previous postdoctoral associates have worked on this project with great success and productivity (3-5 papers each). The research is funded by the National Science Foundation and the US Department of Agriculture. The successful candidate will contribute in some of the following areas: (i) Generalization of software developed for Bayesian and Residual Maximum Likelihood statistical gene mapping to pedigree structures other than granddaughter designs. (ii) Continuation of work on fitting several linked QTLs per chromosome. (iii) Further optimization of linkage analysis software, including parallelization. (iv) Analysis of large genetic marker data sets with above methods and with simpler methods, including nonparametric approaches. (v) Further developments of efficient Markov chain Monte Carlo (including Gibbs sampling) algorithms for gene mapping, models including mapped genes and (non)linear models. (vi) Further developments of tests for the nature of QTLs (single biallelic or multiallelic, polygenic) (vii) Linkage analysis with multiple traits and categorical traits. The successful candidate will be given the opportunity to teach and collaborate on other projects, if desired. He/she will work in a team of two PhD students and two postdoctorals, and probably one or two visiting scientists. The candidate must be (i) well trained in Statistics and Genetics and hold a PhD degree in either field, and (ii) familiar with Fortran programming and Unix environments. The ideal candidate is also (iii) somewhat familiar with major gene and linkage analyses (iv) eager to succeed and exercise initiative (v) reliable and a good team member Please send applications including Curriculum Vitae, list of publications, and three references to email: inah@vt.edu fax: (USA) 406 994 1895 c/o Ina Hoeschele (before Aug.10) (USA) 540 231 5014 c/o Ina Hoeschele (after Aug.10) Feel free to contact me by email first if you like to obtain more information about this position. Virginia Polytechnic Institute and State University provides a productive research environment with 11 animal geneticists (a mix of animal breeders, statistical geneticists and molecular geneticists) and a large Statistics department (offering also classes in statistical genetics and biostatistics), providing the possibility for interaction and auditing of classes. There is also interaction with a Human Genetics group in Richmond, VA, including Drs. Charles MacLean and Mike Neale. Finally, the computing facilities are excellent (departmental high-end, multi-processor workstation, computing center workstations, symmetric multi-processor and SP2 systems, and supercomputing access at the Cornell Theory Center). Ina Hoeschele Associate Professor of Quantitative Genetics From majordom@umnstat.stat.umn.edu Wed Jun 19 14:28:27 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA21029; Wed, 19 Jun 1996 14:28:27 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA21023; Wed, 19 Jun 1996 13:28:23 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa11656; 19 Jun 96 14:28 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA17322; Wed, 19 Jun 1996 14:28:16 -0400 Date: Wed, 19 Jun 1996 14:28:15 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: SWIG and the netCDF library interfaces (fwd) Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Has anyone in Xlisp-Stat land ever looked at this ? How hard would it be to mate to Xlisp ? ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] ---------- Forwarded message ---------- Date: Wed, 19 Jun 1996 11:57:10 +0700 From: Paul Hein Reply-To: netcdfgroup@unidata.ucar.edu To: netcdfgroup@unidata.ucar.edu Cc: hein@olympic.atmos.colostate.edu Subject: SWIG and the netCDF library interfaces With all the discussion about adding the Java interface and with Rob Managan mention about a Python interface. I thought I would bring up the tool SWIG which allows one to take C (and C++) code and interface with Tcl, Perl, Python, or Guile. I have not tried it yet but it seems like a very easy way to add netCDF to these scripting languages. And it might be extendable to Java. More information is available at http://www.cs.utah.edu/~beazley/SWIG/ It is something to consider. Below is part of the web page. Paul Paul F. Hein hein@olympic.atmos.colostate.edu (970) 491-8531 SWIG Simplified Wrapper and Interface Generator Introduction SWIG is an easy to use tool that lets you add Tcl, Perl, Python, or Guile to programs written in C and C++. Normally, this requires you to write special "wrapper" functions---a process that is both tedious and error-prone. SWIG freesyou from this task and allows you to focus on the application at hand while providing all of the benefits of using an interpreted scripting language interface. A brief overview of features Supports Tcl, Perl4, Perl5, Python, and Guile-3. Allows you to add C/C++ functions, variables, and constants. Handles almost all C datatypes and pointers. Run-time type checking of pointers. Automatic documentation generation in ASCII, LaTeX, or HTML. Extends common applications including tclsh, wish, and expect. Can be extended to support new target languages. From majordom@umnstat.stat.umn.edu Wed Jun 26 05:46:02 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27283; Wed, 26 Jun 1996 05:46:02 -0400 Return-Path: Received: from mailb.surrey.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27278; Wed, 26 Jun 1996 04:45:58 -0500 Received: from soc.surrey.ac.uk (actually host boris-gate.soc.surrey.ac.uk) by mailb.surrey.ac.uk with SMTP (PP); Wed, 26 Jun 1996 10:44:46 +0100 Received: from socmac32 by boris.soc.surrey.ac.uk via Ethernet with SMTP id aa25015; 26 Jun 96 10:46 BST X-Sender: scs1ng@boris.soc.surrey.ac.uk Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-Url: http://www.soc.surrey.ac.uk/ X-Url: http://www.socresonline.org.uk/socresonline/ Date: Wed, 26 Jun 1996 10:45:37 +0100 To: stat-lisp-news@umnstat.stat.umn.edu From: Nigel Gilbert Subject: Problem installing xlispstat 3.47 on Solaris 5.4 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I am trying to install xlispstat 3.47 on a Sun under OpenWindows (`uname -a` yields: SunOS kennedy 5.4 Generic_101945-32 sun4m sparc). Configure runs without apparent error. The Makefile compiles and loads without errors until it comes to loading xlisp to compile backquote, when it dumps core with a segmentation fault. The final few lines from a log of the make are: echo "(compile-file \"compiler/backquot.lsp\") (exit)" | ./xlisp XLISP-PLUS version 2.1g Portions Copyright (c) 1988, by David Betz. Modified by Thomas Almy and others. XLISP-STAT 2.1 Release 3.47 (Beta). Copyright (c) 1989-1994, by Luke Tierney. ; loading "init.lsp" ; loading "common.lsp" ; loading "common2.lsp" ; loading "common3.lsp" *** Error code 139 make: Fatal error: Command failed for target `compiler/backquot.fsl' Can anyone provide some clues about the source of the problem? Has anyone else succeeded with an installation under Solaris? Many thanks for any help. _________________________________________________________________________ Prof G. Nigel Gilbert, Dept of Sociology, University of Surrey, Guildford, GU2 5XH, United Kingdom. Tel: +44 (0)1483 259173 Fax: +44 (0)1483 306290 From majordom@nokomis.stat.umn.edu Wed Jun 26 14:17:09 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA25332; Wed, 26 Jun 1996 14:17:09 -0400 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA25326; Wed, 26 Jun 1996 13:16:58 -0500 From: Luke Tierney Message-Id: <9606261816.AA25326@nokomis.stat.umn.edu> Subject: Re: Problem installing xlispstat 3.47 on Solaris 5.4 To: Nigel.Gilbert@soc.surrey.ac.uk (Nigel Gilbert) Date: Wed, 26 Jun 1996 13:16:58 -0500 (CDT) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Nigel Gilbert" at Jun 26, 96 10:45:37 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1820 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > I am trying to install xlispstat 3.47 on a Sun under OpenWindows (`uname > -a` yields: SunOS kennedy 5.4 Generic_101945-32 sun4m sparc). Configure > runs without apparent error. The Makefile compiles and loads without > errors until it comes to loading xlisp to compile backquote, when it dumps > core with a segmentation fault. The final few lines from a log of the make > are: > > echo "(compile-file \"compiler/backquot.lsp\") (exit)" | ./xlisp > XLISP-PLUS version 2.1g > Portions Copyright (c) 1988, by David Betz. > Modified by Thomas Almy and others. > XLISP-STAT 2.1 Release 3.47 (Beta). > Copyright (c) 1989-1994, by Luke Tierney. > > ; loading "init.lsp" > ; loading "common.lsp" > ; loading "common2.lsp" > ; loading "common3.lsp" > *** Error code 139 > make: Fatal error: Command failed for target `compiler/backquot.fsl' > > Can anyone provide some clues about the source of the problem? Has anyone > else succeeded with an installation under Solaris? > > Many thanks for any help. > > _________________________________________________________________________ > Prof G. Nigel Gilbert, Dept of Sociology, University of Surrey, Guildford, > GU2 5XH, United Kingdom. Tel: +44 (0)1483 259173 Fax: +44 (0)1483 306290 > > > 3.46 compiles fine on Solaris 5.3. The problem is that there are a few places where setjmp/longjmp combinations don't fully protect local variables. Usually this only creates problems when code is optimized -- sun's optimizer probably got more aggressive with 5.4. I had fixed these problems in my working version (gcc tells you what the potential problems are) but there are too many other little changes to easily extract these changes. For now, the simplest solution is to compile the three affected files, xlbcode.c xlcont.c xlisp.c without optimization. luke From majordom@umnstat.stat.umn.edu Mon Jul 1 22:27:48 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13423; Mon, 1 Jul 1996 22:27:48 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13418; Mon, 1 Jul 1996 21:27:42 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa17970; 1 Jul 96 22:27 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA13480; Mon, 1 Jul 1996 22:27:39 -0400 Date: Mon, 1 Jul 1996 22:27:38 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU Reply-To: "Steven D. Majewski" To: Xlisp List Subject: directory functions and open-file-dialog Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'm using the code below ( on Macintosh Xlisp 3.47 ) to add common directories to the file menu. It works more or less as it should -- it adds directories to the file menu, and it sets them as the current working directory. If I try to load a filename from that directory, it works as it should. However, when I do an (open-file-dialog), it doesn't start up in the current working directory. Any idea why it doesn't or how it can be forced ? The whole reason for adding these shortcuts was to avoid having to climb in and out of directories with the open-file dialog!! ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] ;; Adds common default directories to File menu ;; ( defproto dir-menu-item-proto () '( dir-items ) menu-item-proto ) ( defmeth dir-menu-item-proto :mark ( &optional ( value nil set )) ( when set ( call-next-method value ) ( if value ( mapcar #'(lambda (x) (send x :mark nil)) ( remove self ( slot-value 'dir-items))))) (call-next-method)) #+macintosh ( defconstant dir-sep ":" ) #-macintosh ( defconstant dir-sep "/" ) ( defun short-dir-name ( path ) ( let (( dname ( pathname-directory ( pathname path )))) ( if ( > ( length dname ) 3 ) ( concatenate 'string "..." dir-sep ( elt dname (- (length dname) 2)) dir-sep ( elt dname (1- (length dname)))) (make-pathname :directory dname)))) ( defun dir-item ( dname ) ( send *FILE-MENU* :find-item ( short-dir-name dname ))) ( defmeth dir-menu-item-proto :isnew ( dirname ) ( if ( dir-item dirname ) ( error "Directory: ~S already installed~%" dirname )) ( let (( name ( short-name dirname ))) ( call-next-method name :action #'(lambda ( &rest args ) ( set-working-directory dirname ) ( send ( dir-item dirname ) :mark T ) (format *standard-output* "SET WORKING DIRECTORY:\n ~S ~\n\n" (get-working-directory))) :style 'italic ) ( send dir-menu-item-proto :slot-value 'dir-items ( cons self (slot-value 'dir-items))) ( send (find-menu "File") :append-items self ) ( send self :mark Nil ))) ( defun add-dir ( &optional dirname ) ( if ( null dirname ) ( setf dirname ( get-working-directory ))) ( send dir-menu-item-proto :new dirname )) ; ; WARNING -- This is hard coded to keep the original 10 ; menu items -- it is mostly needed for testing and reloading ; these definitions. You can remove or replace. ( defun reset-file-menu () (apply #'send *file-menu* :delete-items ( subseq ( send *file-menu* :items ) 11 ))) ( reset-file-menu ) ( send *FILE-MENU* :append-items ( send dash-item-proto :new )) ( send *FILE-MENU* :append-items ( send menu-item-proto :new "add directory" :action #'(lambda () ( let (( file ( set-file-dialog "Choose directory" ))) ( when file (add-dir)))))) ( send *FILE-MENU* :append-items ( send dash-item-proto :new )) From majordom@nokomis.stat.umn.edu Tue Jul 2 06:59:09 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA20201; Tue, 2 Jul 1996 06:59:09 -0400 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA20195; Tue, 2 Jul 1996 05:59:00 -0500 From: Luke Tierney Message-Id: <9607021059.AA20195@nokomis.stat.umn.edu> Subject: Re: directory functions and open-file-dialog To: sdm7g@Virginia.EDU Date: Tue, 2 Jul 1996 05:59:00 -0500 (CDT) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Steven D. Majewski" at Jul 1, 96 10:27:38 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 3813 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > > I'm using the code below ( on Macintosh Xlisp 3.47 ) to add common > directories to the file menu. It works more or less as it should > -- it adds directories to the file menu, and it sets them as > the current working directory. If I try to load a filename from > that directory, it works as it should. However, when I do an > (open-file-dialog), it doesn't start up in the current working > directory. Any idea why it doesn't or how it can be forced ? > The whole reason for adding these shortcuts was to avoid having > to climb in and out of directories with the open-file dialog!! > > > ---| Steven D. Majewski (804-982-0831) |--- > ---| Computer Systems Engineer University of Virginia |--- > ---| Department of Molecular Physiology and Biological Physics |--- > ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- > [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] > > ;; Adds common default directories to File menu > ;; > > ( defproto dir-menu-item-proto () '( dir-items ) menu-item-proto ) > > ( defmeth dir-menu-item-proto :mark ( &optional ( value nil set )) > ( when set > ( call-next-method value ) > ( if value > ( mapcar #'(lambda (x) (send x :mark nil)) > ( remove self ( slot-value 'dir-items))))) > (call-next-method)) > > #+macintosh ( defconstant dir-sep ":" ) > #-macintosh ( defconstant dir-sep "/" ) > > ( defun short-dir-name ( path ) > ( let (( dname ( pathname-directory ( pathname path )))) > ( if ( > ( length dname ) 3 ) > ( concatenate 'string > "..." > dir-sep > ( elt dname (- (length dname) 2)) > dir-sep > ( elt dname (1- (length dname)))) > (make-pathname :directory dname)))) > > ( defun dir-item ( dname ) > ( send *FILE-MENU* :find-item ( short-dir-name dname ))) > > > ( defmeth dir-menu-item-proto :isnew ( dirname ) > ( if ( dir-item dirname ) > ( error "Directory: ~S already installed~%" dirname )) > ( let (( name ( short-name dirname ))) > ( call-next-method name > :action > #'(lambda ( &rest args ) > ( set-working-directory dirname ) > ( send > ( dir-item dirname ) > :mark T ) > (format *standard-output* > "SET WORKING DIRECTORY:\n ~S ~\n\n" > (get-working-directory))) > :style 'italic > ) > ( send dir-menu-item-proto :slot-value 'dir-items > ( cons self (slot-value 'dir-items))) > ( send (find-menu "File") :append-items self ) > ( send self :mark Nil ))) > > > > > ( defun add-dir ( &optional dirname ) > ( if ( null dirname ) > ( setf dirname ( get-working-directory ))) > ( send dir-menu-item-proto :new dirname )) > > > ; > ; WARNING -- This is hard coded to keep the original 10 > ; menu items -- it is mostly needed for testing and reloading > ; these definitions. You can remove or replace. > > ( defun reset-file-menu () > (apply #'send *file-menu* :delete-items > ( subseq ( send *file-menu* :items ) 11 ))) > > ( reset-file-menu ) > > > ( send *FILE-MENU* :append-items ( send dash-item-proto :new )) > > ( send *FILE-MENU* :append-items > ( send menu-item-proto > :new "add directory" > :action #'(lambda () > ( let (( file ( set-file-dialog > "Choose directory" ))) > ( when file (add-dir)))))) > > ( send *FILE-MENU* :append-items ( send dash-item-proto :new )) > > > 3.47 uses the CodwWarrior chdir routine for setting the working directory. THis routine does not do what is needed to make the OpenFileDialog adjust. Messing with where OpenFile Dialog starts up is oe of those things the Mac doesn't make easy. The only way I know of doing it invilves messing with low memory globals, which is considered dangerous for portability. If I find a cleaner way to do it, I may make this change. luke From majordom@umnstat.stat.umn.edu Thu Jul 4 04:03:26 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA21718; Thu, 4 Jul 1996 04:03:26 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA21713; Thu, 4 Jul 1996 03:03:21 -0500 Received: from localhost by gibbs.oit.unc.edu (8.6.4.3/10.1) id EAA04849; Thu, 4 Jul 1996 04:03:14 -0400 Date: Thu, 4 Jul 1996 04:03:12 -0400 From: "Forrest W. Young" To: "Steven D. Majewski" Cc: Xlisp List Subject: Re: directory functions and open-file-dialog In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk To All Lispers: My experience is that on the Mac there is no solution to this problem. Directories seem to work right under MS-DOS and Unix. Not Macs. Forrest Young On Mon, 1 Jul 1996, Steven D. Majewski wrote: > Date: Mon, 1 Jul 1996 22:27:38 -0400 (EDT) > From: Steven D. Majewski > To: Xlisp List > Subject: directory functions and open-file-dialog > > > I'm using the code below ( on Macintosh Xlisp 3.47 ) to add common > directories to the file menu. It works more or less as it should > -- it adds directories to the file menu, and it sets them as > the current working directory. If I try to load a filename from > that directory, it works as it should. However, when I do an > (open-file-dialog), it doesn't start up in the current working > directory. Any idea why it doesn't or how it can be forced ? > The whole reason for adding these shortcuts was to avoid having > to climb in and out of directories with the open-file dialog!! > > > ---| Steven D. Majewski (804-982-0831) |--- > ---| Computer Systems Engineer University of Virginia |--- > ---| Department of Molecular Physiology and Biological Physics |--- > ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- > [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] > > ;; Adds common default directories to File menu > ;; > > ( defproto dir-menu-item-proto () '( dir-items ) menu-item-proto ) > > ( defmeth dir-menu-item-proto :mark ( &optional ( value nil set )) > ( when set > ( call-next-method value ) > ( if value > ( mapcar #'(lambda (x) (send x :mark nil)) > ( remove self ( slot-value 'dir-items))))) > (call-next-method)) > > #+macintosh ( defconstant dir-sep ":" ) > #-macintosh ( defconstant dir-sep "/" ) > > ( defun short-dir-name ( path ) > ( let (( dname ( pathname-directory ( pathname path )))) > ( if ( > ( length dname ) 3 ) > ( concatenate 'string > "..." > dir-sep > ( elt dname (- (length dname) 2)) > dir-sep > ( elt dname (1- (length dname)))) > (make-pathname :directory dname)))) > > ( defun dir-item ( dname ) > ( send *FILE-MENU* :find-item ( short-dir-name dname ))) > > > ( defmeth dir-menu-item-proto :isnew ( dirname ) > ( if ( dir-item dirname ) > ( error "Directory: ~S already installed~%" dirname )) > ( let (( name ( short-name dirname ))) > ( call-next-method name > :action > #'(lambda ( &rest args ) > ( set-working-directory dirname ) > ( send > ( dir-item dirname ) > :mark T ) > (format *standard-output* > "SET WORKING DIRECTORY:\n ~S ~\n\n" > (get-working-directory))) > :style 'italic > ) > ( send dir-menu-item-proto :slot-value 'dir-items > ( cons self (slot-value 'dir-items))) > ( send (find-menu "File") :append-items self ) > ( send self :mark Nil ))) > > > > > ( defun add-dir ( &optional dirname ) > ( if ( null dirname ) > ( setf dirname ( get-working-directory ))) > ( send dir-menu-item-proto :new dirname )) > > > ; > ; WARNING -- This is hard coded to keep the original 10 > ; menu items -- it is mostly needed for testing and reloading > ; these definitions. You can remove or replace. > > ( defun reset-file-menu () > (apply #'send *file-menu* :delete-items > ( subseq ( send *file-menu* :items ) 11 ))) > > ( reset-file-menu ) > > > ( send *FILE-MENU* :append-items ( send dash-item-proto :new )) > > ( send *FILE-MENU* :append-items > ( send menu-item-proto > :new "add directory" > :action #'(lambda () > ( let (( file ( set-file-dialog > "Choose directory" ))) > ( when file (add-dir)))))) > > ( send *FILE-MENU* :append-items ( send dash-item-proto :new )) > > > From majordom@nokomis.stat.umn.edu Thu Jul 11 15:09:48 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA16671; Thu, 11 Jul 1996 15:09:48 -0400 Date: Thu, 11 Jul 1996 15:09:48 -0400 From: Majordomo Owner Return-Path: Message-Id: <9607111909.AA16671@nokomis.stat.umn.edu> Apparently-To: stat-lisp-news-outgoing@nokomis.stat.umn.edu Some of you may know about tcl/tk, an interpreted language plus X11 interface toolkit from Berkeley, very popular. Ousterhout, the father of tcl/tk, now works for Sun, and Sun develops tcl. They just produced a nice plugin for Netscape, so that you can now run tiny tcl/tk programs in the browser window (like applets). If somebody would do the same thing for Xlisp-Stat, we would really be in business. Sender: owner-stat-lisp-news@stat.umn.edu Precedence: bulk See: http://www.sunlabs.com/research/tcl/plugin/index.html -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Wed Jul 24 10:04:15 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29503; Wed, 24 Jul 1996 10:04:15 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29494; Wed, 24 Jul 1996 09:04:10 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id KAA08259; Wed, 24 Jul 1996 10:03:57 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 24 Jul 1996 10:04:03 -0500 To: edstat-l@jse.stat.ncsu.edu, stat-l@vm1.mcgill.ca, stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: ViSta for Unix Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Announcing a Unix Version of ViSta - The Visual Statistics System Version 4.22a, July 22nd, 1996 Copyright (c) 1991-96 by Forrest W. Young & Anthony Rossini ViSta is now available for Unix! ViSta for Unix requires X11R5 or higher (it has been tested on X11R6 and X11R6.1), and a recent version of XLispStat (version 3.44 or higher). This version has been developed and tested on a IBM PC-compatible notebook (486DX4-100) running Linux, a Sun SPARCStation 20/50 running Solaris 2.4 and 2.5, as well as Sun IPC and IPX workstations running SunOS 4.1.3. Version 4.22 of ViSta for MS-Windows and the Macintosh has been previously announced and is currently available from the server described below. The complete code for ViSta, including documentation, test data, examples, help, guidance, and the XLisp-Stat base, is available from our WWW and FTP servers. WWW: http://forrest.psych.unc.edu/ FTP from www.psych.unc.edu, directory pub/forrest/vista. The remainder of this message summarizes earlier messages about ViSta. ======================================= What is ViSta? ======================================= ViSta, is a Visual Statistics system that > performs univariate and multivariate statistical data analysis and >visualization > is open, extensible, and freely redistributable PLATFORMS: ViSta is available for > Microsoft Windows 3.1 > Macintosh and PowerMac microcomputers > Unix with X11R5 or higher AUDIENCE: ViSta is useful for > UNDERGRADUATES: Teaching introductory and multivariate statistics > GRADUATES: Teaching multivariate, graphical and computational statistics > RESEARCHERS: Research and development in dynamic statistical graphics and >in computational statistics ViSta is designed for an audience of users having a very wide range of data analysis sophistication, ranging from novice to expert. ViSta provides seamlessly integrated data analysis environments specifically tailored to the user's level of expertise. Guidance is available for students and novices, and tools are available for teachers and other experts to create guidance for these novices. A structured graphical user interface is available for competent users, and a command line interface is available for sophisticated users. The complete Lisp-Stat (Tierney, 1990) programming environment is available to researchers, programmers and graduate students who wish to extend ViSta's capabilities. ViSta is "freely redistributable". This means that it is not in the public domain, but that it may be freely copied and redistributed, with certain restrictions. See the COPYING and COPYVSTA files included in the distribution for details. ======================================= ViSta's main features ======================================= > WORKMAPS Maps that visualize the overall structure of an on-going data analysis session. These maps are created by ViSta as the data analysis session progresses. They aid novice and competent analysts in remembering their analysis session, and permit the analyst to return to earlier steps of the analysis. > GUIDEMAPS Maps that visually guide students and other novice data analysts through their data analysis sessions. These maps are created by experts using ViSta's Authoring tools. > SPREADPLOTS Multi-window dynamic statistical graphics that visualize data structure and analysis results > MODEL RE-VISION Dynamic interactive graphical tools that permit the user to visually re-estimate model parameters > ViDAL ViSta's Data Analysis Language. ViDAL is an extension of the underlying Lisp-Stat language that can be used by sophisticated data analysts for data analysis and by programmers to enhance ViSta's capabilities. > DATASHEET A graphical datasheet for displaying and editing data. > AUTHORING TOOLS Visual tools for authoring new GuideMaps. ======================================= Data Analysis Capabilities ======================================= ViSta's data analysis capabilities include: > DYNAMIC GRAPHICS Histograms, Boxplots, Diamond Plots, Dotplots, Scatterplots, Biplots, Spinplots, Scatterplot Matrices. Plots support brushing and labeling, and are dynamically linked. > DESCRIPTIVE STATISTICS Means, Standard Deviations, Variances, Ranges, Quartiles, Medians, Correlations, Covariances, Distances >UNIVARIATE ANALYSIS T- and Z-tests and confidence intervals for single sample, paired samples and two independent samples data, with Wilcoxon Signed-Rank and Mann-Whitney tests in appropriate situations. > ANOVA Univariate analysis of variance of orthogonal one- or multi-way data. Model may or may not include two-way interactions. The model visualization is a spreadplot comprised of boxplot, diamond plot, quantile plot, quantile-quantile plot and effects plot. > REGRESSION Simple and multiple OLS regression analysis (multivariate, robust and monotonic regression in preparation). The model visualization is a spreadplot comprised of a biplot, spinplot, histogram and scatterplot-matrix. > PRINCIPAL COMPONENTS Analysis of correlations or covariances. The model visualization is a spreadplot comprised of a biplot, spin-plot, scree-plot and scatterplot-matrix. > MULTIDIMENSIONAL SCALING Analysis of one or more symmetric or asymmetric matrices. The model visualization is a spreadplot comprised of a scatterplot, spin-plot, scree-plot and scatterplot-matrix. The spreadplot supports graphical re-estimation of model parameters. > CORRESPONDENCE ANALYSIS Analysis of two-way contingency tables. The model visualization is a spreadplot comprised of a biplot, spinplot, residuals plot and scree-plot. The spreadplot supports graphical re-estimation of model parameters. ======================================= Hardware Requirements ======================================= MACINTOSH: There are two versions of ViSta for the Macintosh, one for the PowerMac and one for 680xx Macintoshes. For 680xx Macs, we recommend a 68040 processor (at the minimum, a 68030 + 68881 math coprocessor). Both versions require System 7 or later, 3 MB of free hard disk space and at least 6 MB of RAM (8 for the PowerMac). We recommend at least a 17", 8-bit (256 color) screen (14" minimum). For 12-bit or 24-bit screens, a larger memory partition is required (add 1/2 MB for 12-bit, 1 MB for 24-bit). WINDOWS: The Windows version of ViSta is for MS-Windows 3.1 with the 32-bit extension. Users report that this version of ViSta also works under Windows 95, though no testing has been done. This version requires 3 MB of free hard disk space and 8 MB of RAM. We recommend at least a 17", 8-bit (256 color) screen (14" minimum). For 12-bit or 24-bit screens, more memory is required (about 1/2 MB for 12-bit, 1 MB for 24-bit). UNIX: ViSta for Unix requires X11R5 or higher (it has been tested on X11R6 and X11R6.1), and a recent version of XLispStat (version 3.44 or higher). Thus, if you don't have XLispStat you must download it first, then download ViSta. The version has been developed and tested on a IBM PC-compatible notebook (486DX4-100) running Linux, a Sun SPARCStation 20/50 running Solaris 2.4 and 2.5, as well as Sun IPC and IPX workstations running SunOS 4.1.3. ======================================= ViSta is a work-in-progress ======================================= ViSta is a work-in-progress. Some features are incomplete. While quality assurance testing has been done, this is a beta-release system that is not fully mature. At the current stage of development, ViSta is viable as a supplemental statistical software system used for extra credit exercises in introductory statistics, and as the focus of graduate courses on research in computational and graphical methods in statistics. It may be premature to use ViSta in undergraduate statistics classes that need a complete and robust statistic analysis system. Currently, the Delete Data, Delete Model and Simulate Data menu items are not implemented; the ANOVA visualization is incomplete; the univariate analysis visualization is not available; MULREG does not do multivariate tests (it does many separate multiple regressions); transformations are not available with matrix data; model interpretation is unimplemented; and the Help and GuideMap systems are incomplete. A univariate multiple regression module (which complements the current multivariate multiple regression module with more extensive visualizations and with the choice of robust or monotonic regression) is being written but is not yet available. ======================================= Documentation ======================================= The following two monographs describe how to use ViSta: "ViSta, The Visual Statistics System", is available from the server mentioned above. A nicely bound and printed version is also available for $5.00 from the address below. "ViSta Corresp", a description of the Correspondence Analysis procedure in ViSta, is available for $5.00 from Forrest Young at the address below. The following paper has been recently published: Young, F.W. & Lubinsky, D.J. "Guiding Data Analysis with Visual Statistical Strategies". J. Computational and Graphical Statistics, Dec., 1995. In addition, the following paper is available for $5.00 from Forrest Young at the address below. Young, F.W., et al. "Visual Statistical Analysis - Papers presented at the Special Contributed Session on Visual Statistical Analyais at Interface '94. Psychometric Research Memorandum 94-2, July, 1994. Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Mon Jul 29 18:35:17 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13790; Mon, 29 Jul 1996 18:35:17 -0400 Return-Path: Received: from mailb.surrey.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13785; Mon, 29 Jul 1996 17:35:12 -0500 Received: from [131.227.133.26] (actually host annex1-16.surrey.ac.uk) by mailb.surrey.ac.uk with SMTP (PP) with ESMTP; Mon, 29 Jul 1996 23:33:57 +0100 X-Sender: scs1ng@boris.soc.surrey.ac.uk (Unverified) Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-Url: http://www.soc.surrey.ac.uk/ X-Url: http://www.socresonline.org.uk/socresonline/ Date: Mon, 29 Jul 1996 22:19:15 +0000 To: stat-lisp-news@umnstat.stat.umn.edu From: Nigel Gilbert Subject: Can't change cursor Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I have some code roughly like this (where 'w' is an instance of graph-window-proto) : (send w :cursor 'watch) .... some computation... (send w :cursor 'arrow) The intention is that while the computation is proceeding, the user is reminded to wait by the watch cursor. However, the cursor does not change. It only changes when I abort the computation using command-. This is so even if 'w' is the front (active) window. I also have a menu attached to 'w' which contains a menu item, 'Pause' which sets a flag. This flag is checked during the computation and if found to be true, the computation stops. This works. However, the menu item also has an associated key, command-P, and pressing this does not have any effect. What am I doing wrong? ____________________________________________________________________ Prof G. Nigel Gilbert, Department of Sociology, University of Surrey Guildford GU2 5XH, UK. Tel: +44 1483 259173 Fax: +44 1483 306290 From majordom@umnstat.stat.umn.edu Mon Jul 29 20:23:29 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14147; Mon, 29 Jul 1996 20:23:29 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14142; Mon, 29 Jul 1996 19:23:24 -0500 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.6.4.3/10.1) id UAA11052; Mon, 29 Jul 1996 20:23:15 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 29 Jul 1996 20:23:36 -0500 To: Nigel Gilbert , stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: Re: Can't change cursor Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Nigel Gilbert writes: >I have some code roughly like this (where 'w' is an instance of >graph-window-proto) : > > (send w :cursor 'watch) > .... some computation... > (send w :cursor 'arrow) > >The intention is that while the computation is proceeding, the user is >reminded to wait by the watch cursor. However, the cursor does not change. >It only changes when I abort the computation using command-. This is so >even if 'w' is the front (active) window. > >I also have a menu attached to 'w' which contains a menu item, 'Pause' >which sets a flag. This flag is checked during the computation and if >found to be true, the computation stops. This works. However, the menu >item also has an associated key, command-P, and pressing this does not >have any effect. > >What am I doing wrong? > >____________________________________________________________________ >Prof G. Nigel Gilbert, Department of Sociology, University of Surrey > Guildford GU2 5XH, UK. Tel: +44 1483 259173 Fax: +44 1483 306290 I recall having the identical problem a number of years ago. I decided it was a bug and gave up! Forrest Young Professor, Psychometrics CB# 3270 Davie Hall email: mailto://forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software at: http://forrest.psych.unc.edu/research/ViSta.html From majordom@nokomis.stat.umn.edu Thu Aug 1 15:11:34 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA28239; Thu, 1 Aug 1996 15:11:34 -0400 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA28233; Thu, 1 Aug 1996 14:11:25 -0500 From: Luke Tierney Message-Id: <9608011911.AA28233@nokomis.stat.umn.edu> Subject: Re: Can't change cursor To: Nigel.Gilbert@soc.surrey.ac.uk (Nigel Gilbert) Date: Thu, 1 Aug 1996 14:11:25 -0500 (CDT) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Nigel Gilbert" at Jul 29, 96 10:19:15 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1522 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > I have some code roughly like this (where 'w' is an instance of > graph-window-proto) : > > (send w :cursor 'watch) > .... some computation... > (send w :cursor 'arrow) > > The intention is that while the computation is proceeding, the user is > reminded to wait by the watch cursor. However, the cursor does not change. > It only changes when I abort the computation using command-. This is so > even if 'w' is the front (active) window. > > I also have a menu attached to 'w' which contains a menu item, 'Pause' > which sets a flag. This flag is checked during the computation and if > found to be true, the computation stops. This works. However, the menu > item also has an associated key, command-P, and pressing this does not > have any effect. > > What am I doing wrong? > > ____________________________________________________________________ > Prof G. Nigel Gilbert, Department of Sociology, University of Surrey > Guildford GU2 5XH, UK. Tel: +44 1483 259173 Fax: +44 1483 306290 > > > The windows and mac versions make an effort to check for a user interrupt and to give other applications a chance to run during long computations. Whether or not anything else happens is not predictable (other keystrokes, changing cursors, etc.). The current solution is a hack. To properly support doing anything useful while at the same time doing a long computation meas supporting doing two separate thigs at once, i.e. some form of proper multitasking. THis is on my list of things to add. luke From majordom@umnstat.stat.umn.edu Tue Aug 6 12:20:17 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03645; Tue, 6 Aug 1996 12:20:17 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03640; Tue, 6 Aug 1996 11:20:11 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id MAA20839; Tue, 6 Aug 1996 12:19:59 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Tue, 6 Aug 1996 12:19:25 -0500 To: edstat-l@jse.stat.ncsu.edu, stat-l@vm1.mcgill.ca, stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: ViSta for Unix Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Announcing a slightly revised release for the Unix Version of ViSta - The Visual Statistics System (Version 4.22a, July 22nd, 1996). The examples in the examples directory of the recent release of ViSta for Unix were not properly converted from Mac format to Unix format, resulting in problems in using them. This formatting problem has been fixed. The revised version is now available from the ViSta server. Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Mon Aug 12 12:58:00 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24628; Mon, 12 Aug 1996 12:58:00 -0400 Return-Path: Received: from sulu.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24623; Mon, 12 Aug 1996 11:57:55 -0500 Received: from chekov by sulu.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id JAA07046; Mon, 12 Aug 1996 09:57:51 -0700 (PDT) Date: Mon, 12 Aug 1996 09:57:49 -0700 (PDT) From: Thomas Lumley X-Sender: thomas@chekov To: LispStat List Subject: cox regression Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk A preliminary version of a cox regression package is available from http://www.biostat.washington.edu/~thomas/cox.html It fits stratified proportional hazards models with left truncation and right censoring using either the Breslow or Efron approximations for tied failures. Only very limited diagnostics and graphics are as yet available. Some testing has been done against S-PLUS and SPIDA, but it's still alpha code. thomas lumley Biostatistics U Washington. From majordom@umnstat.stat.umn.edu Wed Aug 14 10:19:29 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22528; Wed, 14 Aug 1996 10:19:29 -0400 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22522; Wed, 14 Aug 1996 09:19:14 -0500 Received: by kodakr.kodak.com id AA08465 (5.67b/IDA-1.5 for stat-lisp-news@stat.umn.edu); Wed, 14 Aug 1996 10:16:31 -0400 Date: Wed, 14 Aug 1996 10:16:31 -0400 From: morris@Kodak.COM Message-Id: <199608141416.AA08465@kodakr.kodak.com> Subject: Mult. Comparisons Apparently-To: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Does anyone know of any code for performing multiple comparisons? I am particularly interested in Dunnett's test. Any guidance would be much appreciated. Jeff Morris morris@kodak.com From majordom@umnstat.stat.umn.edu Fri Aug 16 17:41:02 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23661; Fri, 16 Aug 1996 17:41:02 -0400 Return-Path: Received: from cda.mrs.umn.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23656; Fri, 16 Aug 1996 16:40:58 -0500 Received: by cda.mrs.umn.edu; (5.65v3.2/1.1.8.2/15Apr96-0317PM) id AA17025; Fri, 16 Aug 1996 16:40:58 -0500 From: Jon E. Anderson Message-Id: <9608162140.AA17025@cda.mrs.umn.edu> Subject: Access to OS To: stat-lisp-news@umnstat.stat.umn.edu Date: Fri, 16 Aug 1996 16:40:57 -0500 (CDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 403 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Can anyone give me some advice/information on how to invoke other applications from within XLISP-STAT on Macs or pc's? I know how to access system shells from within XLISP-STAT in a UNIX environment to run other processes, but do these other platforms have the ability to perform tasks (like multimedia etc) from within XLISP-STAT? Just curious. Jon Anderson University of Minnesota, Morris From majordom@umnstat.stat.umn.edu Fri Aug 16 18:18:21 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23818; Fri, 16 Aug 1996 18:18:21 -0400 Return-Path: Received: from gateway.ets.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23813; Fri, 16 Aug 1996 17:18:16 -0500 Received: from research.ets.org by gateway.ets.org with SMTP id AA28074 (5.65c/IDA-1.4.4 for ); Fri, 16 Aug 1996 18:18:13 -0400 Received: from comet.research.ets.org by research.ets.org with SMTP id AA12275 (5.67b/IDA-1.4.3 for anderson@cda.mrs.umn.edu); Fri, 16 Aug 1996 18:18:12 -0400 Received: by comet.research.ets.org (SMI-8.6/SMI-SVR4) id SAA20253; Fri, 16 Aug 1996 18:18:18 -0400 Date: Fri, 16 Aug 1996 18:18:18 -0400 Message-Id: <199608162218.SAA20253@comet.research.ets.org> From: Russell Almond To: anderson@cda.mrs.umn.edu Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <9608162140.AA17025@cda.mrs.umn.edu> (anderson@cda.mrs.umn.edu) Subject: Re: Access to OS Reply-To: ralmond@ets.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Can anyone give me some advice/information on how to invoke other applications from within XLISP-STAT on Macs or pc's? I know how to access system shells from within XLISP-STAT in a UNIX environment to run other processes, but do these other platforms have the ability to perform tasks (like multimedia etc) from within XLISP-STAT? Just curious. Jon Anderson University of Minnesota, Morris On the Mac side, there is a mechanism called Apple Events which allows processes to send messages to each other. Standard messages are start, stop, open file and print file, but specific applications can define their own. XLISP-STAT support for apple events is described in the release notes (at least for the Mac version), but you will probably need to read the chapter on apple events in Inside Macintosh to understand what is going on. There is a newer technology for this sort of thing called "OpenDoc" currently in a beta release, but as far as I know, XLISP-STAT is neather an OpenDoc container or component. On the Windows side, I think XLISP-STAT supports the OLE protocols, but I don't do Windows so I can't tell you any more. Check the release notes. --Russell Almond Educational Testing Service Research Statistics Group, 15-T Princeton, NJ 08541 Phone: 609-734-1557 FAX: 609-734-5420 Email: almond@acm.org, ralmond@ets.org http://bayes.stat.washington.edu/almond/almond.html From majordom@umnstat.stat.umn.edu Mon Aug 26 14:00:56 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA08756; Mon, 26 Aug 1996 14:00:56 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA08751; Mon, 26 Aug 1996 13:00:51 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA28573; Mon, 26 Aug 96 11:00:49 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA20055; Mon, 26 Aug 1996 11:00:47 -0700 From: Jan Deleeuw Message-Id: <9608261800.AA20055@galton.math.ucla.edu> Subject: Journal of Statistical Software To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Mon, 26 Aug 1996 11:00:46 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The Journal of Statistical Software is published electronically on the UCLA Statistics Webserver. The URL is http://www.stat.ucla.edu/journals/jss/ Access is free, and unlimited. Articles are peer-reviewed. Articles are both browsable (in HTML) and downloadable (in PS). Code is browsable and downloadable as well, as are the data used in the article. Articles and code can be upgraded by the author: we always keep the version accepted by the journal and a downloadable current version. Each article comes with a database in which readers can leave comments, and authors can leave responses. Some articles come with interactive demos. All these functions can be accessed from the articles homepage. JSS maintains a chatroom on the WWW, used for discussion of statistical software. There is also an archive of documented and reviewed data, basically a subsection of the journal. These functions can be accessed from the journals homepage. There are currently two articles published, and a number under review. The published ones are Issue 01: A Diagnostic to Assess the Fit of a Variogram Model to Spatial Data Ronald Barry, Department of Mathematical Sciences, University of Alaska Fairbanks. ffrpb@aurora.alaska.edu Issue 02: Homogeneity Analysis in Xlisp-Stat Jason Bond and George Michailides, Program in Statistics, University of California Los Angeles. jbond@stat.ucla.edu and gmichail@stat.ucla.edu Both articles come with databases for comments. The second article has an interactive demo. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Thu Aug 29 10:05:58 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23921; Thu, 29 Aug 1996 10:05:58 -0400 Return-Path: Received: from ganymede.stats.gla.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23916; Thu, 29 Aug 1996 09:05:51 -0500 Message-Id: <923.199608291403@ganymede.stats.gla.ac.uk> Date: Thu, 29 Aug 1996 15:03:39 +0100 (BST) From: Stuart G Young Received: (from stuart@localhost sender stuart) by ganymede.stats.gla.ac.uk (8.7.5/UK-2.2a/stats-sparc) id PAA00923; Thu, 29 Aug 1996 15:03:39 +0100 (BST) To: edstat-L@jse.stat.ncsu.edu, jmp-L@wubios.wustl.edu, data-desk@ieunet.ie, stat-lisp-news@umnstat.stat.umn.edu Subject: Dynamic Graphics Workshop, Glasgow, UK, Sep '96 X-Sun-Charset: ISO-8859-1 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Dynamic Graphics Workshop University of Glasgow Tuesday 24th September 1996, 10am-4.30pm Over a number of years the use of graphical presentations as part of data exploration and analysis has been steadily increasing. This has been made possible by the increasingly graphical nature of computer interfaces and is having the effect of making data analysis more interactive. An important point is that pictorial representations of data can help students to see how the data and the underlying problem relate to each other in a way that numerical summaries do not. This is particularly true of those for whom statistics is not their primary interest, but a tool of their main discipline. In short graphics help them to see what is going on more easily than numerical summaries, because most people think visually. The scope of static graphical presentations can be extended by a newer type of graphic called a dynamic graphic. Dynamic graphics are very similar to their static cousins except that certain aspects can be changed very rapidly without a complete redraw. Graphs can be built up sequentially with the process of building the graph being part of the teaching and learning exercise. The CTI Centre for Statistics is organising a workshop to introduce people to the possibilities offered by statistical packages that can perform interactive (or dynamic) graphics. The workshop will feature one free package, XlispStat (available for PCs, Macs and Unix machines), and two commercial packages, namely DataDesk (only available for Macs) and JMP (now available on both PCs and Macs). The workshop will involve a short introduction and 30-45 minute presentations on each of the packages, concentrating on their use in teaching. This will be followed by a hands-on session during which participants will be able to try some simple exercises devised by the presenters, and also to try out some other things for themselves. It will be up to the participants what they spend their practical sessions doing, but all the presenters will be on hand to help and to discuss use of the packages and interactive graphics in general. The registration fee for the workshop is £30 (£60 for those outwith the UK HE sector), which covers lunch. For a fee of £34 (£64 for non-UK HE), attendees will also receive a copy of our "Maths&Stats Guide to Software for Teaching". This book lists several hundred statistics packages, and includes descriptions and contact information. The combined workshop and book fee represents a saving of £2 on the normal Software Guide price. Please note that an administration fee of £5 will be levied for cancellations less than 14 days prior to the workshop. Completed registration forms and enquiries should be addressed to: Stuart G. Young CTI Statistics E-mail: ctistat@gla.ac.uk Department of Statistics Tel: 0141 330 4873 Mathematics Building Fax: 0141 330 4814 University of Glasgow GLASGOW G12 8QW ======================================================================== CTI STATISTICS REGISTRATION FORM Dynamic Graphics Workshop University of Glasgow Tuesday 24 September 1996 10.00 a.m. - 4.30 p.m. Name: ________________________________________ Address: ________________________________________ ________________________________________ ________________________________________ ________________________________________ Phone number: ________________________________________ E-mail address: ________________________________________ Will you require a car parking space? (delete as applicable) YES / NO If YES, please supply the registration number and model of car: ________________________________________ Lunch is included. Please state any special dietary requirements: ________________________________________ I wish to register for: WORKSHOP ONLY (£30) or WORKSHOP AND SOFTWARE GUIDE SPECIAL OFFER (£34) (delete as applicable) The fee is enclosed / please invoice. (delete as applicable) For those outwith the UK Higher Education sector, the fees are £60 and £64 respectively. IMPORTANT: Please make cheques payable to "University of Glasgow". ------------------------------------------------------------------------ Please duplicate this form for multiple use, if required. ======================================================================== -- Stuart G. Young CTI Statistics Project Manager Department of Statistics | CTI Statistics - http://www.stats.gla.ac.uk/cti/ University of Glasgow, UK | CTI Queries to - ctistat@gla.ac.uk stuart@stats.gla.ac.uk | Allstat - http://www.stats.gla.ac.uk/allstat/ From majordom@umnstat.stat.umn.edu Thu Aug 29 10:48:19 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24250; Thu, 29 Aug 1996 10:48:19 -0400 Return-Path: Received: from sp2nfs.power.uni-essen.de by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24244; Thu, 29 Aug 1996 09:48:14 -0500 Received: from [132.252.142.115] by sp2.power.uni-essen.de (AIX 4.1/UCB 5.64/4.03) id AA65166; Thu, 29 Aug 1996 16:48:21 +0200 Date: Thu, 29 Aug 1996 16:48:21 +0200 Message-Id: <9608291448.AA65166@sp2.power.uni-essen.de> X-Sender: tmi0m0@sp2.power.uni-essen.de X-Mailer: Windows Eudora Version 1.4.4 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: lisp-stat-news@umnstat.stat.umn.edu From: tmi0m0@spf1.power.uni-essen.de _ (Johannes Fa\_binder) Subject: installation problems XLISPSTAT on AIX system Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Could anyone help me with a piece of advice on installing XLispStat. In the compilation of xsivint.c, the compiler interprets the evaluations (w == nil) and (w != nil) in line 219 and 341 respectively, as operations between unsigned long and pointer to void, and thus considers them illegal. In this institute we run AIX on an RS6000 machine. The configure statement yields GRAPHSYS = X11WINDOWS X11INCDIR_FLAG = X11LIBDIR_FLAG = UCFLAGS = -O -D_BSD -qmaxmem=3000 -qlanglvl=ansi -Imachines/aix/dynload ULDFLAGS = -bE:xlisp.exp -bM:SRE EXTRALIBS = -lld EXTRAOBJS = dlfcn.o IEEE_FLAG = -DIEEEFP ANSI_FLAG = -DANSI FOREIGN_FILE = sysvr4-foreign.h FOREIGN_FLAG = -DFOREIGNCALL CC = cc Any piece of advice is greatly appreciated as the software looks promising to me. Thanks a lot Johannes Fassbinder From majordom@umnstat.stat.umn.edu Thu Aug 29 12:18:33 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25196; Thu, 29 Aug 1996 12:18:33 -0400 Return-Path: Received: from netcom21.netcom.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25190; Thu, 29 Aug 1996 11:18:27 -0500 Received: from localhost (mayer@localhost) by netcom21.netcom.com (8.6.13/Netcom) id JAA19448; Thu, 29 Aug 1996 09:17:48 -0700 Message-Id: <199608291617.JAA19448@netcom21.netcom.com> To: Stuart G Young Cc: edstat-L@jse.stat.ncsu.edu, jmp-L@wubios.wustl.edu, data-desk@ieunet.ie, stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Dynamic Graphics Workshop, Glasgow, UK, Sep '96 In-Reply-To: Your message of "Thu, 29 Aug 96 15:03:39 BST." <923.199608291403@ganymede.stats.gla.ac.uk> Date: Thu, 29 Aug 96 09:17:42 -0700 From: "Niels P. Mayer" Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >The scope of static graphical presentations can be extended by a newer >type of graphic called a dynamic graphic. Dynamic graphics are very >similar to their static cousins except that certain aspects can be >changed very rapidly without a complete redraw. By the definition above, WINTERP also supports dynamic graphics, via a nice object oriented wrapper to the Xtango high-level graphics/animation library. See http://www.eit.com/software/winterp for details. The current release is a bit old, as I am not permitted to work on "free software" at my current job. I do have a new release in the works, but I will not be able to release it while continuing employment at EIT/Verifone. I hope to remedy this soon by not working at this job sometime in the near future (any pointers, especially on positions outside the USA would be appreciated!). Here's some further info: -------------------- WINTERP: An object-oriented rapid prototyping, development and delivery environment for building extensible applications with the OSF/Motif UI Toolkit and Xtango-based graphics/animation. by Niels Mayer WINTERP Version 2.03 July 24, 1994 ------------------------------------------------------------------------------ For information on the latest version of WINTERP, and other information (under construction), see the World-Wide-Web/Mosaic "WINTERP Home Page": URL -- http://www.eit.com/software/winterp/winterp.html ------------------------------------------------------------------------------ WINTERP is a Widget INTERPreter, an application development environment enabling rapid prototyping of graphical user-interfaces (GUI) through the interactive programmatic manipulation of user interface objects and their attached actions. The interpreter, based on David Betz, Tom Almy, et al's XLISP-PLUS, provides an interface to the X11 toolkit Intrinsics, the OSF/Motif widget set, primitives for collecting data from UN*X processes, and facilities for interacting with other UN*X processes. WINTERP thus supports rapid prototyping of GUI-based applications by allowing the user to interactively change both the UI appearance and application functionality. These features make WINTERP a good tool for learning and experimenting with the capabilities of the OSF/Motif UI toolkit, allowing UI designers to more easily play "what if" games with different interface styles. WINTERP is also an excellent platform for delivering extensible or customizable applications. By embedding a small, efficient language interpreter with UI primitives within the delivered application, users and system integrators can tailor the static and dynamic layout of the UI, UI-to-application dialogue, and application functionality. WINTERP's use of a real programming language for customization allows WINTERP-based applications to be much more flexible than applications using customization schemes provided by the X resource database, OSF/Motif's UIL (user interface language), and resource-based languages such as WCL. An environment similar to WINTERP's already exists in the Gnu-Emacs text editor -- WINTERP was strongly influenced by Gnu-Emacs' successful design. In Gnu-Emacs, a mini-Lisp interpreter is used to extend the editor to provide text-browser style interfaces to a number of UN*X applications (e.g. e-mail user agents, directory browsers, debuggers, etc). Whereas Emacs-Lisp enables the creation of new applications by tying together C-implemented primitives operating on text-buffer UI objects, WINTERP-Lisp ties together operations on graphical UI objects implemented by the Motif widgets. Both achieve a high degree of customizability that is common for systems implemented in Lisp, while still attaining the speed of execution and (relatively) small size associated with C-implemented applications. Other features: * WINTERP is free software -- available via anonymous ftp from ftp.x.org. * Portable -- runs without porting on many Unix systems. * Interface to GNU Emacs' lisp-mode allows code to be developed and tested without leaving the editor. * Interactive programming also available in the "WINTERP Control Panel", with editing taking place in a Motif text widget controlled by WINTERP. * Built-in RPC mechanism for inter-application communications. * XLISP provides a simple Smalltalk-like object system. * OSF/Motif widgets are real XLISP objects -- widgets can be specialized via subclassing, methods added or altered, etc. * Automatic storage management of all Motif/Xt/X data. * Contains facilities for "direct manipulation" of UI components. The WINTERP 2.0 release includes the following new features: * High-level, object-oriented 2.5D graphics/animation package featuring pixel-independent/resizable/scalable rendering, path-transition animation, and high-level animation operators. This is based on Stasko&Hayes' Xtango system. * Using Xtango, the ability to easily create new widget classes employing arbitrary graphical behavior without the tedium of programming in the Xt instrinsics and Xlib. * Asynchronous subprocess facility enables non-blocking GUI interfaces to existing interactive unix commands and interactive network services. Enables use of multiple asynchronous subprocesses. * XmGraph widget for creating directed acyclic graphs, trees, and direct-manipulation displays. * Table widget allows constraint-based GUI static layout using tbl(1)-style specifications. * Uses XLISP-PLUS version 2.1c -- more functionality, better subset of Common Lisp, e.g. good support for sequence operators. * Support for OSF/Motif 1.2.* * GIF Image Support. * Many new examples, improved example WINTERP applications. * Major code cleanups, bug fixes, etc. You may obtain the latest released version of the WINTERP source, documentation, and examples via anonymous ftp from internet host ftp.x.org in directory /contrib/devel_tools, file winterp-2.xx.tar.gz, where 'xx' represents the revision number. Directory /contrib/devel_tools/winterp contains slides, papers, further documentation, contributed code, etc. An automated reply indicating the locations and status of the latest WINTERP source may be obtained by sending an e-mail message to winterp-source@netcom.com. winterp@netcom.com is the mailing list for WINTERP-related announcements and discussions. To get added/removed from the list, send mail to winterp-request@netcom.com. Please do not send junk mail or subscribe/ unsubscribe messages to the mailing list winterp@netcom.com. For discussions about XLISP, see the USENET newsgroup comp.lang.lisp.x. ------------------------------------------------------------------------------ AUTHOR: Niels P. Mayer Enterprise Integration Technologies 800 El Camino Real, Fourth Floor Menlo Park, CA 94025 mayer@eit.com or mayer@netcom.com URL: http://www.eit.com/~mayer ------------------------------------------------------------------------------ DEPENDENCIES and PLATFORMS: WINTERP is portable software, and should run on any machine that supports Motif versions 1.0, 1.1, or 1.2 and Berkeley sockets. WINTERP will not build directly from the X11r6 distribution. YOU MUST HAVE MOTIF INSTALLED ON YOUR SYSTEM. WINTERP 2.0 has been tested on the following platforms: * Sun SunOS 4.1.3 with developer 'cc' compiler, Motif 1.2.3. * Sun Solaris 2.3 with Motif/X11/cc provided by Sun's 2.3 SDK * HPUX 9.0 on HP9000s300 HP9000s400, HP9000s700 * HPUX 8.0. on HP9000s300 and HP9000s800 * SGI Irix 5.1 and 5.2 w/ Motif/X11/cc from Irix IDO. * DEC Ultrix 4.3 and its built-in Motif 1.1 * DEC OSF1, v2.0 and its built-in Motif 1.2 * NeXT NeXT-Step 3.0 with Pencom Co-Xist X/Motif * 386/486 PC running Linux OS and gcc 2.5.8 compiler (reported by Serge Kolgan root@starato.wwb.noaa.gov). * AIX 3.2.5 on IBM RS6000 using IBM's product 'cc'. (reported by Duncan Swain -- duncan@chaussegros.architecture.mcgill.ca) * Sun SunOS 4.1.3 with gcc 2.5.8 (reported by Wolfgang Kechel -- wolfgang@pr-wiesbaden.de). The WINTERP distribution contains Makefiles specific to the above machines and OSs, and also provides Imakefiles for compiling WINTERP on other systems. ------------------------------------------------------------------------------- OBTAINING WINTERP: Here's how to ftp WINTERP 2.0: (your input denoted by ^^^^^^^^^^) jhvh-1-16-~> cd /tmp jhvh-1-17-/tmp> ftp ftp.x.org ^^^^^^^^^^^^^ [...] Name (jhvh-1.eit.com:mayer): anonymous ^^^^^^^^^ Password (jhvh-1.eit.com:anonymous): ^^^^^^^^^^^^^^^^^^^^^^^^^^^ 331 Guest login ok, send ident as password. 230 Guest login ok, access restrictions apply. ftp> cd contrib/devel_tools ^^^^^^^^^^^^^^^^^^^^^^ 200 CWD command okay. ftp> binary ^^^^^^ 200 Type set to I. ftp> get winterp-2.XX.tar.gz ## replace XX with latest version number ^^^^^^^^^^^^^^^^^^^^^^^ 200 PORT command okay. 150 Opening data connection for winterp-2.XX.tar.gz (192.100.58.37,3988) (2303809 bytes). 226 Transfer complete. 2303809 bytes received in 690.63 seconds (1.96 Kbytes/sec) ftp> quit ^^^^ 221 Goodbye. jhvh-1-18-/tmp> su ## become superuser to install in /usr/local/winterp ^^ jhvh-1-ROOT-1-/tmp> cd /usr/local/ ^^^^^^^^^^^^^^ jhvh-1-ROOT-2-/usr/local> gunzip -c /tmp/winterp-2.XX.tar.gz | tar xvf - ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ [... lengthy list of files output by tar ...] jhvh-1-ROOT-3-/usr/local/> rm /tmp/winterp-2.XX.tar.gz ^^^^^^^^^^^^^^^^^^^^^^^^^^^ << If you don't have GNU zip (gzip(1) gunzip(1), retrieve the >> << winterp-2.XX.tar.Z instead of the winterp-2.XX.tar.gz file and use "zcat">> << in place of "gunzip -c" above.... >> ------------------------------------------------------------------------------ COMPILING and RUNNING: * For compilation tips, see winterp/doc/winterp.doc sections: <> or <>. * For configuration and running hints, see winterp/doc/winterp.doc sections: <>, <<~/.winterp -- WINTERP session startup file:>>, <<~/.winterpapp -- WINTERP development session startup file.>>, and finally, <> * For known bugs, see winterp/doc/BUGS =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= == Niels Mayer -- http://www.eit.com/~mayer -- mayer@netcom.com == =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= From majordom@umnstat.stat.umn.edu Thu Aug 29 20:50:02 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28518; Thu, 29 Aug 1996 20:50:02 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28513; Thu, 29 Aug 1996 19:49:57 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA10784; Thu, 29 Aug 96 17:49:54 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA00830; Thu, 29 Aug 1996 17:49:51 -0700 From: Jan Deleeuw Message-Id: <9608300049.AA00830@galton.math.ucla.edu> Subject: To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Thu, 29 Aug 1996 17:49:50 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The Xlisp-Stat archive now has a directory http://www.stat.ucla.edu/archive/xlispstat/code/incoming/ which has code that is not yet classified and moved to its final destination. I am also working on the HTML pages for the sub-directories, but this is a slow process. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Wed Sep 4 14:18:33 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24137; Wed, 4 Sep 1996 14:18:33 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24131; Wed, 4 Sep 1996 13:18:29 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA10006; Wed, 4 Sep 96 11:18:28 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA21048; Wed, 4 Sep 1996 11:18:25 -0700 From: Jan Deleeuw Message-Id: <9609041818.AA21048@galton.math.ucla.edu> Subject: UCLA Xlisp-Stat Archive To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Wed, 4 Sep 1996 11:18:24 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The new setup of the archive is illustrated by following http://www.stat.ucla.edu/archive/xlispstat then go to code, then go to statistics, then go to categorical, then go to glim thus there are no links to actual lsp files (because on our system they would execute instead of display), but the links are to directories showing the files, and you can drag them (netscape, cyberdog) or url-get them or display them (if you do not have Xlisp-Stat as a helper). At the lowest levels of the hierarchy I will try to put README files, because these show up so nicely in netscape directory windows and in fancy ftp clients. If you submit files, it would be nice to take this into account. Bond, Udina, Naras, and I are writing an Xlisp-Stat book, which will probably come with the archive on a CD, which is why it has to be nice. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Sun Sep 8 14:10:03 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22223; Sun, 8 Sep 1996 14:10:03 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22218; Sun, 8 Sep 1996 13:09:57 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA02158; Sun, 8 Sep 96 11:09:55 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA21139; Sun, 8 Sep 1996 11:09:48 -0700 From: Jan Deleeuw Message-Id: <9609081809.AA21139@galton.math.ucla.edu> Subject: Homals in Xlisp-Stat To: albert-gifi@math.ucla.edu (Albert Gifi) Date: Sun, 8 Sep 1996 11:09:47 -0700 (PDT) Cc: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk See http://www.stat.ucla.edu/journals/jss/v01/i02 for a paper an code and demos on the current interactive version of Homals (aka Multiple Correspondence Analysis) in Xlisp-Stat. Also, if yoy want to submit to the Journal of Statistical Software, you can use this as a model. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Wed Sep 11 17:31:32 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA18206; Wed, 11 Sep 1996 17:31:32 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA18201; Wed, 11 Sep 1996 16:31:28 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA10771; Wed, 11 Sep 96 14:31:26 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA24803; Wed, 11 Sep 1996 14:31:24 -0700 Date: Wed, 11 Sep 1996 14:31:24 -0700 From: Jason Bond Message-Id: <9609112131.AA24803@galton.math.ucla.edu> To: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk See ftp://ftp.stat.ucla.edu/pub/lisp/xlisp/xlisp-stat/code/statistics/plotting/stray s/zoom.lsp It is a revised zooming prototype. Below is the header ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;; ;; zoom-overlay (inherits from graph-overlay) and zoom-plot-proto (inherits ;; from scatterplot-proto). When the "+" slider button is pushed ;; in the overlay, a box can be drawn around points (using the mouse ;; mouse) and they will be "zoomed" in on. To incrementally zoom out ;; push the "-" button. The "Out" button completely zooms out. Lines ;; that are drawn in the plot whose linestarts lie completely inside of ;; the box that is drawn will also be drawn in the "zoom". Selecting ;; is handled in the it way it should be. The shift key keeps points ;; selected while selecting others and this is true at any zooming stage. ;; Other niceties inlcude adding/removing zoom-overlays in the plots ;; menu and the option to include a list of 1's and 0's indicating ;; selectedness/de-selectedness of points in the plots menu. ;; The overlay is added by sending the original plot the :add-overlay ;; method. An example is shown at the bottom. ;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; Jason Bond jbond@stat.ucla.edu From majordom@umnstat.stat.umn.edu Thu Sep 12 17:08:02 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04611; Thu, 12 Sep 1996 17:08:02 -0400 Return-Path: Received: from [129.252.16.71] by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04606; Thu, 12 Sep 1996 16:07:57 -0500 Received: (from rossini@localhost) by fhat.stat.sc.edu (8.6.12/8.6.12) id RAA00504; Thu, 12 Sep 1996 17:07:56 -0400 Date: Thu, 12 Sep 1996 17:07:56 -0400 Message-Id: <199609122107.RAA00504@fhat.stat.sc.edu> From: Anthony Rossini To: stat-lisp-news@umnstat.stat.umn.edu Subject: XLispStat for Windows 95 Reply-To: rossini@stat.sc.edu X-Face: -JZ8Fs2;tGK&eaV#. U#%d52]n51?][w92.tu+>13)STQ Received: from davinci.socsci.McMaster.CA by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05154; Thu, 12 Sep 1996 17:18:08 -0500 Received: (from fox@localhost) by davinci.socsci.mcmaster.ca (8.7.4/8.7.3) id SAA01722; Wed, 11 Sep 1996 18:17:24 -0400 Message-Id: X-Mailer: XFMail 0.3 [p0] on Linux Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 8bit Mime-Version: 1.0 In-Reply-To: <199609122107.RAA00504@fhat.stat.sc.edu> Reply-To: jfox@mcmaster.ca Organization: McMaster University Date: Wed, 11 Sep 1996 18:13:23 EDT From: John Fox To: Anthony Rossini Subject: RE: XLispStat for Windows 95 Cc: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On 12-Sep-96 Anthony Rossini wrote: >>Has anyone had any experience with XLispStat on MS-Windows 95? Will the >latest versions (alpha or prior) for Windows (3.xx, I guess) work on >MS Win 95? > I've been using the 32-bit version of Lisp-Stat (version 3.45, I think) with Windows 95, and it has been working just fine. ----------------------------------------------------------- | John Fox jfox@mcmaster.ca | | Department of Sociology McMaster University | ----------------------------------------------------------- From majordom@umnstat.stat.umn.edu Fri Sep 13 09:38:20 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12295; Fri, 13 Sep 1996 09:38:20 -0400 Return-Path: Received: from server.cs.uiowa.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12290; Fri, 13 Sep 1996 08:38:15 -0500 Received: from nuthatch (20985@nuthatch.stat.uiowa.edu [128.255.25.15]) by server.cs.uiowa.edu with SMTP id IAA18464; Fri, 13 Sep 1996 08:38:14 -0500 (CDT) Message-Id: <323963C5.52BFA1D7@stat.uiowa.edu> Date: Fri, 13 Sep 1996 08:38:13 -0500 From: Russ Lenth Organization: University of Iowa X-Mailer: Mozilla 3.0 (X11; U; SunOS 4.1.3 sun4c) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Cc: cdobler@server.cs.uiowa.edu Subject: math errors in Sun version of xlispstat Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk A colleague has asked me about computational errors in xlispstat, running on the Sun version (SunOS 4.1), and I found errors all over the place... XLISP-STAT 2.1 Release 3.44 (Beta). Copyright (c) 1989-1994, by Luke Tierney. > (/ 13.5 17.5) 0.7711285714285777 should be .77142857142857... > (/ 1 7) 0.14225714285714444 should be .142857142857... > (/ 1 17) 0.058888529411764705 should be .0588235294118... > (/ 1 13) 0.07696907692307070 should be .0769230769231... I seem to find more errors than correct answers! Does anybody else get the same errors on SunOS 4.1? I checked these out on the HPUX version also, and got the correct answers there. Russ -- Russell V. Lenth -- Department of Statistics & Actuarial Science The University of Iowa -- Iowa City, IA 52242 USA Tel (319)335-0814 -- FAX (319)335-3017 mailto:Russell-Lenth@uiowa.edu - http://www.stat.uiowa.edu/~rlenth From majordom@umnstat.stat.umn.edu Fri Sep 13 10:10:25 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12641; Fri, 13 Sep 1996 10:10:25 -0400 Return-Path: Received: from hypatia-appli.ethz.ch by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12636; Fri, 13 Sep 1996 09:10:18 -0500 Received: from (maechler@sophie [129.132.58.2]) by hypatia.math.ethz.ch (8.6.12/Main-STAT-mailer) with SMTP id QAA14663; Fri, 13 Sep 1996 16:10:07 +0200 Received: by (4.1/D-MATH-stat-client) id AA11714 for stat-lisp-news@umnstat.stat.umn.edu; Fri, 13 Sep 96 16:10:06 +0200 Date: Fri, 13 Sep 96 16:10:06 +0200 Message-Id: <9609131410.AA11714@> From: Martin Maechler To: rlenth@stat.uiowa.edu Cc: stat-lisp-news@umnstat.stat.umn.edu, cdobler@server.cs.uiowa.edu In-Reply-To: <323963C5.52BFA1D7@stat.uiowa.edu> (message from Russ Lenth on Fri, 13 Sep 1996 08:38:13 -0500) Subject: Re: math errors in Sun version of xlispstat Reply-To: Martin Maechler Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk No, I don't have your problems -- everything is ok, here : XLISP-PLUS version 2.1g Portions Copyright (c) 1988, by David Betz. Modified by Thomas Almy and others. XLISP-STAT 2.1 Release 3.44 (Beta). Copyright (c) 1989-1994, by Luke Tierney. > (/ 13.5 17.5) 0.7714285714285715 > (/ 1 7) 0.14285714285714285 > (system "uname -a") SunOS sophie 4.1.4 2 sun4m 0 > From majordom@umnstat.stat.umn.edu Fri Sep 13 10:24:23 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12939; Fri, 13 Sep 1996 10:24:23 -0400 Return-Path: Received: from relay.pnl.gov by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12934; Fri, 13 Sep 1996 09:24:19 -0500 Received: from pound.pnl.gov by pnl.gov (PMDF V4.3-13 #6012) id <01I9F78MG7TC90N6I8@pnl.gov>; Fri, 13 Sep 1996 07:24:12 -0700 (PDT) Received: by pound.pnl.gov (5.x/SMI-SVR4) id AA08326; Fri, 13 Sep 1996 07:23:17 -0700 Date: Fri, 13 Sep 1996 07:23:17 -0700 From: PD Whitney Subject: Re: math errors in Sun version of xlispstat In-Reply-To: <323963C5.52BFA1D7@stat.uiowa.edu> (message from Russ Lenth on Fri, 13 Sep 1996 08:38:13 -0500) To: rlenth@stat.uiowa.edu Cc: stat-lisp-news@umnstat.stat.umn.edu Reply-To: whitney@pound.pnl.gov Message-Id: <9609131423.AA08326@pound.pnl.gov> Content-Transfer-Encoding: 7BIT Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I recall a discussion of math errors on SunOS 4.1 a few years ago where the final culprit was a math library; and the error could be fixed by applying a library patch (supplied for free by Sun). Unfortunately, I couldn't track down any of the news/email documenting the problem and the patch to apply. -Paul Whitney From majordom@nokomis.stat.umn.edu Fri Sep 13 11:10:39 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA01893; Fri, 13 Sep 1996 11:10:39 -0400 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA01886; Fri, 13 Sep 1996 10:10:29 -0500 From: Luke Tierney Message-Id: <9609131510.AA01886@nokomis.stat.umn.edu> Subject: Re: math errors in Sun version of xlispstat To: rlenth@stat.uiowa.edu (Russ Lenth) Date: Fri, 13 Sep 1996 10:10:29 -0500 (CDT) Cc: stat-lisp-news@nokomis.stat.umn.edu, cdobler@server.cs.uiowa.edu In-Reply-To: <323963C5.52BFA1D7@stat.uiowa.edu> from "Russ Lenth" at Sep 13, 96 08:38:13 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1223 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > A colleague has asked me about computational errors in xlispstat, > running on the Sun version (SunOS 4.1), and I found errors all over the > place... > > XLISP-STAT 2.1 Release 3.44 (Beta). > Copyright (c) 1989-1994, by Luke Tierney. > > > (/ 13.5 17.5) > 0.7711285714285777 should be .77142857142857... > > (/ 1 7) > 0.14225714285714444 should be .142857142857... > > (/ 1 17) > 0.058888529411764705 should be .0588235294118... > > (/ 1 13) > 0.07696907692307070 should be .0769230769231... > > I seem to find more errors than correct answers! Does anybody else get > the same errors on SunOS 4.1? > > I checked these out on the HPUX version also, and got the correct > answers there. > > Russ > -- > Russell V. Lenth -- Department of Statistics & Actuarial Science > The University of Iowa -- Iowa City, IA 52242 USA > Tel (319)335-0814 -- FAX (319)335-3017 > mailto:Russell-Lenth@uiowa.edu - http://www.stat.uiowa.edu/~rlenth > The problem is not in arithmetic but in printing. If you are using SunOS 4.1 you either need to upgrade to 3.45 (ftp://stat.umn.edu/pub/xlispstat/Alpha) or apply the patches for 3.44 and SunOS 4.1 available in (ftp://stat.umn.edu/pub/xlispstat/SunOSPatches) luke From majordom@umnstat.stat.umn.edu Fri Sep 13 13:10:41 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14264; Fri, 13 Sep 1996 13:10:41 -0400 Return-Path: Received: from sulu.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14259; Fri, 13 Sep 1996 12:10:36 -0500 Received: from chekov by sulu.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id KAA04292; Fri, 13 Sep 1996 10:10:34 -0700 (PDT) Date: Fri, 13 Sep 1996 10:10:33 -0700 (PDT) From: Thomas Lumley X-Sender: thomas@chekov To: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: math errors in Sun version of xlispstat In-Reply-To: <323963C5.52BFA1D7@stat.uiowa.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Fri, 13 Sep 1996, Russ Lenth wrote: > A colleague has asked me about computational errors in xlispstat, > running on the Sun version (SunOS 4.1), and I found errors all over the > place... This may explain why my cox regression code (running on SunOS 4.1) gives errors that differ in about the fourth significant figure from those given by S-PLUS. thomas lumley U Washington From majordom@umnstat.stat.umn.edu Fri Sep 13 13:41:04 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14503; Fri, 13 Sep 1996 13:41:04 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14498; Fri, 13 Sep 1996 12:41:00 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id NAA18803; Fri, 13 Sep 1996 13:40:53 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 13 Sep 1996 13:40:54 -0500 To: Luke Tierney From: forrest@unc.edu (Forrest Young) Subject: Re: math errors in Sun version of xlispstat Cc: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >> >> A colleague has asked me about computational errors in xlispstat, >> running on the Sun version (SunOS 4.1), and I found errors all over the >> place... >> >The problem is not in arithmetic but in printing. If you are using >SunOS 4.1 you either need to upgrade to 3.45 >(ftp://stat.umn.edu/pub/xlispstat/Alpha) or apply the patches for 3.44 >and SunOS 4.1 available in >(ftp://stat.umn.edu/pub/xlispstat/SunOSPatches) > >luke I take it this applies to Unix, but not Mac and MSDos? Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@nokomis.stat.umn.edu Sat Sep 14 05:21:56 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA16380; Sat, 14 Sep 1996 05:21:56 -0400 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA16372; Sat, 14 Sep 1996 04:21:50 -0500 From: Luke Tierney Message-Id: <9609140921.AA16372@nokomis.stat.umn.edu> Subject: Re: math errors in Sun version of xlispstat To: forrest@unc.edu (Forrest Young) Date: Sat, 14 Sep 1996 04:21:49 -0500 (CDT) Cc: luke@nokomis.stat.umn.edu, stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Forrest Young" at Sep 13, 96 01:40:54 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1406 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > >> > >> A colleague has asked me about computational errors in xlispstat, > >> running on the Sun version (SunOS 4.1), and I found errors all over the > >> place... > >> > >The problem is not in arithmetic but in printing. If you are using > >SunOS 4.1 you either need to upgrade to 3.45 > >(ftp://stat.umn.edu/pub/xlispstat/Alpha) or apply the patches for 3.44 > >and SunOS 4.1 available in > >(ftp://stat.umn.edu/pub/xlispstat/SunOSPatches) > > > >luke > > I take it this applies to Unix, but not Mac and MSDos? > > Forrest Young > Professor, Psychometrics > > UNC Psychometrics Lab e-mail: forrest@unc.edu > CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu > Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 > > ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html > > > This does not affect the Mac or Windows versions. It could affect some other UNIX systems, but the only one I know of is SunOS 4.1. If your UNIX has a memmove routine then it is definitely not affected. If it has no memmove but its memcpy routine handles overlapping regions properly then it is also unaffected. SunOS 4.1 did not yet include memmove but had changed its memcpy to one that did not work with overlapping source and destination. The source fixes avoid the problem by providing a proper memmove if none is available. luke From majordom@umnstat.stat.umn.edu Mon Sep 16 10:23:56 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24512; Mon, 16 Sep 1996 10:23:56 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24507; Mon, 16 Sep 1996 09:23:52 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id KAA24582; Mon, 16 Sep 1996 10:23:39 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 16 Sep 1996 10:23:31 -0500 To: edstat-l@jse.stat.ncsu.edu, stat-l@vm1.mcgill.ca, stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: New ViSta Release Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Announcing a major new release of ViSta - The Visual Statistics System Version 4.23m, September 15th, 1996 Copyright (c) 1991-96 by Forrest W. Young ViSta is a free statistical system that performs univariate and multivariate statistical data analysis and visualization. ViSta is available for Microsoft Windows, MacOS, and for Unix with XWindows from our WWW and FTP servers at: WWW: http://forrest.psych.unc.edu FTP: www.psych.unc.edu, directory pub/forrest/vista. A complete description of ViSta's capabilities, including examples, is available at the WWW site. A brief description, including a list of new features, is given in this message. AUDIENCE: ViSta is useful for teaching multivariate statistics, statistical graphics and computational statistics; and for research and development in statistical graphics and computational statistics. ViSta is designed for an audience of users having a very wide range of data analysis sophistication, ranging from novice to expert. ViSta provides seamlessly integrated data analysis environments specifically tailored to the user's level of expertise. Guidance is available for students and novices, and tools are available for teachers and other experts to create guidance for these novices. A structured graphical user interface is available for competent users, and a command line interface is available for sophisticated users. The complete Lisp-Stat (Tierney, 1990) programming environment is available to researchers, programmers and graduate students who wish to extend ViSta's capabilities. CAPABILITIES: ViSta's data analysis and visualization capabilities include: exploratory graphics and descriptive statistics; univariate tests and visualizations; analysis of variance; regression analysis; principal components analysis; multidimensional scaling and correspondence analysis. All analysis methods have extensive visualization features. NEW FEATURES: The new version focuses on new (and improved) data analysis and visualization methods. The major new features in this version include: 1) Multiple Regression, including OLS, robust and monotonic regression. 2) Redundancy analysis (regression with multiple responses). 3) Extensive visual regression diagnostics. 4) A visualization method for Univariate Analysis. 5) A completed visualization method for Analysis of Variance. 6) An improved visualization method for Multivariate Regression. 7) A multivariate regression test. Currently, the new version is only available for the MacOS. The new version will be available for MS-Windows soon. The Unix version will follow. Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Mon Sep 16 13:20:17 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26134; Mon, 16 Sep 1996 13:20:17 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26129; Mon, 16 Sep 1996 12:20:05 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA16904; Mon, 16 Sep 96 10:20:03 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA08713; Mon, 16 Sep 1996 10:20:00 -0700 From: Jan Deleeuw Message-Id: <9609161720.AA08713@galton.math.ucla.edu> Subject: changes in archive To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Mon, 16 Sep 1996 10:19:59 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Those of you who use the archive may have noticed a few changes in the last couple of days. (a) The classical Apple directory icons have been replaced by the System 8.0 directory icons. Macintosh will live ! (b) The foo.lsp files have been renamed foo.lsp.txt This is for the benefit of those (such as us) whose WWW browsers are set to execute foo.lsp files and to display foo.txt files. Otherwise changes are minor. Some of the lisp is updated an improved. We'll try to indicate this with server side includes that give the last revision date. Generally, the interface is improving and information about the files in various forms is added. This is a slow process. New submissions are welcome. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon Sep 16 13:34:59 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26207; Mon, 16 Sep 1996 13:34:59 -0400 Return-Path: Received: from geneva.crew.umich.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26199; Mon, 16 Sep 1996 12:34:55 -0500 Received: from zug.crew.umich.edu by crew.umich.edu (8.7.1/2.2) with ESMTP id NAA04265; Mon, 16 Sep 1996 13:34:54 -0400 (EDT) Received: from localhost by zug.crew.umich.edu (8.7.5/dumb-1.0) id NAA21314; Mon, 16 Sep 1996 13:34:53 -0400 (EDT) Date: Mon, 16 Sep 1996 13:34:53 -0400 (EDT) From: Michael Cohen To: Jan Deleeuw Cc: Xlisp List Subject: Re: changes in archive In-Reply-To: <9609161720.AA08713@galton.math.ucla.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Mon, 16 Sep 1996, Jan Deleeuw wrote IN PART: > ... > (b) The foo.lsp files have been renamed foo.lsp.txt This is for the > benefit of those (such as us) whose WWW browsers are set to execute > foo.lsp files and to display foo.txt files. > Jan: Anywhere in the archives -- or elsewhere -- is there material on the steps needed to make lsp a helper for a browser? Just a pointer would be welcome. I don't want you to spend your time writing me a how-to if it doesn't already exist. --michael cohen u. michigan From majordom@umnstat.stat.umn.edu Mon Sep 16 19:10:12 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29095; Mon, 16 Sep 1996 19:10:12 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29090; Mon, 16 Sep 1996 18:10:04 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa18872; 16 Sep 96 19:09 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA16945; Mon, 16 Sep 1996 19:01:42 -0400 Date: Mon, 16 Sep 1996 19:01:41 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Lisp Statistics Cc: Luke Tierney Subject: LispStat XES Application and Framework Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk A first draft of notes on my Xray Energy Spectra analysis package written in LispStat, the LispStat graphics and data class framework used, and notes on the modifications, changes and bug-fixes made to the 3.47 LispStat sources, as well as Macintosh PPC binaries and CW9 project and modified sources are temporarily available at: The application itself is alpha/experimental, but it includes fairly stable version of some code I had previously posted for color-graphics overlays, linkable objects, jitterplots, and some other parts of the basic framework. Although the application is somewhat specialized, the framework classes used to build it are intended to be general extensions to XLispStat. I hope to have more complete documentation, as well as final versions of the code available soon. My usual web server has been down and I've been too busy coding and documenting to take time to get things back as they were -- thus the temporary site above. I don't expect to have time to set up the old site for several weeks or more. ---| Steven D. Majewski (804-982-0831) |--- ---| Computer Systems Engineer University of Virginia |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| Box 449 Health Science Center Charlottesville,VA 22908 |--- [ "The grass is always greener, except at t=0" - Stan Kelly-Bootle ] From majordom@umnstat.stat.umn.edu Tue Sep 17 17:53:55 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13601; Tue, 17 Sep 1996 17:53:55 -0400 Return-Path: Received: from zeus.usq.edu.au by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13594; Tue, 17 Sep 1996 16:53:48 -0500 Received: from po.usq.edu.au (pluto.usq.edu.au [139.86.128.23]) by zeus.usq.edu.au (8.7.5/8.7) with SMTP id HAA21230 for ; Wed, 18 Sep 1996 07:53:11 +1000 (EST) Received: from Microsoft Mail (PU Serial #1166) by po.usq.edu.au (PostalUnion/SMTP(tm) v2.1.8d for Windows NT(tm)) id AA-1996Sep18.074809.1166.506557; Wed, 18 Sep 1996 07:53:10 +1000 From: darnell%scipo.usq.edu.au@usq.edu.au (Ross Darnell) To: stat-lisp-news@umnstat.stat.umn.edu ('stat-lisp-news@stat.umn.edu') Message-Id: <1996Sep18.074809.1166.506557@po.usq.edu.au> X-Mailer: Microsoft Mail via PostalUnion/SMTP for Windows NT Mime-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Organization: University of Southern Queensland, Toowoomba, AUS Date: Wed, 18 Sep 1996 07:53:10 +1000 Subject: Pasting from MS Windows clipboard Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk If anyone could tell me how to place clipboard info into a variable i would be grateful. (msw-paste) "displays" the contents of the clipboard but how do I put that data into a variable? (def x (msw-paste)) doesn't work. Ross Darnell ___________________________________________________________ Ross Darnell Telephone: 61 76 311522 Faculty of Science Fax: 61 76 311775 University of Southern Queensland Email:darnell@usq.edu.au Darling Heights Q 4350. ___________________________________________________________ From majordom@umnstat.stat.umn.edu Fri Sep 27 12:28:42 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04946; Fri, 27 Sep 1996 12:28:42 -0400 Return-Path: Received: from bubba.NMSU.Edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04938; Fri, 27 Sep 1996 11:28:33 -0500 Received: from NMSU.Edu by bubba.NMSU.Edu (8.6.10/NMSU) id KAA23124; Fri, 27 Sep 1996 10:29:02 -0600 Received: from ccserver by NMSU.Edu (8.6.10/NMSU-1.18) id KAA20223; Fri, 27 Sep 1996 10:28:31 -0600 Date: Fri, 27 Sep 1996 10:28:31 -0600 Message-Id: <199609271628.KAA20223@NMSU.Edu> Received: from [128.123.43.11] by ccserver (4.1/NMSU) id AA07885; Fri, 27 Sep 96 10:28:25 MDT X-Sender: ddaniel@dante.nmsu.edu Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: stat-lisp-news@umnstat.stat.umn.edu From: ddaniel@nmsu.edu (David L. Daniel) Subject: histogram oddities... Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On the PowerMac XLispStat Version 2.1 Release 3.44, creating a histogram with: (def hp (histogram (iseq 8))) and then resizing the plot by simply clicking and dragging on the resize button on the bottom right of the histogram window has resulted in four different histogram shapes: * * * * * * * * --------------- (type 1 - uniform as one would expect) * * * * * * * * --------------- (type 2) * * * * * * * * --------------- (type 3) * * * * * * * * --------------- (type 4) In each case, (send hp :num-bins) returns 8. Any comments? David Daniel University Statistics Center New Mexico State University ddaniel@nmsu.edu ################################################################### Degenerate distributions - They're everywhere! They're everywhere! ################################################################### From majordom@umnstat.stat.umn.edu Fri Sep 27 13:18:35 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05251; Fri, 27 Sep 1996 13:18:35 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05246; Fri, 27 Sep 1996 12:18:30 -0500 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.6.4.3/10.1) id NAA04218; Fri, 27 Sep 1996 13:18:25 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 27 Sep 1996 13:19:04 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: Re: histogram oddities... Cc: ddaniel@nmsu.edu (David L. Daniel) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk yep...does it for me too, on PowerMac and under MSWin (3.45). what gives? >On the PowerMac XLispStat Version 2.1 Release 3.44, creating a histogram with: > > (def hp (histogram (iseq 8))) > >and then resizing the plot by simply clicking and dragging on the resize >button on the bottom right of the histogram window has resulted in four >different histogram shapes: > > * * * * * * * * > --------------- (type 1 - uniform as one would expect) > > * > * * * * * * * > --------------- (type 2) > > * * > * * * * * * > --------------- (type 3) > > * * * * > * * * * > --------------- (type 4) > >In each case, (send hp :num-bins) returns 8. > >Any comments? > > > >David Daniel >University Statistics Center >New Mexico State University >ddaniel@nmsu.edu >################################################################### >Degenerate distributions - They're everywhere! They're everywhere! >################################################################### Forrest Young Professor, Psychometrics CB# 3270 Davie Hall email: mailto://forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software at: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Fri Sep 27 18:18:49 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA07732; Fri, 27 Sep 1996 18:18:49 -0400 Return-Path: Received: from geneva.crew.umich.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA07727; Fri, 27 Sep 1996 17:18:43 -0500 Received: from zug.crew.umich.edu by crew.umich.edu (8.7.1/2.2) with ESMTP id SAA06445; Fri, 27 Sep 1996 18:18:42 -0400 (EDT) Received: from localhost by zug.crew.umich.edu (8.7.5/dumb-1.0) id SAA22520; Fri, 27 Sep 1996 18:18:41 -0400 (EDT) Date: Fri, 27 Sep 1996 18:18:41 -0400 (EDT) From: Michael Cohen To: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: histogram oddities... In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Ditto for 2.1 Release 3.47 (Beta) on a Sun. So at least it is the same logic on all platforms.... --michael cohen _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ michael d. cohen 313 763-0098 _/ _/ school of information mdc@umich.edu _/ _/ university of michigan _/ _/ http://geneva.crew.umich.edu:80/~mdc/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ On Fri, 27 Sep 1996, Forrest Young wrote: > yep...does it for me too, on PowerMac and under MSWin (3.45). what gives? > > > >On the PowerMac XLispStat Version 2.1 Release 3.44, creating a histogram with: > > > > (def hp (histogram (iseq 8))) > > > >and then resizing the plot by simply clicking and dragging on the resize > >button on the bottom right of the histogram window has resulted in four > >different histogram shapes: > > > > * * * * * * * * > > --------------- (type 1 - uniform as one would expect) > > > > * > > * * * * * * * > > --------------- (type 2) > > > > * * > > * * * * * * > > --------------- (type 3) > > > > * * * * > > * * * * > > --------------- (type 4) > > > >In each case, (send hp :num-bins) returns 8. > > > >Any comments? > > > > > > > >David Daniel > >University Statistics Center > >New Mexico State University > >ddaniel@nmsu.edu > >################################################################### > >Degenerate distributions - They're everywhere! They're everywhere! > >################################################################### > > Forrest Young > Professor, Psychometrics > > CB# 3270 Davie Hall email: mailto://forrest@unc.edu > UNC Psychometric Lab www: http://forrest.psych.unc.edu > Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 > > ViSta Software at: http://forrest.psych.unc.edu/research/ViSta.html > > > From majordom@umnstat.stat.umn.edu Mon Sep 30 13:45:19 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA18229; Mon, 30 Sep 1996 13:45:19 -0400 Return-Path: Received: from halley.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA18224; Mon, 30 Sep 1996 12:45:02 -0500 Received: from libiya.upf.es by halley.upf.es with SMTP (1.37.109.15/16.2) id AA008179714; Mon, 30 Sep 1996 19:55:14 +0100 Received: by libiya.upf.es (1.38.193.4/15.6) id AA01393; Mon, 30 Sep 1996 18:45:18 +0100 Date: Mon, 30 Sep 1996 18:45:18 +0100 (MET) From: Frederic Udina To: Michael Cohen Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: histogram oddities... In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I wrote some time ago about this bug. I was thinking it has been fixed but... I think that this is a serious bug for a system that pretends to be used by statisticians! There are some packages written in XLS to do right histograms and other... but anyway, I think the bug might be fixed. ftp.upf.es:/pub/stat/anchor-position can be a posibility http://www.stat.ucla.edu/archive/xlispstat/code/statistics/plotting/my-histogram/ is another one Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Ramon Trias Fargas, 25-27 | Compuserve: 100555.603 08005 Barcelona | WWWeb: http://libiya.upf.es/ SPAIN | or http://www.econ.upf.es/~udina ___________________________________/_\_________________________________ From majordom@umnstat.stat.umn.edu Tue Oct 1 14:01:41 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06280; Tue, 1 Oct 1996 14:01:41 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06275; Tue, 1 Oct 1996 13:01:34 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA21245; Tue, 1 Oct 96 11:01:31 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA28502; Tue, 1 Oct 1996 11:01:28 -0700 From: Jan Deleeuw Message-Id: <9610011801.AA28502@galton.math.ucla.edu> Subject: Course 153a To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Tue, 1 Oct 1996 11:01:28 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The project http://www.stat.ucla.edu/~deleeuw/courses/153a/outlines/ may be of interest to some people on this list. It is changing very rapidly each day, and will continue to change at least during this quarter. Technically, it is done with a combination of the pphtml and latex2html preprocessors. The pages are written in LaTeX, using html.sty, and the rawhtml environment. Suggestions are welcome. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Tue Oct 1 16:55:33 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA07526; Tue, 1 Oct 1996 16:55:33 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA07521; Tue, 1 Oct 1996 15:55:29 -0500 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.6.4.3/10.1) id QAA12927; Tue, 1 Oct 1996 16:55:25 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Tue, 1 Oct 1996 16:56:13 -0500 To: stat-lisp-news@umnstat.stat.umn.edu, vista@unc.edu From: forrest@unc.edu (Forrest Young) Subject: Lisp-Stat/ViSta course Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk LispStat and/or ViSta users ... This semester I am putting all of my materials for my LispStat/ViSta graduate course online. I teach the entire course via the web. Check out http://forrest.psych.unc.edu/teaching/p285/p285.html. This link is the syllabus, which itself includes links to class notes, assignments, answers, and links to useful LispStat sites. The information is being added to every day. Forrest Young Professor, Psychometrics CB# 3270 Davie Hall email: mailto://forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software at: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Thu Oct 3 12:10:49 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11883; Thu, 3 Oct 1996 12:10:49 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11878; Thu, 3 Oct 1996 11:10:44 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id MAA24745; Thu, 3 Oct 1996 12:10:41 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 3 Oct 1996 12:09:39 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: Residuals vs. Raw-Residuals ? Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk LispStaters: Does anyone know the difference between the :residuals and :raw-residuals messages for Lukes regression-model-proto? Does it have to do with weighting? If so, what? The only thing I see about this is on page 201 of Luke's book, but there doesn't seem to be any explanation. Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Thu Oct 3 12:42:09 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12237; Thu, 3 Oct 1996 12:42:09 -0400 Return-Path: Received: from sulu.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12229; Thu, 3 Oct 1996 11:42:04 -0500 Received: from chekov by sulu.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id JAA09642; Thu, 3 Oct 1996 09:41:58 -0700 (PDT) Date: Thu, 3 Oct 1996 09:41:57 -0700 (PDT) From: Thomas Lumley X-Sender: thomas@chekov To: Forrest Young Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Residuals vs. Raw-Residuals ? In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Thu, 3 Oct 1996, Forrest Young wrote: > Does anyone know the difference between the :residuals and :raw-residuals > messages for Lukes regression-model-proto? Does it have to do with > weighting? If so, what? :raw-residuals are the observed-fitted values :residuals are multiplied by the square root of the weights (ie approximately constant variance if you ignore the leverage correction) This information is in the object documentation (as well as in the source code) > (send regression-model-proto :help :residuals) loading in help file information - this will take a minute ...done RESIDUALS Message args: () Returns the raw residuals for a model without weights. If the model includes weights the raw residuals times the square roots of the weights are returned. NIL > (send regression-model-proto :help :raw-residuals) RAW-RESIDUALS Message args: () Returns the raw residuals for a model. NIL thomas lumley From majordom@umnstat.stat.umn.edu Thu Oct 3 14:07:12 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13221; Thu, 3 Oct 1996 14:07:12 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13215; Thu, 3 Oct 1996 13:07:04 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id OAA10332; Thu, 3 Oct 1996 14:06:58 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 3 Oct 1996 14:05:56 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: Re: Residuals vs. Raw-Residuals ? Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >LispStaters: > >Does anyone know the difference between the :residuals and :raw-residuals >messages for Lukes regression-model-proto? Does it have to do with >weighting? If so, what? Thanks for your replies. I guess I should have tried the help first (though over the years, I've become conditioned not to, I guess :-) ). Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Fri Oct 4 10:34:55 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29062; Fri, 4 Oct 1996 10:34:55 -0400 Return-Path: Received: from name.ufl.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29057; Fri, 4 Oct 1996 09:34:50 -0500 Received: from muskie (muskie.stat.ufl.edu [128.227.77.74]) by name.ufl.edu (8.6.13/8.6.10) with SMTP id KAA12075 for ; Fri, 4 Oct 1996 10:34:38 -0400 Date: Fri, 4 Oct 1996 10:34:38 -0400 Message-Id: <199610041434.KAA12075@name.ufl.edu> X-Sender: arcs@stat.ufl.edu X-Mailer: Windows Eudora Version 1.4.4 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: stat-lisp-news@umnstat.stat.umn.edu From: arcs@stat.ufl.edu (Art Smith) Subject: hiding name-list windows Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, This is the first time I've posted here and this may be a really stupid question but here goes. I'm trying to create a text window using a function which constructs the list of text from data stored in an object. It then uses the name-list function to create and display the window. I want to hide this window after it's been created. The :hide-window message doesn't seem to work. (defun make-text-window object) . . use name-list to create the window containing text . (setf some-window (make-text-window object)) <----- This works. (send some-window :hide-window) <----- This doesn't Error: bad argument type - NIL Happened in # Since name-list-proto inherits from graph and graph-window-proto(s), I don't understand why this won't work. I'd appreciate any insites including workarounds. Thanks in advance, Art From majordom@umnstat.stat.umn.edu Fri Oct 4 11:54:48 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29915; Fri, 4 Oct 1996 11:54:48 -0400 Return-Path: Received: from geneva.crew.umich.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29910; Fri, 4 Oct 1996 10:54:43 -0500 Received: from zug.crew.umich.edu by crew.umich.edu (8.7.1/2.2) with ESMTP id LAA02267; Fri, 4 Oct 1996 11:54:42 -0400 (EDT) Received: from localhost by zug.crew.umich.edu (8.7.5/dumb-1.0) id LAA11994; Fri, 4 Oct 1996 11:54:41 -0400 (EDT) Date: Fri, 4 Oct 1996 11:54:41 -0400 (EDT) From: Michael Cohen To: stat-lisp-news@umnstat.stat.umn.edu Subject: ps2GIF In-Reply-To: <199610041434.KAA12075@name.ufl.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk A simulation study I am doing inside xlispstat is producing postscript files that I want to convert to GIF or JPEG for insertion in a Web page about the work. Does anyone doing something similar know of publically available method of doing this? (In my case SunOS is more convenient than MacOS which is better than MS-Windows.) Jan deLeeuw has been kind enough to help me figure out how I can have dynamic displays for folks who have xlispstat as a .lsp helper for their browser. But I also need to provide static inline alternatives for the majority of people not willing to set that up. _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ michael d. cohen 313 763-0098 _/ _/ school of information mdc@umich.edu _/ _/ university of michigan _/ _/ http://geneva.crew.umich.edu:80/~mdc/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ From majordom@umnstat.stat.umn.edu Fri Oct 4 12:16:35 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00175; Fri, 4 Oct 1996 12:16:35 -0400 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00169; Fri, 4 Oct 1996 11:16:30 -0500 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA13397 (5.67b/IDA-1.5 for ); Fri, 4 Oct 1996 12:16:13 -0400 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA22519 for ; Fri, 4 Oct 1996 12:15:48 -0400 Message-Id: <32556390.3D37@kodak.com> Date: Fri, 04 Oct 1996 12:20:48 -0700 From: Jeff Morris Organization: Eastman Kodak Company X-Mailer: Mozilla 2.01 (Win16; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: downloading with Netscape Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi...I recently was moved (dragged) from a Mac to a PC and am accessing the WWW using Netscape v2.0. I want to download the Windows version of XLS from the UCLA site. When I double click on the .EXE file it downloads it as nonsense code (binary?) and when I save it and try to unpack it won't work. I can get the windows version using my Mac to download (and have) but would like to know if anyone has any experience with Netscape on a PC. Thanks for any aid!! From majordom@umnstat.stat.umn.edu Fri Oct 4 12:52:09 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00692; Fri, 4 Oct 1996 12:52:09 -0400 Return-Path: Received: from concorde.inria.fr by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00683; Fri, 4 Oct 1996 11:52:02 -0500 Received: from cao.inria.fr (cao.inria.fr [128.93.8.112]) by concorde.inria.fr (8.7.6/8.7.1) with SMTP id SAA05040; Fri, 4 Oct 1996 18:51:51 +0200 (MET DST) Received: from cao by cao.inria.fr (SMI-8.6/SMI-SVR4) id SAA07400; Fri, 4 Oct 1996 18:50:59 +0200 Message-Id: <199610041650.SAA07400@cao.inria.fr> To: Jeff Morris Cc: stat-lisp-news@umnstat.stat.umn.edu, aude@cao.inria.fr Subject: Re: downloading with Netscape In-Reply-To: Your message of "Fri, 04 Oct 1996 12:20:48 PDT." <32556390.3D37@kodak.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Id: <7397.844447859.1@cao> Date: Fri, 04 Oct 1996 18:50:59 +0200 From: Jean-Christophe Aude Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, when you are using netscape hold SHIFT and click on the left button of your mouse, you will be prompt to save this file.. hope this help... jc =========================================================================== Jean-Christophe AUDE | INRIA - Action Genome (thesard) | Institut National de Recherche | Email : jean-christophe.aude@inria.fr en Informatique et Automatique | Phone : +33 1 39 63 56 10 B.P. 105, 78153 Le chesnay Cedex | Fax : +33 1 39 63 53 30 FRANCE | =========================================================================== From majordom@umnstat.stat.umn.edu Fri Oct 4 13:58:24 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA01593; Fri, 4 Oct 1996 13:58:24 -0400 Return-Path: Received: from gateway.ets.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA01588; Fri, 4 Oct 1996 12:58:18 -0500 Received: from research.ets.org by gateway.ets.org with SMTP id AA26969 (5.65c/IDA-1.4.4 for ); Fri, 4 Oct 1996 13:57:20 -0400 Received: from comet.research.ets.org by research.ets.org with SMTP id AA09084 (5.67b/IDA-1.4.3 for mdc@crew.umich.edu); Fri, 4 Oct 1996 13:57:19 -0400 Received: by comet.research.ets.org (SMI-8.6/SMI-SVR4) id NAA03848; Fri, 4 Oct 1996 13:57:39 -0400 Date: Fri, 4 Oct 1996 13:57:39 -0400 Message-Id: <199610041757.NAA03848@comet.research.ets.org> From: Russell Almond To: mdc@crew.umich.edu Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: (message from Michael Cohen on Fri, 4 Oct 1996 11:54:41 -0400 (EDT)) Subject: Re: ps2GIF Reply-To: ralmond@ets.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk A simulation study I am doing inside xlispstat is producing postscript files that I want to convert to GIF or JPEG for insertion in a Web page about the work. Does anyone doing something similar know of publically available method of doing this? (In my case SunOS is more convenient than MacOS which is better than MS-Windows.) Ghostscript is a free (GNU licence) postscript interpreter which I think can be programmed to send its output to a file. I forgotten what kinds of formats are supported, but if it doesn't have the one you want, you should be able to use the package PBMplus to do the conversion. I think that most of these tools are available for both Unix and Windows environments. In the Mac world, there is a shareware program called Graphic Converter which I would recommend. BTW, GIF is probably better than JPEG for most statistical graphics. GIF uses a nearest-neighbor based compression scheme which is very efficient with line drawings (it can compress the large areas of constant color efficiently). JPEG uses a harmonic analysis which is great for photographs with continous color gradations but is not very good at the step functions in most line drawings. --Russell Almond From majordom@umnstat.stat.umn.edu Fri Oct 4 14:01:23 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA01647; Fri, 4 Oct 1996 14:01:23 -0400 Return-Path: Received: from mathfs.math.montana.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA01642; Fri, 4 Oct 1996 13:01:19 -0500 Received: (from jimrc@localhost) by mathfs.math.montana.edu (8.7.5/8.7.3) id MAA28423; Fri, 4 Oct 1996 12:01:11 -0600 Date: Fri, 4 Oct 1996 12:01:10 -0600 (MDT) From: Jim Robison-Cox To: Michael Cohen Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: ps2GIF In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Fri, 4 Oct 1996, Michael Cohen wrote: > A simulation study I am doing inside xlispstat is producing postscript > files that I want to convert to GIF or JPEG for insertion in a Web page > about the work. > > Does anyone doing something similar know of publically available method of > doing this? (In my case SunOS is more convenient than MacOS which is > better than MS-Windows.) An interesting problem, I could also use such a converter. Searching with archie for pstogif I find: Host wcarchive.cdrom.com Location: /.11/netlib/parkbench/gbis/bin FILE -r--r--r-- 1840 May 16 21:40 pstogif And the associated files in that directory which seem to be needed. It uses ghostscript to do the conversion. Has anyone used this? Jim Robison-Cox ____________ Department of Math Sciences | | phone: (406)994-5340 2-214 Wilson Hall \ BZM, MT | FAX: (406)994-1789 Montana State University | *_______| Bozeman, MT 59717 \_| e-mail: jimrc@math.montana.edu From majordom@umnstat.stat.umn.edu Sun Oct 6 05:11:08 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26775; Sun, 6 Oct 1996 05:11:08 -0400 Return-Path: Received: from pub.pub.ro by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26770; Sun, 6 Oct 1996 04:11:00 -0500 Received: (from root@localhost) by pub.pub.ro (8.7.6/8.6.12) id LAA21351 for stat-lisp-news@umnstat.stat.umn.edu; Sun, 6 Oct 1996 11:27:38 +0200 (EET) Received: (from dcorlan@localhost) by ottonel.pub.ro (8.6.12/8.6.9) id JAA04202; Sun, 6 Oct 1996 09:40:03 +0200 Date: Sun, 6 Oct 1996 09:40:00 +0200 (EET) From: Alexandru Dan Corlan To: Michael Cohen Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: ps2GIF In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk o----------------------------------------------------------------o Alexandru Dan Corlan MD, Univ. Hosp. Bucharest Electrocardiognosis 28 Argentina str. 1, 71206 Bucharest Romania tel +401-312.9360 +401-679.3415 email: dcorlan@ottonel.pub.ro o---- Adevarul este neverosimil -- (Truth is not verisimilar) ---o > A simulation study I am doing inside xlispstat is producing postscript > files that I want to convert to GIF or JPEG for insertion in a Web page > about the work. > > Does anyone doing something similar know of publically available method of > doing this? (In my case SunOS is more convenient than MacOS which is > better than MS-Windows.) > Try to find a program like ghostscript. Run it with something like: gs -sDEVICE=gif8 -sOutputFile=x.gif x.ps This will create x.gif. Some new versions of gs do not support gif (because of some unclear patent problems) but you can produce, say, .bmp or .ppm with a similar command and convert that bitmap format to .gif with a program such as xv (un Unix) or alchemy (on DOS) and, surely, tens of others. From majordom@umnstat.stat.umn.edu Mon Oct 7 14:35:39 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12743; Mon, 7 Oct 1996 14:35:39 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12738; Mon, 7 Oct 1996 13:35:35 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id OAA20991; Mon, 7 Oct 1996 14:35:31 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 7 Oct 1996 14:34:17 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: Dialog Box Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk LispStaters: Here's something else I can't figure out: I want to put up a simple dialog box that says "Please wait while computations finish", that stays up until computations are finished and which then disappears. I want it to have no buttons, and to be dismissed automatically at the right time (i.e., I don't want a button that is clicked to dismiss it). The following puts up the desired dialog box on Macs: (setf dialog (send modal-dialog-proto :new (list (send text-item-proto :new "Please wait until computations finish.")))) And then when time is right, do: (send dialog :remove) However, under windows it puts up an empty box (I havn't tried it under unix). But anyway, this doesn't feel right to me, since nowhere are we doing a (send dialog :modal-dialog). I am missing somthing? This sounds simple, but I feel like I can't figure it out. (Have I already figured it out and this is a Windows bug?) Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Mon Oct 7 16:58:44 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA21500; Mon, 7 Oct 1996 16:58:44 -0400 Return-Path: Received: from pv6613.vincent.iastate.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA21495; Mon, 7 Oct 1996 15:58:38 -0500 Received: by pv6613.vincent.iastate.edu with sendmail-5.65 id ; Mon, 7 Oct 1996 15:58:34 -0500 Message-Id: <9610072058.AA23792@pv6613.vincent.iastate.edu> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Duplicating Windows Date: Mon, 07 Oct 1996 15:58:33 CDT From: Mervyn Marasinghe Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi XLispstaters: What is the most efficient way of duplicating graphic windows along with the contents from within a lispstat function ? Thanks for any help. Mervyn Marasinghe From majordom@umnstat.stat.umn.edu Tue Oct 8 16:30:43 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA08085; Tue, 8 Oct 1996 16:30:43 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA08080; Tue, 8 Oct 1996 15:30:37 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id QAA02023; Tue, 8 Oct 1996 16:30:13 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Tue, 8 Oct 1996 16:28:56 -0500 To: vista@unc.edu, edstat-l@jse.stat.ncsu.edu, stat-l@vm1.mcgill.ca, stat-lisp-news@umnstat.stat.umn.edu, sug-l@listserv.oit.unc.edu From: forrest@unc.edu (Forrest Young) Subject: ViSta for Windows and Macs Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Announcing a major new release of ViSta - The Visual Statistics System Version 4.24, October 7th, 1996 Copyright (c) 1991-96 by Forrest W. Young This major new release brings major new (and improved) data analysis and visualization methods to ViSta for MSWindows. This release is also available for MacOS, fixing bugs in release 4.23m. The MSWindows and MacOS versions are identical. The Unix version will follow. ViSta performs univariate and multivariate statistical data analysis and visualization. ViSta is available for free for Microsoft Windows, MacOS, and for Unix with XWindows. ViSta, and its documentation, is available from our WWW and FTP servers at: WWW: http://forrest.psych.unc.edu FTP: www.psych.unc.edu, directory pub/forrest/vista. ViSta is useful for teaching multivariate statistics, statistical graphics and computational statistics; and for research and development in statistical graphics and computational statistics. ViSta is designed for an audience of users having a very wide range of data analysis sophistication, ranging from novice to expert. ViSta provides seamlessly integrated data analysis environments specifically tailored to the user's level of expertise. Guidance is available for students and novices, and tools are available for teachers and other experts to create guidance for these novices. A structured graphical user interface is available for competent users, and a command line interface is available for sophisticated users. The complete Lisp-Stat (Tierney, 1990) programming environment is available to researchers, programmers and graduate students who wish to extend ViSta's capabilities. ViSta's data analysis and visualization capabilities include: exploratory graphics and descriptive statistics; univariate tests and visualizations; analysis of variance; regression analysis; principal components analysis; multidimensional scaling and correspondence analysis. All analysis methods have extensive visualization features. The major new features include: 1) A new OLS, robust and monotonic regression module. 2) Redundancy analysis (new option of the multivariate regression module). 3) Extensive visual regression diagnostics. 4) A visualization method for Univariate Analysis. 5) A completed visualization method for Analysis of Variance. 6) An improved visualization method for Multivariate Regression. 7) A multivariate regression test. Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Wed Oct 9 22:10:13 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27653; Wed, 9 Oct 1996 22:10:13 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27648; Wed, 9 Oct 1996 21:10:07 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA21768; Wed, 9 Oct 96 19:10:02 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA22607; Wed, 9 Oct 1996 19:09:59 -0700 From: Jan Deleeuw Message-Id: <9610100209.AA22607@galton.math.ucla.edu> Subject: Xlisp plots on the WWW To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Wed, 9 Oct 1996 19:09:58 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Consider echo "(plot-points (iseq 10) (normal-rand 10))" | lsp2gif.sh where lsp2gif.sh is the file ----------------------------------------------------------------------------- #!/bin/sh echo $1 | xlispbatch | tek2ps -s 75 > tmp.ps gs -q -dNOPAUSE -dNO_PAUSE -sDEVICE=ppmraw -sOutputFile=tmp.ppm tmp.ps -c quit pnmflip -rotate270 tmp.ppm | ppmtogif 2>/dev/null rm tmp.ps tmp.ppm ------------------------------------------------------------------------------ and where xlispbatch is a version of xlispstat which does not write a banner. This produces a gif file with the plot produced by Xlisp-Stat on stdout, and thus this can be called from a WWW page. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Fri Oct 11 18:28:55 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29424; Fri, 11 Oct 1996 18:28:55 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29419; Fri, 11 Oct 1996 17:28:50 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA14173; Fri, 11 Oct 96 15:28:49 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA14224; Fri, 11 Oct 1996 15:28:47 -0700 From: Jan Deleeuw Message-Id: <9610112228.AA14224@galton.math.ucla.edu> Subject: WWW Xlisp Help To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Fri, 11 Oct 1996 15:28:46 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk See http://www.stat.ucla.edu/~deleeuw/courses/153a/outlines/find.html This is a combination of the Mikkelsen and Almy material, quite a bit is sadly out of date, but the general idea is to upgrade this over time 8-( --- Jan -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Sun Oct 13 20:29:05 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26279; Sun, 13 Oct 1996 20:29:05 -0400 Return-Path: Received: from westnet.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26274; Sun, 13 Oct 1996 19:29:00 -0500 Received: from port7.ts2.westnet.com (port7.ts2.westnet.com [206.28.203.7]) by westnet.com (8.7.6/8.7.3) with SMTP id UAA23940 for ; Sun, 13 Oct 1996 20:28:54 -0400 (EDT) Date: Sun, 13 Oct 1996 20:28:54 -0400 (EDT) Message-Id: <199610140028.UAA23940@westnet.com> X-Sender: lt@westnet.com X-Mailer: Windows Eudora Version 1.4.3 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: stat-lisp-news@umnstat.stat.umn.edu From: lt@westnet.com (Larry Troxler) Subject: MS-Win multiline-edit? Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, new to list. In XLISP-STAT, How does one enable multi-line editing for edit-text-item-proto, under MS Windows? Also, how is the text-item-proto class different from the edit-text-item-proto class, given that the :isnew for both of them have a parameter which sets whether the text is editable or not? I think there's a similar situation with some of the other classes. Larry Troxler lt@westnet.com From majordom@umnstat.stat.umn.edu Mon Oct 14 14:01:21 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10542; Mon, 14 Oct 1996 14:01:21 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10537; Mon, 14 Oct 1996 13:01:16 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id OAA11284; Mon, 14 Oct 1996 14:01:07 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 14 Oct 1996 13:59:29 -0500 To: lt@westnet.com (Larry Troxler) From: forrest@unc.edu (Forrest Young) Subject: Re: MS-Win multiline-edit? Cc: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >Hi, new to list. > >In XLISP-STAT, How does one enable multi-line editing for >edit-text-item-proto, under MS Windows? > >Also, how is the text-item-proto class different from the >edit-text-item-proto class, given that the :isnew for both of them have a >parameter which sets whether the text is editable or not? I think there's a >similar situation with some of the other classes. > >Larry Troxler >lt@westnet.com The last I knew (i.e., version 3.45), this wasn't possible. There is a version 3.47 that I havn't tried. Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Mon Oct 14 16:52:11 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11802; Mon, 14 Oct 1996 16:52:11 -0400 Return-Path: Received: from fullfeed.msn.fullfeed.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11797; Mon, 14 Oct 1996 15:52:06 -0500 Received: by fullfeed.msn.fullfeed.com (8.6.9/MSN-1.4) id PAA12865; Mon, 14 Oct 1996 15:52:05 -0500 Received: from localhost by fullfeed.msn.fullfeed.com (8.6.9/MSN-1.4) id NAB29795; Mon, 14 Oct 1996 13:34:39 -0500 Date: Mon, 14 Oct 1996 13:34:39 -0500 From: Mailer-Daemon@msn.fullfeed.com (Mail Delivery Subsystem) Subject: Returned mail: Host unknown (Name server: smtp.pop.msn.fullfeed.com: host not found) Message-Id: <199610141834.NAB29795@fullfeed.msn.fullfeed.com> To: postmaster@msn.fullfeed.com Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The original message was received at Mon, 14 Oct 1996 13:34:36 -0500 from umnstat.stat.umn.edu [128.101.51.1] ----- The following addresses had delivery problems ----- (unrecoverable error) ----- Transcript of session follows ----- 501 ... 550 Host unknown (Name server: smtp.pop.msn.fullfeed.com: host not found) ----- Message header follows ----- Return-Path: Received: from umnstat.stat.umn.edu by fullfeed.msn.fullfeed.com (8.6.9/MSN-1.4) id NAA29792; Mon, 14 Oct 1996 13:34:36 -0500 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10542; Mon, 14 Oct 1996 14:01:21 -0400 Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10537; Mon, 14 Oct 1996 13:01:16 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id OAA11284; Mon, 14 Oct 1996 14:01:07 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 14 Oct 1996 13:59:29 -0500 To: lt@westnet.com (Larry Troxler) From: forrest@unc.edu (Forrest Young) Subject: Re: MS-Win multiline-edit? Cc: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@stat.umn.edu Precedence: bulk ----- Message body suppressed ----- From majordom@umnstat.stat.umn.edu Tue Oct 15 21:00:20 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00535; Tue, 15 Oct 1996 21:00:20 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00530; Tue, 15 Oct 1996 20:00:14 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA11846; Tue, 15 Oct 96 18:00:13 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA29894; Tue, 15 Oct 1996 18:00:10 -0700 From: Jan Deleeuw Message-Id: <9610160100.AA29894@galton.math.ucla.edu> Subject: Spin Plot in VRML To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Tue, 15 Oct 1996 18:00:10 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk See below for something not directly related to Xlisp-Stat, but will may try to combine it with the spin-proto at some point. ======================================================================= At http://www.stat.ucla.edu/calculators/ I have added a VRML spinplot in CGI (inspired by Webster West of the University of South Carolina). You type in three columns of data (or you paste), and submit the form. If your browser knows about the MIME type x-world/x-vrml it will display the plot as a VRML world, in which you can walk, point, rotate, switch on the light, and so on. Points are represented as little spheres, size and color of the spheres is adjustable in the form. Best representation by Netscape 3.0 and the Live3D plug-in, but there are other options (for UNIX there is a problem, unless you have an SGI, or unless you are willing to set up Liquid Reality). If the browser does not know about x-world/x-vrml, then it will suggest you download the file. You can then view the downloaded file in vrweb, or whurlplug, or whatever. There will be updates to use more VRML capacities. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Thu Oct 17 10:23:24 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29438; Thu, 17 Oct 1996 10:23:24 -0400 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29433; Thu, 17 Oct 1996 09:23:20 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id KAA13598; Thu, 17 Oct 1996 10:23:13 -0400 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 17 Oct 1996 10:21:27 -0500 To: vista@unc.edu, edstat-l@jse.stat.ncsu.edu, stat-l@vm1.mcgill.ca, stat-lisp-news@umnstat.stat.umn.edu, sug-l@listserv.oit.unc.edu From: forrest@unc.edu (Forrest Young) Subject: ViSta for Win95 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Announcing the availability for Windows 95 of ViSta - The Visual Statistics System Version 4.24, October 7th, 1996 Copyright (c) 1991-96 by Forrest W. Young ViSta is now freely available for Windows 95. This version is also freely available for MacOS and Windows 3.1, as previously announced. The Unix version will follow. ViSta performs univariate and multivariate statistical data analysis and visualization. The software and documentation for ViSta is available for free from our WWW and FTP servers at: WWW: http://forrest.psych.unc.edu FTP: www.psych.unc.edu, directory pub/forrest/vista. ViSta is useful for teaching multivariate statistics, statistical graphics and computational statistics; and for research and development in statistical graphics and computational statistics. ViSta is designed for an audience of users having a very wide range of data analysis sophistication, ranging from novice to expert. ViSta provides seamlessly integrated data analysis environments specifically tailored to the user's level of expertise. Guidance is available for students and novices, and tools are available for teachers and other experts to create guidance for these novices. A structured graphical user interface is available for competent users, and a command line interface is available for sophisticated users. The complete Lisp-Stat (Tierney, 1990) programming environment is available to researchers, programmers and graduate students who wish to extend ViSta's capabilities. ViSta's data analysis and visualization capabilities include: exploratory graphics and descriptive statistics; univariate tests and visualizations; analysis of variance; regression analysis; principal components analysis; multidimensional scaling and correspondence analysis. All analysis methods have extensive visualization features. The major new features available in this version include: 1) A new OLS, robust and monotonic regression module. 2) Redundancy analysis (new option of the multivariate regression module). 3) Extensive visual regression diagnostics. 4) A visualization method for Univariate Analysis. 5) A completed visualization method for Analysis of Variance. 6) An improved visualization method for Multivariate Regression. 7) A multivariate regression test. Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Sat Oct 19 23:18:07 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA08120; Sat, 19 Oct 1996 23:18:07 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA08115; Sat, 19 Oct 1996 22:18:00 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA01815; Sat, 19 Oct 96 20:17:58 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA21185; Sat, 19 Oct 1996 20:17:56 -0700 From: Jan Deleeuw Message-Id: <9610200317.AA21185@galton.math.ucla.edu> Subject: Demo To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Sat, 19 Oct 1996 20:17:56 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk ;; The little demo below I use in my class to illustrate (a) simple OOP, (b) ;; simple graphics programming, (c) the power method to compute the dominant ;; eigenvalue. ;; Here is the basic prototype we use. It inherits from the ;; graph-window-proto, and introduces six new slots (defproto power-window-proto '(a c s x done eps) nil graph-window-proto) ;; Assessor methods for the five slots. These either ;; display the value in the slot, if called without an ;; argument, or they put an object in a slot if called ;; with an argument (defmeth power-window-proto :a (&optional (val nil set)) (if set (setf (slot-value 'a) val)) (slot-value 'a)) (defmeth power-window-proto :c (&optional (val nil set)) (if set (setf (slot-value 'c) val)) (slot-value 'c)) (defmeth power-window-proto :s (&optional (val nil set)) (if set (setf (slot-value 's) val)) (slot-value 's)) (defmeth power-window-proto :x (&optional (val nil set)) (if set (setf (slot-value 'x) val)) (slot-value 'x)) (defmeth power-window-proto :done (&optional (val nil set)) (if set (setf (slot-value 'done) val)) (slot-value 'done)) (defmeth power-window-proto :eps (&optional (val nil set)) (if set (setf (slot-value 'eps) val)) (slot-value 'eps)) ;; ;; Some methods that provide some assistance ;; (defmeth power-window-proto :coord-map (x) (let* ( (s (send self :s)) (c (send self :c)) (b (/ s c)) ) (list (round (+ s (* b (first x)))) (round (- s (* b (second x))))) )) (defmeth power-window-proto :a-line (x y) (let ( (u (send self :coord-map x)) (v (send self :coord-map y)) ) (send self :draw-line (first u) (second u) (first v) (second v)) )) (defmeth power-window-proto :e-line (x y) (let ( (u (send self :coord-map x)) (v (send self :coord-map y)) ) (send self :draw-mode 'xor) (send self :draw-line (first u) (second u) (first v) (second v)) (send self :draw-mode 'normal) )) (defmeth power-window-proto :unit-circle () (let ( (u (send self :coord-map '(-1 +1))) (b (* 2 (/ (send self :s) (send self :c)))) ) (send self :frame-oval (first u) (second u) b b) )) (defmeth power-window-proto :redraw () (send self :unit-circle) (send self :a-line '(0 0) (send self :x)) ) (defmeth power-window-proto :an-iter () (let* ( (old-x (send self :x)) (raw-x (update (send self :a) old-x)) (new-x (normalize raw-x)) (pnt-x (send self :coord-map new-x)) ) (send self :erase-window) (send self :unit-circle) (pause 20) (send self :a-line '(0 0) old-x) (pause 20) (send self :a-line '(0 0) raw-x) (pause 20) (send self :e-line '(0 0) raw-x) (send self :a-line '(0 0) new-x) (send self :draw-string (format nil "~8,4f" (elength raw-x)) (first pnt-x) (second pnt-x)) (pause 40) (if (test old-x new-x (send self :eps)) (send self :done t) (send self :x new-x)) )) ;; and the constructor function (defun power-method (a &key (size 300) (c 3) (eps 1e-6)) (let ( (w (send power-window-proto :new :size (repeat size 2) :title "Power Method")) ) (send w :s (/ size 2)) (send w :c c) (send w :a a) (send w :x '(1 0)) (send w :eps eps) (send w :done nil) (loop (if (send w :done) (return) (send w :an-iter))) )) ;; These are some simple helpers (defun update (a x) (matmult a x) ) (defun normalize (x) (/ x (elength x)) ) (defun test (x y eps) (> eps (max (abs (- x y)))) ) (defun elength (x) (sqrt (sum (* x x))) ) ;; And fire away ... ;;(power-method #2a((2 -1)(-1 2))) (power-method #2a((1 .01)(.01 2))) ;;(power-method #2a((2 1)(-2 1))) -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Tue Oct 22 00:19:05 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA09748; Tue, 22 Oct 1996 00:19:05 -0400 Return-Path: Received: from kasuga.cc.chuo-u.ac.jp by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA09743; Mon, 21 Oct 1996 23:18:48 -0500 Received: from isg.indsys.chuo-u.ac.jp (isg.indsys.chuo-u.ac.jp [133.91.120.136]) by kasuga.cc.chuo-u.ac.jp (8.7.4+2.6Wbeta6/3.4W4) with SMTP id NAA15965 for ; Tue, 22 Oct 1996 13:18:28 +0900 (JST) Received: from [133.91.120.87] (mactk) by isg.indsys.chuo-u.ac.jp (4.1/6.4J.6-fps0) id AA02257; Tue, 22 Oct 96 13:21:46 JST Date: Tue, 22 Oct 96 13:21:46 JST Message-Id: <9610220421.AA02257@isg.indsys.chuo-u.ac.jp> To: stat-lisp-news@umnstat.stat.umn.edu From: kamakura@indsys.chuo-u.ac.jp (Toshinari Kamakura) X-Sender: kamakura@indsys.chuo-u.ac.jp (Unverified) Subject: bugs? Mime-Version: 1.0 Content-Type: text/plain; charset=iso-2022-jp X-Mailer: Eudora-J(1.3.8-J13) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, I mostly use the Mac Xlispstat. When I change the color, I cannot use the make-color command. It seems to be a bug for the Mac version. I tried the folloing two cases. Two lines (one is yellow, and the other is violet) shall be drawn if the system works correctly. But on the Macintosh version the case 1 does not work properly. On the windows version the case 1 does work. The 2nd case of course works on the both versions. Does anyone know the color bug-fixed version for the Macintosh? I would like to present the color images on the Lispstat window. ;;; color-change test ;;; (setf w (send graph-window-proto :new)) (send w :use-color t) ;;; case 1 (make-color 'my-color 1 1 0) (send w :draw-color 'my-color) (send w :draw-line 50 50 100 100) (make-color 'my-color 1 0 1) (send w :draw-color 'my-color) (send w :draw-line 100 50 50 100) ;;; case 2 (make-color 'my-color1 1 1 0) (send w :draw-color 'my-color1) (send w :draw-line 50 50 100 100) (make-color 'my-color2 1 0 1) (send w :draw-color 'my-color2) (send w :draw-line 100 50 50 100) +++++++++++++++++++++++++++++++++++++++++++++++ Toshinari Kamakura Chuo University 1-13-27 Kasuga Bunkyo-ku, Tokyo 112 Japan TEL +81-3-3817-1934 FAX +81-3-3817-1943 email: kamakura@indsys.chuo-u.ac.jp +++++++++++++++++++++++++++++++++++++++++++++++ From majordom@nokomis.stat.umn.edu Tue Oct 22 07:42:23 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA10364; Tue, 22 Oct 1996 07:42:23 -0400 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA10357; Tue, 22 Oct 1996 06:41:38 -0500 From: Luke Tierney Message-Id: <9610221141.AA10357@nokomis.stat.umn.edu> Subject: Re: bugs? To: kamakura@indsys.chuo-u.ac.jp (Toshinari Kamakura) Date: Tue, 22 Oct 1996 06:41:37 -0500 (CDT) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: <9610220421.AA02257@isg.indsys.chuo-u.ac.jp> from "Toshinari Kamakura" at Oct 22, 96 01:21:46 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1617 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > Hi, > > I mostly use the Mac Xlispstat. When I change the color, > I cannot use the make-color command. It seems to be a bug > for the Mac version. I tried the folloing two cases. > Two lines (one is yellow, and the other is violet) shall be > drawn if the system works correctly. > But on the Macintosh version the case 1 does not work > properly. On the windows version the case 1 does work. > The 2nd case of course works on the both versions. > > Does anyone know the color bug-fixed version for the Macintosh? > I would like to present the color images on the Lispstat window. > > ;;; color-change test > ;;; > > (setf w (send graph-window-proto :new)) > (send w :use-color t) > > ;;; case 1 > > > (make-color 'my-color 1 1 0) > (send w :draw-color 'my-color) > (send w :draw-line 50 50 100 100) > > (make-color 'my-color 1 0 1) > (send w :draw-color 'my-color) > (send w :draw-line 100 50 50 100) > > > ;;; case 2 > > (make-color 'my-color1 1 1 0) > (send w :draw-color 'my-color1) > (send w :draw-line 50 50 100 100) > > (make-color 'my-color2 1 0 1) > (send w :draw-color 'my-color2) > (send w :draw-line 100 50 50 100) > > > > > +++++++++++++++++++++++++++++++++++++++++++++++ > Toshinari Kamakura > > Chuo University > 1-13-27 Kasuga > Bunkyo-ku, Tokyo 112 > Japan > > TEL +81-3-3817-1934 > FAX +81-3-3817-1943 > > email: kamakura@indsys.chuo-u.ac.jp > +++++++++++++++++++++++++++++++++++++++++++++++ > > Make sure you turned color use on in your window, for example with (send w :use-color t) luke From majordom@umnstat.stat.umn.edu Tue Oct 22 21:28:09 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25012; Tue, 22 Oct 1996 21:28:09 -0400 Return-Path: Received: from kasuga.cc.chuo-u.ac.jp by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25007; Tue, 22 Oct 1996 20:27:31 -0500 Received: from isg.indsys.chuo-u.ac.jp (isg.indsys.chuo-u.ac.jp [133.91.120.136]) by kasuga.cc.chuo-u.ac.jp (8.7.4+2.6Wbeta6/3.4W4) with SMTP id KAA17247; Wed, 23 Oct 1996 10:27:13 +0900 (JST) Received: from [133.91.120.87] (mactk) by isg.indsys.chuo-u.ac.jp (4.1/6.4J.6-fps0) id AA16802; Wed, 23 Oct 96 10:30:32 JST Date: Wed, 23 Oct 96 10:30:32 JST Message-Id: <9610230130.AA16802@isg.indsys.chuo-u.ac.jp> To: Luke Tierney From: kamakura@indsys.chuo-u.ac.jp (Toshinari Kamakura) Subject: Re: bugs? Cc: stat-lisp-news@umnstat.stat.umn.edu Mime-Version: 1.0 Content-Type: text/plain; charset=iso-2022-jp X-Mailer: Eudora-J(1.3.8-J13) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk At 6:41 AM 96.10.22 -0500, Luke Tierney wrote: >> >> Hi, >> >> I mostly use the Mac Xlispstat. When I change the color, >> I cannot use the make-color command. It seems to be a bug >> for the Mac version. I tried the folloing two cases. >> Two lines (one is yellow, and the other is violet) shall be >> drawn if the system works correctly. >> But on the Macintosh version the case 1 does not work >> properly. On the windows version the case 1 does work. >> The 2nd case of course works on the both versions. >> >> Does anyone know the color bug-fixed version for the Macintosh? >> I would like to present the color images on the Lispstat window. >> >> ;;; color-change test >> ;;; >> >> (setf w (send graph-window-proto :new)) >> (send w :use-color t) >> >> ;;; case 1 >> >> >> (make-color 'my-color 1 1 0) >> (send w :draw-color 'my-color) >> (send w :draw-line 50 50 100 100) >> >> (make-color 'my-color 1 0 1) >> (send w :draw-color 'my-color) >> (send w :draw-line 100 50 50 100) >> >> >> ;;; case 2 >> >> (make-color 'my-color1 1 1 0) >> (send w :draw-color 'my-color1) >> (send w :draw-line 50 50 100 100) >> >> (make-color 'my-color2 1 0 1) >> (send w :draw-color 'my-color2) >> (send w :draw-line 100 50 50 100) >> >> >> >> >> >> > >Make sure you turned color use on in your window, for example with > > (send w :use-color t) > >luke Sure! Please check my e-mal which include this line. I am grealty happy to get the Mac new version. +++++++++++++++++++++++++++++++++++++++++++++++ Toshinari Kamakura Chuo University 1-13-27 Kasuga Bunkyo-ku, Tokyo 112 Japan TEL +81-3-3817-1934 FAX +81-3-3817-1943 email: kamakura@indsys.chuo-u.ac.jp +++++++++++++++++++++++++++++++++++++++++++++++ From majordom@umnstat.stat.umn.edu Wed Oct 23 08:18:40 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05914; Wed, 23 Oct 1996 08:18:40 -0400 Return-Path: Received: from BALU.kfunigraz.ac.at by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05907; Wed, 23 Oct 1996 07:18:19 -0500 Received: from pcim82 (BIMIGW.kfunigraz.ac.at [143.50.40.2]) by balu (950413.SGI.8.6.12/950213.SGI.AUTOCF) via SMTP id OAA01944 for ; Wed, 23 Oct 1996 14:17:56 +0200 Message-Id: <326E296D.7F7D@kfunigraz.ac.at> Date: Wed, 23 Oct 1996 14:19:25 -0100 From: Franz Quehenberger Organization: Karl-Franzens-Universitaet Graz X-Mailer: Mozilla 3.0 (WinNT; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: Q: Data Bases and XLispStat? Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Has anybody tried to connect to a data base from XLispStat? At the moment I try Mini SQL, an SQL-Server on a Decstation that is free of charge for universities. However I need to communicate with it with a C library or with a script language like Perl. This approach quite cumbersome to me, as I know none of the two languages. TIA -- franz.quehenberger@kfunigraz.ac.at (++43 316 385-3205, fax -3590) Dept. of Medical Informatics, Statistics, and Documentation Graz University, Auenbruggerplatz 9/III, A-8036 Graz, Austria From majordom@umnstat.stat.umn.edu Wed Oct 23 10:06:19 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06594; Wed, 23 Oct 1996 10:06:19 -0400 Return-Path: Received: from COMPSTAT.WHARTON.UPENN.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06589; Wed, 23 Oct 1996 09:06:14 -0500 Received: from popeye.wharton.upenn.edu (POPEYE.WHARTON.UPENN.EDU [165.123.8.105]) by compstat.wharton.upenn.edu (8.7.1/8.7.1) with SMTP id KAA07078; Wed, 23 Oct 1996 10:05:03 -0400 (EDT) Date: Wed, 23 Oct 1996 09:35:10 -0400 (EDT) From: Russell Almond X-Sender: ralmond@popeye.wharton.upenn.edu Reply-To: almond@acm.org To: Franz Quehenberger Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Q: Data Bases and XLispStat? In-Reply-To: <326E296D.7F7D@kfunigraz.ac.at> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Wed, 23 Oct 1996, Franz Quehenberger wrote: > Has anybody tried to connect to a data base from XLispStat? Have you looked at the comp.lang.lisp FAQ? While I'm willing to bet that nobody has done it specifically for Xlisp, it is a fairly common Lisp problem and their may be some public domain solutions which can be adapted to xlisp. (I know we did a proprietary one between Allegro CL and OOR database while I was at StatSci). Also, somebody at University of Washington wrote some code to link Xlispstat and Splus. It occurs to me that you might be able to adapt that to your needs. > > At the moment I try Mini SQL, an SQL-Server on a Decstation that is free > of charge for universities. However I need to communicate with it with a > C library or with a script language like Perl. This approach quite > cumbersome to me, as I know none of the two languages. > I'm afraid that if you are working with database, then into each life some SQL must fall. I suspect that whatever your solution, you will either need to manipulate or generate SQL, Perl or some similiar data manipulation language statements within Xlisp and send them to you database. The alternative is to read the entire data table into Xlispstat and use the manipulation equations there. This will be much slower as Xlisp simply isn't optimized for database operations like a true database system. --Russell Almond Educational Testing Service almond@acm.org From majordom@umnstat.stat.umn.edu Thu Oct 24 18:39:36 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA09135; Thu, 24 Oct 1996 18:39:36 -0400 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA09130; Thu, 24 Oct 1996 17:39:28 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa24984; 24 Oct 96 18:38 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA21965; Thu, 24 Oct 1996 18:38:47 -0400 Date: Thu, 24 Oct 1996 18:38:46 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU Reply-To: "Steven D. Majewski" To: Toshinari Kamakura Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: bugs? [ And printing on the Mac ] In-Reply-To: <9610220421.AA02257@isg.indsys.chuo-u.ac.jp> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Tue, 22 Oct 1996, Toshinari Kamakura wrote: > Hi, > > I mostly use the Mac Xlispstat. When I change the color, > I cannot use the make-color command. It seems to be a bug > for the Mac version. I tried the folloing two cases. > Two lines (one is yellow, and the other is violet) shall be > drawn if the system works correctly. > But on the Macintosh version the case 1 does not work > properly. On the windows version the case 1 does work. > The 2nd case of course works on the both versions. > > Does anyone know the color bug-fixed version for the Macintosh? > I would like to present the color images on the Lispstat window. > There is one color bug in the 3.47 Mac version that I know of -- see: I don't know if that is the source of your problem, or if that bug exists in earlier versions. A compiled Mac PPC binary with this and some other fixes, and modified CW9 sources and project files are available at: [ I have CW10 projects and 68K binaries that haven't been moved to the server. I haven't had time to test them on a 68K machine yet, but they seem to work (slowly) with SoftwareFPU installed on my PowerMac. If anyone wants to to test them out let me know. ] BTW: Since XlispStat on the Mac has no 'print' function for graphics, I had been cutting and pasting my graphs into other programs - I tried WordPerfect, and a couple of image conversion or paint programs with various levels of success. I've just recently started using Claris Works, and I discovered if you paste you graph from LispStat into a Claris Works DRAW document, the graph can be UNGROUPED and various elements can be edited - you can change the text fonts or the background colors, for example. ( changeing the color or thickness of lines is rather tedious, though, because once you ungroup the graph, each line segment is a separate line. ) This obviously implies that the quickdraw command sequence, and not just the resulting image bitmap is being cut and pasted - which also means that you get full printing resolution. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Thu Oct 24 20:48:53 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA09717; Thu, 24 Oct 1996 20:48:53 -0400 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA09712; Thu, 24 Oct 1996 19:48:47 -0500 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA18969; Thu, 24 Oct 96 17:48:45 PDT Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA11410; Thu, 24 Oct 1996 17:48:43 -0700 From: Jan Deleeuw Message-Id: <9610250048.AA11410@galton.math.ucla.edu> Subject: The Mac, the Clipboard, and PICT To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Thu, 24 Oct 1996 17:48:42 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk This is a brief follow-up to Steven Majewski's earlier note. If you copy an Xlisp-Stat window to the clipboard, it becomes a PICT file, which contains the info in the picture in vector format. It can be edited in SuperPaint, Cricket Draw, MacDraw, Claris Works, and so on. You can also get the information out with DeReZ or ResEdit (see Inside Macintosh, Imaging with QuickDraw, Appendix A, table A3 for the opcodes, and Listing A-7 for a sample version 1 picture). As Steven remarks, this means that the clipboard contains all the information for 600 dpi printing, bypassing the bitmap, and functions that convert to gnuplot or xfig are not needed on the Mac. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Fri Oct 25 10:17:17 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA20757; Fri, 25 Oct 1996 10:17:17 -0400 Return-Path: Received: from rgborafsa.er.usgs.gov by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA20747; Fri, 25 Oct 1996 09:17:11 -0500 Received: from [130.11.54.131] (mcmac.er.usgs.gov [130.11.54.131]) by rgborafsa.er.usgs.gov (Geomail 1.2.3) with SMTP id KAA18464 for ; Fri, 25 Oct 1996 10:17:44 -0400 Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 25 Oct 1996 10:18:01 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: "Richard B McCammon Reston, VA " (Richard McCammon (USGS)) Subject: grapher? Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I would be interested in knowing if there is anyone out there who has crafted in Xlisp or Xlisp-Stat anything analogous to GRAPHER.DCOM that was one of the library functions in the InterLisp-D programming environment. This was an extremely versatile tool for creating, modifying, and incorporating tree structures used for inference nets in AI-related programs. GRAPHER was an integral part of the Prospector program USGS inherited from SRI. __________________________________________________________ snail mail: Richard B. McCammon Eastern Minerals Team U.S. Geological Survey MS-954 12201 Sunrise Valley Drive Reston, VA 20192 USA e-mail: mccammon@rgborafsa.er.usgs.gov or mccammon@130.11.54.100 Phone: 703-648-6150 Fax: 703-648-6383 From majordom@umnstat.stat.umn.edu Sun Oct 27 20:49:39 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA20825; Sun, 27 Oct 1996 20:49:39 -0500 Return-Path: Received: from westnet.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA20820; Sun, 27 Oct 1996 19:49:33 -0600 Received: from port11.ts2.westnet.com (port11.ts2.westnet.com [206.28.203.11]) by westnet.com (8.7.6/8.7.3) with SMTP id UAA25095 for ; Sun, 27 Oct 1996 20:49:26 -0500 (EST) Date: Sun, 27 Oct 1996 20:49:26 -0500 (EST) Message-Id: <199610280149.UAA25095@westnet.com> X-Sender: lt@westnet.com X-Mailer: Windows Eudora Version 1.4.3 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: stat-lisp-news@umnstat.stat.umn.edu From: lt@westnet.com (Larry Troxler) Subject: return-from in defmeth's ? Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Is it possible to use "return-from" to abort a loop inside a method definition (using "defmeth")? If so, what is the block name that I would be returning from? Larry From majordom@umnstat.stat.umn.edu Tue Oct 29 17:28:43 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA21556; Tue, 29 Oct 1996 17:28:43 -0500 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA21551; Tue, 29 Oct 1996 16:28:38 -0600 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA16470 (5.67b/IDA-1.5 for ); Tue, 29 Oct 1996 14:26:58 -0500 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA7278 for ; Tue, 29 Oct 1996 09:52:06 -0400 Message-Id: <327643B8.5612@kodak.com> Date: Tue, 29 Oct 1996 09:49:44 -0800 From: Jeff Morris Organization: Eastman Kodak Company X-Mailer: Mozilla 2.01 (Win16; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: GLIM Null Deviance Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk How does one get the null deviance for fitting a constant only model to say, binomialreg-model data? Thanks! -- Jeffrey D. Morris, Ph.D. Health & Environment Labs Eastman Kodak Company morris@kodak.com From majordom@umnstat.stat.umn.edu Tue Oct 29 21:00:48 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23106; Tue, 29 Oct 1996 21:00:48 -0500 Return-Path: Received: from kasuga.cc.chuo-u.ac.jp by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23096; Tue, 29 Oct 1996 20:00:38 -0600 Received: from isg.indsys.chuo-u.ac.jp (isg.indsys.chuo-u.ac.jp [133.91.120.136]) by kasuga.cc.chuo-u.ac.jp (8.7.4+2.6Wbeta6/3.4W4) with SMTP id LAA01008; Wed, 30 Oct 1996 11:00:20 +0900 (JST) Received: from [133.91.120.87] (mactk) by isg.indsys.chuo-u.ac.jp (4.1/6.4J.6-fps0) id AA10667; Wed, 30 Oct 96 11:03:42 JST Date: Wed, 30 Oct 96 11:03:42 JST Message-Id: <9610300203.AA10667@isg.indsys.chuo-u.ac.jp> To: "Steven D. Majewski" From: kamakura@indsys.chuo-u.ac.jp (Toshinari Kamakura) Subject: Re: bugs? [ And printing on the Mac ] Cc: stat-lisp-news@umnstat.stat.umn.edu Mime-Version: 1.0 Content-Type: text/plain; charset=iso-2022-jp X-Mailer: Eudora-J(1.3.8-J13) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Thank you for Macintosh latest version. >There is one color bug in the 3.47 Mac version that I know of -- >see: > > >I don't know if that is the source of your problem, or if that >bug exists in earlier versions. > > >A compiled Mac PPC binary with this and some other fixes, and modified >CW9 sources and project files are available at: > > I am afraid that this version also include bugs. It is impossible to change the color table assignment. This means that the number of colors handled is restrected to the size of the color table. In my experiences the lates Mac version can manipulate less than 110 colors. In the case of Windows version we can use as many colors as we wish without regard to speed. I need 256 colors. Do you have any idea? +++++++++++++++++++++++++++++++++++++++++++++++ Toshinari Kamakura Chuo University 1-13-27 Kasuga Bunkyo-ku, Tokyo 112 Japan TEL +81-3-3817-1934 FAX +81-3-3817-1943 email: kamakura@indsys.chuo-u.ac.jp +++++++++++++++++++++++++++++++++++++++++++++++ From majordom@umnstat.stat.umn.edu Wed Oct 30 00:45:17 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23997; Wed, 30 Oct 1996 00:45:17 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23992; Tue, 29 Oct 1996 23:45:11 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa18925; 30 Oct 96 0:44 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA12963; Wed, 30 Oct 1996 00:44:19 -0500 Date: Wed, 30 Oct 1996 00:44:18 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Toshinari Kamakura Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: bugs? [ And printing on the Mac ] In-Reply-To: <9610300203.AA10667@isg.indsys.chuo-u.ac.jp> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Wed, 30 Oct 1996, Toshinari Kamakura wrote: > > I am afraid that this version also include bugs. > It is impossible to change the color table assignment. > This means that the number of colors handled is restrected > to the size of the color table. In my experiences the > lates Mac version can manipulate less than 110 colors. > In the case of Windows version we can use as many colors > as we wish without regard to speed. > > I need 256 colors. Do you have any idea? > OK - I didn't understand the problem the first time around... In maciviewwindow2.c are the constants: #define NumBasicColors 8 #define NumRGBColors 128 /* sdm7g up from 100 */ The latter was 100 in the original source code. I had bumped it up to 200 in a previous build, but then reduced it in the current version. You can try setting it to 255. ( I'll try that on my next build, but I'm hoping to finish several other changes before I'm ready to post a new one. ) ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Wed Oct 30 01:56:25 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24156; Wed, 30 Oct 1996 01:56:25 -0500 Return-Path: Received: from kasuga.cc.chuo-u.ac.jp by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24150; Wed, 30 Oct 1996 00:54:27 -0600 Received: from isg.indsys.chuo-u.ac.jp (isg.indsys.chuo-u.ac.jp [133.91.120.136]) by kasuga.cc.chuo-u.ac.jp (8.7.4+2.6Wbeta6/3.4W4) with SMTP id PAA12727; Wed, 30 Oct 1996 15:54:03 +0900 (JST) Received: from [133.91.120.87] (mactk) by isg.indsys.chuo-u.ac.jp (4.1/6.4J.6-fps0) id AA12650; Wed, 30 Oct 96 15:57:26 JST Date: Wed, 30 Oct 96 15:57:26 JST Message-Id: <9610300657.AA12650@isg.indsys.chuo-u.ac.jp> To: "Steven D. Majewski" From: kamakura@indsys.chuo-u.ac.jp (Toshinari Kamakura) Subject: Re: bugs? [ And printing on the Mac ] Cc: stat-lisp-news@umnstat.stat.umn.edu Mime-Version: 1.0 Content-Type: text/plain; charset=iso-2022-jp X-Mailer: Eudora-J(1.3.8-J13) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Thanks. >In maciviewwindow2.c are the constants: > >#define NumBasicColors 8 >#define NumRGBColors 128 /* sdm7g up from 100 */ > >The latter was 100 in the original source code. I had bumped it >up to 200 in a previous build, but then reduced it in the current >version. You can try setting it to 255. >( I'll try that on my next build, but I'm hoping to finish several >other changes before I'm ready to post a new one. ) > Please note that if we can change the color table allocation, the size of the color table does not matter so much. I do not know the size of the color table in the case of the PC Window version. But perhaps we could handle the millions color if the hardware were availabel. Because we could easily change the color tabel allocation, using 'free-color' in case of the table full of colors. I hope that you will be able to release the version with capability of changing the color allocation table. +++++++++++++++++++++++++++++++++++++++++++++++ Toshinari Kamakura Chuo University 1-13-27 Kasuga Bunkyo-ku, Tokyo 112 Japan TEL +81-3-3817-1934 FAX +81-3-3817-1943 email: kamakura@indsys.chuo-u.ac.jp +++++++++++++++++++++++++++++++++++++++++++++++ From majordom@umnstat.stat.umn.edu Wed Oct 30 04:11:11 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03756; Wed, 30 Oct 1996 04:11:11 -0500 Return-Path: Received: from kasuga.cc.chuo-u.ac.jp by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03751; Wed, 30 Oct 1996 03:10:57 -0600 Received: from isg.indsys.chuo-u.ac.jp (isg.indsys.chuo-u.ac.jp [133.91.120.136]) by kasuga.cc.chuo-u.ac.jp (8.7.4+2.6Wbeta6/3.4W4) with SMTP id SAA20049; Wed, 30 Oct 1996 18:10:13 +0900 (JST) Received: from [133.91.120.87] (mactk) by isg.indsys.chuo-u.ac.jp (4.1/6.4J.6-fps0) id AA14668; Wed, 30 Oct 96 18:13:36 JST Date: Wed, 30 Oct 96 18:13:36 JST Message-Id: <9610300913.AA14668@isg.indsys.chuo-u.ac.jp> To: "Steven D. Majewski" From: kamakura@indsys.chuo-u.ac.jp (Toshinari Kamakura) X-Sender: kamakura@indsys.chuo-u.ac.jp (Unverified) Subject: sort-data Cc: stat-lisp-news@umnstat.stat.umn.edu Mime-Version: 1.0 Content-Type: text/plain; charset=iso-2022-jp X-Mailer: Eudora-J(1.3.8-J13) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Dear Sir, I tried the function 'sort-data' using the your modified version. XLISP-PLUS version 2.1g Portions Copyright (c) 1988, by David Betz. Modified by Thomas Almy and others. XLISP-STAT 2.1 Release 3.47 (Beta-2-Mac). Copyright (c) 1989-1994, by Luke Tierney. It took much time for sorting the data sets of size 10000. > (def x (normal-rand 10000)) X > (time (def y (sort-data x))) The evaluation took 60.35 seconds; 0.20 seconds in gc. Y In the 3.44 version it took not so much. The followings are the results: XLISP-PLUS version 2.1g Portions Copyright (c) 1988, by David Betz. Modified by Thomas Almy and others. XLISP-STAT 2.1 Release 3.44 (Beta). Copyright (c) 1989-1994, by Luke Tierney. > (def x (normal-rand 10000)) X > (time (def y (sort-data x))) The evaluation took 0.50 seconds; 0.22 seconds in gc. Y How do you thik of this? +++++++++++++++++++++++++++++++++++++++++++++++ Toshinari Kamakura Chuo University 1-13-27 Kasuga Bunkyo-ku, Tokyo 112 Japan TEL +81-3-3817-1934 FAX +81-3-3817-1943 email: kamakura@indsys.chuo-u.ac.jp +++++++++++++++++++++++++++++++++++++++++++++++ From majordom@umnstat.stat.umn.edu Wed Oct 30 10:53:16 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05321; Wed, 30 Oct 1996 10:53:16 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05316; Wed, 30 Oct 1996 09:53:11 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa26076; 30 Oct 96 10:51 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA27369; Wed, 30 Oct 1996 10:51:20 -0500 Date: Wed, 30 Oct 1996 10:51:19 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU Reply-To: "Steven D. Majewski" To: Toshinari Kamakura Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: sort-data In-Reply-To: <9610300913.AA14668@isg.indsys.chuo-u.ac.jp> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Wed, 30 Oct 1996, Toshinari Kamakura wrote: > I tried the function 'sort-data' using the your > modified version. > > XLISP-PLUS version 2.1g > Portions Copyright (c) 1988, by David Betz. > Modified by Thomas Almy and others. > XLISP-STAT 2.1 Release 3.47 (Beta-2-Mac). > Copyright (c) 1989-1994, by Luke Tierney. > > It took much time for sorting the data sets of size 10000. > In the 3.44 version it took not so much. > How do you thik of this? I haven't made any changes to sort-data, so I have no idea why there is a difference. I'll look into this though. [ I don't have 3.44 source on hand, so I don't know what else may be different in 3.47 ] Any other significant differences when running the test ? For example, amount of memory allocated for the program ( Do a "Get Info" on the executables. ) You might also want to run it several times and average the values. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Wed Oct 30 11:47:31 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06035; Wed, 30 Oct 1996 11:47:31 -0500 Return-Path: Received: from avon.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06029; Wed, 30 Oct 1996 10:47:23 -0600 Received: by avon.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.21 † id IAA02883; Wed, 30 Oct 1996 08:47:13 -0800 (PST) Date: Wed, 30 Oct 1996 08:47:10 -0800 (PST) From: Thomas Lumley X-Sender: thomas@avon To: Jeff Morris Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: GLIM Null Deviance In-Reply-To: <327643B8.5612@kodak.com> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Tue, 29 Oct 1996, Jeff Morris wrote: > How does one get the null deviance for fitting a constant only > model to say, binomialreg-model data? There does not seem to be a null-deviance calculated anywhere. The simplest way is probably to fit a model without an intercept and with a single predictor which is a column of 1s and use its deviance thomas lumley UW biostat From majordom@nokomis.stat.umn.edu Wed Oct 30 13:45:19 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA20275; Wed, 30 Oct 1996 13:45:19 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA20265; Wed, 30 Oct 1996 12:45:01 -0600 From: Luke Tierney Message-Id: <9610301845.AA20265@nokomis.stat.umn.edu> Subject: Re: sort-data To: sdm7g@Virginia.EDU Date: Wed, 30 Oct 1996 12:45:01 -0600 (CST) Cc: kamakura@indsys.chuo-u.ac.jp, stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Steven D. Majewski" at Oct 30, 96 10:51:19 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 2040 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > > On Wed, 30 Oct 1996, Toshinari Kamakura wrote: > > > I tried the function 'sort-data' using the your > > modified version. > > > > XLISP-PLUS version 2.1g > > Portions Copyright (c) 1988, by David Betz. > > Modified by Thomas Almy and others. > > XLISP-STAT 2.1 Release 3.47 (Beta-2-Mac). > > Copyright (c) 1989-1994, by Luke Tierney. > > > > It took much time for sorting the data sets of size 10000. > > > In the 3.44 version it took not so much. > > > How do you thik of this? > > > I haven't made any changes to sort-data, so I have no idea why > there is a difference. I'll look into this though. > [ I don't have 3.44 source on hand, so I don't know what else > may be different in 3.47 ] > > Any other significant differences when running the test ? > For example, amount of memory allocated for the program > ( Do a "Get Info" on the executables. ) > You might also want to run it several times and average the > values. > > > ---| Steven D. Majewski (804-982-0831) |--- > ---| Department of Molecular Physiology and Biological Physics |--- > ---| University of Virginia Health Sciences Center |--- > ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- > By doing just a little every day, you can gradually > let the task completely overwhelm you. > > The code for sort-data did not change between 3.44 and 3.47 -- both use the C library routine qsort (for better or worse). It is possible that Metrowerks changed that routine between the two compiler releases used. Another possibility (maybe related) is that somehow some 68K is being used in the new version instead of native PPC code -- though eventhat usually only accounts for a factor of 10 slowdown. My macs are in a state of disrepair at the moment. Once they get back up I'll see what gives in the newest CodeWarrior and replace qsort if necessary. I've been meaning to replace it by a merge sort sooner or later -- maybe it will have to be sooner. luke From majordom@nokomis.stat.umn.edu Wed Oct 30 13:53:57 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA20349; Wed, 30 Oct 1996 13:53:57 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA20341; Wed, 30 Oct 1996 12:53:52 -0600 From: Luke Tierney Message-Id: <9610301853.AA20341@nokomis.stat.umn.edu> Subject: Re: bugs? To: kamakura@indsys.chuo-u.ac.jp (Toshinari Kamakura) Date: Wed, 30 Oct 1996 12:53:52 -0600 (CST) Cc: luke@nokomis.stat.umn.edu, stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: <9610230130.AA16802@isg.indsys.chuo-u.ac.jp> from "Toshinari Kamakura" at Oct 23, 96 10:30:32 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 906 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > >> > >> ;;; case 1 > >> > >> > >> (make-color 'my-color 1 1 0) > >> (send w :draw-color 'my-color) > >> (send w :draw-line 50 50 100 100) > >> > >> (make-color 'my-color 1 0 1) > >> (send w :draw-color 'my-color) > >> (send w :draw-line 100 50 50 100) > >> > >> > >> ;;; case 2 > >> > >> (make-color 'my-color1 1 1 0) > >> (send w :draw-color 'my-color1) > >> (send w :draw-line 50 50 100 100) > >> > >> (make-color 'my-color2 1 0 1) > >> (send w :draw-color 'my-color2) > >> (send w :draw-line 100 50 50 100) > >> I didn't follow this at first -- a bit too subtle for me :-). The problem is that in case 1 mycolor is redefined without first being free'd. The intent was that you call free-color before reusing a color symbol. I consider this use an error, but currently no error is signalled and the exact results are implementation dependent. I'll signal an error in future versions. luke From majordom@umnstat.stat.umn.edu Wed Oct 30 17:21:59 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA09913; Wed, 30 Oct 1996 17:21:59 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA09908; Wed, 30 Oct 1996 16:21:55 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa25447; 30 Oct 96 17:16 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA39624; Wed, 30 Oct 1996 17:16:39 -0500 Date: Wed, 30 Oct 1996 17:16:38 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Cc: Luke Tierney Subject: tracking current directory on Mac XlispStat Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk A while ago I posted a question about the problems I was having with the current directory used by the Standard File Dialogs and the current directory set and returned by the get|set-working-directory functions not tracking each other. You can control whether the standard file dialogs should update the working directory via an optional parameter. The default is to track, so ( open-file-dialog ) updates the current working directory to the one from which you chose a file ( No change if you Cancel dialog ) and, ( open-file-dialog Nil ) prevents the current-working-directory from being updated by the dialog. However, the other direction did not work -- you could not "preload" the directory where the Standard File Dialog would begin. A couple of lines added to chdir() in dirent.c appears to fix this. ( You don't really need the defines, but I started translating from the Pascal example in the Inside Mac Files volume, and I used the same names as their example before I found the Metrowerks Note to use the LowMem.h functions. ) Not extremely tested ! I only ran it a couple of times on my Mac but it seemed to work properly with no problems. No additional error checking -- If you're going to have a problem from stuffing ints into low memory globals, they are going to show up in some other odd way. But it ought to repeat the check on dd_error before trying to set the values. I'll change that in the final version and post it on with the other mods. ( probably not right away -- as noted in another thread, I'm going to try to finish a couple of other fixes before I package them up. ) The lines commented with /* sdm7g */ are the added code. There is another definition of chdir right above this one, wrapped in a "#ifdef DODO" clause -- with no DODO's defined, apparently. Is this old bypassed code, or is there a case for DODO ? ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. #include /* sdm7g */ #define SetSFCurVol( val ) LMSetSFSaveDisk( (short) val ) /* sdm7g */ #define SetSFCurDir( val ) LMSetCurDirStore( (long) val ) /* sdm7g */ int chdir(char *path) { WDPBRec pb; pb.ioCompletion = nil; pb.ioNamePtr = nil; if ((dd_errno = PBHGetVol(&pb, false)) != noErr) return -1; strcpy(buf, path); CtoPstr(buf); pb.ioNamePtr = (StringPtr) buf; dd_errno = PBHSetVol(&pb, false); SetSFCurVol( pb.ioVRefNum ); /* sdm7g shadow in Standard File Dialogs */ SetSFCurDir( pb.ioWDDirID ); /* sdm7g */ return dd_errno ? -1 : 0; } From majordom@umnstat.stat.umn.edu Wed Oct 30 18:23:19 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10697; Wed, 30 Oct 1996 18:23:19 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10691; Wed, 30 Oct 1996 17:23:07 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa04175; 30 Oct 96 18:22 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA27890; Wed, 30 Oct 1996 18:22:24 -0500 Date: Wed, 30 Oct 1996 18:22:23 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Cc: Luke Tierney Subject: Re: tracking current directory on Mac XlispStat In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Well -- I swear it REALLY worked when I first tested it! It stopped working when I added an additional error test, and I can't seem to get it to work for me again. Maybe there is some additional magic about those low globals -- they are globals afterall -- that lets them work in one case and not another. I'll just sit here and quietly grumble till I figure it out. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Wed Oct 30 22:22:26 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12892; Wed, 30 Oct 1996 22:22:26 -0500 Return-Path: Received: from kasuga.cc.chuo-u.ac.jp by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12887; Wed, 30 Oct 1996 21:22:00 -0600 Received: from isg.indsys.chuo-u.ac.jp (isg.indsys.chuo-u.ac.jp [133.91.120.136]) by kasuga.cc.chuo-u.ac.jp (8.7.4+2.6Wbeta6/3.4W4) with SMTP id MAA21909; Thu, 31 Oct 1996 12:21:38 +0900 (JST) Received: from [133.91.120.87] (mactk) by isg.indsys.chuo-u.ac.jp (4.1/6.4J.6-fps0) id AA01656; Thu, 31 Oct 96 12:25:00 JST Date: Thu, 31 Oct 96 12:25:00 JST Message-Id: <9610310325.AA01656@isg.indsys.chuo-u.ac.jp> To: Luke Tierney From: kamakura@indsys.chuo-u.ac.jp (Toshinari Kamakura) X-Sender: kamakura@indsys.chuo-u.ac.jp (Unverified) Subject: Re: bugs? Cc: stat-lisp-news@umnstat.stat.umn.edu Mime-Version: 1.0 Content-Type: text/plain; charset=iso-2022-jp X-Mailer: Eudora-J(1.3.8-J13) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk At 0:53 PM 96.10.30 -0600, Luke Tierney wrote: >I didn't follow this at first -- a bit too subtle for me :-). > >The problem is that in case 1 mycolor is redefined without first being >free'd. The intent was that you call free-color before reusing a color >symbol. I consider this use an error, but currently no error is >signalled and the exact results are implementation dependent. I'll >signal an error in future versions. > >luke No, this is not the case. Because in the Mac version the situation is not changed even if we use the 'free-color' (setf w (send graph-window-proto :new)) (send w :use-color t) (make-color 'my-color 1 1 0) (send w :draw-color 'my-color) (send w :draw-line 50 50 100 100) (free-color 'my-color) (make-color 'my-color 1 0 1) (send w :draw-color 'my-color) (send w :draw-line 100 50 50 100) But I finally I found that if I issued 'free-color' and then dragged the Graph Window by the mouse to anothe place and the use 'make-color', then it resuted in success in changing the color. (make-color 'my-color 1 1 0) (send w :draw-color 'my-color) (send w :draw-line 50 50 100 100) (free-color 'my-color) ;;;; draggin the Graph Window by the mouse to another place (make-color 'my-color 1 0 1) (send w :draw-color 'my-color) (send w :draw-line 100 50 50 100) Be sure that draggin the window before issuing the 'free-color' results in failure. Do you imagine this? It seems that 'free-color does not work properly. +++++++++++++++++++++++++++++++++++++++++++++++ Toshinari Kamakura Chuo University 1-13-27 Kasuga Bunkyo-ku, Tokyo 112 Japan TEL +81-3-3817-1934 FAX +81-3-3817-1943 email: kamakura@indsys.chuo-u.ac.jp +++++++++++++++++++++++++++++++++++++++++++++++ From majordom@umnstat.stat.umn.edu Thu Oct 31 08:33:20 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23333; Thu, 31 Oct 1996 08:33:20 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23328; Thu, 31 Oct 1996 07:33:13 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id IAA00758; Thu, 31 Oct 1996 08:31:55 -0500 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 31 Oct 1996 08:29:21 -0500 To: "Steven D. Majewski" From: forrest@unc.edu (Forrest Young) Subject: Re: tracking current directory on Mac XlispStat Cc: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >Well -- I swear it REALLY worked when I first tested it! > >It stopped working when I added an additional error test, and >I can't seem to get it to work for me again. > >Maybe there is some additional magic about those low globals -- >they are globals afterall -- that lets them work in one >case and not another. > >I'll just sit here and quietly grumble till I figure it out. Just a note of support saying that there's at least one of us out here who would be very happy if you did figure it out! Don't grumble too much :-) Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Thu Oct 31 11:35:17 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24657; Thu, 31 Oct 1996 11:35:17 -0500 Return-Path: Received: from gateway.ets.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24651; Thu, 31 Oct 1996 10:35:11 -0600 Received: from research.ets.org by gateway.ets.org with SMTP id AA05939 (5.65c/IDA-1.4.4 for ); Thu, 31 Oct 1996 11:34:58 -0500 Received: from comet.research.ets.org by research.ets.org with SMTP id AA21041 (5.67b/IDA-1.4.3 for sdm7g@Virginia.EDU); Thu, 31 Oct 1996 11:34:58 -0500 Received: by comet.research.ets.org (SMI-8.6/SMI-SVR4) id LAA10581; Thu, 31 Oct 1996 11:34:57 -0500 Date: Thu, 31 Oct 1996 11:34:57 -0500 Message-Id: <199610311634.LAA10581@comet.research.ets.org> From: Russell Almond To: sdm7g@Virginia.EDU Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: (sdm7g@Virginia.EDU) Subject: Re: tracking current directory on Mac XlispStat Reply-To: ralmond@ets.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'm not sure what the purpose of this functionality is. (I admit that for certain purposes, especially designing student exercises, it would be very useful to have this under programatic control.) On the other hand, there is a rather nice shareware utility called Default Folder which I've been using for about a month now which adds a menu of the 10 most recently accessed directories to the open dialog. This might solve most of the problem you are having with the open dialogs, it will also work with applications other than xlisp-stat. I've been using for about a month without any problems attirbutable to it. It should be available from all the usual sources for Mac Shareware (I think I got if from the MacUser archive on zdnet). --Russell Almond almond@acm.org From majordom@umnstat.stat.umn.edu Thu Oct 31 13:02:33 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25426; Thu, 31 Oct 1996 13:02:33 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25421; Thu, 31 Oct 1996 12:02:26 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa20172; 31 Oct 96 13:01 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA17990; Thu, 31 Oct 1996 13:01:20 -0500 Date: Thu, 31 Oct 1996 13:01:19 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: ralmond@ets.org Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: tracking current directory on Mac XlispStat In-Reply-To: <199610311634.LAA10581@comet.research.ets.org> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Thu, 31 Oct 1996, Russell Almond wrote: > I'm not sure what the purpose of this functionality is. (I admit that > for certain purposes, especially designing student exercises, it would > be very useful to have this under programatic control.) OK - expressing the problem as it's stated in the subject line is just an artifact of a unix programmer trying to grapple with Mac-isms: "Why doesn't it work the way it's supposed to?" The real problem isn't tracking - it's that I need to manage different sets of defaults -- a default directory for user input data, a default directory for reference files used in fitting the data, a default directory for output & results, a default or three for loading new program modules. The latter is the only one where it makes sense to bypass the standard file dialogs and load directly by name. > On the other hand, there is a rather nice shareware utility called > Default Folder which I've been using for about a month now which > adds a menu of the 10 most recently accessed directories to the open > dialog. This might solve most of the problem you are having with the > open dialogs, it will also work with applications other than > xlisp-stat. I've been using for about a month without any problems > attirbutable to it. Thanks -- I'll take a look at this. It doesn't sound as if it will do what I want for *THIS* case, but it would sure be nice to use when programming in LispStat. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Thu Oct 31 13:42:58 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25954; Thu, 31 Oct 1996 13:42:58 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25949; Thu, 31 Oct 1996 12:42:51 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id NAA07083; Thu, 31 Oct 1996 13:42:39 -0500 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 31 Oct 1996 13:40:04 -0500 To: ralmond@ets.org From: forrest@unc.edu (Forrest Young) Subject: Re: tracking current directory on Mac XlispStat Cc: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >I'm not sure what the purpose of this functionality is. (I admit that >for certain purposes, especially designing student exercises, it would >be very useful to have this under programatic control.) The reason I would like it is so that I can have dialogs open to default directories. This works under windows and unix, but the argument that tells the open dialog when to open doesn't work under MacOS. This is especially true for naive folks such as students who may not know how to navigate arround the directory structure. >On the other hand, there is a rather nice shareware utility called >Default Folder which I've been using for about a month now which >adds a menu of the 10 most recently accessed directories to the open >dialog. This might solve most of the problem you are having with the >open dialogs, it will also work with applications other than >xlisp-stat. I've been using for about a month without any problems >attirbutable to it. This is true (I use now utilities super-boomerang to do this). But it would still be really nice to be able to tell the open dialog to open at a specified point in the directory hierarchy. Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Fri Nov 1 10:16:45 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11752; Fri, 1 Nov 1996 10:16:45 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11747; Fri, 1 Nov 1996 09:16:41 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id KAA28235; Fri, 1 Nov 1996 10:16:31 -0500 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 1 Nov 1996 10:13:55 -0500 To: vista@unc.edu, edstat-l@jse.stat.ncsu.edu, stat-l@vm1.mcgill.ca, stat-lisp-news@umnstat.stat.umn.edu, sug-l@listserv.oit.unc.edu From: forrest@unc.edu (Forrest Young) Subject: New Release of ViSta Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Announcing a maintainence release of ViSta - The Visual Statistics System Version 4.26, October 30th, 1996 This maintainence release fixes a bug which prevented the New Data menu item from working, and introduces minor changes oriented towards those developing new ViSta statistical objects. You should update to this version of ViSta if any of the following apply: 1) Your version of ViSta is older than or 4.24. 2) You want to use the data editor to enter new data. 3) You are developing new statistical objects. Here is brief information about ViSta for those new to the newsgroup: 1) ViSta performs univariate and multivariate statistical data analysis and visualization. 2) ViSta is useful for teaching multivariate statistics, statistical graphics and computational statistics; and for research and development in statistical graphics and computational statistics. 3) ViSta is available for free for Windows 95 and 3.1, for MacOS, and for Unix with XWindows. 4) Further information about ViSta, as well as the software and documentation, are available from our WWW and FTP servers at: WWW: http://forrest.psych.unc.edu FTP: www.psych.unc.edu, directory pub/forrest/vista. Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Fri Nov 1 21:01:29 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17465; Fri, 1 Nov 1996 21:01:29 -0500 Return-Path: Received: from mcmail.CIS.McMaster.CA by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17460; Fri, 1 Nov 1996 20:01:24 -0600 Received: from johnfox (ellen15.slip.yorku.ca [130.63.122.155]) by mcmail.CIS.McMaster.CA (8.7.1/8.7.1) with SMTP id VAA02900 for ; Fri, 1 Nov 1996 21:01:13 -0500 (EST) Message-Id: <1.5.4.32.19961102015814.0069ae74@mcmail.cis.mcmaster.ca> X-Sender: jfox@mcmail.cis.mcmaster.ca X-Mailer: Windows Eudora Light Version 1.5.4 (32) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 01 Nov 1996 20:58:14 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: John Fox Subject: apparent bug in matrix inverse in release 3.47 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I discovered (when some previously functioning code began to misbehave) that inverse in xlisp-stat 2.1 Release 3.47 fails to invert a 1x1 matrix; for example: > (inverse (matrix '(1 1) '(1))) Error: incompatible with access indices - #(0) Happened in: # This works just fine in the previous version that I was using, Release 3.44: > (inverse (matrix '(1 1) '(1))) #2A((1.0)) I'm running the 32-bit Windows version under Windows 95. John |----------------------------------------------------| | John Fox jfox@McMaster.ca | | Department of Sociology McMaster University | |----------------------------------------------------| From majordom@nokomis.stat.umn.edu Tue Nov 5 11:05:28 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA07225; Tue, 5 Nov 1996 11:05:28 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA07217; Tue, 5 Nov 1996 10:05:07 -0600 From: Luke Tierney Message-Id: <9611051605.AA07217@nokomis.stat.umn.edu> Subject: Re: bugs? To: kamakura@indsys.chuo-u.ac.jp (Toshinari Kamakura) Date: Tue, 5 Nov 1996 10:05:07 -0600 (CST) Cc: luke@nokomis.stat.umn.edu, stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: <9610310325.AA01656@isg.indsys.chuo-u.ac.jp> from "Toshinari Kamakura" at Oct 31, 96 12:25:00 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1102 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > No, this is not the case. Because in the Mac version > the situation is not changed even if we use the 'free-color' > > (setf w (send graph-window-proto :new)) > (send w :use-color t) > > (make-color 'my-color 1 1 0) > (send w :draw-color 'my-color) > (send w :draw-line 50 50 100 100) > > (free-color 'my-color) > (make-color 'my-color 1 0 1) > (send w :draw-color 'my-color) > (send w :draw-line 100 50 50 100) > > > But I finally I found that if I issued 'free-color' and then > dragged the Graph Window by the mouse to anothe place > and the use 'make-color', then it resuted in success in changing > the color. > > (make-color 'my-color 1 1 0) > (send w :draw-color 'my-color) > (send w :draw-line 50 50 100 100) > > (free-color 'my-color) > ;;;; draggin the Graph Window by the mouse to another place > (make-color 'my-color 1 0 1) > (send w :draw-color 'my-color) > (send w :draw-line 100 50 50 100) > > Be sure that draggin the window before issuing the 'free-color' > results in failure. > Tat definitely looks like a bug. I'll see if I can reproduce it and track it down. luke From majordom@umnstat.stat.umn.edu Wed Nov 6 08:32:24 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29557; Wed, 6 Nov 1996 08:32:24 -0500 Return-Path: Received: from lenzie.cent.gla.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29552; Wed, 6 Nov 1996 07:32:11 -0600 Message-Id: <25110.199611061329@lenzie.cent.gla.ac.uk> From: James Currall Received: (from jepc1e@localhost sender jepc1e) by lenzie.cent.gla.ac.uk (8.7.6/UK-2.2a/cent-sparc) id NAA25110 for stat-lisp-news@umnstat.stat.umn.edu; Wed, 6 Nov 1996 13:29:33 GMT Subject: Potential Speakers To: stat-lisp-news@umnstat.stat.umn.edu Date: Wed, 6 Nov 1996 13:29:32 +0000 (GMT) X-Mailer: ELM [version 2.4 PL23] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi XlispStatters, Apologies to those outside the UK, but this is really addressedto UK statisticians who write code in XlispStat. The Royal Statistical Society Statistical Computing Section is keen to have a meeting on Object Oriented Programming, and as wondering if there is anyone who has some interesting examples of work that they have been doing in XlispStat that they would be willing to talk about in London in March. Reasonable expenses would be paid, but I am afraid only for travel within the UK. Anyone interested, let me know ASAP. Thanks James -- --------------------- Dr James Currall ----------------------- | E-mail: J.Currall@compserv.gla.ac.uk ~ ~ --- ~ | | Fax: (UK) 0141-330-4808 ~ / ^ / ~ ~ | | Phone: (UK) 0141-339-8855 Ext 4855 ~ /^ ^ --- ~ | | Telex: 777070 / Answerback UNIGLA ~ /^ ^ / ~ | | Post: Computing Service, University of ~ \ * -- ~ | | Glasgow, Glasgow G12 8QQ, Scotland. ~ / ^ ^ | ~ | -------------------------------------------------------------- From majordom@umnstat.stat.umn.edu Tue Nov 12 12:58:23 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA20786; Tue, 12 Nov 1996 12:58:23 -0500 Return-Path: Received: from rgborafsa.er.usgs.gov by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA20781; Tue, 12 Nov 1996 11:58:18 -0600 Received: from [130.11.54.131] (mcmac.er.usgs.gov [130.11.54.131]) by rgborafsa.er.usgs.gov (Geomail 1.2.3) with SMTP id MAA16693 for ; Tue, 12 Nov 1996 12:58:51 -0500 Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Tue, 12 Nov 1996 12:59:16 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: "Richard B McCammon Reston, VA " (Richard McCammon (USGS)) Subject: log10 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk How easy is it to adjust the R-code to transform with log10 instead of loge using the Transform option of the standard regression dialogue? __________________________________________________________ snail mail: Richard B. McCammon Eastern Minerals Team U.S. Geological Survey MS-954 12201 Sunrise Valley Drive Reston, VA 20192 USA e-mail: mccammon@rgborafsa.er.usgs.gov or mccammon@130.11.54.100 Phone: 703-648-6150 Fax: 703-648-6383 From majordom@umnstat.stat.umn.edu Wed Nov 13 08:35:05 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06528; Wed, 13 Nov 1996 08:35:05 -0500 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06522; Wed, 13 Nov 1996 07:34:59 -0600 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA08892 (5.67b/IDA-1.5 for ); Wed, 13 Nov 1996 08:34:29 -0500 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA5069 for ; Wed, 13 Nov 1996 08:33:48 -0400 Message-Id: <3289F819.2899@kodak.com> Date: Wed, 13 Nov 1996 08:32:25 -0800 From: Jeff Morris Organization: Eastman Kodak Company X-Mailer: Mozilla 2.01 (Win16; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: Missing Values Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk One of the concerns I've always had with XLS is its inability to handle missing values. For example, I'm fitting a logistic model to the byssinosis data from Lindsey's "Modeling Frequency and Count Data" which has 5 factors and a binomial response r, n for each combination. Seven of the combinations have no data. When I try to fit the model I get an error because of the zeros. To fix this, I simply deleted the guilty observations and everything fit fine. But, it was tedious to do this by hand. My question is how does one delete the 31st, 34th, 41st, etc observations from each of the 5 variables and the responses at the same time. What would be really nice would be to have a function that identified the observations that had 0 n, and deleted those from all the variables. Any thoughts? And BTW what is the difference between (def x) and (defvar x)? As always, thanks!! Jeff -- Jeffrey D. Morris, Ph.D. Health & Environment Labs Eastman Kodak Company morris@kodak.com From majordom@umnstat.stat.umn.edu Fri Nov 15 11:41:08 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12642; Fri, 15 Nov 1996 11:41:08 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12629; Fri, 15 Nov 1996 10:41:01 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.6.4.3/10.1) id LAA01186; Fri, 15 Nov 1996 11:40:51 -0500 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 15 Nov 1996 11:37:30 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: bug in Mac 3.44 XLispStat? Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I have noticed the following strangness in the Mac 3.44 version of XLS. It looks like the binomial-dens and poisson-dens functions are unbound, but if I do an apropos for them they seem to be there. Here's the listener: > (binomial-dens) Error: The function BINOMIAL-DENS is unbound. > (poisson-dens) Error: The function POISSON-DENS is unbound. > (apropos 'dens) DENS POISSON-DENS BINOMIAL-DENS T-DENS BETA-DENS KERNEL-DENS CONDENSE CHISQ-DENS GAMMA-DENS NORMAL-DENS F-DENS CAUCHY-DENS > How can this be? And where did the density functions go to? Is this really a bug? Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Fri Nov 15 12:09:13 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12896; Fri, 15 Nov 1996 12:09:13 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12891; Fri, 15 Nov 1996 11:09:07 -0600 Received: from galton.math.ucla.edu by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA02764; Fri, 15 Nov 96 09:09:05 PST Received: by galton.math.ucla.edu (Sendmail 5.x/1.13) id AA28737; Fri, 15 Nov 1996 09:09:04 -0800 From: Jan Deleeuw Message-Id: <9611151709.AA28737@galton.math.ucla.edu> Subject: Re: bug in Mac 3.44 XLispStat? To: forrest@unc.edu (Forrest Young) Date: Fri, 15 Nov 1996 09:09:03 -0800 (PST) Cc: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) In-Reply-To: from "Forrest Young" at Nov 15, 96 11:37:30 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk This is just a documentation bug. Binomials and poissons do not have densities, thus the functions are called binomial-pmf and poisson-pmf. > > I have noticed the following strangness in the Mac 3.44 version of XLS. It > looks like the binomial-dens and poisson-dens functions are unbound, but if > I do an apropos for them they seem to be there. Here's the listener: > > > (binomial-dens) > Error: The function BINOMIAL-DENS is unbound. > > (poisson-dens) > Error: The function POISSON-DENS is unbound. > > (apropos 'dens) > DENS > POISSON-DENS > BINOMIAL-DENS > T-DENS > BETA-DENS > KERNEL-DENS > CONDENSE > CHISQ-DENS > GAMMA-DENS > NORMAL-DENS > F-DENS > CAUCHY-DENS > > > > How can this be? And where did the density functions go to? Is this really > a bug? > > Forrest Young > Professor, Psychometrics > > UNC Psychometrics Lab e-mail: forrest@unc.edu > CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu > Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 > > ViSta Software Site: http://forrest.psych.unc.edu/research/ViSta.html > > > -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Fri Nov 15 12:54:42 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13540; Fri, 15 Nov 1996 12:54:42 -0500 Return-Path: Received: from neopath.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13535; Fri, 15 Nov 1996 11:54:37 -0600 Received: from denali.neopath.com by neopath.com (SMI-8.6/SMI-SVR4) id JAA02913; Fri, 15 Nov 1996 09:51:14 -0800 Received: from pilchuck.neopath.com by denali.neopath.com (5.x/SMI-SVR4) id AA17716; Fri, 15 Nov 1996 09:51:16 -0800 Received: by pilchuck.neopath.com (5.x/SMI-SVR4) id AA02606; Fri, 15 Nov 1996 09:51:15 -0800 Date: Fri, 15 Nov 1996 09:51:15 -0800 From: keith@neopath.com (Keith Frost) Message-Id: <9611151751.AA02606@pilchuck.neopath.com> To: deleeuw@math.ucla.edu Cc: forrest@unc.edu, stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <9611151709.AA28737@galton.math.ucla.edu> (message from Jan Deleeuw on Fri, 15 Nov 1996 09:09:03 -0800 (PST)) Subject: Re: bug in Mac 3.44 XLispStat? Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Note that the attempt to call the non-existent functions, has the unintended side-effect of creating symbols with those names, which then show up when "apropos" is called. I've run into this bit of confusion before, when hunting for the name of a function I can't recall. -- Keith Frost From majordom@umnstat.stat.umn.edu Fri Nov 15 23:39:29 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16875; Fri, 15 Nov 1996 23:39:29 -0500 Return-Path: Received: from mcmail.CIS.McMaster.CA by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16870; Fri, 15 Nov 1996 22:39:23 -0600 Received: from johnfox (howie06.slip.yorku.ca [130.63.122.39]) by mcmail.CIS.McMaster.CA (8.7.1/8.7.1) with SMTP id XAA07329; Fri, 15 Nov 1996 23:39:04 -0500 (EST) Message-Id: <1.5.4.32.19961116043544.006a6f4c@mcmail.cis.mcmaster.ca> X-Sender: jfox@mcmail.cis.mcmaster.ca X-Mailer: Windows Eudora Light Version 1.5.4 (32) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 15 Nov 1996 23:35:44 -0500 To: Jeff Morris , stat-lisp-news@umnstat.stat.umn.edu From: John Fox Subject: Re: Missing Values Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk At 08:32 AM 11/13/96 -0800, Jeff Morris wrote: >One of the concerns I've always had with XLS is its inability to >handle missing values. For example, I'm fitting a logistic model >to the byssinosis data from Lindsey's "Modeling Frequency and Count >Data" which has 5 factors and a binomial response r, n for each >combination. Seven of the combinations have no data. When I try to >fit the model I get an error because of the zeros. > To fix this, I simply deleted the guilty observations and everything >fit fine. But, it was tedious to do this by hand. My question is how >does one delete the 31st, 34th, 41st, etc observations from each of the >5 variables and the responses at the same time. What would be really >nice would be to have a function that identified the observations that >had 0 n, and deleted those from all the variables. Any thoughts? It would be nice if missing-data capabilities were built into lisp-stat, but it's not to hard to do this kind of thing. Here's a simple function that gets rid of missing data in a list of variables. I'm sure that someone else could program this more elegantly. (defun delete-missing (variables &key (miss '?) (test #'equal)) "Args: (variables &key (miss '?) (test #'equal)) VARIABLES is a list of sequences, containing the variables; MISS gives the missing data code (default '?); TEST gives the equality test (default #'equal). The function returns a list a sequences containing only observations with valid data for all variables. Example: (delete-missing (list x y z) :miss '*)" (let* ((bad (remove-duplicates (combine (mapcar #'(lambda (x) (which (map-elements test miss x))) variables)))) (all (iseq (length (first variables)))) (good (sort-data (set-difference all bad)))) (mapcar #'(lambda (x) (select x good)) variables))) >And BTW what is the difference between (def x) and (defvar x)? > Def defines variables that are lexically scoped (the default in lisp); defvar defined variables that are dynamically scoped. Here's a lisp-stat dialog that illustrates the difference: > (defvar x 2) X > (def y 1) Y > (defun f (x) (g)) F > (defun g () x) G 2 > (f 3) 3 > (defun h (y) (i)) H > (defun i () y) I > (i) 1 > (h 3) 1 So, calling g from f shadows the global value of the dynamically scoped variable x, but calling i from h doesn't shadow the global value of y because y is a lexically scoped variable defined in the global environment (or something like that). I hope that this helps, John |----------------------------------------------------| | John Fox jfox@McMaster.ca | | Department of Sociology McMaster University | |----------------------------------------------------| From majordom@umnstat.stat.umn.edu Sat Nov 16 12:02:05 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27193; Sat, 16 Nov 1996 12:02:05 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27188; Sat, 16 Nov 1996 11:02:01 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa09966; 16 Nov 96 12:01 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA18004; Sat, 16 Nov 1996 12:01:42 -0500 Date: Sat, 16 Nov 1996 12:01:41 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: John Fox Cc: Jeff Morris , stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Missing Values In-Reply-To: <1.5.4.32.19961116043544.006a6f4c@mcmail.cis.mcmaster.ca> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I've used XLS #.NOT-A-NUMBER to encode missing values. I don't know if that and the POSITYVE and NEGATIVE-INFINITY values are supported on all platforms or just where there is existing IEEE FP support. It is on the Mac. ( 3.47 ) I also often use a reciprocal function, that doesn't give an error on divide by zero, but yields infinity. I had qualms about just changing the bahaviour of '/' : I find that typically, a divide by zero of a single number is usually an error I want to catch, but something of the form ( / 1 vector ) I usually expect to pass, zeros or not. Much better than just using an arbitrary large number. You can also modify the :adjust-to-data methods to take this into account so that the graph functions don't bother to include those asymptotic or missing values in scaling the graph to the window. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Sat Nov 16 22:44:55 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29506; Sat, 16 Nov 1996 22:44:55 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29501; Sat, 16 Nov 1996 21:44:49 -0600 Received: from 164.67.20.130 (ts19-5.wla.ts.ucla.edu) by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA05547; Sat, 16 Nov 96 19:44:45 PST Message-Id: <328E8A06.17A8@stat.ucla.edu> Date: Sat, 16 Nov 1996 19:44:09 -0800 From: Jan de Leeuw Reply-To: deleeuw@stat.ucla.edu Organization: UCLA Statistics X-Mailer: Mozilla 3.01 (Macintosh; I; PPC) Mime-Version: 1.0 Newsgroups: sci.stat.math,sci.stat.consult,sci.stat.edu Cc: ucla-stat@math.ucla.edu, stat-lisp-news@umnstat.stat.umn.edu Subject: UCLA Statistics Electronic Textbook Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The normal distribution, regression, and measures of central tendency pages in the textbook have been updated. Also, quite a bit of introductory textual material has been added. There are new Xlisp-Stat demos to illustrate the variational characterization of mean and median. The easiest way to find these is to go to http://www.stat.ucla.edu/textbook and then hit the [What's New] button. --- Jan From majordom@umnstat.stat.umn.edu Sun Nov 17 15:58:49 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10625; Sun, 17 Nov 1996 15:58:49 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10620; Sun, 17 Nov 1996 14:58:43 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.7.5/8.7.3) id MAA07509 for stat-lisp-news@umnstat.stat.umn.edu; Sun, 17 Nov 1996 12:58:39 -0800 (PST) From: Jan Deleeuw Message-Id: <199611172058.MAA07509@galton.math.ucla.edu> Subject: power-mean To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Sun, 17 Nov 1996 12:58:39 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;; The power-norm is the 1/p power of the sum of the absolute values ;; to the p-th power. Define the power-loss of a number y for a vector ;; x_1,...,x_n to be the power-mean of the x_i - y. Define the power ;; mean of x_1,...,x_n to be the y minimizing the power loss. ;; ;; We all know the median is the power-mean for p = 1, the mean is the ;; power mean for p = 2, the mode is the power-mean for p -> 0, and ;; the mid-range is the power-mean for p -> infty. ;; ;; Here is a demo with a slider for p. You simply say (power-mean x) ;; for some x, select your power, and move the mouse in the window ;; with the point-plot. ;; ;; It flickers, but it works. ;; ;; Jan de Leeuw, 11-17-96 ;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; (defproto point-window-proto '(x power loss offset) nil graph-window-proto) (defmeth point-window-proto :x (&optional (val nil set)) (if set (setf (slot-value 'x) val)) (slot-value 'x)) (defmeth point-window-proto :power (&optional (val nil set)) (if set (setf (slot-value 'power) val)) (slot-value 'power)) (defmeth point-window-proto :offset (&optional (val nil set)) (if set (setf (slot-value 'offset) val)) (slot-value 'offset)) (defmeth point-window-proto :loss (&optional (val nil set)) (if set (setf (slot-value 'loss) val)) (slot-value 'loss)) (defmeth point-window-proto :coord-map (x) (let* ( (o (send self :offset)) (s (- (first (send self :size)) (+ o o 15))) (b (max (send self :x))) (a (min (send self :x))) (c (- b a)) (r (/ s c)) (e (- o (* r a))) (d (second (/ (send self :size) 2))) ) (list (round (+ (* r (first x)) e)) (round (* d (- 1 (second x))))) )) (defmeth point-window-proto :a-line (x y) (let* ( (u (send self :coord-map x)) (v (send self :coord-map y)) ) (send self :draw-line (first u) (second u) (first v) (second v)) )) (defmeth point-window-proto :redraw () (send self :make-points) ) (defmeth point-window-proto :make-points () (let* ((x (send self :x)) (a (min x)) (b (max x))) (send self :a-line (list a 0) (list b 0)) (dolist (z x) (send self :a-line (list z -.5) (list z .5))) )) (defmeth point-window-proto :do-motion (x y) (let ((s (second (send self :size)))) (send self :erase-window) (send self :make-points) (send self :draw-line x 0 x s) (send self :function-line x) )) (defmeth point-window-proto :function-line (x) (let* ( (z (send self :x)) (o (send self :offset)) (s (- (first (send self :size)) (+ o o 15))) (b (max z)) (a (min z)) (c (- b a)) (r (/ s c)) (e (- o (* r a))) (h (send self :loss)) (u (/ (- x e) r)) (v (send h :range 1)) (power (send self :power)) ) (send h :clear-lines) (send h :add-function (power-loss z power) a b) (send h :add-lines (list u u) v) (format t "Argument ~10,3g Value ~20,6g~%" u (funcall (power-loss z power) u)) )) (defun points-on-line (x power &key (size 300) (offset 10)) (let ((g (send point-window-proto :new :size (list size (/ size 4)) :title "Point Window" :has-h-scroll nil :has-v-scroll nil))) (send g :x x) (send g :offset offset) (send g :power power) (send g :loss (plot-function (power-loss x power) (min x) (max x))) (send g :make-points) (send (send g :loss) :location 200 200) )) (defun power-loss (x power) (lambda (y) (expt (sum (expt (abs (- x y)) power)) (/ power))) ) (defun range (x) (- (max x) (min x))) (defun power-mean (x) (sequence-slider-dialog '(.01 .1 .5 1 1.5 2 3 4 5 10 100) :action #'(lambda (p) (points-on-line x p))) ) -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon Nov 18 17:49:43 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27697; Mon, 18 Nov 1996 17:49:43 -0500 Return-Path: Received: from mailb.surrey.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27692; Mon, 18 Nov 1996 16:49:30 -0600 Received: from soc.surrey.ac.uk (actually host boris-gate.soc.surrey.ac.uk) by mailb.surrey.ac.uk with SMTP (PP); Mon, 18 Nov 1996 22:48:05 +0000 Received: from annex1-11.surrey.ac.uk by boris.soc.surrey.ac.uk via Ethernet with SMTP id aa19047; 18 Nov 96 22:47 GMT X-Sender: scs1ng@boris.soc.surrey.ac.uk Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 18 Nov 1996 22:31:17 +0000 To: Xlisp List From: Nigel Gilbert Subject: Bug list for Release 3.47 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I have discovered the following bugs in release 3.47 while coding a project. Wouldn't it be a good idea if there was a generally available list of bugs that users have found, so that we don't all keep rediscovering the same ones? Perhaps a web page at the UCLA site? [NB1: I don't think that LispStat is particularly buggy - quite the contrary (especially as 3.47 is supposed to be an Alpha version!)] [NB2: Some of the following may be bugs in my understanding, rather than in code; if so, let me know]. [NB3: Some of these bugs have been reported previously to stat-lisp-news] 1. The function 'plot-points' fails to plot any points if the y-coordinates of all the points are identical. EXAMPLE: (plot-points '(1 2 3 4 5 6) '(3 3 3 3 3 3)) 2. equalp does not regard two equivalent structures as equal EXAMPLE: > (defstruct p x y) P > (setq a (make-p :x 1 :y 2)) #S(P X 1 Y 2) > (setq b (make-p :x 1 :y 2)) #S(P X 1 Y 2) > (equalp a b) NIL > 3. A DEFSTRUCT print-function will not work with a lambda expression. It will however work with a named function. EXAMPLE: > (defstruct (node (:print-function (lambda (n s d) (format s "#<~A>" (node-elt n))))) elt (l nil) (r nil)) > (setq a-node (make-node :elt 'X )) Error: bad function - # Happened in: # > a-node Error: bad function - # Happened in: # 4. The Mac. version of the compiler intermittently prints voluminous tracing information while it compiles. I haven't been able to track down what sets this off - it may be an uninitialised variable somewhere. It is harmless but irritating. The following are not strictly bugs, but are things which could (probably) be implemented with little extra effort: 1. (compile-file file-name) returns NIL even if the compilation was successful (or T if the resulting file was also loaded). CLtL2 says that compile-file should return three values, the pathname of the compiled file, the number of warnings plus errors, and the number of errors encountered. These would be much more useful. 2. There is no way for a lisp program to discover the pathname of the file to which an editing window is attached (on a Mac). This is a shame becuase it prevents people like me developing a more sophisticated programmer interface than the rather bare bones default LispStat one. ____________________________________________________________________ Prof G. Nigel Gilbert, Department of Sociology, University of Surrey Guildford GU2 5XH, UK. Tel: +44 1483 259173 Fax: +44 1483 306290 From majordom@umnstat.stat.umn.edu Mon Nov 18 18:41:55 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28037; Mon, 18 Nov 1996 18:41:55 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28032; Mon, 18 Nov 1996 17:41:50 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa22782; 18 Nov 96 18:39 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA21496; Mon, 18 Nov 1996 18:39:10 -0500 Date: Mon, 18 Nov 1996 18:39:09 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU Reply-To: "Steven D. Majewski" To: Nigel Gilbert Cc: Xlisp List Subject: Re: Bug list for Release 3.47 In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Mon, 18 Nov 1996, Nigel Gilbert wrote: > I have discovered the following bugs in release 3.47 while coding a > project. Wouldn't it be a good idea if there was a generally available > list of bugs that users have found, so that we don't all keep rediscovering > the same ones? Perhaps a web page at the UCLA site? I have some bugs and patches to 3.47 -- especially the Mac version, at: I'll try to keep that info updated, however, I am only actively working on the Mac version, so I can't (at the moment) do much testing cross platform. ( I have access to the other platforms -- I just don't have the time to try to keep a cross platform distribution updated and documented.) [ I'ld be more than happy to see someone else manage the "official" list and link my notes into it. ] > > 1. The function 'plot-points' fails to plot any points if the > y-coordinates of all the points are identical. > EXAMPLE: (plot-points '(1 2 3 4 5 6) '(3 3 3 3 3 3)) > It *does* plot the points -- the bug must be in :ADJUST-TO-DATA. When I try it, the y-max is set to 1.0. If you ( SEND plot :RANGE 1 0 5 ) you will see the plotted points. [ Sorry -- that's the only one I can answer off the top of my head. But I'll certainly look at your last point about Mac window pathnames, as that would probably be useful to me. ] ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Thu Nov 21 10:28:42 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17033; Thu, 21 Nov 1996 10:28:42 -0500 Return-Path: Received: from zam149.zam.kfa-juelich.de by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17027; Thu, 21 Nov 1996 09:28:31 -0600 Received: from hlrz31-f by kfa-juelich.de (5.65v3.2/DEC-AXP) id AA30347; Thu, 21 Nov 1996 16:28:22 +0100 Received: from hlrz40.hlrz.kfa-juelich.de by hlrz31.hlrz.kfa-juelich.de (SMI-8.6/SMI-4.1) id QAA21105; Thu, 21 Nov 1996 16:28:19 +0100 Reply-To: jmu@hlrz31.hlrz.kfa-juelich.de Received: by hlrz40.hlrz.kfa-juelich.de (SMI-8.6/SMI-4.1) id QAA29797; Thu, 21 Nov 1996 16:28:18 +0100 Date: Thu, 21 Nov 1996 16:28:18 +0100 Message-Id: <199611211528.QAA29797@hlrz40.hlrz.kfa-juelich.de> From: Johannes Mueller-Gerking To: stat-lisp-news@umnstat.stat.umn.edu Subject: Q: Looking for "color-coding" routines in XLS Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, probably my question in the subject isn't that clear... I'm looking for a tool to visualize relations between multiple timeseries over time. I imagine a grid of colored squares where the square with coordinates (i,j) codes some relation between series i and j (for example covariance, coherence etc.) in color. With a time-moving window I could have a color-movie of these relationships for all channels simulataneously and could watch for any kind of regularities. This kind of dynamic graphics should be easy to program in XLS. A first step would be a map from values to colors. Has anybody done something like that and is willing to easen my job? Thank you very much! Johannes Mueller-Gerking HLRZ, KFA Juelich, Germany E-mail: From majordom@umnstat.stat.umn.edu Sat Nov 23 23:00:52 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24983; Sat, 23 Nov 1996 23:00:52 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24978; Sat, 23 Nov 1996 22:00:43 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id UAA15178 for stat-lisp-news@umnstat.stat.umn.edu; Sat, 23 Nov 1996 20:00:38 -0800 (PST) From: Jan Deleeuw Message-Id: <199611240400.UAA15178@galton.math.ucla.edu> Subject: PCA To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Sat, 23 Nov 1996 20:00:37 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The LaTeX below explains, and the Lisp at the end implements, adding three principal axes to a spin-proto. Implemented as a method :pca, which does not need arguments, because everything is sitting right there in the proto. ========================================================================= \documentclass[12pt]{amsart} \usepackage{verbatim} \begin{document} \title{PCA of three variables in Xlisp-Stat} \author{Jan de Leeuw} \date{\today} \maketitle Suppose $u_{i}=(x_{i},y_{i},z_{i})$ are $n$ points in $\mathbb{R}^{3}$. We want to find direction cosines $w=(w_{1},w_{2},w_{3})$ and an intercept $a=(a_{1},a_{2},a_{3})$ such that the line $\mathcal{L}=\{y\mid y=a+\lambda w\}$ approximates the $u_{i}$ as closely as possible.\par % The loss function we use is \begin{equation*} \sigma(a,w)=\sum_{i=1}^{n}\min_{\lambda_{i}}\ (u_{i}-a-\lambda_{i}w)' (u_{i}-a-\lambda_{i}w). \end{equation*} This means that we project the $u_{i}$ perpendicularly on the line, and measure the sum of squared distances of the $u_{i}$ and the projections $\hat u_{i}$. Then choose $a$ and $w$, with $w'w=1$, such that $\sigma(a,w)$ is minimized.\par % It is clear that this problem is identical to the problem of minimizing \begin{equation*} \sigma(a,w,\lambda)=\sum_{i=1}^{n} (u_{i}-a-\lambda_{i}w)'(u_{i}-a-\lambda_{i}w) \end{equation*} over all three sets of variables, with constraint $w'w=1$. This show that we may also require, without loss of generality, that the $\lambda_{i}$ sum to zero.\par % We first minimize of $a$ for given $w$ and $\lambda$. This gives $\hat a=u_{\bullet},$ with superscript $\bullet$ indicating mean. Thus \begin{equation*} \min_{a}\ \sigma(a,w,\lambda)=\sum_{i=1}^{n} (\tilde u_{i}-\lambda_{i}w)' (\tilde u_{i}-\lambda_{i}w), \end{equation*} with tildes over symbols indicating deviations from the mean. Now minimize of the $\lambda_{i}$, which must add up to zero. The solution is $\hat\lambda_{i}=w'\tilde u_{i}$, which indeed adds up to zero. We now find \begin{equation*} \min_{\lambda}\min_{a}\ \sigma(a,w,\lambda)=\sum_{i=1}^{n}\{\tilde u_{i}'\tilde u_{i}^{}-\hat\lambda_{i}^{2}\}. \end{equation*} In the final step of our minimization problem, we \textit{maximize} the sum of squares of the $\hat\lambda_{i}$ over $w$ with $w'w=1$. But \begin{equation*} \sum_{i=1}^{n}\hat\lambda_{i}^{2}=w'\{\sum_{i=1}^{n}\tilde u_{i}^{} \tilde u_{i}'\}w=w'Cw, \end{equation*} where $C=\tilde U'\tilde U$ is the $3\times 3$ cross-product matrix of the $\tilde u_{i}$. It follows that $\hat w$ is the normalized eigenvector corresponding with the dominant eigenvalue of $C$, or equivalently of the covariance matrix of the $u_{i}$. Suppose this dominant eigenvalue is $\omega$. We then draw the line between $\hat a-\frac{1}{2}\sqrt{\omega}\hat w$ and $\hat a+\frac{1}{2}\sqrt{\omega}\hat w$, which has length $\omega$. \par % This procedure can be repeated for the other two eigenvalues and eigenvectors, that capture the other dimensions of variation. \par % In Xlisp-Stat we simply add a PCA method to the spin-proto. Sending an instance of the spin-proto the PCA message will draw the three principal axis, with squared length equal to the eigenvalues. The code is given below % % \begin{verbatim} (defmeth spin-proto :pca () (let* ((n (send self :num-points)) (x (send self :point-coordinate 0 (iseq n))) (y (send self :point-coordinate 1 (iseq n))) (z (send self :point-coordinate 2 (iseq n))) (c (* (1- n) (covariance-matrix x y z))) (m (list (mean x) (mean y) (mean z))) (g (eigen c)) (e (second g)) (f (sqrt (first g)))) (send self :dircos m (first e) (/ (elt f 0) 2)) (send self :dircos m (second e) (/ (elt f 1) 2)) (send self :dircos m (third e) (/ (elt f 2) 2)) (print e) (print f) )) (defmeth spin-proto :dircos (m w u) (send self :abline (+ m (* u w)) (- m (* u w)) ) ) (defmeth spin-proto :abline (a b) (send self :add-lines (make-pairs a b)) ) (defun make-pairs (x y) (let ((n (length x))) (mapcar #'(lambda (z) (list (elt x z) (elt y z))) (iseq n)) )) \end{verbatim} \end{document} -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Sun Nov 24 00:57:52 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25453; Sun, 24 Nov 1996 00:57:52 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25447; Sat, 23 Nov 1996 23:57:46 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id VAA15553; Sat, 23 Nov 1996 21:57:42 -0800 (PST) From: Jan Deleeuw Message-Id: <199611240557.VAA15553@galton.math.ucla.edu> Subject: TeX and Xlisp-Stat To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List), statsoft@julia.math.ucla.edu (Journal of Statistical Software), ucla-stat@julia.math.ucla.edu (ucla-stat mailing list) Date: Sat, 23 Nov 1996 21:57:41 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk If you have (a) a Java capable browser (b) Xlisp-Stat as a helper for the browser (i.e. lsp files are run locally by Xlisp-Stat) then you can go to http://www.stat.ucla.edu/textbook/multivariables/pca/pca3 You will see the dvi file of a small paper, with a hyperlink. Click the hyperlink and the dvi file will run an Xlisp-Stat program. I think this is a nice mix. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Tue Nov 26 12:40:27 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA09875; Tue, 26 Nov 1996 12:40:27 -0500 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA09870; Tue, 26 Nov 1996 11:40:17 -0600 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA22139 (5.67b/IDA-1.5 for ); Tue, 26 Nov 1996 12:39:42 -0500 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA16360 for ; Tue, 26 Nov 1996 12:39:00 -0400 Message-Id: <329B5542.20DF@kodak.com> Date: Tue, 26 Nov 1996 12:38:26 -0800 From: Jeff Morris Organization: Eastman Kodak Company X-Mailer: Mozilla 2.01 (Win16; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: Successive diffs and sums Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I seem to remember the existence of a statement that will return the successive differences and/or sums of the elements of a list. That is the new list will contain ((x[1]-x[0]) (x[2]-x[1])...) and similarly for sums. Do I remember correctly? Thanks! -- Jeffrey D. Morris, Ph.D. Health & Environment Labs Eastman Kodak Company morris@kodak.com From majordom@umnstat.stat.umn.edu Tue Nov 26 14:57:28 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11060; Tue, 26 Nov 1996 14:57:28 -0500 Return-Path: Received: from mcmail.CIS.McMaster.CA by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11054; Tue, 26 Nov 1996 13:57:18 -0600 Received: from johnfox (curly01.slip.yorku.ca [130.63.122.66]) by mcmail.CIS.McMaster.CA (8.7.1/8.7.1) with SMTP id OAA07906; Tue, 26 Nov 1996 14:56:59 -0500 (EST) Message-Id: <1.5.4.32.19961126195615.0066cc30@mcmail.cis.mcmaster.ca> X-Sender: jfox@mcmail.cis.mcmaster.ca X-Mailer: Windows Eudora Light Version 1.5.4 (32) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Tue, 26 Nov 1996 14:56:15 -0500 To: Jeff Morris , stat-lisp-news@umnstat.stat.umn.edu From: John Fox Subject: Re: Successive diffs and sums Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk At 12:38 PM 11/26/96 -0800, you wrote: >I seem to remember the existence of a statement that will return the >successive differences and/or sums of the elements of a list. That is >the new list will contain ((x[1]-x[0]) (x[2]-x[1])...) and similarly for >sums. Do I remember correctly? > Jeff, Difference and cumsum will do what you want. In general, apropos will find functions whose names contain a particular symbol, so (apropos 'diff) and (apropos 'sum) would lead you to what you need. John |----------------------------------------------------| | John Fox jfox@McMaster.ca | | Department of Sociology McMaster University | |----------------------------------------------------| From majordom@umnstat.stat.umn.edu Wed Nov 27 17:53:12 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29986; Wed, 27 Nov 1996 17:53:12 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29978; Wed, 27 Nov 1996 16:53:07 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id OAA21506 for stat-lisp-news@umnstat.stat.umn.edu; Wed, 27 Nov 1996 14:53:04 -0800 (PST) From: Jan Deleeuw Message-Id: <199611272253.OAA21506@galton.math.ucla.edu> Subject: no subject (file transmission) To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Wed, 27 Nov 1996 14:53:03 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;; ;; This adds a :pca method to the spin-proto. Sending the message to an ;; instance will add the three principal axes, in color. If the call is ;; (send p :pca :allaxes nil) then only the first principal axes is drawn ;; (in red, of course). If (send p :pca :erase t) is used, then the ;; axes are erased. ;; ;; Version 0.5: November 20, 1996; Jan de Leeuw ;; Version 0.6: November 24, 1996; John Fox added color and :erase ;; Version 1.0: November 27, 1996; Jan de Leeuw added :allaxes and ;; formatted printoutput ;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; (defmeth spin-proto :pca (&key (colors t) (allaxes t) erase) (let* ((n (send self :num-points)) (x (send self :point-coordinate 0 (iseq n))) (y (send self :point-coordinate 1 (iseq n))) (z (send self :point-coordinate 2 (iseq n))) (c (* (1- n) (covariance-matrix x y z))) (m (list (mean x) (mean y) (mean z))) (g (eigen c)) (e (second g)) (h (first g)) (s (sum h)) (f (sqrt h)) (cols (if colors (list 'red 'yellow 'blue) (list nil nil nil))) (axis-colors (if erase (repeat (send self :back-color) 3) cols))) (send self :dircos m (first e) (/ (elt f 0) 2) :color (select axis-colors 0)) (if allaxes (send self :dircos m (second e) (/ (elt f 1) 2) :color (select axis-colors 1))) (if allaxes (send self :dircos m (third e) (/ (elt f 2) 2) :color (select axis-colors 2))) (format t "~10,6f ~10,6f ~10,6f~%" (elt (first e) 0) (elt (first e) 1) (elt (first e) 2)) (if allaxes (format t "~10,6f ~10,6f ~10,6f~%" (elt (second e) 0) (elt (second e) 1) (elt (second e) 2))) (if allaxes (format t "~10,6f ~10,6f ~10,6f~%" (elt (third e) 0) (elt (third e) 1) (elt (third e) 2))) (format t "~10,6f ~10,6f ~10,6f~%" (/ (elt h 0) s) (/ (elt h 1) s) (/ (elt h 2) s)) )) (defmeth spin-proto :dircos (m w u &key color) (send self :abline (+ m (* u w)) (- m (* u w)) :color color) ) (defmeth spin-proto :abline (a b &key color) (let ((line-color (if color color (send self :draw-color)))) (send self :add-lines (make-pairs a b) :color line-color)) ) (defun make-pairs (x y) (let ((n (length x))) (mapcar #'(lambda (z) (list (elt x z) (elt y z))) (iseq n)) )) From majordom@umnstat.stat.umn.edu Mon Dec 2 09:51:36 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02891; Mon, 2 Dec 1996 09:51:36 -0500 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02886; Mon, 2 Dec 1996 08:51:31 -0600 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA13445 (5.67b/IDA-1.5 for ); Mon, 2 Dec 1996 09:50:47 -0500 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA17714 for ; Mon, 2 Dec 1996 09:02:08 -0400 Message-Id: <32A30B85.51B2@kodak.com> Date: Mon, 02 Dec 1996 09:01:57 -0800 From: Jeff Morris Organization: Eastman Kodak Company X-Mailer: Mozilla 2.01 (Win16; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: Dunnett's Test Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Does anyone have any code for Dunnett's test or any multiple comparison test for that matter? Thanks! Jeff -- Jeffrey D. Morris, Ph.D. Health & Environment Labs Eastman Kodak Company morris@kodak.com From majordom@umnstat.stat.umn.edu Mon Dec 2 14:09:47 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05106; Mon, 2 Dec 1996 14:09:47 -0500 Return-Path: Received: from sulu.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05101; Mon, 2 Dec 1996 13:09:42 -0600 Received: from chekov by sulu.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id LAA11158; Mon, 2 Dec 1996 11:09:31 -0800 (PST) Date: Mon, 2 Dec 1996 11:09:30 -0800 (PST) From: Thomas Lumley X-Sender: thomas@chekov To: Jeff Morris Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Dunnett's Test In-Reply-To: <32A30B85.51B2@kodak.com> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Mon, 2 Dec 1996, Jeff Morris wrote: > Does anyone have any code for Dunnett's test or any multiple > comparison test for that matter? I don't have Dunnett's test but I do have three p-value based corrections (Bonferroni, Holm, and Hochberg). Holm's test is a more powerful version of Bonferroni and Hochberg's test is still more powerful. They work well when the tests are correlated and when several of the null hypotheses are false (the situation when Bonferroni's inequality is conservative). The theory is explained in "Adjusted p-values for simultaneous inference" by SP Wright, Biometrics 48: 1005-1013. Both tests multiple the smallest of k p-values by k, the next smallest by k-1 and so on, but they differ in the way they exploit the original ordering of the p-values. Code follows: thomas lumley UW biostatistics -------------------- (defun holm (p-values) "Args: p-values Adjusts a list of p-values by Holm's method" (let* ((n (length p-values)) (ranks (rank p-values)) (qi (* p-values (- n ranks))) (junk (dotimes (i (- n 1) qi) (let* ( (index (position (+ 1 i) ranks )) (lastindex (position i ranks )) ) (setf (elt qi index) (max (elt qi index) (elt qi lastindex))) ) ))) qi )) (defun hochberg (p-values) "Args: p-values Adjusts a list of p-values by Hochberg's method" (let* ((n (length p-values)) (ranks (rank p-values)) (qi (* p-values (- n ranks))) (junk (dotimes (i (- n 1) qi) (let* ( (index (position (- n i 2) ranks )) (lastindex (position (- n i 1) ranks )) ) (setf (elt qi index) (min (elt qi index) (elt qi lastindex))) ) ))) qi )) (defun bonferroni (p-values) "Args: p-values Adjusts a list of p-values by Bonferroni's method. Use (holm) or (hochberg) instead." (* p-values (length p-values)) ) From majordom@umnstat.stat.umn.edu Mon Dec 2 17:56:36 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA09489; Mon, 2 Dec 1996 17:56:36 -0500 Return-Path: Received: from mailb.surrey.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA09482; Mon, 2 Dec 1996 16:56:31 -0600 Received: from soc.surrey.ac.uk (actually host boris-gate.soc.surrey.ac.uk) by mailb.surrey.ac.uk with SMTP (PP); Mon, 2 Dec 1996 22:56:18 +0000 Received: from annex1-14.surrey.ac.uk by boris.soc.surrey.ac.uk via Ethernet with SMTP id aa25684; 2 Dec 96 22:56 GMT X-Sender: scs1ng@boris.soc.surrey.ac.uk Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 2 Dec 1996 22:54:18 +0000 To: stat-lisp-news@umnstat.stat.umn.edu From: Nigel Gilbert Subject: Bigger, better Inspector Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The Lisp-Stat package comes with a file called 'inspect.lsp' in the Examples. I've taken this and extended it somewhat, so that it inspects most LispStat objects and has a convenient menu interface. I have attached below a fuller description, lifted from the source file. I'm not including the whole file, becuase it would clutter the mail boxes of those who are not interested in it. But if you would like a copy to use, or would like to provide a Web or FTP site where it can be placed, please email me and I will send you a copy. Happy inspecting! ;;;; inspect.lsp version 1.3 Nigel Gilbert December 2, 1996 ;;;; ;;;; XLISP-STAT 2.1 Copyright (c) 1990, by Luke Tierney ;;;; Additions to Xlisp 2.1, Copyright (c) 1989 by David Michael Betz ;;;; You may give out copies of this software; for conditions see the file ;;;; COPYING included with this distribution. ;;;; ;;;; Based on XLISP-STAT Examples "inspect.lsp" with additional ;;;; capablity to inspect objects, closures, structures and hash tables, ;;;; a menu interface, and some bugs removed. ;;; Usage: ;;; ----- ;;; ;;;--> using the Menu ;;; Loading the file will add a new menu to the menu-bar, "Inspector", ;;; with items "Inspect", "Update" and "Close all". ;;; "Inspect" inspects whatever is selected in the front window (e.g. ;;; to inspect the value of the variable "foo", either find the ;;; word "foo" somewhere or type the word "foo" in a window, ;;; select it and then type command-I or choose Inspect from ;;; the Inspector menu). ;;; "Update" redisplays the front Inspector window, to reflect any changes ;;; in the value of the item being inspected (e.g. by programmatic or ;;; command line edits 'behind the scenes') ;;; "Close all" closes all inspector windows that are open. ;;; The titles of all inspect windows are appended to this menu as they ;;; are opened. ;;; Selecting an inspector window title from the menu brings that window ;;; to the top. ;;; ;;;--> using an Inspector window ;;; Double clicking on any item shown in an Inspector window will ;;; open another Inspector window to inspect that item, until ;;; there is nothing more to inspect (you've got down to a string ;;; or atom) ;;; There will often be a button marked "Edit" at the bottom of an ;;; inspector window. If there is, selecting an item in the window ;;; and then pressing the Edit button allows you to change that item. ;;; A dialogue box is displayed and you can enter the new value (which ;;; is evaluated). ;;; ;;;--> using the Command window ;;; ;;; (inspect data &optional (edit nil)) ;;; opens an inspect window on the value of data. If EDIT is T, allows ;;; editing of values of selected component items of data. ;;; Returns the inspector object. ;;; Double-clicking an item opens an inspect window on that item. ;;; ;;; (close-all-inspectors) ;;; Closes all inspector windows ;;; ;;; ;;; Global options: ;;; -------------- ;;; *inspect-edit* Default editing state, T (allow edits) or NIL ;;; ;;; Installation ;;; ------------ ;;; 1. Load this file ;;; 2. Compile it with (compile-inspect) ;;; 3. Move the file "inspect.fsl" to the Autoload directory/folder ;;; 4. Edit the file "autoload.lsp" to add the line: ;;; (autoload inspect "inspect") ;;; 5. Then, whenever you want use the inspector, just evaluate ;;; either (inspect) to load the menu ;;; or (inspect {data}) to inspect {data} ;;; 6. To see it working, load the inspector (see step 5) and then ;;; evaluate ;;; (setq test-data (test-inspect)) ;;; then select the word "test-data" in the XLISP-STAT window ;;; with the mouse and choose the "Inspect" item from the ;;; Inspector menu. Double-click on an item in the Inspect window ;;; see its value, and so on... ;;; ;;; Bugs ;;; ---- ;;; Won't display the values of structure slots which have accessor ;;; functions defined using the :con-name option in defstruct ;;; (Because XLSIP-STAT doesn't make the names of these functions ;;; available to the Inspector). Works with ordinary accessors, ;;; though. ;;; ;;; Developed on a Macintosh with XLISP-STAT 2.1 Release 3.44 (Beta). ;;; Not tested on other other machines ;;; ;;; Please report other bugs, wish-lists etc. to: ;;; ;;; Nigel Gilbert (n.gilbert@soc.surrey.ac.uk) ;;; University of Surrey, UK ;;; ;;; Acknowledgements: To the memory of Interlisp-D. R.I.P. ____________________________________________________________________ Prof G. Nigel Gilbert, Department of Sociology, University of Surrey Guildford GU2 5XH, UK. Tel: +44 1483 259173 Fax: +44 1483 306290 From majordom@umnstat.stat.umn.edu Mon Dec 2 17:56:36 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA09488; Mon, 2 Dec 1996 17:56:36 -0500 Return-Path: Received: from mailb.surrey.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA09478; Mon, 2 Dec 1996 16:56:26 -0600 Received: from soc.surrey.ac.uk (actually host boris-gate.soc.surrey.ac.uk) by mailb.surrey.ac.uk with SMTP (PP); Mon, 2 Dec 1996 22:56:13 +0000 Received: from annex1-14.surrey.ac.uk by boris.soc.surrey.ac.uk via Ethernet with SMTP id aa25680; 2 Dec 96 22:56 GMT X-Sender: scs1ng@boris.soc.surrey.ac.uk Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 2 Dec 1996 20:40:02 +0000 To: stat-lisp-news@umnstat.stat.umn.edu From: Nigel Gilbert Subject: Bug: Histogram Y scale Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk If you evaluate: (def precipitation (list .77 1.74 .81 1.20 1.95 1.20 .47 1.43 3.37 2.20 3.00 3.09 1.51 2.10 .52 1.62 1.31 .32 .59 .81 2.81 1.87 1.18 1.35 4.75 2.48 .96 1.89 .90 2.05)) PRECIPITATION > (setq hist (histogram precipitation)) # > (send hist :y-axis t) (T NIL 4) you'll get a histogram with the Y axis showing the scale (Incidentally, why isn't this the default? When would one want to plot a histogram and _not_ want to know what the heights of the columns represent?) However, the scale shown is wrong! In this specific case, the scale ranges from 0 to 0.6, when the scale is supposed to be showing a count (the number of observations in each bin), which clearly cannot be non-integral. Other examples I've tried are similarly scaled wrongly. If you try to force the issue, by setting the range 'by hand' using (send hist :range 1 0 10) this is accepted, and yields a sensible scale, but any subsequent :adjust-to-data message reverts to the wrong scale again. Anyone any ideas about where the problem is? I'm using release 3.47 on a Mac. ____________________________________________________________________ Prof G. Nigel Gilbert, Department of Sociology, University of Surrey Guildford GU2 5XH, UK. Tel: +44 1483 259173 Fax: +44 1483 306290 From majordom@umnstat.stat.umn.edu Mon Dec 2 18:23:42 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA09756; Mon, 2 Dec 1996 18:23:42 -0500 Return-Path: Received: from mars.stat.wisc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA09751; Mon, 2 Dec 1996 17:23:35 -0600 Received: (from loh@localhost) by mars.stat.wisc.edu (8.7.6/8.7.3) id RAA03130; Mon, 2 Dec 1996 17:21:04 -0600 (CST) Date: Mon, 2 Dec 1996 17:21:04 -0600 (CST) Message-Id: <199612022321.RAA03130@mars.stat.wisc.edu> From: Wei-Yin Loh To: Nigel.Gilbert@soc.surrey.ac.uk Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: (message from Nigel Gilbert on Mon, 2 Dec 1996 20:40:02 +0000) Subject: Re: Bug: Histogram Y scale Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Thanks for finding out how to make hist show the Y axis. I agree that it would be good to make it the default. It looks like the Y axis is showing the density scale. If so, then it is correct, at least according to textbooks such as Freedman et al.'s "Statistics". Wei-Yin Loh Department of Statistics University of Wisconsin Madison, WI 53706 ============ X-Sender: scs1ng@boris.soc.surrey.ac.uk Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 2 Dec 1996 20:40:02 +0000 From: Nigel Gilbert Sender: owner-stat-lisp-news@stat.umn.edu Precedence: bulk If you evaluate: (def precipitation (list .77 1.74 .81 1.20 1.95 1.20 .47 1.43 3.37 2.20 3.00 3.09 1.51 2.10 .52 1.62 1.31 .32 .59 .81 2.81 1.87 1.18 1.35 4.75 2.48 .96 1.89 .90 2.05)) PRECIPITATION > (setq hist (histogram precipitation)) # > (send hist :y-axis t) (T NIL 4) you'll get a histogram with the Y axis showing the scale (Incidentally, why isn't this the default? When would one want to plot a histogram and _not_ want to know what the heights of the columns represent?) However, the scale shown is wrong! In this specific case, the scale ranges from 0 to 0.6, when the scale is supposed to be showing a count (the number of observations in each bin), which clearly cannot be non-integral. Other examples I've tried are similarly scaled wrongly. If you try to force the issue, by setting the range 'by hand' using (send hist :range 1 0 10) this is accepted, and yields a sensible scale, but any subsequent :adjust-to-data message reverts to the wrong scale again. Anyone any ideas about where the problem is? I'm using release 3.47 on a Mac. ____________________________________________________________________ Prof G. Nigel Gilbert, Department of Sociology, University of Surrey Guildford GU2 5XH, UK. Tel: +44 1483 259173 Fax: +44 1483 306290 From majordom@umnstat.stat.umn.edu Mon Dec 2 19:48:24 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10186; Mon, 2 Dec 1996 19:48:24 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10181; Mon, 2 Dec 1996 18:48:17 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id QAA14184 for stat-lisp-news@umnstat.stat.umn.edu; Mon, 2 Dec 1996 16:48:13 -0800 (PST) From: Jan Deleeuw Message-Id: <199612030048.QAA14184@galton.math.ucla.edu> Subject: Wei-Yin Loh, Nigel Gilbert, and my-histogram To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Mon, 2 Dec 1996 16:48:13 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Dissatisfaction with the standard histogram proto produced http://www.stat.ucla.edu/archive/xlispstat/code/statistics/plotting/my-histogram which has a lot of additional goodies in its menu. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon Dec 2 20:25:11 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10434; Mon, 2 Dec 1996 20:25:11 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10429; Mon, 2 Dec 1996 19:25:06 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id RAA14566 for stat-lisp-news@umnstat.stat.umn.edu; Mon, 2 Dec 1996 17:25:02 -0800 (PST) From: Jan Deleeuw Message-Id: <199612030125.RAA14566@galton.math.ucla.edu> Subject: Cleanup Archive To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Mon, 2 Dec 1996 17:25:02 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I managed to clean up the Multivariate Analysis and Plotting directories in http://www.stat.ucla.edu/archive/xlispstat/code/statistics/ -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@nokomis.stat.umn.edu Tue Dec 3 10:44:38 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA14782; Tue, 3 Dec 1996 10:44:38 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA14773; Tue, 3 Dec 1996 09:43:48 -0600 From: Luke Tierney Message-Id: <9612031543.AA14773@nokomis.stat.umn.edu> Subject: Re: Bug: Histogram Y scale To: loh@stat.wisc.edu (Wei-Yin Loh) Date: Tue, 3 Dec 1996 09:43:48 -0600 (CST) Cc: Nigel.Gilbert@soc.surrey.ac.uk, stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: <199612022321.RAA03130@mars.stat.wisc.edu> from "Wei-Yin Loh" at Dec 2, 96 05:21:04 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 800 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > Thanks for finding out how to make hist show the Y axis. I agree that > it would be good to make it the default. > > It looks like the Y axis is showing the density scale. If so, then it > is correct, at least according to textbooks such as Freedman et al.'s > "Statistics". > > Wei-Yin Loh > Department of Statistics > University of Wisconsin > Madison, WI 53706 > My preference is to have no axis as the default, though it would be better if that could be changed easily. Threason I like no axis is that it is the shape that counts, not the normalization. Different people like different normalizations and get confused if the one used doesn't suit their preference. The reason I chose the density scale as the underlying internal scale is to make superposition of densities easy. luke From majordom@umnstat.stat.umn.edu Tue Dec 3 20:20:35 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26822; Tue, 3 Dec 1996 20:20:35 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26816; Tue, 3 Dec 1996 19:20:29 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id RAA25269 for stat-lisp-news@umnstat.stat.umn.edu; Tue, 3 Dec 1996 17:20:26 -0800 (PST) From: Jan Deleeuw Message-Id: <199612040120.RAA25269@galton.math.ucla.edu> Subject: an inspector calls To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Tue, 3 Dec 1996 17:20:25 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Nigel Gilbert's inspector.lsp is in http://www.stat.ucla.edu/archive/xlispstat/code/incoming/ or ftp://ftp.stat.ucla.edu/pub/lisp/xlisp/xlisp-stat/code/incoming -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Thu Dec 5 12:52:17 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27593; Thu, 5 Dec 1996 12:52:17 -0500 Return-Path: Received: from halley.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27561; Thu, 5 Dec 1996 11:45:44 -0600 Received: from libiya.upf.es by halley.upf.es with SMTP (1.37.109.15/16.2) id AA268181776; Thu, 5 Dec 1996 18:49:36 GMT Received: by libiya.upf.es (1.38.193.4/15.6) id AA03096; Thu, 5 Dec 1996 18:46:09 GMT Date: Thu, 5 Dec 1996 18:46:09 +0000 ("WET) From: Frederic Udina To: lisp stat list Subject: global vars Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I've written a quick hack for seeing and editing most of xlisp-stat global vars. Would someone like it enough to improve it? Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Ramon Trias Fargas, 25-27 | Compuserve: 100555.603 08005 Barcelona | WWWeb: http://libiya.upf.es/ SPAIN | or http://www.econ.upf.es/~udina ___________________________________/_\_________________________________ ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;;; file edit-globals.lsp, frederic udina, dec 5 1996 ;;;; ;;;; calling (edit-globals) a dialog window will appear ;;;; making it easy to see the list of global variables ;;;; its values and allowing also to edit them. ;;;; ;;;; history: fu 96/12/5 wrote the first quick version ;;;; waiting for someone to take it on ;;;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; (defvar *varlist* '(*APPLYHOOK* *ASK-ON-REDEFINE* *BAKTRACE-PRINT-ARGUMENTS* *BREAKENABLE* *COMMAND-LINE* *COMPILE-PRINT* *COMPILE-VERBOSE* *DEFAULT-PATH* *ERROR-OUTPUT* *EVALHOOK* *FEATURES* *FLOAT-FORMAT* *GC-FLAG* *KEEP-DOCUMENTATION-STRINGS* *MODULES* *PACKAGE* *PRINT-ARRAY* *PRINT-BASE* *PRINT-CASE* *PRINT-CIRCLE* *PRINT-ESCAPE* *PRINT-GENSYM* *PRINT-LENGTH* *PRINT-LEVEL* *PRINT-PRETTY* *PRINT-RADIX* *PRINT-READABLY* *RANDOM-STATE* *READ-CIRCLE-DATA* *READ-SUPPRESS* *STANDARD-INPUT* *STANDARD-OUTPUT* *STARTUP-FUNCTIONS* *TERMINAL-IO* *TRACE-OUTPUT* *TRACELIMIT* *TRACELIST* *TRACENABLE* *VARIABLES*)) (defun edit-globals () (let* (current (itlist (send list-item-proto :new (mapcar #'symbol-name *varlist*) :columns 1)) (btn1 (send button-item-proto :new "Show")) (btn2 (send button-item-proto :new "Info")) (btn3 (send button-item-proto :new "Edit")) (ti (send text-item-proto :new " ")) (dlg (send dialog-proto :new (list (list itlist) (list ti) (list btn2 btn3)) :modal nil))) (send itlist :slot-value 'action #'(lambda (&optional (double nil)) (send ti :text (format nil "Value: ~a" (eval (select *varlist* (send itlist :selection))))))) (send btn1 :action #'(lambda () (setq current (send itlist :selection)) ;;(format t "~a ~a ~%" current (select *varlist* current)) (message-dialog (format nil "~a is ~a" (symbol-name (select *varlist* current)) (eval (select *varlist* current)))))) (send btn2 :action #'(lambda () (setq current (send itlist :selection)) ;;(format t "~a ~a ~%" current (select *varlist* current)) (message-dialog (format nil "~a: ~a~%~a" (symbol-name (select *varlist* current)) (eval (select *varlist* current)) (documentation (select *varlist* current) 'variable))))) (send btn3 :action #'(lambda () (setq current (send itlist :selection)) (when (setq newval (get-string-dialog "Edit a new nalue (at your own risk!)" :initial (format nil "~s" (eval (select *varlist* current))))) (set (select *varlist* current) (car (eval-string newval)))))) (def edit-globals-dialog dlg) )) (defun read-string (&optional (stream *standard-input*) (error-on-eof t) (end-of-read-char #\Newline)) "Args optional (stream *standard-input*) (error-on-eof t) (end-of-read-char #\Newline) Reads all chars from stream until end-of-read-char is found, and returns them in a string. Max 25600 chars! If error-on-eof is nil and eof is reached, reading is also completed and the string is returned." (let* ((max-len 25600) (theString (make-array max-len)) (len 0) (aChar nil)) (loop (setq aChar (read-char stream)) (when (and (eql aChar nil) error-on-eof) (print "end of file on read-string") (return nil)) (when (or (eql end-of-read-char aChar) (eql aChar nil)) (return (coerce (select theString (iseq len)) 'string))) (setf (elt theString len) aChar) (setq len (+ len 1)) (when (>= len max-len) (print "maximum length string reached") (return nil))))) (defun eval-string (aString) "Returns a list of the expressions obtained on evaluation of the characters in aString" (let ((st (make-string-input-stream aString)) (result nil) (expr nil)) (loop (setq expr (read st nil 'eof)) (when (eql expr 'eof) (return (reverse result))) (setf result (cons expr result))))) From majordom@umnstat.stat.umn.edu Thu Dec 5 13:15:51 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27779; Thu, 5 Dec 1996 13:15:51 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27771; Thu, 5 Dec 1996 12:15:43 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id KAA11937 for stat-lisp-news@umnstat.stat.umn.edu; Thu, 5 Dec 1996 10:15:41 -0800 (PST) From: Jan Deleeuw Message-Id: <199612051815.KAA11937@galton.math.ucla.edu> Subject: archive To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Thu, 5 Dec 1996 10:15:40 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Nigel Gilbert's inspector.lsp and Frederic Udina's edit-globals.lsp have their own directories in http://www.stat.ucla.edu/archive/xlispstat/code/utilities --- Jan -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Thu Dec 5 13:18:07 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27803; Thu, 5 Dec 1996 13:18:07 -0500 Return-Path: Received: from halley.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27790; Thu, 5 Dec 1996 12:16:59 -0600 Received: from libiya.upf.es by halley.upf.es with SMTP (1.37.109.15/16.2) id AA268181776; Thu, 5 Dec 1996 18:49:36 GMT Received: by libiya.upf.es (1.38.193.4/15.6) id AA03096; Thu, 5 Dec 1996 18:46:09 GMT Date: Thu, 5 Dec 1996 18:46:09 +0000 ("WET) From: Frederic Udina To: lisp stat list Subject: global vars Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I've written a quick hack for seeing and editing most of xlisp-stat global vars. Would someone like it enough to improve it? Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Ramon Trias Fargas, 25-27 | Compuserve: 100555.603 08005 Barcelona | WWWeb: http://libiya.upf.es/ SPAIN | or http://www.econ.upf.es/~udina ___________________________________/_\_________________________________ ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;;; file edit-globals.lsp, frederic udina, dec 5 1996 ;;;; ;;;; calling (edit-globals) a dialog window will appear ;;;; making it easy to see the list of global variables ;;;; its values and allowing also to edit them. ;;;; ;;;; history: fu 96/12/5 wrote the first quick version ;;;; waiting for someone to take it on ;;;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; (defvar *varlist* '(*APPLYHOOK* *ASK-ON-REDEFINE* *BAKTRACE-PRINT-ARGUMENTS* *BREAKENABLE* *COMMAND-LINE* *COMPILE-PRINT* *COMPILE-VERBOSE* *DEFAULT-PATH* *ERROR-OUTPUT* *EVALHOOK* *FEATURES* *FLOAT-FORMAT* *GC-FLAG* *KEEP-DOCUMENTATION-STRINGS* *MODULES* *PACKAGE* *PRINT-ARRAY* *PRINT-BASE* *PRINT-CASE* *PRINT-CIRCLE* *PRINT-ESCAPE* *PRINT-GENSYM* *PRINT-LENGTH* *PRINT-LEVEL* *PRINT-PRETTY* *PRINT-RADIX* *PRINT-READABLY* *RANDOM-STATE* *READ-CIRCLE-DATA* *READ-SUPPRESS* *STANDARD-INPUT* *STANDARD-OUTPUT* *STARTUP-FUNCTIONS* *TERMINAL-IO* *TRACE-OUTPUT* *TRACELIMIT* *TRACELIST* *TRACENABLE* *VARIABLES*)) (defun edit-globals () (let* (current (itlist (send list-item-proto :new (mapcar #'symbol-name *varlist*) :columns 1)) (btn1 (send button-item-proto :new "Show")) (btn2 (send button-item-proto :new "Info")) (btn3 (send button-item-proto :new "Edit")) (ti (send text-item-proto :new " ")) (dlg (send dialog-proto :new (list (list itlist) (list ti) (list btn2 btn3)) :modal nil))) (send itlist :slot-value 'action #'(lambda (&optional (double nil)) (send ti :text (format nil "Value: ~a" (eval (select *varlist* (send itlist :selection))))))) (send btn1 :action #'(lambda () (setq current (send itlist :selection)) ;;(format t "~a ~a ~%" current (select *varlist* current)) (message-dialog (format nil "~a is ~a" (symbol-name (select *varlist* current)) (eval (select *varlist* current)))))) (send btn2 :action #'(lambda () (setq current (send itlist :selection)) ;;(format t "~a ~a ~%" current (select *varlist* current)) (message-dialog (format nil "~a: ~a~%~a" (symbol-name (select *varlist* current)) (eval (select *varlist* current)) (documentation (select *varlist* current) 'variable))))) (send btn3 :action #'(lambda () (setq current (send itlist :selection)) (when (setq newval (get-string-dialog "Edit a new nalue (at your own risk!)" :initial (format nil "~s" (eval (select *varlist* current))))) (set (select *varlist* current) (car (eval-string newval)))))) (def edit-globals-dialog dlg) )) (defun read-string (&optional (stream *standard-input*) (error-on-eof t) (end-of-read-char #\Newline)) "Args optional (stream *standard-input*) (error-on-eof t) (end-of-read-char #\Newline) Reads all chars from stream until end-of-read-char is found, and returns them in a string. Max 25600 chars! If error-on-eof is nil and eof is reached, reading is also completed and the string is returned." (let* ((max-len 25600) (theString (make-array max-len)) (len 0) (aChar nil)) (loop (setq aChar (read-char stream)) (when (and (eql aChar nil) error-on-eof) (print "end of file on read-string") (return nil)) (when (or (eql end-of-read-char aChar) (eql aChar nil)) (return (coerce (select theString (iseq len)) 'string))) (setf (elt theString len) aChar) (setq len (+ len 1)) (when (>= len max-len) (print "maximum length string reached") (return nil))))) (defun eval-string (aString) "Returns a list of the expressions obtained on evaluation of the characters in aString" (let ((st (make-string-input-stream aString)) (result nil) (expr nil)) (loop (setq expr (read st nil 'eof)) (when (eql expr 'eof) (return (reverse result))) (setf result (cons expr result))))) From majordom@umnstat.stat.umn.edu Thu Dec 5 18:08:14 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00456; Thu, 5 Dec 1996 18:08:14 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00451; Thu, 5 Dec 1996 17:08:07 -0600 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.6.4.3/10.1) id SAA04227; Thu, 5 Dec 1996 18:08:05 -0500 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 5 Dec 1996 18:10:00 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: LispStat for Be? Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Is anyone out there contemplating porting LispStat to the BeOS? Forrest Young Professor, Psychometrics CB# 3270 Davie Hall email: mailto://forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software at: http://forrest.psych.unc.edu/research/ViSta.html From majordom@umnstat.stat.umn.edu Thu Dec 5 22:16:46 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA01907; Thu, 5 Dec 1996 22:16:46 -0500 Return-Path: Received: from alpha.maths.unsw.EDU.AU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA01902; Thu, 5 Dec 1996 21:16:38 -0600 Received: (from frances@localhost) by alpha.maths.unsw.EDU.AU (8.7.5/8.7.3) id OAA04446 for stat-lisp-news@stat.umn.edu; Fri, 6 Dec 1996 14:16:33 +1100 (EST) Date: Fri, 6 Dec 1996 14:16:33 +1100 (EST) From: Frances Fang Message-Id: <199612060316.OAA04446@alpha.maths.unsw.EDU.AU> To: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, Could you please send the following job advertisement to the members of your news group ? Thanks. DIRECTOR: STATISTICAL CONSULTING ASSOCIATE LECTURER/LECTURER IN STATISTICS SCHOOL OF MATHEMATICS REF. 548SNV ________________________________________________________________________________ FIXED TERM - Salary: Associate Lecture: $30,145-$40,889 per year, Lecturer: $43,042-$51,113 per year depending on qualifications and experience. ________________________________________________________________________________ The successful applicant will be expected to make a strong contribution to the consulting, teaching and research efforts of the Department. The Department of Statistics has secured a UNSW Development grant to establish a Professional Statistical Consulting Service for the UNSW Research Community. Duties include initial contact with clients, assessment of their statistical advice requirements, provision of advice and analysis and directing enquiries to members of the Department. It is intended that the Statistical Consulting Centre rapidly become self funding. Funds will be derived from charging for advice provided as well as from fees for various courses in statistical methods for researchers. The University of New South Wales is one of Australia's leading technological Universities. The School of Mathematics at UNSW, which comprises the Departments of Applied Mathematics, Pure Mathematics and Statistics, is highly successful in Australia in attracting financial support for research. The Department has active research groups in analysis of time series, extreme value theory, stochastic processes, financial modelling statistics, survival analysis, quality management and statistical computing. Applications from candidates in any field of statistical science are invited. Candidates must have at least a good honours degree plus demonstrated experience and ability in statistical consulting; a sound knowledge of the use of computing in statistical practice and a knowledge and understanding of EEO/AA principles. In addition to the requirements outlined in the application procedure applicants should provide a statement of consulting and research interest, publication record and consulting experience. The position is for a fixed term of one year and is available from 3 March, 1997. Further information may be obtained from Dr P J Cooke, Head of Statistics, on telephone (02) 9385 3356, facsimile (02) 9385 1071 or emial: pjc@maths.unds.edu.au. Applications close 15 January 1997. From majordom@umnstat.stat.umn.edu Fri Dec 6 10:59:38 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13434; Fri, 6 Dec 1996 10:59:38 -0500 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13426; Fri, 6 Dec 1996 09:59:31 -0600 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA08523 (5.67b/IDA-1.5 for ); Fri, 6 Dec 1996 10:58:52 -0500 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA5768 for ; Fri, 6 Dec 1996 10:58:09 -0400 Message-Id: <32A86CC6.5E6F@kodak.com> Date: Fri, 06 Dec 1996 10:58:14 -0800 From: Jeff Morris Organization: Eastman Kodak Company X-Mailer: Mozilla 2.01 (Win16; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: gnuplot with Windows Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Before I start plowing through the GNUPLOT manual, has anyone had success with producing nice graphics with gnuplot.lsp in a MS Windows environment? I would like to be able to cut and paste publication quality graphs into a MS Word document. TIA! Jeff -- Jeffrey D. Morris, Ph.D. Health & Environment Labs Eastman Kodak Company morris@kodak.com From majordom@umnstat.stat.umn.edu Fri Dec 6 12:59:04 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14758; Fri, 6 Dec 1996 12:59:04 -0500 Return-Path: Received: from neurocog.lrdc.pitt.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14751; Fri, 6 Dec 1996 11:58:47 -0600 Received: by neurocog.lrdc.pitt.edu (1.37.109.10G/16.2) id AA266815118; Fri, 6 Dec 1996 12:58:38 -0500 Date: Fri, 6 Dec 1996 12:58:38 -0500 (EST) From: Jeff Shrager To: Jeff Morris Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: gnuplot with Windows In-Reply-To: <32A86CC6.5E6F@kodak.com> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk What's gnuplot.lsp if I might ask? I use gnuplot from lisp (gcl) all the time (under Unix) and find that it's HORRIBLE as a call-out application. If anyone can suggest a better call-out plotting package than gnuplot, I (and probably many others) would be in your eternal debt. (Or, at least for a couple of years, anyhow. :) 'Jeff From majordom@umnstat.stat.umn.edu Fri Dec 6 13:31:28 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15041; Fri, 6 Dec 1996 13:31:28 -0500 Return-Path: Received: from halley.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15034; Fri, 6 Dec 1996 12:31:19 -0600 Received: from libiya.upf.es by halley.upf.es with SMTP (1.37.109.15/16.2) id AA100210899; Fri, 6 Dec 1996 19:34:59 GMT Received: by libiya.upf.es (1.38.193.4/15.6) id AA03631; Fri, 6 Dec 1996 19:31:31 GMT Date: Fri, 6 Dec 1996 19:31:30 +0000 ("WET) From: Frederic Udina To: Jeff Shrager Cc: Jeff Morris , stat-lisp-news@umnstat.stat.umn.edu Subject: Re: gnuplot with Windows In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk gnuplot-lsp is a hack made by jan deleeuw anf myself to make easier to call gnuplot from xlisp-stat graphical objects. I agree with you about that HORRIBLE interface but I think thah gnuplot 2.6 has done a giant step to avoid that. Anyone with experience in that arena? On Fri, 6 Dec 1996, Jeff Shrager wrote: > What's gnuplot.lsp if I might ask? I use gnuplot from lisp (gcl) all > the time (under Unix) and find that it's HORRIBLE as a call-out > application. If anyone can suggest a better call-out plotting package > than gnuplot, I (and probably many others) would be in your eternal > debt. (Or, at least for a couple of years, anyhow. :) > > 'Jeff > > > Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Ramon Trias Fargas, 25-27 | Compuserve: 100555.603 08005 Barcelona | WWWeb: http://libiya.upf.es/ SPAIN | or http://www.econ.upf.es/~udina ___________________________________/_\_________________________________ From majordom@umnstat.stat.umn.edu Fri Dec 6 13:50:44 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15247; Fri, 6 Dec 1996 13:50:44 -0500 Return-Path: Received: from neurocog.lrdc.pitt.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15242; Fri, 6 Dec 1996 12:50:37 -0600 Received: by neurocog.lrdc.pitt.edu (1.37.109.10G/16.2) id AA274668218; Fri, 6 Dec 1996 13:50:18 -0500 Date: Fri, 6 Dec 1996 13:50:18 -0500 (EST) From: Jeff Shrager To: Frederic Udina Cc: Jeff Morris , stat-lisp-news@umnstat.stat.umn.edu Subject: Re: gnuplot with Windows In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk > I agree with you about > that HORRIBLE interface but I think thah gnuplot 2.6 has done a giant > step to avoid that. Anyone with experience in that arena? Hmmm. Wel, I think that 3.5 is the latest version, which is what I'm running, and it still bytes. 'Jeff From majordom@umnstat.stat.umn.edu Mon Dec 9 03:10:12 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24296; Mon, 9 Dec 1996 03:10:12 -0500 Return-Path: Received: from hilbert.edv.agrar.tu-muenchen.de by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24291; Mon, 9 Dec 1996 02:09:57 -0600 Received: by hilbert.edv.agrar.tu-muenchen.de; id AA05577; Mon, 9 Dec 1996 09:08:17 +0100 Date: Mon, 9 Dec 1996 09:08:16 +0100 (MET) From: Max Pilgram X-Sender: max@hilbert.edv.agrar.tu-muenchen.de To: Jeff Shrager Cc: Jeff Morris , stat-lisp-news@umnstat.stat.umn.edu Subject: Re: gnuplot with Windows In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Fri, 6 Dec 1996, Jeff Shrager wrote: > What's gnuplot.lsp if I might ask? I use gnuplot from lisp (gcl) all > the time (under Unix) and find that it's HORRIBLE as a call-out > application. If anyone can suggest a better call-out plotting package > than gnuplot, I (and probably many others) would be in your eternal > debt. (Or, at least for a couple of years, anyhow. :) > > 'Jeff Bernhard Walter wrote a method to export figures derived from graph-proto to XFIG 3.1. XFIG is an interactive, menu driven graphic package available for Unix machines. I think the file "xlisp.lsp" should be available on the stat-lisp server. Max Max Pilgram; Technische Universitaet Muenchen-Weihenstephan Lehreinheit fuer Mathematik u. Statistik, Datenverarbeitungsstelle phone: 08161-714040; e-mail: max@edv.agrar.tu-muenchen.de www: http://www.edv.agrar.tu-muenchen.de/dvs/persons/max/max.html From majordom@umnstat.stat.umn.edu Mon Dec 9 14:10:46 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27406; Mon, 9 Dec 1996 14:10:46 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27401; Mon, 9 Dec 1996 13:10:40 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id LAA17687 for stat-lisp-news@umnstat.stat.umn.edu; Mon, 9 Dec 1996 11:10:11 -0800 (PST) From: Jan Deleeuw Message-Id: <199612091910.LAA17687@galton.math.ucla.edu> Subject: archive updates To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Mon, 9 Dec 1996 11:10:10 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk In http://www.stat.ucla.edu/archive/xlispstat/ I added a link to the Scales and Smith "Surfer's Guide to Lisp-Stat". In http://www.stat.ucla.edu/archive/xlispstat/code/statistics/regression/ I installed Bob Stine's Variance Inflation Factor package. In http://www.stat.ucla.edu/archive/xlispstat/code/utilities/foreign/ I installed Bob Stine's customization packages to load code fragments at runtime (translated to Code Warrior 10) -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon Dec 9 14:34:27 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27686; Mon, 9 Dec 1996 14:34:27 -0500 Return-Path: Received: from ts011p15.pop3b.netvision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27676; Mon, 9 Dec 1996 13:33:50 -0600 Received: (from hjstein@localhost) by blinky.bfr.co.il (8.7.6/8.6.9) id SAA02615; Mon, 9 Dec 1996 18:05:40 +0200 Date: Mon, 9 Dec 1996 18:05:40 +0200 Message-Id: <199612091605.SAA02615@blinky.bfr.co.il> From: "Harvey J. Stein" To: stat-lisp-news@umnstat.stat.umn.edu Subject: xlispstat seg faults when setting frame size of sequence-slider-dialogs. Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I can get xlispstat to seg fault in 3 commands: hjstein@blinky:~$ xlispstat XLISP-PLUS version 2.1g Portions Copyright (c) 1988, by David Betz. Modified by Thomas Almy and others. XLISP-STAT 2.1 Release 3.47 (Beta). Copyright (c) 1989-1994, by Luke Tierney. > (setq p (sequence-slider-dialog '(-1 0 1) :action (lambda (s) (format t "hi\n")))) # > (send p :frame-size) (210 116) > (send p :frame-size 405 200) Segmentation fault (core dumped) Any ideas? -- Dr. Harvey J. Stein Berger Financial Research abel@netvision.net.il From majordom@umnstat.stat.umn.edu Mon Dec 9 14:34:27 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27688; Mon, 9 Dec 1996 14:34:27 -0500 Return-Path: Received: from ts011p15.pop3b.netvision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27680; Mon, 9 Dec 1996 13:34:20 -0600 Received: (from hjstein@localhost) by blinky.bfr.co.il (8.7.6/8.6.9) id SAA02629; Mon, 9 Dec 1996 18:07:15 +0200 Date: Mon, 9 Dec 1996 18:07:15 +0200 Message-Id: <199612091607.SAA02629@blinky.bfr.co.il> From: "Harvey J. Stein" To: "Harvey J. Stein" Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: xlispstat seg faults when setting frame size of sequence-slider-dialogs. In-Reply-To: <199612091605.SAA02615@blinky.bfr.co.il> References: <199612091605.SAA02615@blinky.bfr.co.il> Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Harvey J. Stein writes: > > I can get xlispstat to seg fault in 3 commands: > > hjstein@blinky:~$ xlispstat > XLISP-PLUS version 2.1g > Portions Copyright (c) 1988, by David Betz. > Modified by Thomas Almy and others. > XLISP-STAT 2.1 Release 3.47 (Beta). > Copyright (c) 1989-1994, by Luke Tierney. > > > (setq p (sequence-slider-dialog '(-1 0 1) :action (lambda (s) (format t "hi\n")))) > # > > (send p :frame-size) > (210 116) > > (send p :frame-size 405 200) > Segmentation fault (core dumped) > > Any ideas? > Oops - I forgot to mention that this is under Linux, kernel 2.0.26, libc 5.2.18, libm 5.0.5, gcc 2.7.2. Thanks, -- Dr. Harvey J. Stein Berger Financial Research abel@netvision.net.il From majordom@umnstat.stat.umn.edu Mon Dec 9 15:04:36 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28059; Mon, 9 Dec 1996 15:04:36 -0500 Return-Path: Received: from halley.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28054; Mon, 9 Dec 1996 14:04:25 -0600 Received: from libiya.upf.es by halley.upf.es with SMTP (1.37.109.15/16.2) id AA262885701; Mon, 9 Dec 1996 21:08:21 GMT Received: by libiya.upf.es (1.38.193.4/15.6) id AA05150; Mon, 9 Dec 1996 21:04:37 GMT Date: Mon, 9 Dec 1996 21:04:37 +0000 ("WET) From: Frederic Udina To: "Harvey J. Stein" Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: xlispstat seg faults when setting frame size of sequence-slider-dialogs. In-Reply-To: <199612091605.SAA02615@blinky.bfr.co.il> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk me too, using a HP-UX box. My experience says that you must no try to modifiy a xls dialog window once it has been created. $ xlispstat XLISP-PLUS version 2.1g Portions Copyright (c) 1988, by David Betz. Modified by Thomas Almy and others. XLISP-STAT 2.1 Release 3.44 (Beta). Copyright (c) 1989-1994, by Luke Tierney. > (setq p (sequence-slider-dialog '(-1 0 1) :action (lambda (s) (format t "hi\n")))) # > (send p :frame-size) (210 116) > (send p :frame-size 405 200) Bus error (core dumped) Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Ramon Trias Fargas, 25-27 | Compuserve: 100555.603 08005 Barcelona | WWWeb: http://libiya.upf.es/ SPAIN | or http://www.econ.upf.es/~udina ___________________________________/_\_________________________________ From majordom@umnstat.stat.umn.edu Mon Dec 9 15:53:04 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28671; Mon, 9 Dec 1996 15:53:04 -0500 Return-Path: Received: from mailb.surrey.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28666; Mon, 9 Dec 1996 14:52:52 -0600 Received: from soc.surrey.ac.uk (actually host boris-gate.soc.surrey.ac.uk) by mailb.surrey.ac.uk with SMTP (PP); Mon, 9 Dec 1996 20:52:40 +0000 Received: from annex1-10.surrey.ac.uk by boris.soc.surrey.ac.uk via Ethernet with SMTP id aa15302; 9 Dec 96 20:52 GMT X-Sender: scs1ng@boris.soc.surrey.ac.uk Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 9 Dec 1996 20:51:27 +0000 To: Xlisp List From: Nigel Gilbert Subject: Spin-proto in Windows Lisp-stat Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Having developed a program using Release 3.47 on a Mac, I wanted to demonstrate it at a meeting where there is only a Windows 95 PC with a LCD tablet. However, when I tried to run the program using the Win32 version of Lisp-Stat, I find that the spin-proto code fails in graph-overlay-proto. Even evaluating (setq p (send spin-proto :new 3)) in a clean image at the top level fails. Has anyone tripped over this before and found the bug? I need an answer within the next few hours if I am to get the demo to work in time - otherwise it's back to transparencies of screenshots for the talk! Many thanks. ____________________________________________________________________ Prof G. Nigel Gilbert, Department of Sociology, University of Surrey Guildford GU2 5XH, UK. Tel: +44 1483 259173 Fax: +44 1483 306290 From majordom@nokomis.stat.umn.edu Mon Dec 9 16:08:40 1996 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA01697; Mon, 9 Dec 1996 16:08:40 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA01690; Mon, 9 Dec 1996 15:08:26 -0600 From: Luke Tierney Message-Id: <9612092108.AA01690@nokomis.stat.umn.edu> Subject: Re: xlispstat seg faults when setting frame size of sequence-slider-dialogs. To: abel@netvision.net.il (Harvey J. Stein) Date: Mon, 9 Dec 1996 15:08:26 -0600 (CST) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: <199612091605.SAA02615@blinky.bfr.co.il> from "Harvey J. Stein" at Dec 9, 96 06:05:40 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 979 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > > I can get xlispstat to seg fault in 3 commands: > > hjstein@blinky:~$ xlispstat > XLISP-PLUS version 2.1g > Portions Copyright (c) 1988, by David Betz. > Modified by Thomas Almy and others. > XLISP-STAT 2.1 Release 3.47 (Beta). > Copyright (c) 1989-1994, by Luke Tierney. > > > (setq p (sequence-slider-dialog '(-1 0 1) :action (lambda (s) (format t "hi\n")))) > # > > (send p :frame-size) > (210 116) > > (send p :frame-size 405 200) > Segmentation fault (core dumped) > > Any ideas? > > -- > Dr. Harvey J. Stein > Berger Financial Research > abel@netvision.net.il > In X11graph.c:StWSetSize change adjust_scroll_bars(gwinfo, TRUE); to if (gwinfo != NULL) adjust_scroll_bars(gwinfo, TRUE); That should cure the core dump. The end result probably won't be all that satisfying though -- the dialogs don't know how to adjust themselves to a new size. luke From majordom@umnstat.stat.umn.edu Mon Dec 9 17:56:46 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00340; Mon, 9 Dec 1996 17:56:46 -0500 Return-Path: Received: from atlas.socsci.umn.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00335; Mon, 9 Dec 1996 16:56:40 -0600 Received: from pc4.hinet.net ([202.39.153.124]) by atlas.socsci.umn.edu (8.7.5/8.7.3) with SMTP id QAA19224 for ; Mon, 9 Dec 1996 16:56:38 -0600 (CST) Message-Id: <30CA1357.7EF2@atlas.socsci.umn.edu> Date: Sun, 10 Dec 1995 06:53:11 +0800 From: harry Reply-To: harry@atlas.socsci.umn.edu Organization: gis X-Mailer: Mozilla 3.0 (Win95; I) Mime-Version: 1.0 To: Xlisp List Subject: WHere to get XLISP-STAT for LINUX Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, Netters. Is there any ftp site that I can download XLISP-STAT binary code for Linux? 5 years ago, when I was still a ph.d. student in MPLS, I download the source code from School of Statistics, UMN and compiled on a Linux at home. Now I wonder if there is any binary code for linux available that I can use on a Pentium running Linux 2.0? Many thanks in advance! Best, Harry -- Dr. Harry KC Chang Associate Prof. Land Management Dept. FCU harry@atlas.socsci.umn.edu SSRFC, UMN, MPLS, MN55455 (612)625-8556 kchang@cc-sun.fcu.edu.tw FCU Taichung Taiwan (04)381-5514 Homepage http://www.globalnet.com.tw/~kchang (FAX)04-381-6533 *********************************************************************** * Don't take life too seriously! You will never get out of it alive! * *********************************************************************** From majordom@umnstat.stat.umn.edu Tue Dec 10 23:26:15 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19167; Tue, 10 Dec 1996 23:26:15 -0500 Return-Path: Received: from super.super.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19162; Tue, 10 Dec 1996 22:26:08 -0600 Received: by super.super.org; id XAA11784; Tue, 10 Dec 1996 23:26:06 -0500 (EST) Received: from gotham.super.org(192.239.79.2) by super.super.org via smap (3.2) id xma011729; Tue, 10 Dec 96 23:25:53 -0500 Received: from hume.super.org (hume [192.239.79.11]) by gotham.super.org (8.6.12/8.6.12.1) with ESMTP id XAA17008 for ; Tue, 10 Dec 1996 23:25:53 -0500 From: pcolsen@super.org (Peter C Olsen) Received: (pcolsen@localhost) by hume.super.org (8.6.12/8.6.12.client) id XAA04135 for stat-lisp-news@umnstat.stat.umn.edu; Tue, 10 Dec 1996 23:25:52 -0500 Date: Tue, 10 Dec 1996 23:25:52 -0500 Message-Id: <199612110425.XAA04135@hume.super.org> To: stat-lisp-news@umnstat.stat.umn.edu Subject: Call for entries: 1997 Mathematical Contest in Modeling Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Call For Entries --- Mathematical Contest in Modeling 1997 ------------------------- The Basics ------------------------- The Mathematical Contest in Modeling (MCM), a national contest for college undergraduates, will hold its thirteenth competition in February, 1997. The MCM is designed to stimulate and improve problem-solving and writing skills in a team setting. Last year, over 393 teams submitted solution papers, representing 235 schools in nine countries. Outstanding solution papers will be published October, 1997, in a special issue of the "The UMAP Journal." Previous winners have included a broad range of colleges and outstanding secondary schools. Contest Date and Time: The 1997 contest will begin at 12:01 AM local time, Friday, February 7 and will officially end at 5:00 PM on Monday, February 10. Registration of Teams: Teams must be registered in advance. Each department may sign up either one team of three members or two teams of three members each. No more than four teams may be registered from any one institution. Team members may be changed at the last minute without notifying COMAP. If a department preregisters one team, a second team may be added, but a control number must be obtained from COMAP before a second team can be entered. To request a Registration Packet: By mail: write to COMAP Incorporated, Suite 210, 57 Bedford Street, Lexington, MA 02173-9684 USA By telephone: Call Clarice Callahan at 800-77-COMAP (800-77-6627) or (617)-862-7878 (ext 37) By FAX: (617)-863-1202 By email: primary: mcm@comap.com secondary: pcolsen@super.org REGISTRATION PACKETS MUST BE RECEIVED AT COMAP BY FRIDAY, 3 JANUARY, 1997. Further information can be found at the following URLs. Several of them contain sample problems from previous years. http://www.informs.org/Edu/MCM.html http://www.stetson.edu/~mathcs/modeling/ http://math.smith.edu/Local/guide/node83.html http://buzzard.ups.edu/mcm.html http://mickey.lcsc.edu/~sas/ns/news/mathteam Sponsorship: Major funding for the Mathematical Contest in Modeling is provided by the National Security Agency. Mathematics is of central importance to the mission of the National Security Agency, an agency of the Department of Defense. NSA proudly supports numerous mathematics education and research activities such as the MCM Additional support for this project is provided by: The Institute for Operations Research and Management Sciences, The Society for Industrial and Applied Mathematics, and The Mathematical Association of America. ------------------------------- More Details ----------------------- The Two Problems: Each team will choose one of two problems. Sample problems are available from the URLs listed above or from pcolsen@super.org. After the envelope containing the contest problem is opened, the team may not discuss any aspect of the problem with the advisor or anyone else. The competition intends to allow participants to use all information that is currently available, but to disqualify any teams that discuss the problems with people in a position to supply them with information reflecting experience or professional expertise. The relevant issue is one of intent; each team is expected to develop all of its substantive analysis without the help of others. Participants may use computers, software packages, libraries, or any other inanimate sources. The MCM packet will provide requirements and suggested procedures. Problems tend to be open-ended and are unlikely to have a unique solution. Attention must be paid to clarity, analysis, and design in attempting a solution. The narrative part of the solution papers must be typed and in English. Partial solutions are acceptable. Two copies of each paper must be submitted. Student teams are to complete all work in English by 5:00 PM on Monday, February 10, 1997. Faculty advisors must ensure that no alterations of any type are made after the 5:00 PM cutoff, and that papers are received at COMAP by Friday, February 14. Papers mailed from other than the continental United States follow the same procedures, but have until Monday, February 17 to arrive at COMAP. Foreign Entries: Students must submit their work in English. No work is to be done after 5:00 PM on Monday, February 10. In order to assure timely judging, no additional time is provided for translation apart from the contest. MCM Results: Judging will take place three weeks after the contest. The solutions will be recognized as Successful Participants, Honorable Mention, Meritorious, or Outstanding. Advisors and teams will be notified of the results in April, 1997. News releases will be prepared and there will be announcements in college and professional publications. Recognition: All successful participants will receive a certificate. Outstanding teams will receive bronze plaques and their solution papers will be published in The UMAP Journal. Prizes: The Institute for Operations Research and Management Sciences (INFORMS) will designate one Outstanding team for each problem as an INFORMS winner, and will award cash prizes to each team. In addition, INFORMS will give free memberships to all Meritorious teams. The Society for Industrial and Applied Mathematics (SIAM) will designate one Outstanding team for each problem as a SIAM winner. Each SIAM winning team will receive a certificate, a cash prize, and partial expenses for a trim to the SIAM annual meeting. From majordom@umnstat.stat.umn.edu Wed Dec 11 14:00:02 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA01864; Wed, 11 Dec 1996 14:00:02 -0500 Return-Path: Received: from mcmail.CIS.McMaster.CA by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA01853; Wed, 11 Dec 1996 12:59:55 -0600 Received: from johnfox (larry05.slip.yorku.ca [130.63.122.219]) by mcmail.CIS.McMaster.CA (8.7.1/8.7.1) with SMTP id NAA29806 for ; Wed, 11 Dec 1996 13:59:48 -0500 (EST) Message-Id: <1.5.4.32.19961211185830.0068dcc8@mcmail.cis.mcmaster.ca> X-Sender: jfox@mcmail.cis.mcmaster.ca X-Mailer: Windows Eudora Light Version 1.5.4 (32) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 11 Dec 1996 13:58:30 -0500 To: Lisp-Stat list From: John Fox Subject: NAN and tests for equality Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I've been working with a dataset in which missing data are encoded as not-a-number and encountered what seems to me strange behaviour for tests of equality: > (equal not-a-number 9.1) T > (equal not-a-number 9.) NIL > (equal not-a-number 9) NIL > (equalp not-a-number 9.1) T > (eql not-a-number 9.1) T > (eq not-a-number 9.1) NIL > (= not-a-number 9.1) T That is, all of the equality predicates seem to distinguish between NAN and integers (including integer reals), but only eq distinguishes between NAN and a real number. Although I can use eq to get what I want, I found all this pretty confusing. Does this behaviour conform to the common lisp standard? John |----------------------------------------------------| | John Fox jfox@McMaster.ca | | Department of Sociology McMaster University | |----------------------------------------------------| From majordom@umnstat.stat.umn.edu Fri Dec 13 16:08:45 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05705; Fri, 13 Dec 1996 16:08:45 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05700; Fri, 13 Dec 1996 15:08:40 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id NAA29238 for stat-lisp-news@umnstat.stat.umn.edu; Fri, 13 Dec 1996 13:08:36 -0800 (PST) From: Jan Deleeuw Message-Id: <199612132108.NAA29238@galton.math.ucla.edu> Subject: JSS, volume I, issue III To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Fri, 13 Dec 1996 13:08:34 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk XLISP-Stat tools for building Generalised Estimating Equation models by Thomas Lumley. Find it at http://www.stat.ucla.edu/journals/jss/v01/i03 -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon Dec 16 08:17:22 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16475; Mon, 16 Dec 1996 08:17:22 -0500 Return-Path: Received: from [134.94.80.3] by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16470; Mon, 16 Dec 1996 07:16:53 -0600 Received: from hlrz31-f by kfa-juelich.de (5.65v3.2/DEC-AXP) id AA19230; Mon, 16 Dec 1996 14:16:39 +0100 Received: from hlrz24.hlrz.kfa-juelich.de by hlrz31.hlrz.kfa-juelich.de (SMI-8.6/SMI-4.1) id OAA08496; Mon, 16 Dec 1996 14:16:37 +0100 Reply-To: jmu@hlrz31.hlrz.kfa-juelich.de Received: by hlrz24.hlrz.kfa-juelich.de (SMI-8.6/SMI-4.1) id OAA28343; Mon, 16 Dec 1996 14:16:35 +0100 Date: Mon, 16 Dec 1996 14:16:35 +0100 Message-Id: <199612161316.OAA28343@hlrz24.hlrz.kfa-juelich.de> From: Johannes Mueller-Gerking To: stat-lisp-news@umnstat.stat.umn.edu Subject: Q: XLS and SAPA Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi all, I would like to ask for some hints and advice concerning the use of the SAPA package within XLISP-STAT (3-44): - Compiling: I got everything to compile without problems. However, when loading the compiled files, XLS complains about "multf not defined" (or similar, I forgot). The "define-modify-macro" function used in utilities.lsp to define the multf macro does not exist in XLS, but this doesn't trigger even a warning by the compiler. Replacing the (define-modify-macro call by a (defmacro similar to incf in common.lsp solves this problem. Q: why does the compiler not complain about "define-modify-macro"? (Maybe beginner's LISP question...) - The compiled code does not run. First there are name conflicts: > (in-package :CL-USER) # > (use-package :SAPA) Name conflict of SAPA:ACVS and ACVS when using # in # Name conflict of SAPA:INTERQUARTILE-RANGE and INTERQUARTILE-RANGE when using # in # Name conflict of SAPA:DIFFERENCE and DIFFERENCE when using # in # Name conflict of SAPA:QUANTILE-PLOT and QUANTILE-PLOT when using # in # Error: name conflicts Ok, SAPA is designed for CL systems and needs to defun a lot of things XLS already has. Q: Is there a "downsized" SAPA package for XLS with the functionally doubly defined code replaced by calls to XLS functions? Q: ACVS seems to be defined in USER, but no help is available, the function ACVS is unboud, the variable ACVS is unbound. What is USER::ACVS? Second problem with the compiled code: if I call, say (sapa:acvs some-vector), I get Error: too few arguments ??? there is only one required argument to acvs ??? - Without compiler: The name conflicts remain. Didn't expect otherwise. (sapa:acvs some-vector) seems to work correctly. Two "Warning: special declarations are not supported", but nothing else. Q: What can go wrong with the unsupported special declaration? Could simply ignoring the warning introduce spurious results? General question: could you recommand going "alpha", i.e. upgrading to 3-47? (I'm on SPARC Solaris 2.5) Could upgrading help in the problems above? Thank's a lot for any hints! Johannes From majordom@umnstat.stat.umn.edu Mon Dec 16 12:52:30 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA18214; Mon, 16 Dec 1996 12:52:30 -0500 Return-Path: Received: from maya.medctr.luc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA18209; Mon, 16 Dec 1996 11:52:23 -0600 Received: (from ael@localhost) by maya.medctr.luc.edu (8.7.3/8.6.9) id LAA01536 for stat-lisp-news@stat.umn.edu; Mon, 16 Dec 1996 11:49:36 -0600 (CST) Date: Mon, 16 Dec 1996 11:49:36 -0600 (CST) From: Andy Long Message-Id: <199612161749.LAA01536@maya.medctr.luc.edu> Content-Type: text Apparently-To: Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk stat-lisp-news@stat.umn.edu Is there any way to find out the actual name of (or some information about) the subroutine which has been "violated" in the following type of error? ... Iteration 2. Error: illegal zero argument Happened in: # .... Thanks, Andy Long ael@maya.medctr.luc.edu From majordom@umnstat.stat.umn.edu Wed Dec 18 19:39:47 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19964; Wed, 18 Dec 1996 19:39:47 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19959; Wed, 18 Dec 1996 18:39:38 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id QAA21303 for stat-lisp-news@umnstat.stat.umn.edu; Wed, 18 Dec 1996 16:39:29 -0800 (PST) From: Jan Deleeuw Message-Id: <199612190039.QAA21303@galton.math.ucla.edu> Subject: CW 10 To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Wed, 18 Dec 1996 16:39:28 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk See http://www.stat.ucla.edu/archive/xlispstat/systems/macintosh/cw10/ This has two files -rw-r--r-- 1 deleeuw faculty 1029434 Dec 18 16:28 my-xlisp-stat.sit.hqx -rw-r--r-- 1 deleeuw faculty 1808240 Dec 18 16:22 xlisp-mac.sit.hqx The largest one is a CodeWarrior 10 project file for building Xlisp-Stat, together with all the necessary code (some minor changes were necessary since Apple mucked with the universal headers again). The project file is set for PowerPC, 604 instruction scheduling, speed optimized at level 4. Minimum heap at 5MB, preferred heap at 10MB, stack at 256K. But all that is easily adjusted. All the code is in one large segment, which you would have to change for a 68K version. The smallest one has the Xlisp-Stat binary for the Mac from this project, plus a workspace file with compiler and help preloaded, plus a statinit.lsp which automatically loads inspector.fsl and compile-menu.fsl (which I like a lot), plus a preferences file with some set-ups. I have had no bombs with this one, I am using this as my personal version. It is Xlisp-Stat 3.47, by the way. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Sun Dec 22 11:38:44 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10167; Sun, 22 Dec 1996 11:38:44 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10162; Sun, 22 Dec 1996 10:38:39 -0600 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.6.4.3/10.1) id LAA27635; Sun, 22 Dec 1996 11:38:36 -0500 X-Sender: uluru@gibbs.oit.unc.edu (Unverified) Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sun, 22 Dec 1996 11:39:23 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: Unix Fonts Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi LispStaters: I am trying to figure out how to control the size of the font used in graphics windows under X11. What I know is: 1) In appendix B of the Lisp-Stat book (p 365) Luke says: "You can specify alternate fonts ... by using the X11 resource management facilities. Resources controling appearance ... (include) xlisp*graph*font" 2) In "XLISP-STAT 2.1 Release 3 Beta Release Notes", dated September 19, 1994 (p. 5) Luke also says: "Upon startup of the X11 version ... the system merges resources found in file /usr/lib/X11/app-defaults/Xlisp then ones found in the file Xlisp in the directory named by the environment variable XAPPLRESDIR, and finally the resources of the display. Resources can thus be specified in all three places, with the display's resources having highest precedence." So, from this I assume that I would specify the xlisp*graph*font resource in one of those three place (probably in the resources for the display). What I don't know is how to do this so that font size comes out as I wish. Can anyone help? Thanks ... ----- Forrest Young -- at home Professor, Psychometrics CB# 3270 Davie Hall email: mailto://forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software sites: http://forrest.psych.unc.edu/research/ViSta.html USA http://www.psych.unc.edu/fwy/research/ViSta.html USA http://sunsite.univie.ac.at/ViSta/ViSta.html Austria http://micros.hensa.ac.uk/mirrors/vista/ViSta.html Great Britain From majordom@umnstat.stat.umn.edu Mon Dec 23 13:17:50 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23031; Mon, 23 Dec 1996 13:17:50 -0500 Return-Path: Received: from rgborafsa.er.usgs.gov by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23026; Mon, 23 Dec 1996 12:17:44 -0600 Received: from [130.11.54.131] (mcmac.er.usgs.gov [130.11.54.131]) by rgborafsa.er.usgs.gov (Geomail 1.2.3) with SMTP id NAA03257 for ; Mon, 23 Dec 1996 13:18:20 -0500 Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 23 Dec 1996 13:19:26 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: "Richard B McCammon Reston, VA " (Richard McCammon (USGS)) Subject: Plot objects Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk question: Once I have created a plot object using Plot..., how can I recover it or assign it a variable name if, for instance, I want to plot the y-axis on the histogram I have created. Thanks __________________________________________________________ snail mail: Richard B. McCammon Eastern Minerals Team U.S. Geological Survey MS-954 12201 Sunrise Valley Drive Reston, VA 20192 USA e-mail: mccammon@rgborafsa.er.usgs.gov or mccammon@130.11.54.100 Phone: 703-648-6150 Fax: 703-648-6383 From majordom@umnstat.stat.umn.edu Thu Dec 26 19:54:14 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27826; Thu, 26 Dec 1996 19:54:14 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27821; Thu, 26 Dec 1996 18:54:10 -0600 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.6.4.3/10.1) id TAA10176; Thu, 26 Dec 1996 19:54:07 -0500 X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 26 Dec 1996 19:55:06 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: Unix Fonts Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I send this message 4 days ago and still havn't seen it appear in the news group. Is there a problem? Did I fall off the group? I'll try again. >Date: Sun, 22 Dec 1996 11:39:24 -0500 >To:Lisp-Stat >From:forrest@unc.edu (Forrest Young) >Subject:Unix Fonts > >Hi LispStaters: > >I am trying to figure out how to control the size of the font used in >graphics windows under X11. > >What I know is: >1) In appendix B of the Lisp-Stat book (p 365) Luke says: "You can specify >alternate fonts ... by using the X11 resource management facilities. >Resources controling appearance ... (include) xlisp*graph*font" >2) In "XLISP-STAT 2.1 Release 3 Beta Release Notes", dated September 19, >1994 (p. 5) Luke also says: "Upon startup of the X11 version ... the >system merges resources found in file /usr/lib/X11/app-defaults/Xlisp then >ones found in the file Xlisp in the directory named by the environment >variable XAPPLRESDIR, and finally the resources of the display. Resources >can thus be specified in all three places, with the display's resources >having highest precedence." > >So, from this I assume that I would specify the xlisp*graph*font resource >in one of those three place (probably in the resources for the display). >What I don't know is how to do this so that font size comes out as I wish. > >Can anyone help? Thanks ... > > ----- Forrest Young -- at home Professor, Psychometrics CB# 3270 Davie Hall email: mailto://forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software sites: http://forrest.psych.unc.edu/research/ViSta.html USA http://www.psych.unc.edu/fwy/research/ViSta.html USA http://sunsite.univie.ac.at/ViSta/ViSta.html Austria http://micros.hensa.ac.uk/mirrors/vista/ViSta.html Great Britain From majordom@umnstat.stat.umn.edu Sun Dec 29 15:45:19 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02895; Sun, 29 Dec 1996 15:45:19 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02890; Sun, 29 Dec 1996 14:45:14 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id MAA20067; Sun, 29 Dec 1996 12:45:08 -0800 (PST) From: Jan Deleeuw Message-Id: <199612292045.MAA20067@galton.math.ucla.edu> Subject: Archive To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Sun, 29 Dec 1996 12:45:07 -0800 (PST) Cc: ael@maui.medctr.luc.edu, naras@Playfair.Stanford.EDU (Balasubramanian Narasimhan) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I added http://www.stat.ucla.edu/archive/xlispstat/code/mathematics/teach/ to the archive, with the calculus modules by Andy Long, and also http://www.stat.ucla.edu/archive/xlispstat/systems/linux/ with the linux binary of 3-44 by Naras. By the way, the Mac version of 3-47 in http://www.stat.ucla.edu/archive/xlispstat/systems/macintosh/cw10/ is up to 20% faster than the official one, and it seems pretty stable (on my 9500/200 (it uses 604 instruction scheduling). -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon Dec 30 11:26:12 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15171; Mon, 30 Dec 1996 11:26:12 -0500 Return-Path: Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15165; Mon, 30 Dec 1996 10:26:09 -0600 From: Robert J. Granvin Message-Id: <9612301626.AA15165@umnstat.stat.umn.edu> Subject: Please ignore (test) To: stat-lisp-news@umnstat.stat.umn.edu Date: Mon, 30 Dec 1996 10:26:08 -0600 (CST) Reply-To: rjg@umnstat.stat.umn.edu X-Favorite-Animal-Sound: Moo X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 380 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Testing the distribution of stat-lisp-news. Please ignore this message. -- \\ Robert J. "Una-Bobber" Granvin rjg@stat.umn.edu // School of Statistics - University of Minnesota ICBM: 44.924 -93.212 http://www.stat.umn.edu/~rjg The graveyards are full of indispensable men. --DeGaulle From majordom@umnstat.stat.umn.edu Mon Dec 30 11:38:26 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15295; Mon, 30 Dec 1996 11:38:26 -0500 Return-Path: Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15289; Mon, 30 Dec 1996 10:38:23 -0600 From: Robert J. Granvin Message-Id: <9612301638.AA15289@umnstat.stat.umn.edu> Subject: stat-lisp-news glitch To: stat-lisp-news@umnstat.stat.umn.edu Date: Mon, 30 Dec 1996 10:38:23 -0600 (CST) Reply-To: rjg@umnstat.stat.umn.edu X-Favorite-Animal-Sound: Moo X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 916 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Dunno about this one, but the distribution for stat-lisp-news got interrupted without any symptoms. Now fixed. Since that time, four messages have been received but may not have been distributed to the list. The next four messages will be copies from the archives. If you've seen them already, my apologies of sending them again. On the other hand, if you do respond to them, please note the email address as the forwarding won't preserve them for automatic replies if you email directly back to the sender. Thanks for the report of a potential problem. Got it fixed quickly once it was spotted! /bob -- \\ Robert J. "Una-Bobber" Granvin rjg@stat.umn.edu // School of Statistics - University of Minnesota ICBM: 44.924 -93.212 http://www.stat.umn.edu/~rjg The graveyards are full of indispensable men. --DeGaulle From majordom@umnstat.stat.umn.edu Mon Dec 30 11:39:42 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15313; Mon, 30 Dec 1996 11:39:42 -0500 Return-Path: Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15307; Mon, 30 Dec 1996 10:39:39 -0600 From: Robert J. Granvin Message-Id: <9612301639.AA15307@umnstat.stat.umn.edu> Subject: CW 10 [RE-SENT] To: stat-lisp-news@umnstat.stat.umn.edu Date: Mon, 30 Dec 1996 10:39:39 -0600 (CST) Reply-To: rjg@umnstat.stat.umn.edu X-Favorite-Animal-Sound: Moo X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1509 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Forwarded message: From: Jan Deleeuw Subject: CW 10 Date: Wed, 18 Dec 1996 16:39:28 -0800 (PST) See http://www.stat.ucla.edu/archive/xlispstat/systems/macintosh/cw10/ This has two files -rw-r--r-- 1 deleeuw faculty 1029434 Dec 18 16:28 my-xlisp-stat.sit.hqx -rw-r--r-- 1 deleeuw faculty 1808240 Dec 18 16:22 xlisp-mac.sit.hqx The largest one is a CodeWarrior 10 project file for building Xlisp-Stat, together with all the necessary code (some minor changes were necessary since Apple mucked with the universal headers again). The project file is set for PowerPC, 604 instruction scheduling, speed optimized at level 4. Minimum heap at 5MB, preferred heap at 10MB, stack at 256K. But all that is easily adjusted. All the code is in one large segment, which you would have to change for a 68K version. The smallest one has the Xlisp-Stat binary for the Mac from this project, plus a workspace file with compiler and help preloaded, plus a statinit.lsp which automatically loads inspector.fsl and compile-menu.fsl (which I like a lot), plus a preferences file with some set-ups. I have had no bombs with this one, I am using this as my personal version. It is Xlisp-Stat 3.47, by the way. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon Dec 30 11:41:58 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15342; Mon, 30 Dec 1996 11:41:58 -0500 Return-Path: Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15336; Mon, 30 Dec 1996 10:41:55 -0600 From: Robert J. Granvin Message-Id: <9612301641.AA15336@umnstat.stat.umn.edu> Subject: Unix Fonts [RE-SENT] To: stat-lisp-news@umnstat.stat.umn.edu Date: Mon, 30 Dec 1996 10:41:55 -0600 (CST) Reply-To: rjg@umnstat.stat.umn.edu X-Favorite-Animal-Sound: Moo X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1680 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Forwarded message: Date: Sun, 22 Dec 1996 11:39:23 -0500 From: forrest@unc.edu (Forrest Young) Subject: Unix Fonts Hi LispStaters: I am trying to figure out how to control the size of the font used in graphics windows under X11. What I know is: 1) In appendix B of the Lisp-Stat book (p 365) Luke says: "You can specify alternate fonts ... by using the X11 resource management facilities. Resources controling appearance ... (include) xlisp*graph*font" 2) In "XLISP-STAT 2.1 Release 3 Beta Release Notes", dated September 19, 1994 (p. 5) Luke also says: "Upon startup of the X11 version ... the system merges resources found in file /usr/lib/X11/app-defaults/Xlisp then ones found in the file Xlisp in the directory named by the environment variable XAPPLRESDIR, and finally the resources of the display. Resources can thus be specified in all three places, with the display's resources having highest precedence." So, from this I assume that I would specify the xlisp*graph*font resource in one of those three place (probably in the resources for the display). What I don't know is how to do this so that font size comes out as I wish. Can anyone help? Thanks ... ----- Forrest Young -- at home Professor, Psychometrics CB# 3270 Davie Hall email: mailto://forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software sites: http://forrest.psych.unc.edu/research/ViSta.html USA http://www.psych.unc.edu/fwy/research/ViSta.html USA http://sunsite.univie.ac.at/ViSta/ViSta.html Austria http://micros.hensa.ac.uk/mirrors/vista/ViSta.html Great Britain From majordom@umnstat.stat.umn.edu Mon Dec 30 11:43:03 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15379; Mon, 30 Dec 1996 11:43:03 -0500 Return-Path: Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15373; Mon, 30 Dec 1996 10:43:01 -0600 From: Robert J. Granvin Message-Id: <9612301643.AA15373@umnstat.stat.umn.edu> Subject: Plot objects [RE-SENT] To: stat-lisp-news@umnstat.stat.umn.edu Date: Mon, 30 Dec 1996 10:43:00 -0600 (CST) Reply-To: rjg@umnstat.stat.umn.edu X-Favorite-Animal-Sound: Moo X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 657 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Forwarded message: Date: Mon, 23 Dec 1996 13:19:26 -0500 From: "Richard B McCammon Reston, VA " (Richard McCammon (USGS)) Subject: Plot objects question: Once I have created a plot object using Plot..., how can I recover it or assign it a variable name if, for instance, I want to plot the y-axis on the histogram I have created. Thanks __________________________________________________________ snail mail: Richard B. McCammon Eastern Minerals Team U.S. Geological Survey MS-954 12201 Sunrise Valley Drive Reston, VA 20192 USA e-mail: mccammon@rgborafsa.er.usgs.gov or mccammon@130.11.54.100 Phone: 703-648-6150 Fax: 703-648-6383 From majordom@umnstat.stat.umn.edu Mon Dec 30 11:43:46 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15404; Mon, 30 Dec 1996 11:43:46 -0500 Return-Path: Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15398; Mon, 30 Dec 1996 10:43:44 -0600 From: Robert J. Granvin Message-Id: <9612301643.AA15398@umnstat.stat.umn.edu> Subject: Archive [RE-SENT] To: stat-lisp-news@umnstat.stat.umn.edu Date: Mon, 30 Dec 1996 10:43:43 -0600 (CST) Reply-To: rjg@umnstat.stat.umn.edu X-Favorite-Animal-Sound: Moo X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 879 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Forwarded message: From: Jan Deleeuw Subject: Archive Date: Sun, 29 Dec 1996 12:45:07 -0800 (PST) I added http://www.stat.ucla.edu/archive/xlispstat/code/mathematics/teach/ to the archive, with the calculus modules by Andy Long, and also http://www.stat.ucla.edu/archive/xlispstat/systems/linux/ with the linux binary of 3-44 by Naras. By the way, the Mac version of 3-47 in http://www.stat.ucla.edu/archive/xlispstat/systems/macintosh/cw10/ is up to 20% faster than the official one, and it seems pretty stable (on my 9500/200 (it uses 604 instruction scheduling). -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon Dec 30 14:13:43 1996 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16251; Mon, 30 Dec 1996 14:13:43 -0500 Return-Path: Received: from nessie.mcc.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16240; Mon, 30 Dec 1996 13:13:36 -0600 Received: from Tedistan.annex.umist.ac.uk by nessie.mcc.ac.uk with SMTP (PP); Mon, 30 Dec 1996 19:13:27 +0000 Message-Id: From: Ted.Harding@nessie.mcc.ac.uk (Ted Harding) Subject: Re: Unix Fonts [RE-SENT] To: stat-lisp-news@umnstat.stat.umn.edu (Stat-Lisp News) Date: Mon, 30 Dec 1996 19:02:47 +0000 (GMT) Cc: forrest@unc.edu X-Mailer: ELM [version 2.4 PL23] Content-Type: text Content-Length: 3573 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk > Forwarded message: > Date: Sun, 22 Dec 1996 11:39:23 -0500 > From: forrest@unc.edu (Forrest Young) > Subject: Unix Fonts > > Hi LispStaters: > > I am trying to figure out how to control the size of the font used in > graphics windows under X11. > > What I know is: > 1) In appendix B of the Lisp-Stat book (p 365) Luke says: "You can specify > alternate fonts ... by using the X11 resource management facilities. > Resources controling appearance ... (include) xlisp*graph*font" > 2) In "XLISP-STAT 2.1 Release 3 Beta Release Notes", dated September 19, > 1994 (p. 5) Luke also says: "Upon startup of the X11 version ... the system > merges resources found in file /usr/lib/X11/app-defaults/Xlisp then ones > found in the file Xlisp in the directory named by the environment variable > XAPPLRESDIR, and finally the resources of the display. Resources can thus > be specified in all three places, with the display's resources having > highest precedence." > > So, from this I assume that I would specify the xlisp*graph*font resource > in one of those three place (probably in the resources for the display). > What I don't know is how to do this so that font size comes out as I wish. > > Can anyone help? Thanks ... > > Forrest Young -- at home I don't know how you would change the fonts xlispstat uses "on the fly"; maybe someone else can suggest the way? You can choose the font you want for the xterm in which xlispstat is running in the usual way for xterm by starting a new xterm with that font. For instance, xterm -fn 9x15bold -e xlispstst & xterm -fn 12x24 -e xlispstst & give two different fonts. The font names used here ("9x15bold", "12x24") are among those found in /usr/X11/lib/fonts/misc/fonts.dir on my system; other font directories under .../fonts offer a range of other possibilities. However, the font set in this way only applies to the window in which xlispstat is running (its "system window", to which it sends its messages) and not to the windows started up by xlispstat itself (e.g. if you run "tourdemo.lsp" from the Examples directory: xterm -fn 12x24 -e xlispstst tourdemo.lsp & the primary xterm will be in 12x24 but the font shown in the plot window will be the same, "9x15", as it always is, regardless of the xterm font). The xlisp binary contains a number of font-related strings: XCreateFontCursor XUnloadFont XLoadQueryFont XSetFont xlisp.graph.font xlisp.font fixed xlisp.menu.font 9x15 xlisp.dialog.font and maybe there are xlisp[stat] commands which can manipulate these. The README file has a few (but barely sufficient) relevant words: ** X11 FEATURES/BUGS ** ** Under X11 where plots and modeless dialogs show up on your screen, and ** how large they end up being, is up to the window manager. Requests you ** make may be honored under one manager but not under another. ** ** Several options can be set using the X11 resource management ** facilities. Options controlling appearance are ** ** xlisp*menu*titles: on for a title on a menu, off otherwise ** xlisp*menu*font: ** xlisp*dialog*font: ** xlisp*graph*font: and, indeed, if you put into /usr/lib/X11/app-defaults/Xlisp a line which says xlisp*graph*font: 12x24 (using a TAB for the spacing) you will see your graphs annotated in the 12x24 font. However, this is global and for all runs: I am still wondering how you change this on-the-fly for (say) different graphs in the same run. Happy New Year to all, Ted. (Ted.Harding@nessie.mcc.ac.uk) From majordom@umnstat.stat.umn.edu Fri Jan 3 19:23:35 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11245; Fri, 3 Jan 1997 19:23:35 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11240; Fri, 3 Jan 1997 18:23:30 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa11522; 3 Jan 97 19:23 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA31415; Fri, 3 Jan 1997 19:23:26 -0500 Date: Fri, 3 Jan 1997 19:23:25 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: Macintosh XlispStat: tracking working-directory Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The following version of chdir is a slight modification of the one in the file dirent.c in the mac xlispstat3.47 distribution. This version will set the low memory global to allow ( if you have the right option set in the General Control panel ) the get|set file-dialog starting directory to track the directory set by (set-working-directory path). I read thru the Inside Mac doc's and Mac Programmer's FAQ once more and discovered that a more complicated hack is only supposed to be required in the case that you want the directory to be manually settable no matter what the control panel settings. The simple version using LMSetSFSaveDisk and LMSetCurDirStore *should* work if the control panel allows application to set the directory. However, this only seems to work if I set the directory and not the volume. If you uncomment the line: /* LMSetSFSaveDisk( -vRefNum ); /* sdm7g */ it ceases to work. -- I'm a bit puzzled by this behaviour! [ I have problems using set-working-directory across volumes, but there were problems with that before -- I don't think it's any different. ] ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. #include /* sdm7g */ int chdir(char *path) { DIR *d; short vRefNum; long dirID; WDPBRec pb; /* Open the directory */ d = opendir(path); if (d == nil) return -1; /* Get the Mac FS identification for this directory */ vRefNum = d->dd_volume; dirID = d->dd_fd; /* Close the directory */ closedir(d); /* CD to the new directory */ pb.ioCompletion = nil; pb.ioNamePtr = nil; pb.ioVRefNum = vRefNum; pb.ioWDDirID = dirID; dd_errno = PBHSetVol(&pb, false); /* LMSetSFSaveDisk( -vRefNum ); /* sdm7g */ LMSetCurDirStore( dirID ); /* sdm7g */ return dd_errno?-1:0; } From majordom@umnstat.stat.umn.edu Mon Jan 6 13:09:12 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25648; Mon, 6 Jan 1997 13:09:12 -0500 Return-Path: Received: from ns.research.att.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25643; Mon, 6 Jan 1997 12:09:09 -0600 Received: from research.att.com by ns; Mon Jan 6 13:08:02 EST 1997 Received: from ulysses.att.com by research; Mon Jan 6 13:05:55 EST 1997 Received: by hera; Mon Jan 6 12:06 EST 1997 Message-Id: <32D13170.3D@ulysses.att.com> Date: Mon, 06 Jan 1997 12:08:00 -0500 From: Walt Paczkowski X-Mailer: Mozilla 3.0 (Win95; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: Formating Output Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I seem to recall seeing a message about formating floating point numbers using *float-format*. The number of digits to the right of the decimal can be specified. Does anyone recall or know how to do this? Walt Paczkowski AT&T Labs wrp@ulysses.att.com From majordom@umnstat.stat.umn.edu Wed Jan 8 13:41:40 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27018; Wed, 8 Jan 1997 13:41:40 -0500 Return-Path: Received: from mars.itc.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27013; Wed, 8 Jan 1997 12:41:36 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa01303; 8 Jan 97 13:41 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA16569; Wed, 8 Jan 1997 13:41:33 -0500 Newsgroups: comp.lang.lisp,comp.lang.lisp.x Date: Wed, 8 Jan 1997 13:41:28 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: Package Madness! Message-Id: Followup-To: comp.lang.lisp Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I had quite a few problems mastering the intricacies of Lisp Packages -- particularly when they involve using packages to redefine and shadow existing functions, until I discovered ( in CLtL1 ) the mnemonic "Put in seven extremely random interface commands" - which corresponds to the sequence: Provide IN-package Shadow EXport Require USE-package Import Commands==Contents of package/module Which is the order you (mostly) have to use those statements to be able to Require your package without tripping over any name clashes. ( Actually, I've been sticking "Provide" at the *end* of my packages so that if it is provided, that indicates that it did load successfully. That shouldn't make any difference with a well tested module, but it helps with something that I'm interactively editing and testing. If PROVIDE executes and *then* the module bombs out, I have to manually LOAD the module before reloading the module which REQUIREd it. ) That technique has mostly worked for me, however, I have still seen some dependencies on the order in which my modules are loaded. I ignored or worked around most of these glitches until I was trying to compile some of the more finished modules and save them into my workspace. I found that some of the name clashes had to do with using a symbol in a non defining position, such as an argument list. These problems were with XlispStat 3.47 ( for the Macintosh ). Since Xlisp is not quite Common Lisp -- it includes a lot of functions for Common Lisp compatibility, but things like packages and symbol binding are where I would suspect the most compatibility problems -- I also tried simulating a similar case in KCL ( the particular code where I noticed the problem had too many XlispStat and intermodule dependencies for me to easily load that code ), and I get a similar error. The two sessions below both generate errors from a name conflict on the symbol 'SEQ. In both, SEQ is defined and exported in package "X". In the first case, #'SEQ is used in a definition before the ( USE-PACKAGE "X" ) statement. In the second, SEQ is merely used in the argument list in defining another function. This is basically the same cases that are generating my errors in XlispStat, although the exact error message is slight different. It's basically the same error. On both systems, ( FIND-SYMBOL "SEQ" ) returns NIL before the use of the symbol in the USER package, and returns the values SEQ, :INTERNAL after it's use. ( The only significant difference I've noticed is that in some cases, the symbol XlispStat complains about is :SEQ -- presumably in the keyword package, even thought the string ":SEQ" is never used in my code. ) * Am I doing something wrong here ? * Is this error normal Lisp behavior for other implementations or is this behavior according to Common Lisp standard ? [ And if it *is* normal or standard, isn't it terribly inconvenient? ( I can swallow the first case, but the second being an error is pretty extreme! ) I have been finding Lisp packages a horribly intricate maze compared to the simple module system in Python. ( I haven't yet tried to do anything complex in Java, so I can't contrast its packages. ) ] $ kcl AKCL (Austin Kyoto Common Lisp) Version(1.505) Sat Dec 22 02:25:20 EST 1990 Contains Enhancements by W. Schelter Changes in version 1-455 definitely require recompilation of user files. >( defun a () ( seq a )) A >( setf a ( list 1 2 3 4 )) (1 2 3 4) >( make-package "X" ) #<"X" package> >( in-package "X" ) #<"X" package> X>( EXPORT '( SEQ )) T X>( defun seq ( &rest args ) args ) SEQ X>( in-package "USER" ) #<"USER" package> >( use-package "X" ) Error: Cannot use #<"X" package> from #<"USER" package>, because X:SEQ and SEQ will cause a name conflict. Error signalled by USE-PACKAGE. Broken at USE-PACKAGE. Type :H for Help. >> AKCL (Austin Kyoto Common Lisp) Version(1.505) Sat Dec 22 02:25:20 EST 1990 Contains Enhancements by W. Schelter Changes in version 1-455 definitely require recompilation of user files. >( defun MOST ( &rest seq ) ( cdr seq )) MOST >( make-package "X" ) #<"X" package> >( in-package "X" ) #<"X" package> X>( export '( seq )) T X>( defun seq ( &rest args ) args ) SEQ X>( in-package "USER" ) #<"USER" package> >( use-package "X" ) Error: Cannot use #<"X" package> from #<"USER" package>, because X:SEQ and SEQ will cause a name conflict. Error signalled by USE-PACKAGE. Broken at USE-PACKAGE. Type :H for Help. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Wed Jan 8 16:05:32 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28482; Wed, 8 Jan 1997 16:05:32 -0500 Return-Path: Received: from ZIMA.BBN.COM by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28477; Wed, 8 Jan 1997 15:05:28 -0600 Received: from [128.33.229.246] ([128.33.229.246]) by zima.bbn.com (8.7.3/8.7.3) with SMTP id QAA14507; Wed, 8 Jan 1997 16:04:59 -0500 (EST) X-Sender: kanderso@zima.bbn.com Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 8 Jan 1997 16:09:06 -0500 To: "Steven D. Majewski" From: KAnderso@bbn.com (Ken Anderson) Subject: Re: Package Madness! Cc: Xlisp List Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk At 13:41 1/8/97, Steven D. Majewski wrote: >I had quite a few problems mastering the intricacies of Lisp >Packages -- particularly when they involve using packages to >... > * Am I doing something wrong here ? > > * Is this error normal Lisp behavior for other implementations > or is this behavior according to Common Lisp standard ? Yes, this is normal. > [ And if it *is* normal or standard, isn't it terribly inconvenient? > ( I can swallow the first case, but the second being an error > is pretty extreme! ) > I have been finding Lisp packages a horribly intricate maze > compared to the simple module system in Python. ( I haven't > yet tried to do anything complex in Java, so I can't contrast > its packages. ) ] I've added some comments to explain what is happening. > >$ kcl >AKCL (Austin Kyoto Common Lisp) Version(1.505) Sat Dec 22 02:25:20 EST 1990 >Contains Enhancements by W. Schelter >Changes in version 1-455 definitely require recompilation of user files. > >>( defun a () ( seq a )) >A ;;; Here the USER package has 2 symbols, USER::A and USER::SEQ. The Lisp READ function creates these. >>( setf a ( list 1 2 3 4 )) >(1 2 3 4) >>( make-package "X" ) >#<"X" package> >>( in-package "X" ) >#<"X" package> >X>( EXPORT '( SEQ )) ;;; Here the X package has a symbol, X:SEQ >T >X>( defun seq ( &rest args ) args ) >SEQ ;;; Here the reader creates X::ARGS >X>( in-package "USER" ) >#<"USER" package> >>( use-package "X" ) > >Error: Cannot use #<"X" package> > from #<"USER" package>, > because X:SEQ and SEQ will cause > a name conflict. >Error signalled by USE-PACKAGE. When the USER package uses the package X, all the exported symbols, only X::SEQ here, are visible to the user package without needing an explicit package prefix. Since there is already a USER::SEQ symbol, the unprefixed symbol, SEQ, is ambiguous. You should first define the X package and then the packages that will use it. Something like: (make-package "X") (export 'X::SEQ) (defun seq ...) (in-package "USER") (use-package "X") (defun a ...) The CLTL2 way is to use DEFPACKAGE that lets you define the package in one form. From majordom@umnstat.stat.umn.edu Wed Jan 8 21:25:53 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00388; Wed, 8 Jan 1997 21:25:53 -0500 Return-Path: Received: from westnet.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00383; Wed, 8 Jan 1997 20:25:51 -0600 Received: from port17.ts2.westnet.com (port17.ts2.westnet.com [206.28.203.17]) by westnet.com (8.8.4/8.7.3) with SMTP id VAA12369; Wed, 8 Jan 1997 21:25:51 -0500 (EST) Date: Wed, 8 Jan 1997 21:25:51 -0500 (EST) Message-Id: <199701090225.VAA12369@westnet.com> X-Sender: lt@westnet.com (Unverified) X-Mailer: Windows Eudora Version 1.4.3 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: Xlisp List From: lt@westnet.com (Larry Troxler) Subject: Code walker? (and other CL porting issues) Cc: "Tobias Kunze" , taube@uiuc.edu (Rick Taube) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Is there a code walker ( walk-form ) version that works with Xlisp? I'm trying to look into the feasability of porting some Common Lisp code (Common Music) to XLISP-STAT, but it does use "walk-form" for some things. A related question having to do with porting Common Lisp code: Is the closette implementation of CLOS that's in the archives, as complete as PCL? If not, then not what is it missing? Larry Troxler lt@westnet.com From majordom@umnstat.stat.umn.edu Thu Jan 9 09:33:19 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10923; Thu, 9 Jan 1997 09:33:19 -0500 Return-Path: Received: from [144.81.225.2] by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10918; Thu, 9 Jan 1997 08:33:14 -0600 Received: from research.ets.org by gateway.ets.org with SMTP id AA25859 (5.65c/IDA-1.4.4 for ); Thu, 9 Jan 1997 09:32:13 -0500 Received: from comet.research.ets.org by research.ets.org with SMTP id AA23222 (5.67b/IDA-1.4.3 for lt@westnet.com); Thu, 9 Jan 1997 09:32:11 -0500 Received: by comet.research.ets.org (SMI-8.6/SMI-SVR4) id JAA11831; Thu, 9 Jan 1997 09:32:10 -0500 Date: Thu, 9 Jan 1997 09:32:10 -0500 Message-Id: <199701091432.JAA11831@comet.research.ets.org> From: Russell Almond To: lt@westnet.com Cc: stat-lisp-news@umnstat.stat.umn.edu, tkunze@ulysses.stanford.edu, taube@uiuc.edu In-Reply-To: <199701090225.VAA12369@westnet.com> (lt@westnet.com) Subject: Re: Code walker? (and other CL porting issues) Reply-To: ralmond@ets.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Is there a code walker ( walk-form ) version that works with Xlisp? I'm trying to look into the feasability of porting some Common Lisp code (Common Music) to XLISP-STAT, but it does use "walk-form" for some things. I don't think walk-form is a standard Common-Lisp function. (It certainly doesn't appear in the Second Edition of Steele). It must either be defined somewhere in the Common Music code or in a specific implementation. A related question having to do with porting Common Lisp code: Is the closette implementation of CLOS that's in the archives, as complete as PCL? If not, then not what is it missing? No, closette is a mini implementation of CLOS which is used in the Art of the Meta-Object protocol to illustrate the ideas. It is hard to say what does and does not work. PCL was a complete implementation of the then proposed CLOS, but the standard changed and I don't think the work necessary to bring PCL up to standard has ever been done (I could be wrong here). I never managed to get Graphical-Belief compiled using PCL (it currently runs under Allegro CL). Good Luck. --Russell Almond almond@acm.org From majordom@umnstat.stat.umn.edu Thu Jan 9 10:19:18 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11311; Thu, 9 Jan 1997 10:19:18 -0500 Return-Path: Received: from [144.81.225.2] by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11306; Thu, 9 Jan 1997 09:19:15 -0600 Received: from research.ets.org by gateway.ets.org with SMTP id AA28031 (5.65c/IDA-1.4.4 for ); Thu, 9 Jan 1997 10:19:12 -0500 Received: from comet.research.ets.org by research.ets.org with SMTP id AA24096 (5.67b/IDA-1.4.3 for sdm7g@Virginia.EDU); Thu, 9 Jan 1997 10:19:11 -0500 Received: by comet.research.ets.org (SMI-8.6/SMI-SVR4) id KAA11855; Thu, 9 Jan 1997 10:19:11 -0500 Date: Thu, 9 Jan 1997 10:19:11 -0500 Message-Id: <199701091519.KAA11855@comet.research.ets.org> From: Russell Almond To: sdm7g@Virginia.EDU Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: (sdm7g@Virginia.EDU) Subject: Re: Package Madness! Reply-To: ralmond@ets.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I had quite a few problems mastering the intricacies of Lisp Packages -- particularly when they involve using packages to redefine and shadow existing functions, until I discovered ( in CLtL1 ) the mnemonic "Put in seven extremely random interface commands" - which corresponds to the sequence: Provide IN-package Shadow EXport Require USE-package Import Commands==Contents of package/module Which is the order you (mostly) have to use those statements to be able to Require your package without tripping over any name clashes. ( Actually, I've been sticking "Provide" at the *end* of my packages so that if it is provided, that indicates that it did load successfully. That shouldn't make any difference with a well tested module, but it helps with something that I'm interactively editing and testing. If PROVIDE executes and *then* the module bombs out, I have to manually LOAD the module before reloading the module which REQUIREd it. ) I eventually found this proscription to be useless. First, as you note, provide needs to come at the end of the file not the begining. Second, I found it was much more useful to have a single package definition file for each package I created. (In most of my code a package corresponds to a directory and it has a file "package.lisp" which contains its package definition). Thus, I can load all the package information without loading all of the functions (this is often useful for letting the program know where dependencies will lie). Third, I carefully graph dependencies between packages and load the corresponding packages in the required order. Thus, if the "DISTRIBUTIONS" package uses symbols from the "PARAMETERS" package, I make sure I load that one first. If at all possible, you package dependencies should form an Acyclic Directed Graph. Fourth, I rely on defpackage rather than the Seven Random Functions. (There is a public domain version of Defpackage kicking around the net. I can try and dig it out if people are interested.) Fifth, I try to avoid explicit dependencies on packages, but rather refer to them with package qualified symbols. Thus X:SEQ as opposed to SEQ (with an implicit use-package above defining the dependency). There are some exceptions, but usually this works. Hope this helps. --Russell ALmond From majordom@umnstat.stat.umn.edu Sun Jan 12 19:40:19 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25367; Sun, 12 Jan 1997 19:40:19 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25363; Sun, 12 Jan 1997 19:40:13 -0600 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.7.5/10.1) id UAA19188; Sun, 12 Jan 1997 20:39:31 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sun, 12 Jan 1997 20:40:54 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: HTML Notes for LispStat Course Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi Lisp-Stat'ers: The entire notes for my LispStat course are available in HTML on line at http://forrest.psych.unc.edu/teaching/p285/p285.html These notes include programming notes and assignments with answers. The material covers essentially of the Luke Tierney's book on Lisp-Stat. Enjoy ... ----- Forrest Young -- at home Professor, Psychometrics CB# 3270 Davie Hall email: mailto://forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software sites: http://forrest.psych.unc.edu/research/ViSta.html USA http://www.psych.unc.edu/fwy/research/ViSta.html USA http://sunsite.univie.ac.at/ViSta/ViSta.html Austria http://micros.hensa.ac.uk/mirrors/vista/ViSta.html Great Britain From majordom@umnstat.stat.umn.edu Tue Jan 14 18:29:54 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26857; Tue, 14 Jan 1997 18:29:54 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26853; Tue, 14 Jan 1997 18:29:48 -0600 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.7.5/10.1) id TAA13749; Tue, 14 Jan 1997 19:29:42 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Tue, 14 Jan 1997 19:30:30 -0500 To: vista@unc.edu, edstat-l@jse.stat.ncsu.edu, stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: ViSta Mirrors Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk To: Those interested in ViSta, the Visual Statistics system: The original ViSta site is now mirrored at two locations in Europe as well as at a new location at the University of North Carolina. The mirror addresses are given in the signature of this message. I am seeking additional mirrors in Australia and Japan and would appreciate any suggestions. ----- Forrest Young Professor, Psychometrics CB# 3270 Davie Hall email: mailto://forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software and Documentation sites: http://forrest.psych.unc.edu/research/ViSta.html USA http://www.psych.unc.edu/fwy/research/ViSta.html USA http://sunsite.univie.ac.at/ViSta/ViSta.html Austria http://micros.hensa.ac.uk/mirrors/vista/ViSta.html United Kingdom From majordom@umnstat.stat.umn.edu Thu Jan 23 05:55:30 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00314; Thu, 23 Jan 1997 05:55:30 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00310; Thu, 23 Jan 1997 05:55:24 -0600 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.7.5/10.1) id GAA19595; Thu, 23 Jan 1997 06:55:21 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 23 Jan 1997 06:55:31 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: Mime Type for Lisp-Stat Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi Lispers: I am preparing to publically serve Lisp-Stat applets from my site and am wondering if anyone has experience doing this. (I've been doing this privately for a while). I recall that Jan did something like this but remember that he set up his site as a Lisp-Stat provider (users would net into his site which would then do the computations on his remote machine), whereas I plan to set up mine as a server (users would net into my site, download code onto their local machine, which then does the computations locally). Do I recall correctly? I plan on using "application/xlispstat" as the mime-type unless someone else is already using a different mime-type. Anyone out there already doing this? Is there an already existing mime-type for this? Forrest ----- Forrest Young Professor, Psychometrics CB# 3270 Davie Hall email: mailto://forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software and Documentation sites: http://forrest.psych.unc.edu/research/ViSta.html USA http://www.psych.unc.edu/fwy/research/ViSta.html USA http://sunsite.univie.ac.at/ViSta/ViSta.html Austria http://micros.hensa.ac.uk/mirrors/vista/ViSta.html United Kingdom From majordom@umnstat.stat.umn.edu Thu Jan 23 07:11:11 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00546; Thu, 23 Jan 1997 07:11:11 -0600 Return-Path: Received: from fhat.stat.sc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00541; Thu, 23 Jan 1997 07:11:06 -0600 Received: (from rossini@localhost) by fhat.stat.sc.edu (8.6.12/8.6.12) id IAA16022; Thu, 23 Jan 1997 08:11:04 -0500 Date: Thu, 23 Jan 1997 08:11:04 -0500 Message-Id: <199701231311.IAA16022@fhat.stat.sc.edu> From: Anthony Rossini To: forrest@unc.edu (Forrest Young) Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Mime Type for Lisp-Stat In-Reply-To: References: Reply-To: rossini@stat.sc.edu Mime-Version: 1.0 (generated by tm-edit 7.101) Content-Type: text/plain; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >>>>> "Forrest" == Forrest Young writes: Forrest> Hi Lispers: I am preparing to publically serve Lisp-Stat Forrest> applets from my site and am wondering if anyone has Forrest> experience doing this. (I've been doing this privately Forrest> for a while). Forrest> I recall that Jan did something like this but remember Forrest> that he set up his site as a Lisp-Stat provider (users Forrest> would net into his site which would then do the Forrest> computations on his remote machine), whereas I plan to Forrest> set up mine as a server (users would net into my site, Forrest> download code onto their local machine, which then does Forrest> the computations locally). Forrest> Do I recall correctly? Yes and no. Both local and remote things are done there. In particular, from my .mailcap file (personal mime configuration): ## Statistical information, PSU/AJR style application/x-stat-xlisp; xlispstat.wrap %s ## Statistical information, UCLA style. application/x-lisp; xlispstat.wrap %s where xlispstat.wrap is just a script set up to preview the code before dumping it into xlispstat (I'm slightly paranoid about my account). Forrest> I plan on using "application/xlispstat" as the mime-type Forrest> unless someone else is already using a different Forrest> mime-type. Anyone out there already doing this? Is there Forrest> an already existing mime-type for this? See above. Technically, you should use x-xlispstat, to be RFC-XXXX (I forget which) compliant, since it isn't a mandated mime type. I prefer something like: x-stat- to denote that the information is statistical in nature (with program name being R, or xlisp (short for xlispstat), etc. In the long run, it will matter, which is why I was a bit annoyed at Jan for choosing "x-lisp" after we'd set up "x-stat-xlisp" at Penn State long ago, but it also isn't really important -- just another line to the mime types (or .mailcap, if you use Unix and customize things yourself). I believe that Iowa State Stats uses yet another mime type, but I never did add it to my list, sorry... If anyone else knows of people doing this as well (besides me at Penn State/South Carolina, UCLA, Iowa State, etc), I'd be interested in hearing about it. best, ---- -tony (Anthony Rossini) Asst. Prof and General Nuisance Department of Statistics rossini@stat.sc.edu University of South Carolina http://www.stat.sc.edu/~rossini/ Columbia, SC 29208 803-777-3578(O) ..7-4048 (fax) From majordom@umnstat.stat.umn.edu Thu Jan 23 08:01:05 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00715; Thu, 23 Jan 1997 08:01:05 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00711; Thu, 23 Jan 1997 08:00:59 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.7.5/10.1) id JAA26759; Thu, 23 Jan 1997 09:00:56 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 23 Jan 1997 09:00:44 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: Re: Mime Type for Lisp-Stat Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'm forwarding this reply to the StatLisp news group since I think it is of interest to all --- Forrest >From: Erich Neuwirth >Subject: Re: Mime Type for Lisp-Stat >To: forrest@unc.edu (Forrest Young) >Date: Thu, 23 Jan 1997 13:34:18 +0100 (MET) >Cc: konrad@smc.univie.ac.at (Konrad Neuwirth) >Mime-Version: 1.0 >Status: > >hi forrest, > >on our sunsite, where we are mirroring you, >we are also offereing >some xlispstat examples. > >just go into my multemdia and iteractive apps over the internet >item. > >i am using mime type >x-xlispstat > >the x- part indicates that this is not an >officially registered mime type. >si i think this is better since it is >complying with the net rules. > >you can contact > >konrad@smc.univie.ac.at > >who set up your mirror here, >and he is much more knowleadgeable than me >on the topic of naming mime types. > >he is gettinga cc of this message anyhow. > >greetings from vienna > >erich > > > > >-- >Email: neuwirth@smc.univie.ac.at >PGP-key at http://www.smc.univie.ac.at/PGP-Keys/neuwirth.asc >Visit my spreadsheet home page at http://sunsite.univie.ac.at/Spreadsite/ > --- Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Sites: http://forrest.psych.unc.edu/research/ViSta.html USA http://www.psych.unc.edu/fwy/research/ViSta.html USA http://sunsite.univie.ac.at/ViSta/ViSta.html Austria http://micros.hensa.ac.uk/mirrors/vista/ViSta.html Great Britain From majordom@nokomis.stat.umn.edu Thu Jan 23 09:56:05 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA17505; Thu, 23 Jan 1997 09:56:05 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA17500; Thu, 23 Jan 1997 09:55:52 -0600 From: Luke Tierney Message-Id: <9701231555.AA17500@nokomis.stat.umn.edu> Subject: Re: Mime Type for Lisp-Stat To: forrest@unc.edu (Forrest Young) Date: Thu, 23 Jan 1997 09:55:52 -0600 (CST) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Forrest Young" at Jan 23, 97 06:55:31 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 158 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk I use the mime type Anthony Rossini chose, application/x-stat-xlisp On our server, files with the .xlsp extension are served up with this mime type. luke From majordom@umnstat.stat.umn.edu Thu Jan 23 10:32:13 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02130; Thu, 23 Jan 1997 10:32:13 -0600 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02122; Thu, 23 Jan 1997 10:32:03 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id IAA19664; Thu, 23 Jan 1997 08:30:42 -0800 (PST) From: Jan Deleeuw Message-Id: <199701231630.IAA19664@galton.math.ucla.edu> Subject: Re: Mime Type for Lisp-Stat To: rossini@stat.sc.edu Date: Thu, 23 Jan 1997 08:30:41 -0800 (PST) Cc: forrest@unc.edu, stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <199701231311.IAA16022@fhat.stat.sc.edu> from "Anthony Rossini" at Jan 23, 97 08:11:04 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Tony says: In the long run, it will matter, which is why I was a bit annoyed at Jan for choosing "x-lisp" after we'd set up "x-stat-xlisp" at Penn State long ago, but it also isn't really important -- just another line to the mime types (or .mailcap, if you use Unix and customize things yourself). Well -- I am not so sure about the "after" in there. But this shows again that if you are annoyed, you should let the person know what you are annoyed with. Actually, it is a very good idea to change to x-stat-xlisp for scripts that are supposed to execute on the clients. I now have the problem that if I try to download a foo.lsp file from the archive, it will execute, so I have to do something anyway. I'll change matters up here, according to the Tony/Luke rules. --- Jan -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Thu Jan 23 10:51:33 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02419; Thu, 23 Jan 1997 10:51:33 -0600 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02414; Thu, 23 Jan 1997 10:51:27 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id IAA19941 for stat-lisp-news@umnstat.stat.umn.edu; Thu, 23 Jan 1997 08:51:26 -0800 (PST) From: Jan Deleeuw Message-Id: <199701231651.IAA19941@galton.math.ucla.edu> Subject: mime -- again To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Thu, 23 Jan 1997 08:51:25 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Let's agree on foo.xlsp gets mime type "x-stat-xlisp" and (on our clients) gets executed foo.lsp gets mime type "x-lisp" and (on our clients) gets downloaded This means the scripts in the UCLA Textbook and Toolbox will change to foo.xlsp, and it means that people wanting to use this need to change their mimecaps (which means that they configure netscape appropriately). I'll put instructions in appropriate places, and I'll add icons for foo.lsp and foo.xlsp to our Apache servers. An answer for Forrest: this is the way most of our stuff works. But there are quite a few others which use xlispstat via CGI (i.e. the program executes on the server). This is done for the power calculator, the regression calculator, and the Xlisp-Stat interface in http://www.stat.ucla.edu/cgi-bin/Xlisp-Stat.cgi The CGI programs are all written in C and they use popen/fork/pipe to start the appropriate xlispstat processes. The Xlisp-Stat interface forks a special version of Xlisp-Stat with system etc disabled. Since we recently changed both the machine and the httpd server for www.stat.ucla.edu, there will be some things that are broken at the moment. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Thu Jan 23 10:54:25 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02442; Thu, 23 Jan 1997 10:54:25 -0600 Return-Path: Received: from post3.inre.asu.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02438; Thu, 23 Jan 1997 10:54:21 -0600 Received: from smtp1.asu.edu by asu.edu (PMDF V5.0-6 #7723) id <01IEJQXO3UIO90RY82@asu.edu> for stat-lisp-news@umnstat.stat.umn.edu; Thu, 23 Jan 1997 09:53:48 -0700 (MST) Received: from seamonkey.ed.asu.edu (seamonkey.ed.asu.edu [129.219.88.110]) by smtp1.asu.edu (8.6.12/8.6.12) with SMTP id JAA10186 for ; Thu, 23 Jan 1997 09:53:51 -0700 Received: from 129.219.90.134 ([129.219.90.134]) by seamonkey.ed.asu.edu (5.x/SMI-SVR4-CKG 8/29/95) id AA12978; Thu, 23 Jan 1997 09:50:47 -0700 Date: Thu, 23 Jan 1997 09:53:48 +0000 From: John Behrens Subject: Re: Mime Type for Lisp-Stat To: stat-lisp-news@umnstat.stat.umn.edu Message-Id: <32E73525.FBD@asu.edu> Organization: Arizona State University Mime-Version: 1.0 X-Mailer: Mozilla 3.01Gold (Macintosh; I; PPC) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit References: Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Lispers: At ASU we use x-application/x-xlisp, but would be glad to change it to x-xlispstat if this is the emerging standard. John Behrens From majordom@umnstat.stat.umn.edu Thu Jan 23 11:05:25 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02542; Thu, 23 Jan 1997 11:05:25 -0600 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02538; Thu, 23 Jan 1997 11:05:20 -0600 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA02505 (5.67b/IDA-1.5 for ); Thu, 23 Jan 1997 12:04:24 -0500 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA15552 for ; Thu, 23 Jan 1997 12:03:42 -0400 Message-Id: <32E7C4D7.78AD@kodak.com> Date: Thu, 23 Jan 1997 12:06:47 -0800 From: Jeff Morris Reply-To: morris@Kodak.COM Organization: Eastman Kodak Company X-Mailer: Mozilla 3.0 (Win16; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: Sorting a Matrix Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I am interested in writing an application where we will import an text file containing a column of concentrations and a column of responses. We will perform one-way anovas on datasets but don't want to force the user into a given type of format so need to get the columns into a matrix and sort the rows according to one of the columns. How does one sort the responses by the concentrations? I want to be able to do Dunnett's test on each concentration vs. a control. I've asked about XLS code for Dunnett's test before, but no one seemed to have it. I'm not much of a programmer so if anyone would like take this on as a project I'd sure like it :>) Thanks for any help! -- Jeffrey D. Morris, Ph.D. Health & Environment Labs Eastman Kodak Company morris@kodak.com From majordom@umnstat.stat.umn.edu Thu Jan 23 11:34:39 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02846; Thu, 23 Jan 1997 11:34:39 -0600 Return-Path: Received: from fhat.stat.sc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02842; Thu, 23 Jan 1997 11:34:33 -0600 Received: (from rossini@localhost) by fhat.stat.sc.edu (8.6.12/8.6.12) id MAA16738; Thu, 23 Jan 1997 12:34:08 -0500 Date: Thu, 23 Jan 1997 12:34:08 -0500 Message-Id: <199701231734.MAA16738@fhat.stat.sc.edu> From: Anthony Rossini To: Jan Deleeuw Cc: rossini@stat.sc.edu, forrest@unc.edu, stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Mime Type for Lisp-Stat In-Reply-To: <199701231630.IAA19664@galton.math.ucla.edu> References: <199701231311.IAA16022@fhat.stat.sc.edu> <199701231630.IAA19664@galton.math.ucla.edu> Reply-To: rossini@stat.sc.edu Mime-Version: 1.0 (generated by tm-edit 7.101) Content-Type: text/plain; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >>>>> "Jan" == Jan Deleeuw writes: Jan> Tony says: In the long run, it will matter, which is why I Jan> was a bit annoyed at Jan for choosing "x-lisp" after we'd set Jan> up "x-stat-xlisp" at Penn State long ago, but it also isn't Jan> really important -- just another line to the mime types (or Jan> .mailcap, if you use Unix and customize things yourself). Jan> Well -- I am not so sure about the "after" in there. But this Jan> shows again that if you are annoyed, you should let the Jan> person know what you are annoyed with. Actually, it is a very Jan> good idea to change to x-stat-xlisp for scripts that are Jan> supposed to execute on the clients. I probably should've emphasized "was", Jan. And, I might be wrong about the time frame (since they were both worked on around the same time). However, I overgrew (SOME OF) my childishness 1-2 years ago :-), and really don't care one way or another now. The name for the mime type really isn't important as long as people are aware of what they are, where they are... In fact, I sort of like the idea of x-xlispstat version, these days. x-stat-xlisp would only make sense if more people used x-stat-R or x-stat-sas as client side helpers, for example... (then we get into the grand unification theory argument :-). BTW, one of the things I've been considering is construction of an XLispStat (and/or R) Plug-in for Netscape, using the SDK. The code would be a bit strange, since ideally there would be a base "SDK" device which did the appropriate things on all three platforms? Has anyone done much more than just go over the Netscape SDK specs and jot down some of the needed additions to the code (i.e. studied the routing of listener IO and graphics devices, etc?) best, ---- -tony (Anthony Rossini) Asst. Prof and General Nuisance Department of Statistics rossini@stat.sc.edu University of South Carolina http://www.stat.sc.edu/~rossini/ Columbia, SC 29208 803-777-3578(O) ..7-4048 (fax) From majordom@umnstat.stat.umn.edu Thu Jan 23 11:42:09 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02907; Thu, 23 Jan 1997 11:42:09 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02903; Thu, 23 Jan 1997 11:42:05 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.7.5/10.1) id MAA23854; Thu, 23 Jan 1997 12:42:02 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 23 Jan 1997 12:41:50 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: mime - reprise Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk So Lispers: Do we have an emerged (emerging) standard, as proposed by Jan? >foo.xlsp gets mime type "x-stat-xlisp" and (on our clients) gets executed >foo.lsp gets mime type "x-lisp" and (on our clients) gets downloaded The only problem I can see with this is that windoze based servers can't deal with .xlxp extensions. Either we agree there's not enough of them to matter, or we might go with foo.xls for client executable files. I think I prefer foo.xls, if it doesn't mean too much reworking on other folks parts. Any thoughts? --- Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Sites: http://forrest.psych.unc.edu/research/ViSta.html USA http://www.psych.unc.edu/fwy/research/ViSta.html USA http://sunsite.univie.ac.at/ViSta/ViSta.html Austria http://micros.hensa.ac.uk/mirrors/vista/ViSta.html Great Britain From majordom@umnstat.stat.umn.edu Thu Jan 23 11:51:02 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03015; Thu, 23 Jan 1997 11:51:02 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03011; Thu, 23 Jan 1997 11:50:57 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.7.5/10.1) id MAA24717; Thu, 23 Jan 1997 12:50:38 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 23 Jan 1997 12:50:26 -0500 To: morris@Kodak.COM From: forrest@unc.edu (Forrest Young) Subject: Re: Sorting a Matrix Cc: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >I am interested in writing an application where we will import an text >file containing a column of concentrations and a column of responses. >We will perform one-way anovas on datasets but don't >want to force the user into a given type of format so need to get the >columns into a matrix and sort the rows according to one of the >columns. How does one sort the responses by the concentrations? I >want to be able to do Dunnett's test on each concentration vs. a >control. I've asked about XLS code for Dunnett's test before, but >no one seemed to have it. I'm not much of a programmer so if anyone >would like take this on as a project I'd sure like it :>) > >Thanks for any help! > You might be able to do this with the ViSta extension of lispstat that I've developed. ViSta will read in text files of data where each line contains an observation (a concentration and a response). These are converted to ViSta data objects. It can sort observations into order so that all levels of a concentration are in adjacent records. It can perform one-way anova, but it must be a balanced design. So, this is a potential problem. It can't do a Dunnett's test. How many observations are there in the data? (ViSta can be pretty slow for large data). Let me know how this sounds, and if looks possible, I'll give it a try. --- Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Sites: http://forrest.psych.unc.edu/research/ViSta.html USA http://www.psych.unc.edu/fwy/research/ViSta.html USA http://sunsite.univie.ac.at/ViSta/ViSta.html Austria http://micros.hensa.ac.uk/mirrors/vista/ViSta.html Great Britain From majordom@umnstat.stat.umn.edu Thu Jan 23 12:31:18 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03335; Thu, 23 Jan 1997 12:31:18 -0600 Return-Path: Received: from fhat.stat.sc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03331; Thu, 23 Jan 1997 12:31:12 -0600 Received: (from rossini@localhost) by fhat.stat.sc.edu (8.6.12/8.6.12) id NAA17031; Thu, 23 Jan 1997 13:30:14 -0500 Date: Thu, 23 Jan 1997 13:30:14 -0500 Message-Id: <199701231830.NAA17031@fhat.stat.sc.edu> From: Anthony Rossini To: forrest@unc.edu (Forrest Young) Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: mime - reprise In-Reply-To: References: Reply-To: rossini@stat.sc.edu Mime-Version: 1.0 (generated by tm-edit 7.101) Content-Type: text/plain; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >>>>> "Forrest" == Forrest Young writes: Forrest> So Lispers: Do we have an emerged (emerging) standard, as Forrest> proposed by Jan? >> foo.xlsp gets mime type "x-stat-xlisp" and (on our clients) >> gets executed >> foo.lsp gets mime type "x-lisp" and (on our >> clients) gets downloaded This could also be just plain "text/plain", as well, unless there is a need to think about future routing of the information. For example, we might also think about foo.fsl as being type "x-lisp" as well (which would imply that it would be bad to use the "text/plain" category)? Is this a valid thing to consider? Forrest> The only problem I can see with this is that windoze Forrest> based servers can't deal with .xlxp extensions. Forrest> I think I prefer foo.xls, if it doesn't mean too much Forrest> reworking on other folks parts. I know nearly nothing about MS W95 or MS WNT, but don't they allow for long extensions? We can assume that MS W3.xx will go away in the next year or 3... later, ---- -tony (Anthony Rossini) Asst. Prof and General Nuisance Department of Statistics rossini@stat.sc.edu University of South Carolina http://www.stat.sc.edu/~rossini/ Columbia, SC 29208 803-777-3578(O) ..7-4048 (fax) From majordom@umnstat.stat.umn.edu Thu Jan 23 13:38:59 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04082; Thu, 23 Jan 1997 13:38:59 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04078; Thu, 23 Jan 1997 13:38:52 -0600 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.7.5/10.1) id OAA07818; Thu, 23 Jan 1997 14:38:45 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 23 Jan 1997 14:38:56 -0500 To: rossini@stat.sc.edu From: forrest@unc.edu (Forrest Young) Subject: Re: mime - reprise Cc: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >>>>>> "Forrest" == Forrest Young writes: > > Forrest> So Lispers: Do we have an emerged (emerging) standard, as > Forrest> proposed by Jan? > > >> foo.xlsp gets mime type "x-stat-xlisp" and (on our clients) > >> gets executed > > >> foo.lsp gets mime type "x-lisp" and (on our > >> clients) gets downloaded > >This could also be just plain "text/plain", as well, unless there is a >need to think about future routing of the information. For example, >we might also think about foo.fsl as being type "x-lisp" as well >(which would imply that it would be bad to use the "text/plain" >category)? Is this a valid thing to consider? yes, I too was wondering about .fsl files. I think your suggestion is the right one. > > Forrest> The only problem I can see with this is that windoze > Forrest> based servers can't deal with .xlxp extensions. > > Forrest> I think I prefer foo.xls, if it doesn't mean too much > Forrest> reworking on other folks parts. > >I know nearly nothing about MS W95 or MS WNT, but don't they allow for >long extensions? We can assume that MS W3.xx will go away in the next >year or 3... I've no clue either about W95 or WNT (and OS/2, as well), but I'll bet your assumption is right about W3.xx ... ----- Forrest Young Professor, Psychometrics CB# 3270 Davie Hall email: mailto://forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software and Documentation sites: http://forrest.psych.unc.edu/research/ViSta.html USA http://www.psych.unc.edu/fwy/research/ViSta.html USA http://sunsite.univie.ac.at/ViSta/ViSta.html Austria http://micros.hensa.ac.uk/mirrors/vista/ViSta.html United Kingdom From majordom@nokomis.stat.umn.edu Thu Jan 23 15:25:43 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA17673; Thu, 23 Jan 1997 15:25:43 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA17668; Thu, 23 Jan 1997 15:25:37 -0600 From: Luke Tierney Message-Id: <9701232125.AA17668@nokomis.stat.umn.edu> Subject: Re: Sorting a Matrix To: morris@Kodak.COM Date: Thu, 23 Jan 1997 15:25:37 -0600 (CST) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: <32E7C4D7.78AD@kodak.com> from "Jeff Morris" at Jan 23, 97 12:06:47 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 884 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > I am interested in writing an application where we will import an text > file containing a column of concentrations and a column of responses. > We will perform one-way anovas on datasets but don't > want to force the user into a given type of format so need to get the > columns into a matrix and sort the rows according to one of the > columns. How does one sort the responses by the concentrations? I > want to be able to do Dunnett's test on each concentration vs. a > control. I've asked about XLS code for Dunnett's test before, but > no one seemed to have it. I'm not much of a programmer so if anyone > would like take this on as a project I'd sure like it :>) > > Thanks for any help! > > -- > Jeffrey D. Morris, Ph.D. > Health & Environment Labs > Eastman Kodak Company > morris@kodak.com > sorting y according to x can be done as (select y (order x)) luke From majordom@umnstat.stat.umn.edu Thu Jan 23 17:38:29 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05832; Thu, 23 Jan 1997 17:38:29 -0600 Return-Path: Received: from nessie.mcc.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05827; Thu, 23 Jan 1997 17:38:22 -0600 Received: from Tedistan.annex.umist.ac.uk by nessie.mcc.ac.uk with SMTP (PP); Thu, 23 Jan 1997 23:37:22 +0000 Message-Id: From: Ted.Harding@nessie.mcc.ac.uk (Ted Harding) Subject: Re: mime - reprise To: stat-lisp-news@umnstat.stat.umn.edu (Stat-Lisp News) Date: Thu, 23 Jan 1997 22:51:05 +0000 (GMT) X-Mailer: ELM [version 2.4 PL23] Content-Type: text Content-Length: 2055 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk ( Re Message From: Forrest Young ) > > So Lispers: > > Do we have an emerged (emerging) standard, as proposed by Jan? > > >foo.xlsp gets mime type "x-stat-xlisp" and (on our clients) gets executed > >foo.lsp gets mime type "x-lisp" and (on our clients) gets downloaded > > The only problem I can see with this is that windoze based servers can't > deal with .xlxp extensions. > > Either we agree there's not enough of them to matter, or we might go with > foo.xls for client executable files. > > I think I prefer foo.xls, if it doesn't mean too much reworking on other > folks parts. > > Any thoughts? > > Forrest Young Much as I could go for the idea that there aren't enough of them to matter (even if it's millions ... ), I'm wondering if it matters anyway. Namely, if you try giving a 4+ letter extension to DOS, it only seems to look at the first 3 letters anyway. See the following (*.bak vs *.bake). No doubt some experimentation would be needed to see if it mattered in the application being discussed. But, assuming it continues to work the same way, then anyone can call for "foo.xlsp" and a DOS server will see "foo.xls" and execute it (it could get more complicated if it were the "*.xlsp" files for downloading, for then the filename that the client got would presumably try to be "foo.xls"; but as you propose it the downloadable files have 3-letter extensions anyway). However, this is just a shot in the dark. Experts please judge. Best wishes, Ted. (Ted.Harding@nessie.mcc.ac.uk) =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= [Novell DOS] C:\>DIR *.BAK Volume in drive C does not have a label Directory of C:\ AUTOEXEC BAK 389 24/06/96 17:22 CONFIG BAK 267 27/06/96 15:30 2 File(s) 32407552 bytes free [Novell DOS] C:\>DIR *.BAKE Volume in drive C does not have a label Directory of C:\ AUTOEXEC BAK 389 24/06/96 17:22 CONFIG BAK 267 27/06/96 15:30 2 File(s) 32407552 bytes free From majordom@umnstat.stat.umn.edu Thu Jan 23 17:49:05 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05916; Thu, 23 Jan 1997 17:49:05 -0600 Return-Path: Received: from zeus.usq.edu.au by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05910; Thu, 23 Jan 1997 17:48:58 -0600 Received: from po.usq.edu.au (pluto.usq.edu.au [139.86.128.23]) by zeus.usq.edu.au (8.7.5/8.7) with ESMTP id JAA28837; Fri, 24 Jan 1997 09:47:39 +1000 (EST) Received: from Microsoft Mail (PU Serial #1166) by po.usq.edu.au (PostalUnion/SMTP(tm) v2.1.9c for Windows NT(tm)) id AA-1997Jan24.094300.1166.67691; Fri, 24 Jan 1997 09:47:36 +1100 From: darnell@scipo.usq.edu.au (Ross Darnell) To: forrest@unc.edu (Forrest Young), rossini@stat.sc.edu (rossini) Cc: stat-lisp-news@umnstat.stat.umn.edu (stat-lisp-news) Message-Id: <1997Jan24.094300.1166.67691@po.usq.edu.au> X-Mailer: Microsoft Mail via PostalUnion/SMTP for Windows NT Mime-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Organization: University of Southern Queensland, Toowoomba, QLD, AU Date: Fri, 24 Jan 1997 09:47:36 +1100 Subject: RE: mime - reprise Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >I know nearly nothing about MS W95 or MS WNT, but don't they allow for long extensions? Yes > We can assume that MS W3.xx will go away in the next year or 3... Bill Gates will guaranty it later, ---- -tony (Anthony Rossini) Asst. Prof and General Nuisance Department of Statistics rossini@stat.sc.edu University of South Carolina http://www.stat.sc.edu/~rossini/ Columbia, SC 29208 803-777-3578(O) ..7-4048 (fax) From majordom@umnstat.stat.umn.edu Thu Jan 23 18:05:23 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06055; Thu, 23 Jan 1997 18:05:23 -0600 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06051; Thu, 23 Jan 1997 18:05:19 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id QAA23302; Thu, 23 Jan 1997 16:05:06 -0800 (PST) From: Jan Deleeuw Message-Id: <199701240005.QAA23302@galton.math.ucla.edu> Subject: Re: mime - reprise To: Ted.Harding@nessie.mcc.ac.uk (Ted Harding) Date: Thu, 23 Jan 1997 16:05:05 -0800 (PST) Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: from "Ted Harding" at Jan 23, 97 10:51:05 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I hate to be a spoil-sport, but in the standard mime setup that comes with netscape or mimencode or whatever, xls is used for excel spreadsheets. Now we can collectively all wish that DOS, Windoze, Microsoft, Bill G, and Excel did not exist, but if we want to make Xlisp-Stat even more generally accepted than it already is, I think we should acknowledge that the evil empire is here to stay. Thus xls and xlsp are out, as far as I am concerned. But foo.xlisp could be used, because msdog will truncate this to foo.xli. On the other hand, having extensions with more than three letters is not very elegant, I think, so why not use foo.xli right away. Thus my current proposal is foo.xli => application/x-xlispstat => xlispstat %s foo.lsp => text/plain => browser handles this > > Much as I could go for the idea that there aren't enough of them to > matter (even if it's millions ... ), I'm wondering if it matters anyway. > Namely, if you try giving a 4+ letter extension to DOS, it only seems to > look at the first 3 letters anyway. See the following (*.bak vs *.bake). > No doubt some experimentation would be needed to see if it mattered in the > application being discussed. > > But, assuming it continues to work the same way, then anyone can call for > "foo.xlsp" and a DOS server will see "foo.xls" and execute it (it could > get more complicated if it were the "*.xlsp" files for downloading, for > then the filename that the client got would presumably try to be > "foo.xls"; but as you propose it the downloadable files have 3-letter > extensions anyway). > > However, this is just a shot in the dark. Experts please judge. > > Best wishes, > Ted. (Ted.Harding@nessie.mcc.ac.uk) > =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= > [Novell DOS] C:\>DIR *.BAK > > Volume in drive C does not have a label > Directory of C:\ > > AUTOEXEC BAK 389 24/06/96 17:22 > CONFIG BAK 267 27/06/96 15:30 > 2 File(s) 32407552 bytes free > > [Novell DOS] C:\>DIR *.BAKE > > Volume in drive C does not have a label > Directory of C:\ > > AUTOEXEC BAK 389 24/06/96 17:22 > CONFIG BAK 267 27/06/96 15:30 > 2 File(s) 32407552 bytes free > -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Thu Jan 23 18:55:34 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06274; Thu, 23 Jan 1997 18:55:34 -0600 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06255; Thu, 23 Jan 1997 18:55:28 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id QAA23591 for stat-lisp-news@umnstat.stat.umn.edu; Thu, 23 Jan 1997 16:55:25 -0800 (PST) From: Jan Deleeuw Message-Id: <199701240055.QAA23591@galton.math.ucla.edu> Subject: MIME, try-this To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Thu, 23 Jan 1997 16:55:24 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk See http://www.stat.ucla.edu:7070/~deleeuw/lisp/ this is a version of apache that serves .xli and .fsl as application/x-xlispstat thus if your client starts Xlisp-Stat if application/x-xlispstat is served, this should work .lsp files are served as text/plain, so they are just displayed also see the icons -- which are also in the Xlisp-Stat archive if you use this server (i.e. the one on port 7070) our basic server is not so configured yet, so if you go to http://www.stat.ucla.edu/~deleeuw/lisp/ the .xli and .fsl and .lsp files have ? icons and are all just displayed as text calculator.lsp is David Betz' TI-30, by the way -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Fri Jan 24 08:25:24 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA18226; Fri, 24 Jan 1997 08:25:24 -0600 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA18222; Fri, 24 Jan 1997 08:25:17 -0600 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA16868 (5.67b/IDA-1.5 for ); Fri, 24 Jan 1997 09:24:21 -0500 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA25882 for ; Fri, 24 Jan 1997 09:23:40 -0400 Message-Id: <32E8F0D8.316A@kodak.com> Date: Fri, 24 Jan 1997 09:26:48 -0800 From: Jeff Morris Reply-To: morris@Kodak.COM Organization: Eastman Kodak Company X-Mailer: Mozilla 3.0 (Win16; I) Mime-Version: 1.0 To: xlispstatnews Subject: anova question Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I have used the one-way.lsp to compare series of factors each a list or vector, but how do I compare factors when one list contains the grouping indicators and another list contains all the responses? I'm am sure there is an easy lispish way to create the first situation from the second, but I don't know how. Thanks to all for previous help on sorting a matrix. -- Jeffrey D. Morris, Ph.D. Health & Environment Labs Eastman Kodak Company morris@kodak.com From majordom@umnstat.stat.umn.edu Fri Jan 24 08:34:01 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA18294; Fri, 24 Jan 1997 08:34:01 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA18290; Fri, 24 Jan 1997 08:33:56 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.7.5/10.1) id JAA17587; Fri, 24 Jan 1997 09:33:52 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 24 Jan 1997 09:33:37 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: Re: mime Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi Lispers: So it looks like the result of the first ever, International Virtual Conference on Internet LispStat Standards ... IVCILS ;-) ... is, according to Jan's suggestion: foo.xli => application/x-xlispstat => xlispstat foo.fsl => application/x-xlispstat => xlispstat foo.lsp => text/plain => browser handles this I'll set up my server to work this way. Ya'll do the same, hear! (as they say in these parts) ... --- Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Sites: http://forrest.psych.unc.edu/research/ViSta.html USA http://www.psych.unc.edu/fwy/research/ViSta.html USA http://sunsite.univie.ac.at/ViSta/ViSta.html Austria http://micros.hensa.ac.uk/mirrors/vista/ViSta.html Great Britain From majordom@nokomis.stat.umn.edu Fri Jan 24 10:44:07 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA00919; Fri, 24 Jan 1997 10:44:07 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA00914; Fri, 24 Jan 1997 10:44:03 -0600 From: Luke Tierney Message-Id: <9701241644.AA00914@nokomis.stat.umn.edu> Subject: Re: mime - reprise To: deleeuw@math.ucla.edu (Jan Deleeuw) Date: Fri, 24 Jan 1997 10:44:02 -0600 (CST) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: <199701240005.QAA23302@galton.math.ucla.edu> from "Jan Deleeuw" at Jan 23, 97 04:05:05 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 469 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > could be used, because msdog will truncate this to foo.xli. On the > other hand, having extensions with more than three letters is > not very elegant, I think, so why not use foo.xli right away. I do seem to recall occasionally seeing a few extensions with more than three letters on the web -- something like .html or .java :-) > > foo.xli => application/x-xlispstat => xlispstat %s > foo.lsp => text/plain => browser handles this > but this seems fine. luke From majordom@nokomis.stat.umn.edu Sun Jan 26 10:36:51 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA26793; Sun, 26 Jan 1997 10:36:51 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA26788; Sun, 26 Jan 1997 10:36:46 -0600 From: Luke Tierney Message-Id: <9701261636.AA26788@nokomis.stat.umn.edu> Subject: bug in evalhook (3.44 and 3.47) To: stat-lisp-news@nokomis.stat.umn.edu Date: Sun, 26 Jan 1997 10:36:46 -0600 (CST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 698 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk There is a bug in the internal evalhook function code that can cause a segment violation if a garbage collection occurs at a particular point because a location the GC checks has not been properly initialized. This affects both versions 3.44 and 3.47 -- it is already fixed in 3.48 which will be out in one or two weeks I hope. To fix the bug, in the function xevalhook in the file xlbfun.c, the code #if 0 /* old way (see below) */ xlsave(oldenv); xlsave(oldfenv); xlsave(newenv); #else /* TAA MOD -- see below */ xlprotect(oldenv); xlprotect(oldfenv); xlprotect(newenv); #endif should be changed to xlsave(oldenv); xlsave(oldfenv); xlsave(newenv); luke From majordom@umnstat.stat.umn.edu Mon Jan 27 12:14:06 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03385; Mon, 27 Jan 1997 12:14:06 -0600 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03381; Mon, 27 Jan 1997 12:13:59 -0600 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA22099 (5.67b/IDA-1.5 for ); Mon, 27 Jan 1997 13:13:01 -0500 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA19284 for ; Mon, 27 Jan 1997 13:12:20 -0400 Message-Id: <32ED1AFC.51B9@kodak.com> Date: Mon, 27 Jan 1997 13:15:40 -0800 From: Jeff Morris Reply-To: morris@Kodak.COM Organization: Eastman Kodak Company X-Mailer: Mozilla 3.0 (Win16; I) Mime-Version: 1.0 To: xlispstatnews Subject: Reducing a List to Unique Values Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hell-o again! Today's problem is trying to reduce a list of values with possibly different numbers of repeats, to a list of unique values. to wit... (unknow-command '(0 0 1 2 2 2 3 4 4)) should return (0 1 2 3 4) Thanks as always! -- Jeffrey D. Morris, Ph.D. Health & Environment Labs Eastman Kodak Company morris@kodak.com From majordom@umnstat.stat.umn.edu Mon Jan 27 12:42:25 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03676; Mon, 27 Jan 1997 12:42:25 -0600 Return-Path: Received: from sulu.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03660; Mon, 27 Jan 1997 12:42:18 -0600 Received: from gorn.YP.biostat by sulu.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id KAA11195; Mon, 27 Jan 1997 10:42:15 -0800 (PST) Received: from localhost by gorn.YP.biostat (SMI-8.6/SMI-SVR4) id KAA00846; Mon, 27 Jan 1997 10:42:11 -0800 Date: Mon, 27 Jan 1997 10:42:11 -0800 (PST) From: Thomas Lumley X-Sender: thomas@gorn To: Jeff Morris Cc: xlispstatnews Subject: Re: Reducing a List to Unique Values In-Reply-To: <32ED1AFC.51B9@kodak.com> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Mon, 27 Jan 1997, Jeff Morris wrote: > Hell-o again! Today's problem is trying to reduce a list of values with > possibly different numbers of repeats, to a list of unique values. > > to wit... > (unknow-command '(0 0 1 2 2 2 3 4 4)) > > should return (0 1 2 3 4) remove-duplicates [note: I don't think the *ordering* of the result is guaranteed -- it might return (0 3 2 1 4)]. It also takes an optional argument specifying how to test for equality so that (eq, eql, equalp or something of your choice). Thomas Lumley ------------------------------------------------------+------ Biostatistics : "Never attribute to malice what : Uni of Washington : can be adequately explained by : Box 357232 : incompetence" - Hanlon's Razor : Seattle WA 98195-7232 : : ------------------------------------------------------------ From majordom@umnstat.stat.umn.edu Tue Jan 28 07:27:07 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA18471; Tue, 28 Jan 1997 07:27:07 -0600 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA18467; Tue, 28 Jan 1997 07:27:01 -0600 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AB00229 (5.67b/IDA-1.5 for ); Tue, 28 Jan 1997 08:26:07 -0500 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA27723 for ; Tue, 28 Jan 1997 08:25:26 -0400 Message-Id: <32EE2941.7E69@kodak.com> Date: Tue, 28 Jan 1997 08:28:49 -0800 From: Jeff Morris Reply-To: morris@Kodak.COM Organization: Eastman Kodak Company X-Mailer: Mozilla 3.0 (Win16; I) Mime-Version: 1.0 To: xlispstatnews Subject: Thanks Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Although some of you might not believe it, I AM reluctant to post simple questions about Lisp programming to this group. But I get so much help whenever I do, and that will only encourage me ;>) So I wanted to thank everyone for answering my last question about removing duplicates. Thanks! -- Jeffrey D. Morris, Ph.D. Health & Environment Labs Eastman Kodak Company morris@kodak.com From majordom@nokomis.stat.umn.edu Thu Jan 30 12:54:51 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA29162; Thu, 30 Jan 1997 12:54:51 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA29156; Thu, 30 Jan 1997 12:54:44 -0600 From: Luke Tierney Message-Id: <9701301854.AA29156@nokomis.stat.umn.edu> Subject: New alpha release of xlispstat is now available To: stat-lisp-news@nokomis.stat.umn.edu Date: Thu, 30 Jan 1997 12:54:44 -0600 (CST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 575 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk A new alpha release of xlispstat, 3.48, is now available for ftp from ftp://ftp.stat.umn.edu/pub/xlispstat/Alpha. This release fixes a number of bugs in the previous alpha release, 3.47, and incorporates changes in XLISP-PLUS 3.0. Unix sources (including code for Macintosh and MSWindows versions) and binaries macintosh and windows binaries are provided. Please let me know if you run into any problems. If this release is reasonably stable, it will become the standard release in a few weeks. Luke Tierney School of Statistics University of Minnesota luke@stat.umn.edu From majordom@umnstat.stat.umn.edu Fri Jan 31 07:44:22 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA07066; Fri, 31 Jan 1997 07:44:22 -0600 Return-Path: Received: from zam149.zam.kfa-juelich.de by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA07062; Fri, 31 Jan 1997 07:43:59 -0600 Received: from hlrz9 by kfa-juelich.de (5.65v3.2/DEC-AXP) id AA26384; Fri, 31 Jan 1997 14:43:48 +0100 Received: from hlrz4.hlrz.kfa-juelich.de by hlrz9.hlrz.kfa-juelich.de (SMI-8.6/SMI-4.1) id OAA22027; Fri, 31 Jan 1997 14:43:44 +0100 Reply-To: jmu@hlrz9.hlrz.kfa-juelich.de Received: by hlrz4.hlrz.kfa-juelich.de (SMI-8.6/SMI-4.1) id OAA25525; Fri, 31 Jan 1997 14:43:46 +0100 Date: Fri, 31 Jan 1997 14:43:46 +0100 Message-Id: <199701311343.OAA25525@hlrz4.hlrz.kfa-juelich.de> From: Johannes Mueller-Gerking To: stat-lisp-news@umnstat.stat.umn.edu Subject: xlispstat-3-48: rank command? Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi all, I found a minor problem with the new 3-48 version of xlispstat: > (rank '(9 8 7 6 5)) Error: The function LORDER is unbound. Happened in: # System: SunOS 5.5.1 Generic_103640-03 sun4u sparc SUNW,Ultra-2 Installed by: configure; make; make install BTW, I recognize with joy the moves towards BLAS et al. done since 3-44. Great! Johannes From majordom@nokomis.stat.umn.edu Fri Jan 31 10:42:08 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA12430; Fri, 31 Jan 1997 10:42:08 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA12425; Fri, 31 Jan 1997 10:41:59 -0600 From: Luke Tierney Message-Id: <9701311641.AA12425@nokomis.stat.umn.edu> Subject: Re: xlispstat-3-48: rank command? To: jmu@hlrz9.hlrz.kfa-juelich.de Date: Fri, 31 Jan 1997 10:41:58 -0600 (CST) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: <199701311343.OAA25525@hlrz4.hlrz.kfa-juelich.de> from "Johannes Mueller-Gerking" at Jan 31, 97 02:43:46 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 781 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > Hi all, > > I found a minor problem with the new 3-48 version of > xlispstat: > > > (rank '(9 8 7 6 5)) > Error: The function LORDER is unbound. > Happened in: # > I did say it was an alpha level release :-). Here is a fix: If you are compiling from the sources change the last line in stats.lsp from (defun rank (x) (xlisp::make-compound x (lorder (lorder x)))) to (defun rank (x) (xlisp::make-compound x (order (order x)))) On the mac or windows versions, the easiest fix is to Make a backup copy of xlisp.wks -- just to be safe Then in xlispstat do (defun rank (x) (xlisp::make-compound x (order (order x)))) (save-workspace "xlisp.wks") Copy the new workspace to the startup direcory if it isn't already there. luke From majordom@nokomis.stat.umn.edu Sat Feb 1 14:30:45 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA25356; Sat, 1 Feb 1997 14:30:45 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA25351; Sat, 1 Feb 1997 14:30:40 -0600 From: Luke Tierney Message-Id: <9702012030.AA25351@nokomis.stat.umn.edu> Subject: ramblings on plug-ins To: stat-lisp-news@nokomis.stat.umn.edu Date: Sat, 1 Feb 1997 14:30:40 -0600 (CST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 10124 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk Anthony Rossini raised the possibility of making a Netscape plug-in for XLISP-STAT or R. I've been thinking about the issues for a while, so I thought I'd write down some notes and send them out for reactions. This is kind of long and rambling - you know what key to hit if you're not interested :-). A couple of preliminaries. There is a book on plug-ins available, Netscape Developer's Guide to Plug-Ins, 1/e Doug Young, Mountain View, California http://www.prenhall.com/allbooks/ptr_0132709929.html It seems pretty decent. A plug-in that may be useful as model is a Tcl/Tk plug-in, http://www.sunlabs.com/research/tcl/plugin/index.html There are also two plug-in's available for Python (http://www.python.org). The Oberon Juice system (http://www.ics.uci.edu/~juice/) is also based on a plug-in. There are probably a number of others that can serve as models -- if you know of any I'd like to hear about them. There are two choices that need to be made: Where do the graphs get presented (and textual input/output)? 1) In separate windows, more or less as they would be by a helper application. 2) In a subregion of the browser assigned by the EMBED tag. Where does the code execute? A) In a separate process. B) In the browser process. All four combinations are possible (as are various intermediate positions, but I'll stick to these four). 2B is the standard design for a plug-in, and is the hardest to implement. 1A is most un-plugin-like, but easiest to implement, so let's start there. 1A looks very much like using XLISP-STAT or R as a helper application, but there is an important difference: Plug-in's can allow themselves to be controlled from Java and JavaScript, so this allows a Java/SavaScript page to control the statistical system. The simplest version of this would be very generic -- just start up the application and control it with some interprocess communication mechanism -- AppleEvents on Mac, DDE or OLE Automation on Windows, X IPC or sockets or some such on UNIX. This would require developing a protocol and adding a little code to the application, but is fairly generic. The same plug-in could be made to handle both XLISP-STAT and R, and lots of other things. This is so generic, that there probably is a plug-in available already that does this. The Hamlet system (http://www.cs.utah.edu:80/~hamlet/hamlet/hamlet/hamlet.html) has some of these ideas based just on an intermediate helper application -- a starting point might be to make a plug-in that allows Java control of Hamlet. A slightly fancier version of 1A would be to allow the plug-in to provide a listener window embedded in the browser that can send commands to the statistical application and display output. Commands could also be disabled and the window used only for output. The code for this could be part of the plug-in (which means it would have to be done separately for each platform) or it could be done portably in Java. Again this is generic enough that a solution may already exist, and if not, at least the solution can be made to work for XLISP-STAT, R, and lots of other systems. 2A and 2B, where at least the graphics appear in the browser window, require more thought. There two issues: design and implementation. The design issue involves matching up the user interface model of the statistical application with the model of the browser. For Tcl/Tk this is fairly easy: Tk has essentially a one-window view of the world. Usually that window is a top-level window managed by a window manager, but having it embedded in another window is not a big step. R currently has a two-window view: a text window and a graphics window. Given a rectangular region in a browser, one could allow the region to be split into a text pane and a graphics pane by the user or the program. XLISP-STAT is harder. It currently wants to open a new window for everything it does. Translating that directly into a subregion of a browser means implementing a little window manager, in order not to be too messy. Restricting the plug-in to just one window eliminates plot linking. It might be possible to allow multiple plug-in instances (multiple EMBED tags in a page) to correspond to the same XLISP-STAT process in order to support linking -- that is not in the spirit of the Netscape plug-in model and may or may not be hard to do -- I'm not sure. If XLISP-STAT allowed multiple graphs along with dialogs to be placed in a single top-level window, then a single window restriction would work -- but maybe that isn't all that different from having a tiled mini window manager. This aspect needs to be thougt through very carefully before jumping into the implementation phase. For 2A the implementation should be much easier for R -- "all" one has to do is implement a device driver. Again it could be done natively in the plug-in for each platform, or the plug-in could simply be a conduit to a Java device. I am fairly sure John Chambers and Mark Hansen have already written a Java device driver of some sort for S. In XLISP-STAT it involves essentially the equivalent of a port to a new window system. Again it might make sense to do the core in Java so it is portable, but performance may be a problem. Some performance-critical parts could be identified and be implemented as native methods by the plug-in. 2B is the standard model for a plug-in. Following this approach allows the plug-in to use more standard installation methods and eliminates the need to use inter-process communication. It does potentially involve much more work than any other approach. At a minimum it involves making the core of the base system into a library, which is not too hard. To be effective, it probably also requires tuning the system to use less memory than the stand-alone systems do -- that is more work. It would also require a careful examination of how resources are released. For a stand-alone application this is not an issue since the system releases all resources when the program exits (at least it is supposed to). But a plug-in might get loaded and unloaded many times within a session -- if it does not properly release its resources when it is unloaded then this will eventually bring down the browser. One issue that affects all approaches has to do with the relationship between different plug-in instances active at the same time. The simplest implementation would have all plugin instances run in the same process, which currently means they would share all global variables. This is not ideal since two plug-ins on a page could be written by two different people and hence interfere with each other. For example, each might define a function of the same name that does different things. If XLISP-STAT or R is in a different process, then at least on UNIX and Win32 it is possible (though wasteful) to avoid this problem by running a separate process for each plug-in instance. If the statistical system is run inside the browser process, then separation is not easy. Tcl/Tk supports the idea of separate interpreters with their own global worlds running in the same process, so this can be handled fairly easily. XLISP-STAT currently uses masses of global variables internally -- I think R does a bit of this too. These would have to be moved into a single structure which can then be replicated (at least the modifiable parts) for each plug-in instance. With luck this could perhaps be accomplished "just" by writing the right set of preprocessor macros, but it would take some serious effort. On the other hand, it may also be useful at times to allow different instances -- two plots on the same page -- to run in the same global space in order to support linking. Another issue is security. Automatically loading and running code from who-knows-where is clearly dangerous -- Java has spent a lot of time worrying about this. The plug-in should provide at least two levels of protection. It should allow users to select the sites from which plug-in's (either particular ones or any one) will be accepted. It should also allow the resources a plug-in can access to be controlled. So for example a plug-n from the local machine might have unlimited access, but one from outside would not be allowed to use local files in any way. A lot has been written on this for Java (http://www.yahoo.com/Computers_and_Internet/Programming_Languages/Java/Security/); the Tcl plug-in developers have also thought about this quite a bit (http://www.sunlabs.com/research/tcl/plugin/safetcl.html). Another idea worth considering is signing plug-in code so that code with a trusted creator can be loaded even if it is at an unknown site. The MMM browser (http://pauillac.inria.fr/~rouaix/mmm/) does something like this; it is probably under discussion elsewhere too. In addition to Netscape plug-ins there are also ActiveX and OpenDoc components. The 2B architecture would be best for supporting all three, though a Netscape component can already be incorporated in ActiveX and OpenDoc (though I don't know if a Java interface is available at this time -- it probably will be before too long). One approach to making a system more easily embeddable is to design it around a small virtual machine and a basic run time system. The current VM is a bit too large and too heavily entwined with the Lisp interpreter. A good virtual machine might be able to run both Lisp and R or S code (maybe also matlab or Gauss code) with the right compiler. With a good design it might be possible to have several different VM/run-time implementations, including perhaps one that uses the Java VM. This could be useful if the JIT compiler technology really takes off. There is already one example of using the Java VM for a Lisp dialect, the Kawa Scheme system (Kawa http://www.copsol.com/kawa/index.html). Those are some of the issues. 1A could be done fairly quickly. 2B would, to make it a good fit, require some internal redesigning. Some of that redesigning is a good idea anyway, like allowing multiple graphs in a window and reorganizing the internal globals, but when it will be done is not clear. luke From majordom@umnstat.stat.umn.edu Sun Feb 2 10:21:11 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06051; Sun, 2 Feb 1997 10:21:11 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06047; Sun, 2 Feb 1997 10:21:05 -0600 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.7.5/10.1) id LAA06646; Sun, 2 Feb 1997 11:21:01 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sun, 2 Feb 1997 11:21:38 -0500 To: Luke Tierney , stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: Re: ramblings on plug-ins Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Luke: I'm very glad to see that you are thinking about plug-ins. Clearly, there is a lot to be gained from doing something along these lines. The primative helper approach that I'm using for teaching stat with the web and XLISP-STAT/ViSta could be greatly improved by having plug-in capability available, whether or not graphical windows worked inside the browser. Recent activity in the news group makes it look like there are several others of us that would also be interested. I have two points in response to your ramblings: 1) Wouldn't it be desireable to adopt a two-step strategy? a) First, implement the "quick and dirty" 1A approach b) Second, implement the "slow but elegant" 2B approach? Your ramblings make it sound like 1A could be implemented without much effort in a relatively short time, with a very high probability of useful results. So... I think you should "go for 1A". Then, with that experience, and with additional time to think out the issues underlying what to do about graphical windows, you could decide whether it would be worth the effort to attack the 2B approach. 2) I am, as you can imagine, very interested in the issue of multiple graphs in a single window, since this would be particularly suited to portions of ViSta. If I had such a capability, I would want to be able to have multiple windows that had multiple graph capability [I don't want much, do I ;-) ]. The main ViSta GUI consists of 5 graphical windows: WorkMap, GuideMap, Datasheet, and two namelist windows for variable names and observation labels. I wouldn't want to combine these into a single, multiple pane window. However, in addition to the ViSta GUI, there are numerous spreadplots that present visualizations of data and analyses. Each spreadplot consists of multiple graphs that are presented in multiple windows. It would definitely be better if they were presented as panes in a single window. There are about a dozen different spreadplots, so I would want about a dozen different multiple graph windows. Thus, what I would like is to be able to have many different windows, some multi-pane and some not! I wouldn't want to be forced to have all graphs in the same window. -=-=-=-=-=-= It seems to me that there are two separate issues here: One has to do with how to have multi-pane windows in XLISP-STAT (ignoring questions of the web). The other has to do with how multiple windows are interface to the web (whether or not the window is a multi-pane XLS graphic). Just looking at XLISP-STAT as a stand-alone application, it would be desirable to rearrange the way things work so that the XLISP-STAT programmer could create an instance of a, say, multi-pane-graph-window-proto object. This object would then be identified as the window where the instance of a graph-proto appears, being able to identify what portion of the multi-pane-graph-window-proto is the viewport for the instance of the graph-proto. If this were done within XLISP-STAT it would be independent of any thought about the web. Then there would have to be a second step to work out how the plug-in would handle many different graphics windows some of which might contain multiple graphics. However, the plug-in wouldn't have to know if a graphics window contains one or several graphics. -=-=-=-=- My vote would be for your suggestion to implement: >A slightly fancier version of 1A would be to allow the plug-in to >provide a listener window embedded in the browser that can send >commands to the statistical application and display output. Graphics windows would be >In separate windows, more or less as they would be >by a helper application. If you independently gave XLISP-STAT the multi-pane graphics capabilities, this would work with the above. Then, after doing the above, decide whether to do the 2B approach. ----- Forrest Young Professor, Psychometrics CB# 3270 Davie Hall email: mailto://forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software and Documentation sites: http://forrest.psych.unc.edu/research/ViSta.html USA http://www.psych.unc.edu/fwy/research/ViSta.html USA http://sunsite.univie.ac.at/ViSta/ViSta.html Austria http://micros.hensa.ac.uk/mirrors/vista/ViSta.html United Kingdom From majordom@umnstat.stat.umn.edu Sun Feb 2 12:53:30 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06675; Sun, 2 Feb 1997 12:53:30 -0600 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06671; Sun, 2 Feb 1997 12:53:24 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id KAA27195 for stat-lisp-news@umnstat.stat.umn.edu; Sun, 2 Feb 1997 10:53:23 -0800 (PST) From: Jan Deleeuw Message-Id: <199702021853.KAA27195@galton.math.ucla.edu> Subject: XLS on the WWW To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Sun, 2 Feb 1997 10:53:22 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Excellent ramblings ! Some quick reactions: (a) Netscape is rapidly losing market share, and it may be in trouble because it is in direct competition with Microsoft. Although IE and Cyberdog support Netcape plugins, this may present a problem in the longer run. (b) It seems that Java is not in direct competition with anything produced by Microsoft. Also, IBM and certainly Apple are firmly committed to Java. It seems to me that JIT compilers are a fairly safe bet as well. (c) Plugins need to be installed, and Netscape must be configured to recognize the corresponding MIME. Many people (students) are not able to install anything, unless it comes with an installer. Because of (a) and (b) and (c) I would rather bet on Java than on plug-ins. Because of (c) I would rather bet on the server-side than on the client-side for many instructional applications. One possibility, not mentioned by Luke, is to use CGI in the way php and metahtml do it, i.e. you write Xlisp code in HTML comments directly on the page, and it gets replaced by the Xlisp-Stat output. This approach (not unlike the UCLA WWW interface to Xlisp-Stat) does not do graphics, although the gnuplot type of graphics that now writes tektronix could easily be adapted to write gif. Another possibility, also server-side, is to make an Apache module (as done for php, and even for perl). The system is then compiled into the HTTP server, and CGI is no longer needed. Of course this does not exclude the client side, but for most of our work at the moment (writing a textbook with demos and calculators) leans to the server side. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Sun Feb 2 13:47:39 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06867; Sun, 2 Feb 1997 13:47:39 -0600 Return-Path: Received: from sulu.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06863; Sun, 2 Feb 1997 13:47:34 -0600 Received: from chekov by sulu.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id LAA02191; Sun, 2 Feb 1997 11:47:29 -0800 (PST) Date: Sun, 2 Feb 1997 11:47:28 -0800 (PST) From: Thomas Lumley X-Sender: thomas@chekov To: Xlisp List Subject: Re: XLS on the WWW In-Reply-To: <199702021853.KAA27195@galton.math.ucla.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Sun, 2 Feb 1997, Jan Deleeuw wrote: > Because of [installation difficulties] > I would rather bet on the server-side than on the > client-side for many instructional applications. > > One possibility, not mentioned by Luke, is to use CGI in the > way php and metahtml do it, i.e. you write Xlisp code in HTML > comments directly on the page, and it gets replaced by the > Xlisp-Stat output. This approach (not unlike the UCLA WWW interface > to Xlisp-Stat) does not do graphics, although the gnuplot type > of graphics that now writes tektronix could easily be adapted > to write gif. One problem with this approach, and to some extent with server-side in general is that all these processes can bog down your server really quickly. C[++] CGI programs aren't too bad, but if you are spawning an XLISP-Stat and a gnuplot as well as a shell or whatever is needed to glue it together, you can paralyse quite powerful machines pretty fast. R is even slower to load. It may be possible to reduce the overhead by having the statistics package running all the time and just spawning little C++ processes to feed commains to it -- I believe this is done for perl and databases by some web servers. Even so, the processing requirements will be greater than for serving HTML or JAVA programs. Server-side graphics (especially dynamic ones) will also be very slow for people communicating by modem (the same people who may have difficulty installing a plug-in, since experts will presumably be available to do it for networked university machines). I remember trying to use a server-side demonstration of Penrose tiling from the wrong side of the Pacific. The Pacific link is rather faster than your average modem, but it was still painfully slow. While server-side has lots of advantages I think it would be really useful to have a technology that doesn't require so much from the Web publisher. A major project such as the UCLA textbook can justify this sort of processing power but it might easily be out of the range of a department producing web-based course materials. thomas lumley From majordom@nokomis.stat.umn.edu Sun Feb 2 14:16:18 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA08480; Sun, 2 Feb 1997 14:16:18 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA08475; Sun, 2 Feb 1997 14:16:12 -0600 From: Luke Tierney Message-Id: <9702022016.AA08475@nokomis.stat.umn.edu> Subject: Re: XLS on the WWW To: deleeuw@math.ucla.edu (Jan Deleeuw) Date: Sun, 2 Feb 1997 14:16:12 -0600 (CST) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: <199702021853.KAA27195@galton.math.ucla.edu> from "Jan Deleeuw" at Feb 2, 97 10:53:22 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 2292 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > (a) Netscape is rapidly losing market share, and it may be in trouble > because it is in direct competition with Microsoft. Although IE and > Cyberdog support Netcape plugins, this may present a problem in the > longer run. May well be, but the API might survive, or not. But components of one sort or another are clearly here to stay. It would be a good idea to keep the structure of ActiveX controls and OpenDoc parts in mind while doing anything for Netscape. > > (b) It seems that Java is not in direct competition with anything > produced by Microsoft. Also, IBM and certainly Apple are firmly > committed to Java. It seems to me that JIT compilers are a fairly > safe bet as well. Speed may be a problem, also it would be nice not to have to translate lots of existing C code (though some form of C2J is either out already or around the corner). For stand-alone's you can use native methods to do hook in C code. For applets it is officialy not possible to use native methods. But in fact the Plug-in/LiveWire combination lets you use a hidden plug-in as a library of native methods (essentially at least). At the moment this is the only method for doing this I am aware of. There may be others, and no doubt there eventually will be -- moving from Netscape to BrandX should not be too hard if things are implementad with this posibility in mind. The inner classes to be included in Java 1.1 should make mapping high-level languages to Java easier than it is in 1.02. > > (c) Plugins need to be installed, and Netscape must be configured > to recognize the corresponding MIME. Many people (students) are not > able to install anything, unless it comes with an installer. > I haven't looked at the scriptability of Netscape recently, but I suspect it is actually possible to build an installer that does the right thing. For what it is worth, the intro PolySci class here at Mn is making videatapes of lectures available on the web in a form that requires the VIVO plug-in. I can ask at the end of the quarter how that went. > > Because of (c) I would rather bet on the server-side than on the > client-side for many instructional applications. > Designing any graphics interface/device driver to support server-side applications as well is certainly a good idea. luke From majordom@umnstat.stat.umn.edu Sun Feb 2 14:19:19 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06993; Sun, 2 Feb 1997 14:19:19 -0600 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06989; Sun, 2 Feb 1997 14:19:16 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id MAA27452 for stat-lisp-news@umnstat.stat.umn.edu; Sun, 2 Feb 1997 12:19:15 -0800 (PST) From: Jan Deleeuw Message-Id: <199702022019.MAA27452@galton.math.ucla.edu> Subject: Re: XLS on the WWW To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Sun, 2 Feb 1997 12:19:15 -0800 (PST) In-Reply-To: <9702022016.AA08475@nokomis.stat.umn.edu> from "Luke Tierney" at Feb 2, 97 02:16:12 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk > > > (a) Netscape is rapidly losing market share, and it may be in trouble > > because it is in direct competition with Microsoft. Although IE and > > Cyberdog support Netcape plugins, this may present a problem in the > > longer run. > > May well be, but the API might survive, or not. But components of one > sort or another are clearly here to stay. It would be a good idea to > keep the structure of ActiveX controls and OpenDoc parts in mind while > doing anything for Netscape. > In this connection, it is useful to remember that OpenDoc will be completely rewritten in Java. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Tue Feb 4 11:17:15 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA07594; Tue, 4 Feb 1997 11:17:15 -0600 Return-Path: Received: from suntan.ccs.yorku.ca by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA07590; Tue, 4 Feb 1997 11:17:05 -0600 Received: from nexus.yorku.ca (nexus.yorku.ca [130.63.236.20]) by suntan.ccs.yorku.ca (8.8.4/8.6.11) with SMTP id MAA14568 for ; Tue, 4 Feb 1997 12:17:04 -0500 (EST) Received: from hotspur.psych.yorku.ca (hotspur.psych.yorku.ca [130.63.134.26]) by nexus.yorku.ca (8.6.11/8.6.11) with SMTP id MAA28935 for ; Tue, 4 Feb 1997 12:17:04 -0500 Received: by hotspur.psych.yorku.ca (NX5.67c/NX3.0S) id AA06979; Tue, 4 Feb 97 12:17:03 -0500 Date: Tue, 4 Feb 97 12:17:03 -0500 From: Michael Friendly Message-Id: <9702041717.AA06979@hotspur.psych.yorku.ca> Received: by NeXT.Mailer (1.87.1) Received: by NeXT Mailer (1.87.1) To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Subject: Re: XLS on the WWW Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk A lot of the recent discussion about how to serve .lsp files on the web, security problems, and choice of mime type reinstates the same things I grappled with over SAS. Rather than forcing recipients choose between having *all* .sas files viewed as plain/text or *all* executed locally (opening themselves up to security problems), one alternative is use a cgi program that would display the file as text first, then let the user decide to run it. I sketched this and related ideas in http://www.math.yorku.ca/SCS/Online/sasweb/ which may be of use in the present discussion. -Michael --- Michael Friendly Internet: friendly@hotspur.psych.yorku.ca (NeXTmail OK) Psychology Dept York University Voice: 416 736-5118 Fax: 416 736-5814 4700 Keele Street http://www.math.yorku.ca/SCS/friendly.html Toronto, ONT M3J 1P3 CANADA From majordom@umnstat.stat.umn.edu Wed Feb 5 08:20:19 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22085; Wed, 5 Feb 1997 08:20:19 -0600 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22081; Wed, 5 Feb 1997 08:20:13 -0600 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA03254 (5.67b/IDA-1.5 for ); Wed, 5 Feb 1997 09:19:16 -0500 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA364 for ; Wed, 5 Feb 1997 09:18:35 -0400 Message-Id: <32F8C1D4.740F@kodak.com> Date: Wed, 05 Feb 1997 09:22:28 -0800 From: Jeff Morris Reply-To: morris@Kodak.COM Organization: Eastman Kodak Company X-Mailer: Mozilla 3.0 (Win16; I) Mime-Version: 1.0 To: xlispstatnews Subject: Naming one-way levels Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'm trying to replace the Group 0, Group 1, etc designations on the one-way output with names from a list. It's seems to me that I should just be able to use :predictor-names, but I can't seem to get it to work. I've searched The text but couldn't find the part that I was sure contained some information on this. Thanks as always!! Jeff -- Jeffrey D. Morris, Ph.D. Health & Environment Labs Eastman Kodak Company morris@kodak.com From majordom@nokomis.stat.umn.edu Wed Feb 5 22:39:18 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA18668; Wed, 5 Feb 1997 22:39:18 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA18663; Wed, 5 Feb 1997 22:39:13 -0600 From: Luke Tierney Message-Id: <9702060439.AA18663@nokomis.stat.umn.edu> Subject: problems with 3.48? To: stat-lisp-news@nokomis.stat.umn.edu Date: Wed, 5 Feb 1997 22:39:13 -0600 (CST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 237 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk If you have tried release 3.48 and have run into any problems please let me know. I have addressed one reported problem and, unless others are reported, I hope to put the new version up as the main release early next week. Thanks luke From majordom@umnstat.stat.umn.edu Fri Feb 7 02:25:19 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24111; Fri, 7 Feb 1997 02:25:19 -0600 Return-Path: Received: from kasuga.cc.chuo-u.ac.jp by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24103; Fri, 7 Feb 1997 02:25:08 -0600 Received: from isg.indsys.chuo-u.ac.jp (isg.indsys.chuo-u.ac.jp [133.91.120.136]) by kasuga.cc.chuo-u.ac.jp (8.7.6+2.6Wbeta7/3.4W4) with SMTP id RAA25730; Fri, 7 Feb 1997 17:25:36 +0900 (JST) Received: from [202.213.132.51] by isg.indsys.chuo-u.ac.jp (4.1/6.4J.6-fps0) id AA25236; Fri, 7 Feb 97 17:29:13 JST Date: Fri, 7 Feb 97 17:29:11 JST Message-Id: <9702070829.AA25236@isg.indsys.chuo-u.ac.jp> To: stat-lisp-news@umnstat.stat.umn.edu From: kamakura@indsys.chuo-u.ac.jp (Toshinari Kamakura) X-Sender: BY9T-KMKR@po.asahi-net.or.jp Subject: Re: problems with 3.48? Cc: Luke Tierney Mime-Version: 1.0 Content-Type: text/plain; charset=iso-2022-jp X-Mailer: Eudora-J(1.3.8.5-J13) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, New version is nice and seems to be approximatetly 10% faster than the previous one for numerical calculation. But the bug of 'free-color' command is still observed on the mac version. 'free-color' command does not work properly except the case of draggin the graphic window by the mouse, which I previously reported. Thanks Toshinari Kamakura From majordom@umnstat.stat.umn.edu Fri Feb 7 11:27:33 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26204; Fri, 7 Feb 1997 11:27:33 -0600 Return-Path: Received: from COMPSTAT.WHARTON.UPENN.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26199; Fri, 7 Feb 1997 11:27:23 -0600 Received: (from bob@localhost) by compstat.wharton.upenn.edu (8.7.1/8.7.1) id MAA13263; Fri, 7 Feb 1997 12:26:36 -0500 (EST) From: "Robert A. Stine" Message-Id: <199702071726.MAA13263@compstat.wharton.upenn.edu> Subject: Shared libraries for Power Mac To: stat-lisp-news@umnstat.stat.umn.edu Date: Fri, 7 Feb 1997 12:26:36 -0500 (EST) X-Mailer: ELM [version 2.4 PL24 PGP3 *ALPHA*] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi all, I have assembled a version (modifying verion 3.48) of XLispStat for Mac users that lets you use shared libraries rather than code resources (as used to import C functions, for example). Shared libraries allow you to avoid the special headers needed to work with code resources, making the C code generic enough to work with Unix systems as well. You can also have multiple entry points. There's a web page at http://compstat.wharton.upenn.edu:8001/~bob/dynload.html that gives more details. Bob -------------------------------------------------------------------- Robert A. Stine (phone) 215-898-3114 Department of Statistics ( fax ) 215-898-1280 3620 Locust Walk, Steinburg Hall-Dietrich Hall The Wharton School, University of Pennsylvania Philadelphia, PA 19104-6302 USA stine@stat.wharton.upenn.edu http://compstat.wharton.upenn.edu:8001/~bob/ From majordom@umnstat.stat.umn.edu Fri Feb 7 11:55:12 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26475; Fri, 7 Feb 1997 11:55:12 -0600 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26471; Fri, 7 Feb 1997 11:55:07 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id JAA27630; Fri, 7 Feb 1997 09:55:04 -0800 (PST) From: Jan Deleeuw Message-Id: <199702071755.JAA27630@galton.math.ucla.edu> Subject: Re: Shared libraries for Power Mac To: bob@compstat.wharton.upenn.edu (Robert A. Stine) Date: Fri, 7 Feb 1997 09:55:04 -0800 (PST) Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <199702071726.MAA13263@compstat.wharton.upenn.edu> from "Robert A. Stine" at Feb 7, 97 12:26:36 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'll put the links in the archive, and I cant wait to try this. If it does its work well, then this fully deserves to be the standard version, I guess. I hate to bring this up, but does it work with DLL as well, for those who live on the other side of town ?> > Hi all, > > I have assembled a version (modifying verion 3.48) of XLispStat > for Mac users that lets you use shared libraries rather than > code resources (as used to import C functions, for example). > Shared libraries allow you to avoid the special headers needed > to work with code resources, making the C code generic enough > to work with Unix systems as well. You can also have multiple > entry points. > > There's a web page at > > http://compstat.wharton.upenn.edu:8001/~bob/dynload.html > > that gives more details. > > Bob > > -------------------------------------------------------------------- > Robert A. Stine (phone) 215-898-3114 > Department of Statistics ( fax ) 215-898-1280 > 3620 Locust Walk, Steinburg Hall-Dietrich Hall > The Wharton School, University of Pennsylvania > Philadelphia, PA 19104-6302 > USA stine@stat.wharton.upenn.edu > http://compstat.wharton.upenn.edu:8001/~bob/ > -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Fri Feb 7 12:11:53 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26677; Fri, 7 Feb 1997 12:11:53 -0600 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26669; Fri, 7 Feb 1997 12:11:33 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id KAA27789 for stat-lisp-news@umnstat.stat.umn.edu; Fri, 7 Feb 1997 10:11:31 -0800 (PST) From: Jan Deleeuw Message-Id: <199702071811.KAA27789@galton.math.ucla.edu> Subject: concerned ? To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Fri, 7 Feb 1997 10:11:31 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk This is something which sometimes bothers me. There are now many versions of Xlisp(Stat) with a lot of overlap, but with unique features that would be useful to have in all versions. a. Tom Almy's Xlisp-Plus has rationals and bignums. b. Luke's Xlisp-Stat has a byte-compiler, graphics, statstics, linear algebra, and optimization. And Apple Events etc. c. Jeff Protero's Slisp is a library, which means you can use an Xlisp interpreter in your C programs. d. Bob Stine's XLSPPC-SL has support for shared libraries. e. Dave Betz is back with YAXLS (yet another Xlisp 3.0) -- I haven't looked at it yet. It would be nice to have a choice to compile some of these features in conditionally. I dont like to have all these versions. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@nokomis.stat.umn.edu Fri Feb 7 12:50:28 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA15366; Fri, 7 Feb 1997 12:50:28 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA15360; Fri, 7 Feb 1997 12:50:20 -0600 From: Luke Tierney Message-Id: <9702071850.AA15360@nokomis.stat.umn.edu> Subject: Re: concerned ? To: deleeuw@math.ucla.edu (Jan Deleeuw) Date: Fri, 7 Feb 1997 12:50:20 -0600 (CST) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: <199702071811.KAA27789@galton.math.ucla.edu> from "Jan Deleeuw" at Feb 7, 97 10:11:31 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1001 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > a. Tom Almy's Xlisp-Plus has rationals and bignums. Actually, 3.48 has these too, they are just slightly hidden because / and expt produce floats. For now there are two back doors: divide produces rationals on integer division iexpt produces rationals for integer powers and setting the variable *standard-division* to true makes /, expt, and ^ behave this way also. Both these measures are only temporary -- once I move the statistical stuff into a package the standard (and non-vectorized) behavior will be available from the common-lisp package. A couple of examples: XLISP-PLUS version 3.0 Portions Copyright (c) 1988, by David Betz. Modified by Thomas Almy and others. XLISP-STAT Release 3.48 (Beta). Copyright (c) 1989-1994, by Luke Tierney. (expt> 2 -1) 0.5 > (iexpt 2 -1) 1/2 > (divide 1 2) 1/2 > *standard-division* NIL > (setf *standard-division* t) T > (/ 1 2) 1/2 > (expt 2 -1) 1/2 > (* 100000000 100000000000) 10000000000000000000 > (type-of 10000000000000000000) BIGNUM luke From majordom@umnstat.stat.umn.edu Fri Feb 7 13:38:12 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27488; Fri, 7 Feb 1997 13:38:12 -0600 Return-Path: Received: from stieltjes.smc.univie.ac.at by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27484; Fri, 7 Feb 1997 13:38:05 -0600 Received: from polya.univie.ac.at (plato-slip-157.univie.ac.at [131.130.99.157]) by stieltjes.smc.univie.ac.at (8.7.5/8.6.12) with SMTP id UAA19399; Fri, 7 Feb 1997 20:39:53 +0100 (MET) Date: Fri, 7 Feb 97 20:25:51 W. Europe Standard Time From: Erich Neuwirth Subject: xlispstat mirror To: stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Chameleon ATX 6.0, Standards Based IntraNet Solutions, NetManage Inc. X-Face: Ww0y#m#gfCw8pV9jF$;}^l_CqkSs+!2YJ[yJyyb-1aE?tp~5|GtEti,{1f~,Pt?)qcVs2-UjY#WCI6a3t)LZ=qpl>M8b{%j5NTX8l-nhJ&s+^a X-Priority: 3 (Normal) Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk i just want you to know that we are mirroring the umnstat xlispstat repository here in vienna at the url http://sunsite.univie.ac.at/pub/statistics/ucla/xlispstat/ umnstat.stat.umn.edu/ thanks to a generously donated server from sun microsystems sunsite vienna at http://sunsite.univie.ac.at/ is dedicated to computer support for mathematics, statistics, and science education for people in europe getting xlispstat realted stuff is probably easier to get from our site the from across the atlantic. we also mirror vista and a growing collection of math and stats software -- Erich Neuwirth Computer Supported Didactics Working Group, University of Vienna PGP-key at http://www.smc.univie.ac.at/PGP-Keys/neuwirth.asc Visit my Spreadsheet home page at http://sunsite.univie.ac.at/Spreadsite From majordom@umnstat.stat.umn.edu Fri Feb 7 15:04:48 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28255; Fri, 7 Feb 1997 15:04:48 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28246; Fri, 7 Feb 1997 15:04:40 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa03716; 7 Feb 97 16:04 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA14887; Fri, 7 Feb 1997 16:04:34 -0500 Date: Fri, 7 Feb 1997 16:04:33 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Jan Deleeuw Cc: "Robert A. Stine" , stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Shared libraries for Power Mac In-Reply-To: <199702071755.JAA27630@galton.math.ucla.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I also would like to see this merged into the standard distribution. I just happened to be hacking around with Mac Shared Libraries when I got this note: my interface to our 4pi multichannel analyzer ( which I will make freely available ) will be using shared libraries - I need several entry points to the driver interface. I'ld rather not have to maintain & distribute another version of XlispStat. ( Especially since Luke has fixed all my other complaints - directory function and color restrictions - in the latest alpha. ) I'm trying to figure out if there is a better linkage mechanism than the one currently used. I believe that if some of the xlisp support functions were spun off into a sharable library, then other shared library code could reference those routines and they could all use the same calling conventions. This might be a solution to merging features from various versions back again. However, using XlispStat calling conventions in library code is mutually contradictory with my other goal, which is to be able to use the same library code with XlispStat, Python, Java, [ name your one entry here... ] I don't believe there is a solution to the above short of converting everything to use SOM ( and I'm not sure how difficult that would be! ) but I'm going to think about it a bit more. In the short term, I'm considering putting XlispStat, Python and Java native method wrappers around the same core code in the same library, and use a naming convention for the programs to decide which external to call. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Fri Feb 7 15:36:14 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28679; Fri, 7 Feb 1997 15:36:14 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28670; Fri, 7 Feb 1997 15:36:06 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa17119; 7 Feb 97 16:35 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA17660; Fri, 7 Feb 1997 16:35:51 -0500 Date: Fri, 7 Feb 1997 16:35:50 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: "Robert A. Stine" Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Shared libraries for Power Mac In-Reply-To: <199702071726.MAA13263@compstat.wharton.upenn.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The other advantage of shared libraries is that they can have their own globals, and automatic initialization and termination routines. This is the main reason I chose to use them for this project. To talk to hardware drivers, I need to find which slot the board is in, open the driver and keep a reference number and another bit of info around for the other calls. I don't want the Lisp code to be responsible for keeping this info and passing it on to each call, so the init routine does the default setup and stores the refNo in it's static global. The other routines verify that there is a valid refNo before doing anything. However, it might be worth a namechange to keep "open-resource-file" for something that really open's a resource file. There may be other future needs to be able to load resources. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Fri Feb 7 17:45:40 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29934; Fri, 7 Feb 1997 17:45:40 -0600 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29930; Fri, 7 Feb 1997 17:45:33 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id PAA00444 for stat-lisp-news@umnstat.stat.umn.edu; Fri, 7 Feb 1997 15:45:31 -0800 (PST) From: Jan Deleeuw Message-Id: <199702072345.PAA00444@galton.math.ucla.edu> Subject: Power Calculator To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Fri, 7 Feb 1997 15:45:30 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Jason Bond's Power Calculator is back at http://www.stat.ucla.edu/cgi-bin/textbook/powercalc/ with a new and happy interface. It computes power for given sample size and sample size for given power for normal, binomial, poisson, exponential, and correlation models; both for one-sample and two-sample situations. You click a button, and fill in and submit a form, which returns the answer. In the background a CGI program in perl forks an Xlisp-Stat process on our server -- so you even get your CPU cycles for free. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Sun Feb 9 12:39:00 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24090; Sun, 9 Feb 1997 12:39:00 -0600 Return-Path: Received: from rs6000.cmp.ilstu.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24086; Sun, 9 Feb 1997 12:38:56 -0600 Received: from line8043.ras.ilstu.edu by rs6000.cmp.ilstu.edu (AIX 4.1/UCB 5.64/4.03) id AA123644; Sun, 9 Feb 1997 12:38:50 -0600 Message-Id: <2.2.32.19970209183844.006ce280@rs6000.cmp.ilstu.edu> X-Sender: garyqiyu@rs6000.cmp.ilstu.edu X-Mailer: Windows Eudora Pro Version 2.2 (32) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sun, 09 Feb 1997 12:38:44 -0600 To: stat-lisp-news@umnstat.stat.umn.edu From: "guo (gary) qiyu" Subject: lisp question Cc: Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk > >>Dear Sir.: >> >> > I would like to learn more about lisp. [1]How I can plot several group data on one diagram with several lines matching different group data. [Below is one line matches one group data.] So I can pull each line to get each line equation. [2]For further is there any lisp program, when I change the line equation parameters the position of the line will correspondingly be changed too on the diagram. >> >>I shall appreciate if you can give me some help. >> >> >> >>Gary Guo >> >> > > >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; >;;;; >;;;; A subset of the pull-and-move.lsp program. Shows the projection >;;;; onto a line that the user may move around. The projection is >;;;; based on a specific loss function where the independant variable >;;;; is normally distributed. The regression line may also be inserted. >;;;; >;;;; >;;;; >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; > >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; >;; >;; Prototype Definitions >;; >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; > >(defproto move-line-mix '(plot info reg reg-coefs reg-ss xproj yproj inprod > your-coefs your-rss k-value xx yy xbar ybar > sdx sdy prin-dir)) > >(defproto move-line-proto '(parent) nil scatterplot-proto) >(defproto info-proto '(parent) () graph-window-proto) > >(defun move-line (&optional (x (normal-rand 5)) (y (normal-rand 5))) > (let ((move (send move-line-mix :new x y))) > move)) > >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; >;; >;; :isnew method >;; >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; > >(defmeth move-line-mix :isnew (x y) > (let* ( > (plot (send move-line-proto :new 2 :title "Pull My Line" > :location > #+unix (list 50 50) > #-unix (list 50 50) > )) > (info-window (send info-proto :new :location > #+unix (list 500 50) > #-unix (list 300 50) > :size > #+unix (list 450 150) > #-unix (list 450 150) > )) > (plot-save-image (send menu-item-proto :new "Save to File" > :action #'(lambda () (send plot :project-points) > (send plot :ask-save-image)))) > (info-save-image (send menu-item-proto :new "Save to File" > :action #'(lambda () (send info-window :ask-save-image)))) > (which-type (send menu-item-proto :new "Display Which Projections?" > :action #'(lambda () (get-type self) > (if (send self :reg) > (send self :set-regression)) > (send info-window :redraw) > (send plot :project-points)))) > (plot-menu (send menu-proto :new "Point Pull")) > (info-window-menu (send menu-proto :new "Info Display")) > (xbar (mean x)) > (ybar (mean y)) > (svd (sv-decomp (matrix (list 2 2) > (list (sum (^ (- x xbar) 2)) (sum (* (- x xbar) (- y ybar))) > (sum (* (- x xbar) (- y ybar))) (sum (^ (- y ybar) 2)))))) > (pos (position (max (elt svd 1)) (elt svd 1))) > (prin-dir (elt (column-list (elt svd 2)) pos)) > ) > (send plot :parent self) > (send info-window :parent self) > (send plot :add-points (list x y)) > (send plot :margin 0 25 0 0) > (send plot :add-overlay (send pull-overlay-proto :new)) > (send self :xx (send plot :scaled-range 0)) > (send self :yy (send plot :scaled-range 1)) > (send self :your-rss 0) > (send self :reg-ss 0) > (send plot-menu :append-items which-type plot-save-image) > (send info-window-menu :append-items info-save-image) > (send plot :menu plot-menu) > (send info-window :menu info-window-menu) > (send plot :add-mouse-mode 'pulling :title "Pulling Mode" > :click :where-am-i :cursor 'hand) > (send self :xbar xbar) > (send self :ybar ybar) > (send self :sdx (standard-deviation x)) > (send self :sdy (standard-deviation y)) > (send self :inprod (inner-product (- x xbar) (- y ybar))) > (send self :prin-dir (/ (elt prin-dir 1) (elt prin-dir 0))) > (send plot :adjust-to-data) > (send self :plot plot) > (send self :info info-window) > (send self :begin-characteristics) > (send plot :mouse-mode 'pulling))) > > >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; >;; >;; Line Animation Junk >;; >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; > >(defmeth move-line-proto :where-am-i (x y m1 m2) >(send self :while-button-down > #'(lambda (x y) > (let* ( > (parent (send self :parent)) > (xx (send parent :xx)) > (yy (send parent :yy)) > (cc (send self :canvas-to-scaled x y)) > (bg (list (first xx) (first yy))) > (ed (list (second xx) (second yy))) > (d1 (dist bg cc)) > (d2 (dist ed cc)) > ) > (cond ((< d1 d2) (setf (first xx) (first cc)) > (setf (first yy) (second cc))) > (t (setf (second xx) (first cc)) > (setf (second yy) (second cc)))) > (send self :project-points))))) > > >(defmeth move-line-proto :project-points () >(let* ( > (parent (send self :parent)) > (xx (send parent :xx)) > (yy (send parent :yy)) > (n (send self :num-points)) > (x (send self :point-coordinate 0 (iseq n))) > (y (send self :point-coordinate 1 (iseq n))) > (coefs (send parent :line-calculate xx yy)) > (regtest (send parent :reg)) > ) > (unless regtest (send self :redraw)) > (send parent :calc-params coefs) > (send self :clear-lines) > (when regtest (send self :draw-regline) > (send self :linestart-width (iseq 2) (repeat 2 2)) > (send self :redraw)) > (send self :add-lines xx yy) > (let ((xproj (send parent :xproj)) > (yproj (send parent :yproj))) > (dotimes (i n) > (send self :add-lines (list (elt x i) (elt xproj i)) > (list (elt y i) (elt yproj i)) :type 'dashed))) > (send (send parent :info) :update-info))) > > > >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; >;; >;; Sets initial conditions. Calculates and inserts regression line. >;; >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; > >(defmeth move-line-mix :set-regression () > (send self :calc-params nil) > (send (send self :info) :redraw)) > >(defmeth move-line-mix :begin-characteristics () > (send self :k-value 0) > (send self :calc-params nil) > (let ((xrange (send (send self :plot) :range 0)) > (yrange (send (send self :plot) :range 1)) > (coefs (send self :reg-coefs))) > (send self :xx xrange) > (send self :yy (+ (elt coefs 0) (* (elt coefs 1) xrange))) > (send self :reg nil) > (send (send self :info) :update-info) > (send (send self :plot) :project-points))) > >(defmeth move-line-mix :line-calculate (xx yy) > (let* ( > (m (/ (- (second yy) (first yy)) (- (second xx) (first xx)))) > (a (- (first yy) (* m (first xx)))) > ) > (send self :your-coefs (list a m)) > (list a m) > ) >) > > >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; >;; >;; Calculates Parameters >;; >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; > >(defmeth move-line-mix :calc-params (coefs) > (let* ( > (plot (send self :plot)) > (n (send plot :num-points)) > (x (send plot :point-coordinate 0 (iseq n))) > (y (send plot :point-coordinate 1 (iseq n))) > (projections (send self :find-proj x y coefs)) > ) > (if coefs (progn > (send self :your-coefs coefs) > (send self :xproj (elt (elt projections 1) 0)) > (send self :yproj (elt (elt projections 1) 1)) > (send self :your-rss (elt projections 2))) > (progn > (send self :reg-coefs (elt projections 0)) > (send self :reg-ss (elt projections 2)))))) > >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; >;; >;; Finds Projections >;; K=0 Vertical >;; K=1 Horizontal >;; K=2 Orthogonal >;; >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; > > >(defmeth move-line-mix :find-proj (x y coefs) > (let* ( > (n (length x)) > (k (send self :k-value)) > (xbar (send self :xbar)) > (ybar (send self :ybar)) > (sdx (send self :sdx)) > (sdy (send self :sdy)) > (inprod (send self :inprod)) > (sgn (if (< inprod 0) -1 1)) > (beta (if coefs (elt coefs 1) > (case k (0 (/ inprod (* (1- n) (^ sdx 2)))) > (1 (/ (* (1- n) (^ sdy 2)) inprod)) > (2 (send self :prin-dir))))) > (alpha (if coefs (elt coefs 0) > (case k (0 (- ybar (* beta xbar))) > (1 (* (- (- xbar (* (/ beta) ybar))) beta)) > (2 (- ybar (* beta xbar)))))) > (xproj (case k (0 x) > (1 (- (/ y beta) (/ alpha beta))) > (2 (* (/ beta (1+ (^ beta 2))) > (+ y (/ x beta) (- alpha)))))) > > (yproj (case k (0 (+ alpha (* beta x))) > (1 y) > (2 (+ alpha (* beta xproj))))) > (ss (sum (+ (^ (- x xproj) 2) (^ (- y yproj) 2)))) > ) > (list (list alpha beta) (list xproj yproj) ss))) > > >(defmeth move-line-proto :draw-regline () > (let* ( > (parent (send self :parent)) > (coefs (send parent :reg-coefs)) > (xrange (send self :range 0)) > (yrange (+ (elt coefs 0) (* xrange (elt coefs 1)))) > ) > (send self :add-lines xrange yrange))) > >(defun get-type (move-line) > (let* ( > (val (send move-line :k-value)) > (label (send text-item-proto :new "Which Type of distance?")) > (pick (send choice-item-proto :new (list "Vertical Distances" > "Horizontal Distances" "Orthogonal Distances") > :value val)) > (ok (send modal-button-proto :new "OK" > :action #'(lambda () (setf val (send pick :value))))) > (pick-dialog (send modal-dialog-proto :new (list label pick ok))) > ) > (send pick-dialog :modal-dialog) > (send move-line :k-value val))) > >(defun dist (x y) (sqrt (sum (^ (- x y) 2)))) > > >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; >;; >;; Macro Definitions and slot-makers >;; >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; > >(defmacro normal-assessor (key slot prototype) >`(defmeth ,prototype ,key (&optional (content nil set)) > (when set (setf (slot-value ',slot) content)) > (slot-value ',slot))) > >(normal-assessor :reg reg move-line-mix) >(normal-assessor :pull pull move-line-mix) >(normal-assessor :plot plot move-line-mix) >(normal-assessor :info info move-line-mix) >(normal-assessor :reg-coefs reg-coefs move-line-mix) >(normal-assessor :reg-ss reg-ss move-line-mix) >(normal-assessor :your-coefs your-coefs move-line-mix) >(normal-assessor :your-rss your-rss move-line-mix) >(normal-assessor :xx xx move-line-mix) >(normal-assessor :yy yy move-line-mix) >(normal-assessor :k-value k-value move-line-mix) >(normal-assessor :xproj xproj move-line-mix) >(normal-assessor :yproj yproj move-line-mix) >(normal-assessor :inprod inprod move-line-mix) >(normal-assessor :xbar xbar move-line-mix) >(normal-assessor :ybar ybar move-line-mix) >(normal-assessor :sdx sdx move-line-mix) >(normal-assessor :sdy sdy move-line-mix) >(normal-assessor :prin-dir prin-dir move-line-mix) > >(normal-assessor :parent parent info-proto) >(normal-assessor :parent parent move-line-proto) > > >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; >;; >;; Reg-Line Overlay Prototype >;; >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; > >(defproto pull-overlay-proto nil nil graph-overlay-proto) > >(defmeth pull-overlay-proto :redraw () > (let* ( > (graph (send self :graph)) > (parent (send graph :parent)) > ) > (send self :bitmap (send parent :reg)) > (send graph :draw-string "+/- Reg Line" 25 20) > ) >) > >(defmeth pull-overlay-proto :do-click (x y m1 m2) > (let* ( > (graph (send self :graph)) > (parent (send graph :parent)) > ) > (cond ((and (< 10 x 20) (< 10 y 20)) > (send parent :reg (not (send parent :reg))) > (if (send parent :reg) > (progn (send parent :set-regression) > (send self :bitmap t)) > (send self :bitmap nil)) > (send graph :redraw) > (send (send parent :info) :redraw) > (send graph :project-points))))) > > >(defmeth pull-overlay-proto :bitmap (selected) > (let* ((check-bitmap '#2a((1 1 1 1 1 1 1 1 1 1) > (1 1 0 0 0 0 0 0 1 1) > (1 0 1 0 0 0 0 1 0 1) > (1 0 0 1 0 0 1 0 0 1) > (1 0 0 0 1 1 0 0 0 1) > (1 0 0 0 1 1 0 0 0 1) > (1 0 0 1 0 0 1 0 0 1) > (1 0 1 0 0 0 0 1 0 1) > (1 1 0 0 0 0 0 0 1 1) > (1 1 1 1 1 1 1 1 1 1))) > (blank-bitmap '#2a((1 1 1 1 1 1 1 1 1 1) > (1 1 1 1 1 1 1 1 0 1) > (1 0 0 0 0 0 0 0 0 1) > (1 0 0 0 0 0 0 0 0 1) > (1 0 0 0 0 0 0 0 0 1) > (1 1 1 1 1 1 1 1 1 1))) > (map (if selected check-bitmap blank-bitmap)) > (graph (send self :graph))) > (send graph :draw-bitmap map 10 10))) > > >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; >;; >;; Methods to update info-window >;; >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; > >(defmeth info-proto :update-info () >(let* ( > (parent (send self :parent)) > (regtest (send parent :reg)) > ) > (send self :erase-rect 200 25 250 10) > (send self :erase-rect 100 47 150 10) > (send self :draw-string (format nil "~3,6f + ~3,6fx" > (first (send parent :your-coefs)) > (second (send parent :your-coefs))) 200 35) > (send self :draw-string > (format nil "~3,6f" (send parent :your-rss)) 100 57))) > > > >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; >;; >;; Updates info window >;; >;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; > >(defmeth info-proto :redraw () >(call-next-method) >(send self :erase-window) >(let* ( > (parent (send self :parent)) > (regtest (send parent :reg)) > (type (cond ((= (send parent :k-value) 0) "Vertical") > ((= (send parent :k-value) 1) "Horizontal") > (t "Orthogonal"))) > ) >(when regtest > (send self :draw-string (format nil "Regression Projection: ~a" type) 10 85) > (send self :draw-string (format nil > "The Least Squares Line: ") 10 105) > (send self :draw-string (format nil "~3,6f + ~3,6fx" > (first (send parent :reg-coefs)) > (second (send parent :reg-coefs))) 230 105) > (send self :draw-string (format nil "The Least Squares RSS: ") 10 130) > (send self :draw-string (format nil " ~3,6f" (send parent :reg-ss)) 220 130) > (send self :draw-string (format nil > "_______________") 10 65)) > > (send self :draw-string (format nil "Projection Displayed: ~a" type) 10 15) > (send self :draw-string > (format nil "Your Line Equation: ") 10 35) > (send self :draw-string (format nil "~3,6f + ~3,6fx" > (first (send parent :your-coefs)) > (second (send parent :your-coefs))) 200 35) > (send self :draw-string (format nil "Your RSS: ") 10 57) > (send self :draw-string (format nil "~3,6f" (send parent :your-rss)) > 100 57))) > > > > > > > > > >(defmeth info-proto :close () (exit)) >(defmeth move-line-proto :close () (exit)) > > > > > (def x0 (list 1 4 7 2 9 2 4 1 10 6)) > (def y0 (list 0 -1 3 2 5 5 3 6 14 9)) > >(move-line x0 y0) > > > From majordom@umnstat.stat.umn.edu Mon Feb 10 14:20:18 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10144; Mon, 10 Feb 1997 14:20:18 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10140; Mon, 10 Feb 1997 14:20:12 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.7.5/10.1) id PAA09130; Mon, 10 Feb 1997 15:20:08 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 10 Feb 1997 15:18:58 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: forrest@unc.edu (Forrest Young) Subject: loading files Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi group: I am beginning to prepare XLS/ViSta applets (as my previous mails have indicated) and have run into a problem. Lets say I'm using Netscape and I click on a button that is supposed to get me a file called foo.xli from a server that is configured to serve it up as an application/x-xlispstat mime type file. Lets say that Netscape is configured properly to know what to do with this mime type. Then, the first time I click that button the file comes across as foo.xli and everything works right (i.e., XLS starts and loads that file and does whatever with it). HOWEVER, with Netscape, the second time I click that button the file comes across as foo.xli-1234 or somesuch, and XLS complains that: Error: File needs .lsp or .fsl extension - "foo.xli-1234" I've looked into this a bit more and I discover the following (all on a Mac): 1) If I use the load function, or if I double-click or drag-and-drop foo.xli that's OK --- it doesn't have to be .lsp or .fsl. 2) If I do any of above with a file named foo.xli1234, thats OK too! 3) If I do any of above with a file named foo.dingbat, thats OK too! 4) However, foo (with no extension) doesn't work, giving me the .lsp or .fsl error message. So, clearly the rules aren't what the error message says. What are they? More importantly: How can I get it to be OK with foo.xli-1234 or how can I strip off -1234, or how can I get Netscape to not do this (IE asks if I want to replace the prexisting foo.xli with the second one). --- Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Telephone: 919-962-5038 ViSta Software Sites: http://forrest.psych.unc.edu/research/ViSta.html USA http://www.psych.unc.edu/fwy/research/ViSta.html USA http://sunsite.univie.ac.at/ViSta/ViSta.html Austria http://micros.hensa.ac.uk/mirrors/vista/ViSta.html Great Britain ||| (@ @) -------------------------------------------------------ooO--(_)--Ooo-- From majordom@umnstat.stat.umn.edu Mon Feb 10 19:18:06 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13074; Mon, 10 Feb 1997 19:18:06 -0600 Return-Path: Received: from stat.cs.wvu.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13070; Mon, 10 Feb 1997 19:17:59 -0600 Received: from church.wvu.edu by cs.wvu.edu (SMI-8.6/SMI-SVR4) id UAA07114; Mon, 10 Feb 1997 20:20:46 -0500 Received: from church by church.wvu.edu (SMI-8.6/SMI-SVR4) id UAA12726; Mon, 10 Feb 1997 20:15:32 -0500 Message-Id: <32FFC834.5D1E@cs.wvu.edu> Date: Mon, 10 Feb 1997 20:15:32 -0500 From: "E.James Harner" Organization: West Virginia University X-Mailer: Mozilla 2.01 (X11; I; SunOS 5.5 sun4m) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Cc: ejh@cs.wvu.edu Subject: Re: loading files References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Forrest Young wrote: > > Hi group: > > I am beginning to prepare XLS/ViSta applets (as my previous mails have > indicated) and have run into a problem. > > Lets say I'm using Netscape and I click on a button that is supposed to get > me a file called foo.xli from a server that is configured to serve it up as > an application/x-xlispstat mime type file. > > Lets say that Netscape is configured properly to know what to do with this > mime type. > > Then, the first time I click that button the file comes across as foo.xli > and everything works right (i.e., XLS starts and loads that file and does > whatever with it). > > HOWEVER, with Netscape, the second time I click that button the file comes > across as foo.xli-1234 or somesuch, and XLS complains that: > Error: File needs .lsp or .fsl extension - "foo.xli-1234" We got around this problem by writing an AppleScript to remove the downloaded version from the desktop after it is loaded. I have not advertised that we are developing an introductory stat course based on Lisp-Stat and Java. It is under rapid development since two sections of our Stat 101 (algebra-based) are using the material now. You can see what we are doing if you point to: http://srs.cs.wvu.edu or: http://www.stat.wvu.edu Running Lisp-Stat requires you to download our Lisp-Stat code. We have not changed .lsp to .lxi and probably will not do so until Spring break (the first week in March). Currently, analyses are done in Alpha-Stat and simulations/distributions are done in IntroStat. At this time only the Mac is supported, but this will be expanded within the next several months. The above-mentioned AppleScript is part of the code you will download. Actually The script is the helper app, and it responds to whatever you want., i.e, .lsp or .xli. However, to demo our stuff you will need to use .lsp as an extension (until we cahnge the machines across campus during Spring break). The Script needs to be in the same folder as the XLISP-STAT application and certain scripting extensions (included) need to be in your System folder. Read through the examples to see how to use the software, but a simplified version is on the way. During Spring break we will also replace Alpha-Stat by Omega-Stat a much larger development effort including extensive multivariate analyses. It also include projection tours, dynamic coplots, etc. However, these capabilities will not be included in the initial release for introductory statistics and in fact the underlying structure (a M/V/C paradigm) will be hidden from the students. The multivariate materials will be released during the summer as I development my multivariate course for the Fall. The Modeling components will be released in the Spring when I teach my generalized linear model course. I will provide more details over time. Jim -- E. James Harner Professor of Statistics Dept. of Statistics & Computer Science Phone: 304/293-3607 x3519 P.O. Box 6330 FAX: 304/293-2272 West Virginia University E-mail: ejh@cs.wvu.edu Morgantown, WV 26506 From majordom@umnstat.stat.umn.edu Mon Feb 10 19:49:31 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13197; Mon, 10 Feb 1997 19:49:31 -0600 Return-Path: Received: from stat.cs.wvu.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13193; Mon, 10 Feb 1997 19:49:24 -0600 Received: from church.wvu.edu by cs.wvu.edu (SMI-8.6/SMI-SVR4) id UAA07539; Mon, 10 Feb 1997 20:52:03 -0500 Received: from church by church.wvu.edu (SMI-8.6/SMI-SVR4) id UAA12733; Mon, 10 Feb 1997 20:46:47 -0500 Message-Id: <32FFCF87.5C9A@cs.wvu.edu> Date: Mon, 10 Feb 1997 20:46:47 -0500 From: "E.James Harner" Organization: West Virginia University X-Mailer: Mozilla 2.01 (X11; I; SunOS 5.5 sun4m) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: IntroStat Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk If you try our IntroStat Web pages (the Interactive Workbook is the focal point), please do not submit your answers to our server. Our graders will go crazy. We are writing a server-side Java database which will chech to see whether or not the sender is registered in our courses. If not, you will receive a Web page immediately with your answers and you will receive your answers via e-mail, but you will not have your answers stored on the server nor will you receive grader comments and the correct answers. However, it will take several months to complete this software. We are writing a client-side applet which works with the database to allow persistence of student answers. As of now the student can submit answers more than once prior to the due date, but we only grade the last submission. However, the student must start over again each time since we cannot maintain the answers in a Netscape form. Jim -- E. James Harner Professor of Statistics Dept. of Statistics & Computer Science Phone: 304/293-3607 x3519 P.O. Box 6330 FAX: 304/293-2272 West Virginia University E-mail: ejh@cs.wvu.edu Morgantown, WV 26506 URL: http://www.cs.wvu.edu/~ejh From majordom@nokomis.stat.umn.edu Mon Feb 10 20:17:51 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA27586; Mon, 10 Feb 1997 20:17:51 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA27581; Mon, 10 Feb 1997 20:17:38 -0600 From: Luke Tierney Message-Id: <9702110217.AA27581@nokomis.stat.umn.edu> Subject: Re: loading files To: ejh@stat.cs.wvu.edu (E.James Harner) Date: Mon, 10 Feb 1997 20:17:38 -0600 (CST) Cc: stat-lisp-news@nokomis.stat.umn.edu, ejh@stat.cs.wvu.edu In-Reply-To: <32FFC834.5D1E@cs.wvu.edu> from "E.James Harner" at Feb 10, 97 08:15:32 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 428 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > HOWEVER, with Netscape, the second time I click that button the file comes > > across as foo.xli-1234 or somesuch, and XLS complains that: > > Error: File needs .lsp or .fsl extension - "foo.xli-1234" > The probelm Forrest mentioned is fixed in 3.48 -- 3.49 the rank problem reported to the list fixed and a few other fvery minor adjustments should be the standard release by the end of the week or early next week. luke From majordom@umnstat.stat.umn.edu Fri Feb 14 13:54:08 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14975; Fri, 14 Feb 1997 13:54:08 -0600 Return-Path: Received: from galton.mathstat.uidaho.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14971; Fri, 14 Feb 1997 13:54:02 -0600 Message-Id: <9702141954.AA14971@umnstat.stat.umn.edu> Received: by galton.mathstat.uidaho.edu (1.38.193.4/16.2) id AA07740; Fri, 14 Feb 1997 11:40:40 -0800 From: Steve Lee Subject: Promoting XLSIP-STAT in the State of Idaho To: stat-lisp-news@umnstat.stat.umn.edu Date: Fri, 14 Feb 97 11:40:40 PST Cc: steve@galton.mathstat.uidaho.edu Mailer: Elm [revision: 70.85] Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi all, I would like to promote the use of xlisp-stat for statistics instruction in the State of Idaho here. Recently the Idaho Board of Education is inviting proposals to promote creation and use of alternative methods of instruction: To increase the rate and quality of student learning; To increase faculty productivity; To increase access to educational programs; and To control costs. What I have in mind is to develop an internet-based instruction approach. I've been using Splus and SAS since 1986 and I am not sure how they could be used to do so. I heard people have been using XLISP (and Java) to develop worksheets and applets on the WWW. I would like to hear the pros and cons of using XLISP in statistics instructions. How feasible is putting XLISP on the web? What are the costs? What hardwares are needed to support 1000+ students enrolled in statistics courses per year. Any help would be greatly appreciated and thank you in advance. Stephen Lee Division of Statistics University of Idaho From majordom@umnstat.stat.umn.edu Sat Feb 15 14:22:37 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29414; Sat, 15 Feb 1997 14:22:37 -0600 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29410; Sat, 15 Feb 1997 14:22:31 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id MAA19177; Sat, 15 Feb 1997 12:22:29 -0800 (PST) From: Jan Deleeuw Message-Id: <199702152022.MAA19177@galton.math.ucla.edu> Subject: GLM interface in PHP/FI To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Sat, 15 Feb 1997 12:22:29 -0800 (PST) Cc: statcompute@galton.math.ucla.edu X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I finally had some time to look at PHP, and I wrote a simple interface to the Xlisp-Stat GLM Module. Result is http://www.stat.ucla.edu/calculators/glm/ Computations are in Xlisp-Stat, but CGI is not used, because PHP is build into our ahttpd daemon. Also, integration of PHP and HTML is much tighter. Finally, our PHP has gdbm, msql, postgres, and gd support. All pages ending in php are automatically parsed for php commands -- given this, it seems quite unnecessary to use SSI, CGI, and JavaScript. PHP also has file upload, and I guess animations can be done by client pull/server push. So maybe I dont need Java either. =================== this is index.html ================================ Regression Calculator logo


Regression Calculator

Input or Paste Y and X

Y, on the left, should be a column vector, X, on the right, can be a matrix with more than one entry in each row. Make sure X and Y have the same number of rows, and each row of X has the same number of entries.
YX


What Type of Regression ?Does the Regression have an intercept ?
Yes No



Jan de Leeuw
Email me at: deleeuw@stat.ucla.edu
Document: http://www.stat.ucla.edu
There were visitors since
Last revision: Sat Feb 15 12:02:09 PST 1997
======================= This is glm.php =========================================== Regression Calculator logo


Regression Calculator

Linear Regression Model"; break; case 1; $lf = "poissonreg-model"; Echo "

Poisson Regression Model

"; break; case 2; $lf = "binomialreg-model"; Echo "

Binomial Regression Model

"; break; case 3; $lf = "gammareg-model"; Echo "

Gamma Regression Model

"; break; default; $lf = "regression-model"; break; endswitch; $ap = "/misc2/galton/bin/APPS/xlispstat"; if (intval($intercept) == 1); $ic = ""; else; $ic = ":intercept NIL"; endif; > 1); $xt = $xt + "\(" + $tok + "\)"; endif; $tok = strtok (10); endwhile; Return ($xt + "\)"); ); > "; while ($i < $n - 2) { Echo "%s
" $a[$i]; $i++; } Echo ""; > -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon Feb 17 04:18:10 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23437; Mon, 17 Feb 1997 04:18:10 -0600 Return-Path: Received: from marraco.udl.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23433; Mon, 17 Feb 1997 04:17:46 -0600 Received: from caragol.udl.es ([192.168.1.220]) by marraco.udl.es (8.7.5/8.7.2) with ESMTP id LAA25209 for ; Mon, 17 Feb 1997 11:17:50 -0100 (GMT) Received: from caragol-1878 (caragol-1878 [10.100.7.86]) by caragol.udl.es (8.7.5/8.7.3) with SMTP id LAA00276 for ; Mon, 17 Feb 1997 11:10:24 -0100 (GMT) Date: Mon, 17 Feb 1997 11:10:24 -0100 (GMT) From: Jaume March Message-Id: <199702171210.LAA00276@caragol.udl.es> To: stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Pronto E-Mail [version 2.01] Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello! 1st contact from CATALONIA (spain)- Jaume March I want to post a message asking the followin question: I have a pentium 133 MHz, 32Mb RAM, 200Mb free disck-space and running a Xlisp-stat GEE procedure with matrix 12000 row data x 7 variables (5 within, 1 between and 1 binary responses). And tne trouble is that program shows the message: ":error insufficient node space" The reason for that error, is the machine features? I think there is a very "basic" question, but in the technical reports or web pages on xlisp I can't find any solution myself. Thanks!!!!!!!! Jaume March School of Medicine University of Lleida Lleida, 25198 Jaume.March@cmb.udl.es From majordom@umnstat.stat.umn.edu Mon Feb 17 11:49:45 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25544; Mon, 17 Feb 1997 11:49:45 -0600 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25540; Mon, 17 Feb 1997 11:49:38 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id JAA06888; Mon, 17 Feb 1997 09:49:04 -0800 (PST) From: Jan Deleeuw Message-Id: <199702171749.JAA06888@galton.math.ucla.edu> Subject: Re: your mail To: Jaume.March@cmb.udl.es (Jaume March) Date: Mon, 17 Feb 1997 09:49:03 -0800 (PST) Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <199702171210.LAA00276@caragol.udl.es> from "Jaume March" at Feb 17, 97 11:10:24 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk In The Book, appendix B, section 4. Information about the room and expand functions. --- J. > > Hello! > 1st contact from CATALONIA (spain)- Jaume March > > I want to post a message asking the followin question: > > I have a pentium 133 MHz, 32Mb RAM, 200Mb free disck-space and running a > Xlisp-stat GEE procedure with matrix 12000 row data x 7 variables (5 > within, 1 between and 1 binary responses). > > And tne trouble is that program shows the message: > ":error insufficient node space" > > The reason for that error, is the machine features? > I think there is a very "basic" question, but in the technical reports or > web pages on xlisp I can't find any solution myself. > Thanks!!!!!!!! > > Jaume March > School of Medicine > University of Lleida > Lleida, 25198 > Jaume.March@cmb.udl.es > > > > -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon Feb 17 23:47:28 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29707; Mon, 17 Feb 1997 23:47:28 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29703; Mon, 17 Feb 1997 23:47:22 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa29588; 18 Feb 97 0:47 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA24635; Tue, 18 Feb 1997 00:47:20 -0500 Date: Tue, 18 Feb 1997 00:47:19 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: no multiple-inheritence of hardware objects ? Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Am I correct in the belief that there is no way to combine dialog-proto's and graph-window-proto's due to their both being implemented as hardware-objects ? All of the example code I've seen of graphs with controls build their control widgets "by hand". Is there any way to embed dialog items in a graph-window ? [ The next best solution I can think of is to use pop-up menus triggered from clicking on text in my graph window, and having a menu-item at the end which can call up a model dialog to add new items to the menu. ] ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Tue Feb 18 16:21:16 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12906; Tue, 18 Feb 1997 16:21:16 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12901; Tue, 18 Feb 1997 16:21:09 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa01809; 18 Feb 97 17:21 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA14877; Tue, 18 Feb 1997 17:20:53 -0500 Newsgroups: comp.lang.lisp.x Date: Tue, 18 Feb 1997 17:20:50 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: (get-decoded-time) or equiv. ? Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk There does not seem to be an equivalent of Common Lisp's (GET-DECODED-TIME) in XlispStat/Xlisp-Plus -- only (GET-INTERNAL-REAL-TIME). Does anyone have a function to convert that int value to ascii date/time ( or a list of values as GET-DECODED-TIME returns) on the Mac ? I've got my data-acquisition hooks working and I would like to give my data-acquisition graph windows & data structs default time-stamped names. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Mon Feb 24 12:40:29 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10887; Mon, 24 Feb 1997 12:40:29 -0600 Return-Path: Received: from maui.medctr.luc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10883; Mon, 24 Feb 1997 12:40:23 -0600 Received: (from ael@localhost) by maui.medctr.luc.edu (8.7.3/8.6.9) id MAA19219 for stat-lisp-news@umnstat.stat.umn.edu; Mon, 24 Feb 1997 12:40:03 -0600 (CST) Date: Mon, 24 Feb 1997 12:40:03 -0600 (CST) From: Andy Long Message-Id: <199702241840.MAA19219@maui.medctr.luc.edu> Content-Type: text Apparently-To: Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk stat-lisp-news@umnstat.stat.umn.edu Hi Folks, I'm interested in accessing all previous commands, rather than just *, **, and ***. What's the easiest way of doing this? It looks like I have to change a read-eval-print loop somewhere, but where? Thanks, Andy From majordom@umnstat.stat.umn.edu Mon Feb 24 13:24:17 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11251; Mon, 24 Feb 1997 13:24:17 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11247; Mon, 24 Feb 1997 13:24:08 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa23305; 24 Feb 97 14:24 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA21359; Mon, 24 Feb 1997 14:24:02 -0500 Date: Mon, 24 Feb 1997 14:23:57 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Andy Long Cc: Xlisp List Subject: command recall In-Reply-To: <199702241840.MAA19219@maui.medctr.luc.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Mon, 24 Feb 1997, Andy Long wrote: > > I'm interested in accessing all previous commands, rather than just *, **, and > ***. What's the easiest way of doing this? It looks like I have to change a > read-eval-print loop somewhere, but where? > If you want an unbounded stack of previous commands, it might be better to turn on the transcripting ( via 'dirbble' command or menu ) and write a function to search the transcript file for the command you want. ( Or, you could just run xlisp from emacs or alpha, and use it's recall mechanism. ) You would only need to change the read-eval-print loop if you wanted to add a shortcut for the recall function, like using emacs control chars or arrow keys. ( If you need that, you might consider trying to hook the gnu readline libraries into lisp-stat -- that would probably work on Unix, but I have no idea if it would port to Mac of MSWin. ) ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Mon Feb 24 18:13:46 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13704; Mon, 24 Feb 1997 18:13:46 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13700; Mon, 24 Feb 1997 18:13:39 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa16185; 24 Feb 97 19:13 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA27600; Mon, 24 Feb 1997 19:13:36 -0500 Date: Mon, 24 Feb 1997 19:13:35 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: graphic overlay classes Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Is anyone out there using my xlisp-stat graphic-overlay classes ? I'm considering making some major modifications -- not so much to the functionality, but to the class hierarchy and names. All of the current overlay objects use real-space coordinates. I have need of some other objects -- either using canvas coordinates, or some combination of constraints(*) to position objects, so I was going to factor out some of the common methods and restructure the classes somewhat. ( & how radically may depend on whether anyone else is going to scream! ;-) I now need to add some more complex annotation capability, and it makes sense to do most of this in canvas coordinates, rather than real. ( Specifically, what I need to add are text labels in rectangle boxes which are modifiable -- clicking in the box will bring down a popup menu of alternative tags, double clicking will bring up a dialog box -- unless I find a way to add dialog controls to a graph window, which would be a better way. ) [ (*) Text labels currently use a bastardized combination of real & coordinate space: x,y position is in real graph coordinates, but positive or negative infinity is taken to mean top|right or bottom| left edge of the graph, so I can, for example, position a label over a peak, but at the top or bottom of the graph where it is not likely to be obscured by data points. ] ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Tue Feb 25 19:06:37 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29672; Tue, 25 Feb 1997 19:06:37 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29668; Tue, 25 Feb 1997 19:06:30 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa21718; 25 Feb 97 20:06 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA18871; Wed, 26 Feb 1997 01:06:26 GMT X-Received: from mail.Virginia.EDU by elvis.med.Virginia.EDU (5.65c/1.34) id AA20115; Tue, 25 Feb 1997 20:01:04 -0500 X-Received: from elvis.med.virginia.edu by mail.virginia.edu id aa16480; 25 Feb 97 19:56 EST X-Received: by elvis.med.Virginia.EDU (5.65c/1.34) id AA16241; Tue, 25 Feb 1997 19:56:09 -0500 Date: Tue, 25 Feb 1997 19:56:09 -0500 From: "Steven D. Majewski" Message-Id: <199702260056.AA16241@elvis.med.Virginia.EDU> To: sdm7g@Virginia.EDU Newsgroups: comp.sys.mac.programmer.codewarrior Subject: CW11 internal compiler error: File 'Operands.c' Line: 828 Organization: University of Virginia Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk For several source files ( in xlisp-stat 3.47 ) which all compiled without problem in CW10, I get with CW11 C, the error: Internal compiler error: File 'Operands.c' Line: 828 and a line number from my sources. The same error occurs with several functions and files, and the line marked is always the final bracket after a return statement: } return; ==> } Anyone have a clue ? ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Wed Feb 26 05:25:23 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11156; Wed, 26 Feb 1997 05:25:23 -0600 Return-Path: Received: from melia.qut.edu.au by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11152; Wed, 26 Feb 1997 05:25:16 -0600 Received: from pigeon.qut.edu.au (lunney@pigeon.qut.edu.au) by melia.qut.edu.au (PMDF V5.0-7 #13254) id <01IFVX2H5Q0G00HE2Z@melia.qut.edu.au>; Wed, 26 Feb 1997 21:27:35 +1000 Received: from localhost (lunney@localhost) by pigeon.qut.edu.au (8.8.5/8.8.5) with SMTP id VAA10532; Wed, 26 Feb 1997 21:23:50 +1000 (EST) Date: Wed, 26 Feb 1997 21:23:49 +1000 (EST) From: Paul Lunney Subject: S-plus and xlisp-stat To: s-news@utstat.toronto.edu, stat-lisp-news@umnstat.stat.umn.edu Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Content-Transfer-Encoding: 7BIT Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Could anyone who is familiar with S-plus (Win) and xlisp-stat (Win) give me some idea of what they might use each for? ie what are the strengths and weaknesses for different types of tasks. My impression is that xlisp-stat is superior graphically and may be ultimately more flexible, but that S-plus has got oodles of built-in functions to reduce some of the analytic programing effort. Please tell me if this couldn't be more wrong... With just one fast-disappearing lifetime, I appreciate any insights... Many thanks, Paul Lunney Queensland University of Technology Brisbane Australia From majordom@nokomis.stat.umn.edu Wed Feb 26 10:39:56 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA02829; Wed, 26 Feb 1997 10:39:56 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA02822; Wed, 26 Feb 1997 10:39:51 -0600 From: Luke Tierney Message-Id: <9702261639.AA02822@nokomis.stat.umn.edu> Subject: New alpha release To: stat-lisp-news@nokomis.stat.umn.edu Date: Wed, 26 Feb 1997 10:39:51 -0600 (CST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1329 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk A new alpha-level release of xlispstat, 3.49, is now available from ftp://ftp.stat.umn.edu/pub/xlispstat/Alpha. There were enough little bugs and I did enough tinkering to warrant another alpha release. If this one proves to be stable enough, it will become the primary release. Mostly this release just fixes bugs. There is, however, one substantive change, in reponse to prompting from Steve Majewski and Sandy Weisberg: The protocol used by the standard :redraw-content methods has been augmented to support maintaining a background for plot contents. Instead of erasing the content area before drawing the content, these methods now send the :clear-content message. The default method for this method clears the content area. By overriding this method, you can draw any background you want to maintain. Unless you want to handle clearing the background yourself, you should begin your method with a call to call-next-method. A simple example: The code (setf p (plot-points (normal-rand '(10 10)))) (defmeth p :clear-content () (call-next-method) (send self :frame-rect 50 50 100 100)) sets up a plot that contains a rectangle in its background. This new protocol is still experimental and may change if a better design is found. Luke Tierney School of Statistics Univerity of Minnesota luke@stat.umn.edu From majordom@umnstat.stat.umn.edu Wed Feb 26 10:55:25 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12975; Wed, 26 Feb 1997 10:55:25 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12970; Wed, 26 Feb 1997 10:55:19 -0600 Received: from bootp-17-17.bootp.virginia.edu by mail.virginia.edu id aa10389; 26 Feb 97 11:54 EST Date: 26 Feb 97 11:59:56 +0000 Subject: Re: S-plus and xlisp-stat From: sdm7g@Virginia.EDU To: s-news@utstat.toronto.edu, stat-lisp-news@umnstat.stat.umn.edu Cc: Paul Lunney X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Wed, Feb 26, 1997 11:23 AM, Paul Lunney wrote: > >Could anyone who is familiar with S-plus (Win) and xlisp-stat (Win) give >me some idea of what they might use each for? ie what are the strengths >and weaknesses for different types of tasks. Well, I'm not very familiar with them on Windows, but I've used both. >My impression is that xlisp-stat is superior graphically and may be >ultimately more flexible, but that S-plus has got oodles of built-in >functions to reduce some of the analytic programing effort. Please tell me >if this couldn't be more wrong... I think your impression is basically correct. S-plus is a higher-level, user-oriented language, with a lot of statistical procedures built in -- it's an interactive statictics application. LispStat is more flexible, but not as complete or polished. It's a language for writing interactive statictical applications or for experimenting with new types of interactive statictical graphics. If you want to do something "standard", it's going to be easier in S-plus. If you want to try to do something new or not supported in S-plus, it's probably going to be easier in LispStat. S-plus has several good books about using it for doing statistics. But the documentation on how to really customize it is a bit thin. LispStat documentation is thinner in general, but it's more programmer oriented. Plus, it's Lisp - for which there is plenty of books, tutorials, & documentation. The above is a wild and subjective generalization, but I don't think it's too far off the mark. I did quite a bit of evaluation of them both before launching on a project -- I can't swear that what I wanted to do was impossible to do in S-plus, but if it was possible, there were no examples or documentation to guide me in that sort of low level programming. Some examples of my XlispStat graphics are on: http://galen.med.virginia.edu/~sdm7g/XES/Docs/xesapp.html http://galen.med.virginia.edu/~sdm7g/XES/Docs/jitterplot-examples.html ( I've now got interface code to our spectrometer, so I can graph and analyze data real-time.) ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Wed Feb 26 11:21:52 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13165; Wed, 26 Feb 1997 11:21:52 -0600 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13161; Wed, 26 Feb 1997 11:21:46 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id JAA06289 for stat-lisp-news@umnstat.stat.umn.edu; Wed, 26 Feb 1997 09:21:45 -0800 (PST) From: Jan Deleeuw Message-Id: <199702261721.JAA06289@galton.math.ucla.edu> Subject: CW11 To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Wed, 26 Feb 1997 09:21:44 -0800 (PST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk In article , sdm7g@elvis.med.Virginia.EDU (Steven D. Majewski) wrote: >For several source files ( in xlisp-stat 3.47 ) which all compiled >without problem in CW10, I get with CW11 C, the error: > >Internal compiler error: File 'Operands.c' Line: 828 There is a bug in the C compiler when implicitly converting a float to a double as happens in an assignment or other operations. This causes this internal compiler error. If you use the native doubles instead of floats this should solve your problem. A fix is at hand and should be out any time now. Metrowerks apologizes for this error, Ron -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Wed Feb 26 11:55:07 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13493; Wed, 26 Feb 1997 11:55:07 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13489; Wed, 26 Feb 1997 11:55:00 -0600 Received: from bootp-17-17.bootp.virginia.edu by mail.virginia.edu id aa15958; 26 Feb 97 12:54 EST Date: 26 Feb 97 12:59:50 +0000 Subject: CW11 patches From: sdm7g@Virginia.EDU To: Xlisp List X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I just got a patch emailed from Metrowerks for the CW11 compiler bugs I reported. This is not yet posted on their web site, so if anyone else wants to try it out, let me know and I'll mail you a copy. I'll probably be giving it a test Real Soon, so if you're in now hurry, you can wait for my report. I'm only sending this note now because I see Luke has just released a new alpha 3.49, and someone else may need it right away. - Steve Majewski From majordom@umnstat.stat.umn.edu Wed Feb 26 12:45:02 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14110; Wed, 26 Feb 1997 12:45:02 -0600 Return-Path: Received: from kani.wwa.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14106; Wed, 26 Feb 1997 12:44:58 -0600 Received: from lynd by kani.wwa.com with smtp (Smail3.1.29.1 #2) id m0vzoK2-003oFHC; Wed, 26 Feb 97 12:43 CST Message-Id: <3.0.1.32.19970226124042.00614850@pop.wwa.com> X-Sender: lyndb@pop.wwa.com X-Mailer: Windows Eudora Pro Version 3.0.1 (32) Date: Wed, 26 Feb 1997 12:40:42 -0600 To: Paul Lunney , stat-lisp-news@umnstat.stat.umn.edu From: Lynd Bacon Subject: Re: S-plus and xlisp-stat In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk At 09:23 PM 2/26/97 +1000, Paul Lunney wrote: > >Could anyone who is familiar with S-plus (Win) and xlisp-stat (Win) give >me some idea of what they might use each for? ie what are the strengths >and weaknesses for different types of tasks. > I've been an SPlus user for several years, but have only started working with xlisp-stat recently. What attracted me to xlisp-stat was: (1) the ease with which you can use it to develop and prototype dynamic graphics. (2) Its open and extensible architecture; (3) The price is right; (4) I once took a course in LISP (you don't need to know LISP to use xlisp-stat. I'm just fond of OOP languages!) >My impression is that xlisp-stat is superior graphically and may be >ultimately more flexible, but that S-plus has got oodles of built-in >functions to reduce some of the analytic programing effort. Please tell me >if this couldn't be more wrong... > Splus has a _huge_ number of objects, and the pile seems to grow daily. xlisp-stat is not exactly lacking in the area of built-in fuctions: check out Luke Tierney's 1990 book (Wiley), and you'll see what I mean! I'd say the learning curve for Splus is steeper than for xlisp-stat, but some would disagree. I'm not sure what you mean by "more flexible." Since Splus has so many functions, it may reduce programming effort if the functions are what you need. A 32-bit MS Windows version of SPlus is still being beta-tested. It should be available later this year (June, maybe?). A 32 bit version of xlisp-stat has been available for a bit. Hope the above is some help, -lynd bacon >With just one fast-disappearing lifetime, I appreciate any insights... > >Many thanks, > >Paul Lunney >Queensland University of Technology >Brisbane Australia > > > > From majordom@umnstat.stat.umn.edu Wed Feb 26 15:08:45 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15796; Wed, 26 Feb 1997 15:08:45 -0600 Return-Path: Received: from gateway.ets.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15788; Wed, 26 Feb 1997 15:08:36 -0600 Received: from research.ets.org by gateway.ets.org with SMTP id AA04949 (5.65c/IDA-1.4.4 for ); Wed, 26 Feb 1997 16:08:06 -0500 Received: from comet.research.ets.org by research.ets.org with SMTP id AA03300 (5.67b/IDA-1.4.3 for p.lunney@qut.edu.au); Wed, 26 Feb 1997 16:08:04 -0500 Received: by comet.research.ets.org (SMI-8.6/SMI-SVR4) id QAA08359; Wed, 26 Feb 1997 16:08:04 -0500 Date: Wed, 26 Feb 1997 16:08:04 -0500 Message-Id: <199702262108.QAA08359@comet.research.ets.org> From: Russell Almond To: p.lunney@qut.edu.au Cc: s-news@utstat.toronto.edu, stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: (message from Paul Lunney on Wed, 26 Feb 1997 21:23:49 +1000 (EST)) Subject: Re: S-plus and xlisp-stat Reply-To: ralmond@ets.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Here are a couple of the major differences: 1) Graphics Engine. One of the biggest drawbacks of S is that it still has a plotter based graphics package. I find S easier to use for fancy static graphics with lots of overlays. Xlisp-stat's big advantage comes from its ability to support dynamic graphics. 2) Memory model S stores data objects on disc while XLS stores them in memory. One consequence is that S storage is persistent. XLS should also be faster with some operations as a consequence, but S will handle larger data sets (although with the price of memory, thats not as much of a problem as it used to be). 3) Programming Ease I'd say this one is a toss-up. S has a more C-like syntax for those who have a mental block about parentheses. S also has a truely bizarre object model (class identity and method dispatch are completely divorced from object encapsulation.) Also, I find the S lazy evaluation harder to understand than the Lisp eager evaluation rules. Unfortunately, xlisp lacks many of the debugging tools which makes Lisp so easy to use. (Which reminds me, did I ever post my debugging code to the list?) 4) Functionality S-Plus definetly has more, and more thorough testing. Both collections of functions reflect the research interestsof their respective authors: S tends to favor robust procedures and XLS Bayesian. On the other hand, if you really need a specific function, try SAS. 5) Support Luke does a great job for a one man show, but part of the price of S-Plus is the ability to call somebody on the phone to hold your hand while you go through the mysteries of configuring S-Plus to work on your PC with your printer. Also, there are training couses available. 6) Price XLS costs fewer dollars, but morein sweat equity. Thus, it tends to be favored by researchers and teachers who are willing to put in the effort to get it to do what they want. --Russell Almond From majordom@umnstat.stat.umn.edu Wed Feb 26 15:32:23 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16008; Wed, 26 Feb 1997 15:32:23 -0600 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16004; Wed, 26 Feb 1997 15:32:16 -0600 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id NAA07471; Wed, 26 Feb 1997 13:31:28 -0800 (PST) From: Jan Deleeuw Message-Id: <199702262131.NAA07471@galton.math.ucla.edu> Subject: Re: S-plus and xlisp-stat To: ralmond@ets.org Date: Wed, 26 Feb 1997 13:31:28 -0800 (PST) Cc: p.lunney@qut.edu.au, s-news@utstat.toronto.edu, stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <199702262108.QAA08359@comet.research.ets.org> from "Russell Almond" at Feb 26, 97 04:08:04 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Generally, I think the comparisons of S-plus and XLS have been fair. Also, there is one in Statistical Science some time ago. Although we have both, the five major and overriding reasons why I would not even consider using S-plus are perhaps worth repeating. (a) XLS comes with complete source code -- in ANSI C. Generally it is a breeze to port it to a new flavor of UNIX. It is quite easy to extend it at compile time, and by now it is also easy to extend it at runtime with object code. One can always go back to the source in case of doubt. One can look at the source and marvel. (b) XLS runs on the Mac, in fact that is where both XLS and I prefer to be most of the time. The other two are just necessary evils. (c) I can give it away to my students and my family, in unlimited quantities, and for their own home computers, no matter what hardware and OS they have. I can provide WWW interfaces to the program which run for free from the web servers -- I dont think the S-plus people would appreciate a similar service. (d) Porting from Common Lisp is now easy. Porting from FORTRAN to CL, and then to XL is now easy. Lisp is a general purpose language. There are lots of books on Lisp. Lisp is a beautiful language. Because of XLS I now read books such as "On Lisp" or Norvig or Steele. (e) Dynamic graphics for demos, instructional modules, and research tools cannot really be done in S. Computing and making static plots is perhaps marginally easier in S, but this is not what I do. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Wed Feb 26 19:25:08 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17907; Wed, 26 Feb 1997 19:25:08 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17903; Wed, 26 Feb 1997 19:24:58 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id ab11014; 26 Feb 97 20:24 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA16146; Wed, 26 Feb 1997 20:24:41 -0500 Date: Wed, 26 Feb 1997 20:24:40 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Cc: Jan Deleeuw , Luke Tierney Subject: CW11 changes to compile Mac XLispStat 3.49 (probably the same for prev. versions) Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk [1] Install the "970217 Compilers/Linkers" CW11 patch. ( I expect this will show up on their web page soon, if not you can ask them for it, or ask me and I'll forward a copy. If one does show up on their web, it may be a more final and tested version than the one I have. ) [2] replace #define in Transkel1.c: #include with: #include [3] replace #define in xlisp.h #include with: #include [4] in CWHeader.h, remove: #define EACCESS [5] for PPC, need to change project settings for prefix file from "CWHeaderPPC" to "CWHerder.h" [6] in MacHeader.pch, change OLDROUTINENAMES define from 0 to 1: #define OLDROUTINENAMES 1 ( I also always add the prefix "::" to the target so that the executable will be in the same parent directory where the init.lsp and other .lsp files are, rather than in the :macintosh subdirectory. ) It compiles and builds PPC version for me with those changes. Minimal test that I can load and run my application -- no problems seen. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Wed Feb 26 19:34:04 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA18000; Wed, 26 Feb 1997 19:34:04 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17996; Wed, 26 Feb 1997 19:33:55 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa15858; 26 Feb 97 20:33 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA21296; Wed, 26 Feb 1997 20:33:09 -0500 Date: Wed, 26 Feb 1997 20:33:06 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Cc: Jan Deleeuw , Luke Tierney Subject: Re: CW11 changes to compile Mac XLispStat 3.49 (probably the same for prev. versions) In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Oops -- one more step I forgot to write. remove ANSI lib and replace with MSL C lib. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Thu Feb 27 01:34:29 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19601; Thu, 27 Feb 1997 01:34:29 -0600 Return-Path: Received: from hilbert.edv.agrar.tu-muenchen.de by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19597; Thu, 27 Feb 1997 01:34:21 -0600 Received: from localhost (max@localhost) by hilbert.edv.agrar.tu-muenchen.de (8.8.5/8.8.5) with SMTP id IAA04157 for ; Thu, 27 Feb 1997 08:34:17 +0100 X-Authentication-Warning: hilbert.edv.agrar.tu-muenchen.de: max owned process doing -bs Date: Thu, 27 Feb 1997 08:34:17 +0100 (CET) From: Max Pilgram X-Sender: max@hilbert.edv.agrar.tu-muenchen.de To: Liste StatLisp Subject: Compiling Xlisp-stat Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, I try to compile xlispstat-3-44 on a DEC Alpha running Linux. At the end of the compilation process I get the following error message: make: *** [compiler/backquot.fsl] Error 139 Could anybody explain me that error message, which supposably does not come from the C-compiler. Max Max Pilgram Technische Universitaet Muenchen-Weihenstephan Lehreinheit fuer Mathematik, Statistik und Datenverarbeitung phone: 08161-714040 e-mail: max@edv.agrar.tu-muenchen.de http://www.edv.agrar.tu-muenchen.de/dvs/persons/max/max.html From majordom@umnstat.stat.umn.edu Thu Feb 27 06:06:27 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29069; Thu, 27 Feb 1997 06:06:27 -0600 Return-Path: Received: from fhat.stat.sc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29065; Thu, 27 Feb 1997 06:06:21 -0600 Received: (from rossini@localhost) by fhat.stat.sc.edu (8.6.12/8.6.12) id HAA02506; Thu, 27 Feb 1997 07:06:13 -0500 Date: Thu, 27 Feb 1997 07:06:13 -0500 Message-Id: <199702271206.HAA02506@fhat.stat.sc.edu> From: Anthony Rossini Mime-Version: 1.0 To: Max Pilgram Cc: Liste StatLisp Subject: Compiling Xlisp-stat In-Reply-To: References: Reply-To: rossini@stat.sc.edu X-Face: -JZ8Fs2;tGK&eaV#. U#%d52]n51?][w92.tu+>13)STQ>>>> "Max" == Max Pilgram writes: Max> Hi, I try to compile xlispstat-3-44 on a DEC Alpha running Max> Linux. At the end of the compilation process I get the Max> following error message: Max> make: *** [compiler/backquot.fsl] Error 139 Max> Could anybody explain me that error message, which supposably Max> does not come from the C-compiler. I used to have that all the time on a PC with a bad memory card. (similar to the random errors which might occur while recompiling the kernel or gcc, or any other intensive use of memory). They (the errors) went away when I replaced the memory. Maybe the same holds for you? (YMMV). best, ---- -tony (Anthony Rossini) Asst. Prof and General Nuisance Department of Statistics rossini@stat.sc.edu University of South Carolina http://www.stat.sc.edu/~rossini/ Columbia, SC 29208 803-777-3578(O) ..7-4048 (fax) From majordom@umnstat.stat.umn.edu Thu Feb 27 06:58:31 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29229; Thu, 27 Feb 1997 06:58:31 -0600 Return-Path: Received: from ns.ge.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29225; Thu, 27 Feb 1997 06:58:25 -0600 Received: from thomas.ge.com (thomas.ge.com [3.47.28.21]) by ns.ge.com (8.8.4/8.7.3) with ESMTP id HAA05830; Thu, 27 Feb 1997 07:56:28 -0500 (EST) Received: from gemb1.ae.ge.com (gemb1.ae.ge.com [129.202.7.246]) by thomas.ge.com (8.8.4/8.7.5) with SMTP id HAA24684; Thu, 27 Feb 1997 07:59:19 -0500 (EST) Received: from ss09.ae.ge.com by gemb1.ae.ge.com; (5.65/1.1.8.2/04Nov94-0158PM) id AA16450; Thu, 27 Feb 1997 07:55:50 -0500 Message-Id: <9702271255.AA16450@gemb1.ae.ge.com> Received: from c0230.ae.ge.com by ss09.ae.ge.com with ESMTP (1.37.109.16/15.6) id AA297638266; Thu, 27 Feb 1997 07:57:47 -0500 Received: by c0230.ae.ge.com (1.37.109.16/16.2) id AA106698264; Thu, 27 Feb 1997 07:57:44 -0500 Date: Thu, 27 Feb 1997 07:57:44 -0500 From: U-E59264-Osman Buyukisik To: deleeuw@math.ucla.edu Cc: ralmond@ets.org, p.lunney@qut.edu.au, s-news@utstat.toronto.edu, stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <199702262131.NAA07471@galton.math.ucla.edu> (message from Jan Deleeuw on Wed, 26 Feb 1997 13:31:28 -0800 (PST)) Subject: Re: S-plus and xlisp-stat Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >>>>> "Jan" == Jan Deleeuw writes: Jan> (a) XLS comes with complete source code -- in ANSI Jan> C. Generally it is a breeze to port it to a new flavor of Jan> UNIX. It is quite easy to extend it at compile time, and by Jan> now it is also easy to extend it at runtime with object Jan> code. One can always go back to the source in case of Jan> doubt. One can look at the source and marvel. Even though S and S-plus are commercial products, there is GPL'd free S like program called "R." Authors suggest the S "Blue book" for a manual. Except for complex numbers, it is mostly compatible. INcludes free source code. Jan> (b) XLS runs on the Mac, in fact that is where both XLS and I Jan> prefer to be most of the time. The other two are just Jan> necessary evils. I know there is a windoze version. Mac may not be supported yet :( Jan> (c) I can give it away to my students and my family, in Jan> unlimited quantities, and for their own home computers, no Jan> matter what hardware and OS they have. I can provide WWW Jan> interfaces to the program which run for free from the web Jan> servers -- I dont think the S-plus people would appreciate a Jan> similar service. You can do the same with R. Jan> (d) Porting from Common Lisp is now easy. Porting from Jan> FORTRAN to CL, and then to XL is now easy. Lisp is a general Jan> purpose language. There are lots of books on Lisp. Lisp is a Jan> beautiful language. Because of XLS I now read books such as Jan> "On Lisp" or Norvig or Steele. I think the next version may be written in some functional language. Could be better than Lisp :) Jan> (e) Dynamic graphics for demos, instructional modules, and Jan> research tools cannot really be done in S. Computing and Jan> making static plots is perhaps marginally easier in S, but Jan> this is not what I do. Here xlisp-stat shines. Just my $0.02 Osman From majordom@umnstat.stat.umn.edu Thu Feb 27 07:53:14 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29450; Thu, 27 Feb 1997 07:53:14 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29445; Thu, 27 Feb 1997 07:53:08 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa20125; 27 Feb 97 8:53 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA21519; Thu, 27 Feb 1997 08:53:05 -0500 Date: Thu, 27 Feb 1997 08:53:03 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Cc: Jan Deleeuw , Luke Tierney Subject: Re: CW11 changes to compile Mac XLispStat 3.49 (probably the same for prev. versions) In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk > [6] in MacHeader.pch, change OLDROUTINENAMES define from 0 to 1: > #define OLDROUTINENAMES 1 Apparently, if you also define OLDROUTINELOCATION on the next line, you should not require these two changes: > > [2] replace #define in Transkel1.c: > #include > with: > #include > > > [3] replace #define in xlisp.h > #include > with: > #include > ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Thu Feb 27 08:39:41 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29681; Thu, 27 Feb 1997 08:39:41 -0600 Return-Path: Received: from ns.ge.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29677; Thu, 27 Feb 1997 08:39:35 -0600 Received: from thomas.ge.com (thomas.ge.com [3.47.28.21]) by ns.ge.com (8.8.4/8.7.3) with ESMTP id JAA13922; Thu, 27 Feb 1997 09:37:39 -0500 (EST) Received: from gemb1.ae.ge.com (gemb1.ae.ge.com [129.202.7.246]) by thomas.ge.com (8.8.4/8.7.5) with SMTP id JAA05626; Thu, 27 Feb 1997 09:40:30 -0500 (EST) Received: from ss09.ae.ge.com by gemb1.ae.ge.com; (5.65/1.1.8.2/04Nov94-0158PM) id AA11801; Thu, 27 Feb 1997 09:37:01 -0500 Message-Id: <9702271437.AA11801@gemb1.ae.ge.com> Received: from c0230.ae.ge.com by ss09.ae.ge.com with ESMTP (1.37.109.16/15.6) id AA056081901; Thu, 27 Feb 1997 08:58:21 -0500 Received: by c0230.ae.ge.com (1.37.109.16/16.2) id AA108651850; Thu, 27 Feb 1997 08:57:30 -0500 Date: Thu, 27 Feb 1997 08:57:30 -0500 From: U-E59264-Osman Buyukisik To: Osman.Buyukisik@ae.ge.com Cc: deleeuw@math.ucla.edu, ralmond@ets.org, p.lunney@qut.edu.au, s-news@utstat.toronto.edu, stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <9702271255.AA16450@gemb1.ae.ge.com> (message from U-E59264-Osman Buyukisik on Thu, 27 Feb 1997 07:57:44 -0500) Subject: Re: S-plus and xlisp-stat Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk OOPS, I just found out that my reply was in error (about R): Jan> (b) XLS runs on the Mac, in fact that is where both XLS and I Jan> prefer to be most of the time. The other two are just Jan> necessary evils. I know there is a windoze version (of R). Mac may not be supported yet :( ^ |_______________ this is wrong. Mac versions of R are complete, but not windoze ! Sorry ! Osman From majordom@umnstat.stat.umn.edu Thu Feb 27 08:59:27 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29852; Thu, 27 Feb 1997 08:59:27 -0600 Return-Path: Received: from ns1.rapidnet.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29848; Thu, 27 Feb 1997 08:59:23 -0600 Received: from zip-s-166 (ts2-08.mth.cyberhighway.net [206.26.243.48]) by rapidnet.com (8.8.5/8.6.12) with SMTP id HAA29561 for ; Thu, 27 Feb 1997 07:59:15 -0700 (MST) Message-Id: <199702271459.HAA29561@rapidnet.com> Comments: Authenticated sender is From: bdudas@rapidnet.com To: Xlisp List Date: Thu, 27 Feb 1997 07:59:13 +0000 Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7BIT Subject: STOP THE MADNESS!!!!! X-Mailer: Pegasus Mail for Win32 (v2.52) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Please remove my name from the recipient list of this newsgroup.... bdudas@rapidnet.com From majordom@umnstat.stat.umn.edu Fri Feb 28 04:29:27 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15474; Fri, 28 Feb 1997 04:29:27 -0600 Return-Path: Received: from marraco.udl.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15460; Fri, 28 Feb 1997 04:28:06 -0600 Received: from caragol.udl.es ([192.168.1.220]) by marraco.udl.es (8.8.5/8.8.5) with ESMTP id LAA23080 for ; Fri, 28 Feb 1997 11:31:47 -0100 (GMT) Received: from caragol-1878 (caragol-1878 [10.100.7.86]) by caragol.udl.es (8.7.5/8.7.3) with SMTP id LAA08727 for ; Fri, 28 Feb 1997 11:23:57 -0100 (GMT) Date: Fri, 28 Feb 1997 11:23:57 -0100 (GMT) From: Jaume March Message-Id: <199702281223.LAA08727@caragol.udl.es> To: stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Pronto E-Mail [version 2.01] Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi Folks, First thanks for the solution on my problem about ":error insufficient node space" In Luke words " I don't know how much memory GEE uses. But task manager shows there is big process. And I increased my virtual memory. Now there is another question. After 25 hours running a GEE in the P133MHz (now 1'5 GB free memory) the system shows de message: "Complex Cholesky not available yet Happened in #" I'm following a report about GEE in Xlisp-stat and I'm doing (I think) the same procedure. What's does the message mean? It looks like I need to load another ".lsp" , but which? The loaded lsp are: glim.lsp and .fsl modelformula.lsp and gee_1_0.lsp or gee_3_0.lsp Thanks, Jaume :):):) Jaume March :):):) School of Medicine :):):) University of Lleida :):):) Lleida, 25198 :):):) Jaume.March@cmb.udl.es From majordom@umnstat.stat.umn.edu Fri Feb 28 10:19:21 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16706; Fri, 28 Feb 1997 10:19:21 -0600 Return-Path: Received: from sirio.stat.unipd.it by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16702; Fri, 28 Feb 1997 10:19:01 -0600 Received: (qmail 103 invoked by uid 501); 28 Feb 1997 16:15:53 -0000 Message-Id: <19970228161553.102.qmail@sirio.stat.unipd.it> From: guido@sirio.stat.unipd.it Subject: IASC Summer School To: stat-lisp-news@umnstat.stat.umn.edu Date: Fri, 28 Feb 1997 17:15:52 +0100 (GMT+0100) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk ------------------------------------------------------------------------------ IV INTERNATIONAL SUMMER SCHOOL ON COMPUTATIONAL STATISTICS organized by International Association for Statistical Computing and Department of Statistical Science, Padua, Italy ------------------------------------------------------------------------------- June, 23 - July 4, 1997 ------------------------------------------------------------------------------- GENERAL INFORMATION The International Association for Statistical Computing (IASC) and the Department of Statistical Science of the University of Padua, Italy, will host an International Summer School on Computational Statistics at the University of Padua, Italy. The school is intended for young researchers and PhD students. Participants are required to have a good background in statistics and a basic knowledge of computer usage. The school will last two weeks and the courses will include practical sessions using computers. The official language of the school will be English. The courses will be held from Monday to Friday from 9am to 6pm. ------------------------------------------------------------------------------- COURSES * 1. Non-parametric smoothing methods * 2. Monte Carlo Markov Chains * 3. Neural networks and classification trees LECTURERS * topic 1 A.W. Bowman (University of Glasgow, United Kingdom), W. Haerdle (Humboldt-University of Berlin, Germany) * topic 2 C. Berzuini (University of Pavia, Italy) A. Gelman (Columbia University, U.S.A.), P. Volf ( Czech Academy of Sciences, Czech Republic), * topic 3 J. Nakache (INSERM, France) B. Ripley (University of Oxford, U.K.), SCIENTIFIC COMMITTEE * Adelchi Azzalini (University of Padua) * Nicola Torelli (University of Padua) * David Firth (Nuffield College, Oxford, U.K) * Jaromir Antoch (University of Praha, Czech Republic) * Nanny Wermuth (University of Mainz, Germany) Local Organizing Committee: N. Torelli, M. Chiogna, C. Gaetan, G. Masarotto. ------------------------------------------------------------------------------- APPLICATIONS The number of participants is limited to 45. If the number of applicants exceeds this limit, applicants will be accepted on the basis of their curriculum vitae which must be submitted. In order to have a balanced attendance from different countries, regional distribution will be also taken into consideration. The application form, as well as curriculum vitae and certificate of Academic position, must be sent to the local organizing committee no later than May 15. Applicants will be immediately notified of their acceptance. ------------------------------------------------------------------------------- Registration fee Graduate and PhD Students 350,000 Lit Researchers from academic institutions 500,000 Lit Researchers from private institutions 1,500,000 Lit Travel information Padua and its Province are among the liveliest and most dynamic areas in North East Italy. The city has a great artistic and cultural tradition and possesses famous monuments such as the University, which is among the oldest and renowned in Europe. It can be reached by train or by plain. Travellers by air may use the airports of Venice (less than 1 hour by coach), Bologna (1 hour and half by train) and Milan (3 hour by train). Accomodation Low budget accomodation in shared rooms is available for a limited number of participants. Participants will be accomodated on a first-registered first-served basis. The local organizing committee will arrange room reservations for the other participants. Approximate prices for hotels located nearby the Department are: * 4-stars: 120,000 - 170,000 (single room); 150,000 - 200,000 (double room) * 3-stars: 90,000 - 120,000 (single room); 100,000 - 150,000 (double room) * 2-stars: 60,000 - 90,000 (single room); 70,000 - 100,000 (double room) Details about room reservation will be given after acceptance. Important information for accepted participants 1. For registration fees received after June 15, 1997 a 20% surcharge is applicable. 2. Hotel room reservations should be made before May 30, 1997 to ensure availability. 3. Payment must be made by Lit check or Lit bank draft. Postal Money Orders are not acceptable. Make checks or bank drafts payable to Dipartimento di Scienze Statistiche, Universita' di Padova. 4. Fees and deposits are non-refundable after June 15, 1997. Before June 15, 1997 a 20% handling fee will be charged for cancellation. ------------------------------------------------------------------------------ School Office Please use the following address for correspondence: International Summer School on Computational Statistics Dipartimento di Scienze Statistiche, Universita' di Padova via San Francesco, 33 I35121 PADOVA ITALY E-mail: ssiasc@gauss.stat.unipd.it Web-page: http://www.stat.unipd.it/dip/ssiasc/iasc.html From majordom@umnstat.stat.umn.edu Fri Feb 28 10:47:48 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16862; Fri, 28 Feb 1997 10:47:48 -0600 Return-Path: Received: from sulu.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16854; Fri, 28 Feb 1997 10:47:41 -0600 Received: from chekov by sulu.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id IAA20207; Fri, 28 Feb 1997 08:44:26 -0800 (PST) Date: Fri, 28 Feb 1997 08:44:20 -0800 (PST) From: Thomas Lumley X-Sender: thomas@chekov To: Jaume March Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: your mail In-Reply-To: <199702281223.LAA08727@caragol.udl.es> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Fri, 28 Feb 1997, Jaume March wrote: > > Now there is another question. After 25 hours running a GEE in the P133MHz > (now 1'5 GB free memory) the system shows de message: > > "Complex Cholesky not available yet > Happened in #" > > I'm following a report about GEE in Xlisp-stat and I'm doing (I think) the > same procedure. I think I missed your earlier question, so I'm not sure exactly what GEE model and data you're using that takes so much time and space. However, if you're using my GEE code, as it appears, this means that a matrix you're trying to invert contains complex numbers. This is due to an earlier correlation matrix not being positive definite. That is, the problem is with the correlation matrix, which is not being estimated as positive definite. What working correlation are you using and what is your data structure like? Two solutions: try a different working correlation structure (if you can't get reliabily positive-definite estimates of the one you're using then the GEE asymptotics probably don't apply very well) or replace the call to (inverse) in the :estimate and :rinv methods of the correlation structure by calls to the following function which provides an inverse to the "nearest" positive definite matrix instead of to the matrix you supplied. (defun positive-inverse (mat) (let ((Linv (inverse (first (chol-decomp mat))))) (matmult (transpose Linv) Linv) )) Thomas Lumley -----------------------------------------------------+------ Biostatistics : "Never attribute to malice what : Uni of Washington : can be adequately explained by : Box 357232 : incompetence" - Hanlon's Razor : Seattle WA 98195-7232 : : ------------------------------------------------------------ From majordom@umnstat.stat.umn.edu Sat Mar 1 15:30:55 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03035; Sat, 1 Mar 1997 15:30:55 -0600 Return-Path: Received: from melia.qut.edu.au by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03029; Sat, 1 Mar 1997 15:30:46 -0600 Received: from pigeon.qut.edu.au (lunney@pigeon.qut.edu.au) by melia.qut.edu.au (PMDF V5.0-7 #20429) id <01IG0P40BPF4003U46@melia.qut.edu.au>; Sun, 02 Mar 1997 07:33:42 +1000 Received: from localhost (lunney@localhost) by pigeon.qut.edu.au (8.8.5/8.8.5) with SMTP id HAA18701; Sun, 02 Mar 1997 07:30:29 +1000 (EST) Date: Sun, 02 Mar 1997 07:30:29 +1000 (EST) From: Paul Lunney Subject: S-plus and xlisp-stat - thankyou To: s-news@utstat.toronto.edu, stat-lisp-news@umnstat.stat.umn.edu Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Content-Transfer-Encoding: 7BIT Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Sincere thanks to all who responded to my query asking for comments about the respective strengths of S-plus and xlisp-stat. Your observations were much appreciated and have given me a good idea of where I should be devoting my efforts. Best wishes to all, Regards, Paul Lunney Queensland University of Technology Brisbane Australia From majordom@umnstat.stat.umn.edu Mon Mar 3 05:17:24 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27093; Mon, 3 Mar 1997 05:17:24 -0600 Return-Path: Received: from marraco.udl.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27089; Mon, 3 Mar 1997 05:17:03 -0600 Received: from caragol.udl.es ([192.168.1.220]) by marraco.udl.es (8.8.5/8.8.5) with ESMTP id MAA04680 for ; Mon, 3 Mar 1997 12:20:12 -0100 (GMT) Received: from caragol-1878 (caragol-1878 [10.100.7.86]) by caragol.udl.es (8.7.5/8.7.3) with SMTP id MAA05094 for ; Mon, 3 Mar 1997 12:12:13 -0100 (GMT) Date: Mon, 3 Mar 1997 12:12:13 -0100 (GMT) From: Jaume March Message-Id: <199703031312.MAA05094@caragol.udl.es> To: stat-lisp-news@umnstat.stat.umn.edu Subject: once more GEE X-Mailer: Pronto E-Mail [version 2.01] Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi and Thanks for messages! Sorry, the problem "is" with the correlation matrix. > Thomas Lumley: > which is not being estimated as positive definite. When I selected only the :error poisson, :link is the default and :correlation exchangeable-corr, it works. Perhaps the trouble is a bad combination logit-link or log-link with binomial-error??? > Thomas Lumley: >What working correlation are you using and what is your data structure >like? About working correlation, I'm, for the moment, only learning or testing. And that's the explanation for the Cholesky's message! The data structure is: 9600 binary responses, 4 binary predictor variables and 2 categorical predictor variable. And finally, yes I'm using your GEE code, and I have another two problems. The first is that as-formula only runs with a approximately 500 observations, with the entire sample (9600) a "stack overflow" message appears. Is there a limitation? And the second is when there is a variable like subject with multiple observations (rows) and a second grouping variable that means "nesting" and there's no way of doing something? Jaume March From majordom@umnstat.stat.umn.edu Mon Mar 3 11:46:16 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28939; Mon, 3 Mar 1997 11:46:16 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28935; Mon, 3 Mar 1997 11:46:03 -0600 Received: from bootp-17-17.bootp.virginia.edu by mail.virginia.edu id aa17625; 3 Mar 97 12:44 EST Date: 3 Mar 97 12:49:16 +0000 Subject: more CW11/Mac XlispStat From: sdm7g@Virginia.EDU To: Xlisp List X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Some of the discrepancy about which MacHeaders mods are required to compile have to do with whether or not the Project Stationary patch has been applied. I hadn't looked closely at the release notes because I thought it just affected new projects created from the Project Stationary, however the patch makes some changes in the defaults for precompiled MacHeaders. If you haven't applied that patch, you may not require some of the changes. - Steve Majewski From majordom@umnstat.stat.umn.edu Mon Mar 3 12:30:29 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29231; Mon, 3 Mar 1997 12:30:29 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29226; Mon, 3 Mar 1997 12:30:22 -0600 Received: from bootp-17-17.bootp.virginia.edu by mail.virginia.edu id aa03482; 3 Mar 97 13:28 EST Date: 3 Mar 97 13:34:11 +0000 Subject: :CLEAR-CONTENT [Re: New alpha release] From: sdm7g@Virginia.EDU To: stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Wed, Feb 26, 1997 4:39 PM, Luke Tierney wrote: >Mostly this release just fixes bugs. There is, however, one >substantive change, in reponse to prompting from Steve Majewski and >Sandy Weisberg: > >The protocol used by the standard :redraw-content methods has been >augmented to support maintaining a background for plot >contents. Instead of erasing the content area before drawing the >content, these methods now send the :clear-content message. The >default method for this method clears the content area. By overriding >this method, you can draw any background you want to maintain. > Well -- the good news is that this change doesn't affect any of my modifications. I can still run my code unchanged. However, when I tried to modify my code to use the new feature, I found it didn't quite work as I expected. I had the following two redefinitions: :REDRAW-BACKGROUND was modified to draw a x/y grid on the background behind the graph. ( defmeth scatterplot-proto :redraw-background () (call-next-method) ( let (( color ( send self :grid-color ))) ( if color (send self :draw-grid :color color))) ( let (( color ( send self :axis-color ))) ( if color (send self :draw-zero-axis :color color)))) and :REDRAW-CONTENT was redefined to redraw the content without erasing the content-region. ( defmeth scatterplot-proto :redraw-content () ( let (( vars ( send self :current-variables ))) (send self :draw-data-lines (car vars) (cadr vars) 0 (send self :num-lines)) (send self :draw-data-points (car vars) (cadr vars) 0 (send self :num-points)))) I first tried replacing by :redraw-background with a similar, but renamed :clear-content, and eliminating the redefinition of :REDRAW-CONTENT, but it didn't have the same effect, so I did a bit of testing to discover where that call fits into the protocol: :REDRAW-BACKGROUND :REDRAW-OVERLAYS :REDRAW-CONTENT ( which calls :CLEAR-CONTENT ) In the case of my code, which uses overlays drawn in the content rect, this clears and overwrites the overlays. The solution seems to be to keep the background/ grid drawing in :REDRAW-BACKGROUND, to cut the redefinition of :REDRAW-CONTENT, and to add a null :CLEAR-CONTENT method that *doesn't* (call-next-method). ( defmeth scatterplot-proto :clear-content () ) [ Note: I got rid of my initial separate/parallel implementation of data-overlays once I realized that LispStat graphic overlays are really graphically, underlays ( under the graph content ) but on top of the graph content in their event/click dispatching -- which turned out to be exactly what I wanted. It just wasn't initially clear that that was the case. ] Results: The new way is slightly simpler and shorter that the old way, but it's not a significant change either way. The main problem I had was and is still understanding the whole flow of the redraw protocols, and which methods trigger which other methods. More/better documentation is probably more important that more changes to the protocol. With both the before and after versions of my modified scatterplot classes loaded, Tourdemo and the other standard examples appear to run correctly without change. The sort of changes that would be REALLY NICE are ones that I know to be not trivial: better. more general linking mechanism; a way to mix dialog controls and text regions in a graphics window. ( Tk and Java's AWT both have a Canvas item as a widget that can be placed in a window container. Their Windows are just generic containers. So you can create a window with a graphics canvas and controls and text fields mixed together. ) - Steve Majewski From majordom@umnstat.stat.umn.edu Mon Mar 3 15:09:20 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00756; Mon, 3 Mar 1997 15:09:20 -0600 Return-Path: Received: from bremmer.Mines.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00752; Mon, 3 Mar 1997 15:09:14 -0600 Received: (from kasper@localhost) by bremmer.Mines.EDU (8.6.12/8.6.9) id OAA17170; Mon, 3 Mar 1997 14:09:12 -0700 Date: Mon, 3 Mar 1997 14:09:12 -0700 Message-Id: <199703032109.OAA17170@bremmer.Mines.EDU> From: Kasper van Wyk To: stat-lisp-news@umnstat.stat.umn.edu Subject: text in plots Reply-To: kasper@dix.mines.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I have a problem with writing text in my graph windows. I believe my window WITH text, is being overwritten by one without. How do I assign text to my graph object that will "stick" ? Thank you, -- Kasper van Wijk +(303) 273 3464 (work) +(303) 384 9602 (home) -- From majordom@umnstat.stat.umn.edu Mon Mar 3 17:06:48 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02194; Mon, 3 Mar 1997 17:06:48 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02190; Mon, 3 Mar 1997 17:06:34 -0600 Received: from bootp-17-17.bootp.virginia.edu by mail.virginia.edu id aa12182; 3 Mar 97 18:06 EST Date: 3 Mar 97 18:11:29 +0000 Subject: Re: text in plots From: sdm7g@Virginia.EDU To: stat-lisp-news@umnstat.stat.umn.edu Cc: Kasper van Wyk X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Mon, Mar 3, 1997 9:09 PM, Kasper van Wyk wrote: > >I have a problem with writing text in my graph windows. >I believe my window WITH text, is being overwritten by one >without. > >How do I assign text to my graph object that will "stick" ? > > I assume you are just writing text with :DRAW-TEXT or :DRAW-STRING Those methods, as well as the low level graphic methods like :FRAME-RECT, et. al. is that they are immediate, not retained mode. They draw when you give the command, but they don't create a permanent object that is redraw when the graph is redraw or resized. You need your low level draw methods to be called by one of the redraw methods. A graph :REDRAW causes the sequence: :REDRAW-BACKGROUND :REDRAW-OVERLAYS :REDRAW-CONTENT to be send to itself. You can either redefine one of those methods to redraw your text, or you can create a text overlay which will get called by :REDRAW-OVERLAYS. You can see my graph overlay methods for an example: http://galen.med.virginia.edu/~sdm7g/XES/Docs/SourceIndex.html especially realgraf.lsp. [ See the comments in the source and my previous message to this list about :CLEAR-CONTENT for some more tips. ] Note: The code at that URL is a bit out of date. I'll have a newer version posted real soon. All of the overlays in that example use real space coordinates. I'm going to add some other canvas based coordinate overlays, including, I hope, a composite rectangle-with-text-field-and-drop-down-menu-or-edit-dialog... I haven't finished the changes or gotten the documentation up to sync yet. ] -- Steve Majewski From majordom@umnstat.stat.umn.edu Tue Mar 4 14:25:51 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA20015; Tue, 4 Mar 1997 14:25:51 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA20011; Tue, 4 Mar 1997 14:25:42 -0600 Received: from bootp-17-17.bootp.virginia.edu by mail.virginia.edu id aa05218; 4 Mar 97 15:25 EST Date: 4 Mar 97 15:30:47 +0000 Subject: graph overlays [was: :CLEAR-CONTENT ] From: sdm7g@Virginia.EDU To: Luke Tierney Cc: stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Luke, I hope you don't mind me replying to the whole list -- this fills in some more details touched on in my reply to Kasper van Wyk 's query about "text in plots". On Tue, Mar 4, 1997 4:25 PM, Luke Tierney wrote: >I'm confused by your description. My intention with overlays was that >they would only draw in the margin area. I'm sure that wasn't clear, >but that was my intention :-). If you are drawing in the content area >of an overlay, it seems to me you should be able to take the code in >:redraw-background and in the overlay methods that draws in the >content area, move that into :clear-content after (call-next-method) >and then use the standard :redraw-content methods. What am I missing? We've had some discussion about this before, and I've posted some code examples, but I've gone thru a couple of implementations before I settled on the current one, so I'll try to summarize the history: Yes: your intentions were not clear from the book or other documentation, and it took some experimenting to figure out what the graph/update protocol actually is. The sequence initiated by a :REDRAW of a graph is: :REDRAW-BACKGROUND :REDRAW-OVERLAYS :REDRAW-CONTENT however, :REDRAW-CONTENT clears the content rect before drawing the content, so anything draw in the background or overlays in the content-rect is lost. ( The only difference in the new version is that now :REDRAW-CONTENT calls the graph's :CLEAR-CONTENT method, which can be overridden. ) We had some discussion about this on the list where you stated your intention about the limitations of overlays and that you had not expected them to be used in the content region of the graph. I started, in my next implementation, to construct a parallel to GRAPH-OVERLAY-PROTO's called CONTENT-OVERLAY-PROTO, which would be redrawn after the content had been cleared and redrawn. I posted a version of this code, but I was not happy with it's performance. Although the standard overlays are actually UNDER- lays as graphical elements -- they are draw before the content, they ARE overlays with respect to their order in event processing. Overlays get a chance to :DO-CLICK before the graph itself catches the event. Except for the erase of the content area, the standard overlays actually behaved exactly as I wanted. I really wanted graphic underlays -- the graph content should always be visible -- but I wanted event priority. So in the next and current implementation, I threw out all of the duplication of code to construct CONTENT-OVERLAYs that did basically the same thing as the builtin overlays, and instead rewrote scatterplot 's :REDRAW-CONTENT to just redraw the data points and lines, without clearing the content area. ( What has changed in the latest version is that instead of overriding :REDRAW- CONTENT, I can override :CLEAR-CONTENT to do nothing. ) This produced the desired behaviour, with better performance, with less modification to the standard behaviour of SCATTERPLOT-PROTO and the code was much easier to read and understand. I'm still not even sure that the :CLEAR-CONTENT call is required in most plots. Even with my mods to make it a noop, all of the animated demo graphs I've tried seem to work the same. However, this is with minimal testing on a Mac only. I don't know if the call to :CLEAR-CONTENT is required on other platforms to get the same behaviour. ( I've tested my code on the Windows version -- which caused me to rename some of the files with 8+3 filenames. That was the only problem I noticed. ) [ You may recall, the other problem I had complained about was that the plot linking mechanism had built into it assumptions about the implementation of the linked objects: they had to be instances of HARDWARE-OBJECT-PROTO with a HARDWARE-ADDRESS slot that pointed to the correct internal implementation object. In this case, I had to retain a complete reimplementation of the linking protocol -- XLINKABLE-PROTO & STATEFUL-PROTO -- so that I could link either overlay items on a graph to points on a name-list, or points on a summary plot to plots of the raw-data plot from which that point was derived. ] One of my goals was to minimize the changes to the built in graph prototypes, however modification, rather than inheritance, was necessary because there is not another abstraction layer in between SCATTERPLOT-PROTO and the standard graph functions: plot-points, plot-lines, plot-function. Perhaps it would be better if those functions created an instance of USERPLOT-PROTO, where USERPLOT-PROTO is initially inherited from SCATTERPLOT-PROTO with no changes. This would user modification of USERPLOT-PROTO without having to change the default built-in behaviour of SCATTERPLOT-PROTO. You can see the screen dump at http://galen.med.virginia.edu/~sdm7g/XES/Docs/xesapp.html for an example of my overlays and linking. ( The linking is displayed by the blue overlay peaks being selected on both the main graph and the name-list graph of elemental peak references. ) Source code, class hierarchy diagrams and some other notes are at: http://galen.med.virginia.edu/~sdm7g/XES/Docs/SourceIndex.html -- sorry, one generation out of date. I'll be posting a newer version soon. To more directly answer your question of why not move these methods into :CLEAR-CONTENT -- that would again get me into reimplementing the existing overlay mechanism -- I need complex hierarchical objects. ( for example, a peak overlay with elemental lines and text labels attached. ) I need a sequential ordering of objects, both grapically and for event processing. ( for example, the lines and labels for the above will only be displayed if that peak is selected. ) I need to dynamically change the overlay sequence (:BRING-TO-FRONT), add and delete overlays, and change their state, color, etc. and link them to other objects. (dynamic behaviour not obvious in the screen dump. When a peak is selected by clicking under the peak, it is brought to the front, redrawn with a thicker line, text labels and elemental lines are made visible and the proper linking messages are sent to the name-list-proto with a list of elemental peak reference names. You can also select a peak from the linked name-list with the same effect. ) Except for the fact that by default, they get erased if they are in the content rect, the existing overlay-proto and their protocol/interaction with scatterplots is exactly what I need. ( i.e. IMNSHO I think it's the assumption that overlays exist only outside the content region that is the thing that needs to be "fixed" :-) For more detail on the current class hierarchy, below is the comments from the current source snapshot of realgraf.lsp. I am reorganizing the hierarchy somewhat from the version on my web pages, and I'm going to be adding some new classes that use canvas coordinates rather than real space, and some composite graph objects that create text fields with drop-down menus to change the text, for example. The web pages will be updated with the latest version when that is complete. If anyone wants the current working snapshot before that, please let me know. -- Steve Majewski -------- comments from realgraf.lsp ------------- ;;; Color graphic overlay objects using real-space coordinates ;;; ;;; Steven D. Majewski ;;; ;; ;; XlispStat graphics "overlays" are ;; [1] really "underlays" -- the :redraw sequence is ;; :REDRAW-BACKGROUND ;; :REDRAW-OVERLAYS ;; :REDRAW-CONTENT ;; i.e., the overlays are drawn on top of the background ;; and beneath the content of the plot. (However, they ARE ;; overlays in the sense that the :DO-CLICK protocol checks ;; for an overlay handler before calling the graphs handler.) ;; [2] But there is an assumption that the overlays are not drawn ;; within the content region, and an erase of the content area ;; is done as the initial step in :REDRAW-CONTENT. ;; #1 above is good -- we usually want the content visible, but #2 is an annoyance. ;; It keeps us from using overlays to represent data elements. If we ;; redefine :redraw-content, then overlays become really useful graphic ;; objects capable of rendering complex hierarchical graphs. ; ; object-proto-hierarchy related functions ; ; GRAPH-OBJ-PROTO: ; [parent-classes: GRAPH-OVERLAY-PROTO, XLINKABLE-PROTO, STATEFUL-PROTO] ; ; REAL-GRAPH-OBJ-PROTO: ; ; RECT-PROTO RECT ; OVAL-PROTO (& with optional args, ARCs) OVAL | ARC ; LINE-PROTO LINE ; POINTS-PROTO POINTS ; LABEL-PROTO (text) LABEL ; POLY-PROTO POLY ; ZPOLY-PROTO ZPOLY ; DATA-OVERLAY-PROTO ; ; COMPOSIT-GRAPH-OBJ-PROTO GROUP ; ; Classes were reorganized due to some multiple-inheritance problems - the base class ; now has most of the functionality of COMPOSIT-GRAPH-OBJ-PROTO - thus any of the other ; classes can now have associated parts. COMPOSIT-GRAPH-OBJ-PROTO is different in ; having no "base part" of it's own, and in having some more of it's methods automatically ; propagate slot values ( :LINE-COLOR, :FILL-COLOR, etc. ) ; ; ; ZPOLY-PROTO is a polygon that has it's end points "tied" to the Y=zero axis, so that ; it fills the area under (or above is negative) the curve. Since it is used to plot ; data or functions, it's :ISNEW method differs from the other methods in expecting ; a pair of X Y sequences, rather than a sequence of point pairs: ((x0 y0) (x1 y1) ... ). ; ; Typically, ZPOLY is used for creating a data-content overlay, and it may have some ; associated annotation of marker lines, which should not necessarily change color with ; the basic object, but which may need their state set separately ( for example, when ; the base object is not selected, the part objects may be set :STATE == 'INVISIBLE. ) ; ; ZPOLY has the most complete implementation -- It's the primary class I need for my ; application -- lines and labels are required mainly for annotation of ZPOLY overlays ; and the base scatterplot, I added ovals & rects to do box-plots,etc., so methods like ; :IN-RECT & :IN-OVAL ( or a general :IN-REGION ) are not implemented. ; ; Added for real time display: DATA-OVERLAY-PROTO : a ZPOLY-PROTO with ; :REAL-TO-CANVAS modified to track the graphs data points, does fill ; area under the points. ( I ought to remove :DO-CLICK for this class ) ;; ;;; From majordom@umnstat.stat.umn.edu Wed Mar 5 03:52:51 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04203; Wed, 5 Mar 1997 03:52:51 -0600 Return-Path: Received: from ts038p6.pop3b.netvision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04198; Wed, 5 Mar 1997 03:52:41 -0600 Received: (from hjstein@localhost) by blinky.bfr.co.il (8.7.6/8.6.9) id LAA02730; Wed, 5 Mar 1997 11:50:12 +0200 Date: Wed, 5 Mar 1997 11:50:12 +0200 Message-Id: <199703050950.LAA02730@blinky.bfr.co.il> From: "Harvey J. Stein" To: sdm7g@Virginia.EDU Cc: Luke Tierney , stat-lisp-news@umnstat.stat.umn.edu Subject: graph overlays [was: :CLEAR-CONTENT ] In-Reply-To: References: Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Regarding graph overlays - I'm not sure I've completely followed all the discussions, but there are things that I've been doing or have wanted to do which I've found difficult under the current implementation. My hope is that any changes made would help to support these sorts of actions. For example, I've wanted boxes around 3d plots instead of axes. Under the current implementation, what I did was turn off the axes and add lines to the plot. However, I think this is an ugly solution, and can interfere with plot manipulation (because of the extra elements in the plot). It's also a pain because one needs to keep track of the lines added so that one can turn off the axes again, (if they haven't already been erased by a :clear-lines message). I suppose an alternative would have been to use the overlay facilities, but then I'd have to go and do the 3d -> screen mappings myself, which is a pain. Add to this all the additional things one would want from a box around one's data, such as perhaps a grid on the bottom & two back sides of the box, or tick marks, or labels with values, and both of the above seem like poor solutions. Another thing that I've found to be awkward is working with colored plots. For example, if I want to plot 3 functions in 3 different colors, I have to, for each function, record the number of lines, plot the function, get the number of lines, and change the color of the new lines. I'd be nice if there was an easier way to do this - perhaps default colors that wouldn't affect the entire graph, but would just be used for all new elements added to a graph. One solution which comes to mind would be if one could construct a set of graph protos which all draw in the same window. Another possibility might be for a graph proto to have a list of children graph protos, which again all draw in the same window. I guess one would have to be careful about which messages go just to the parent, and which also go to the children. But, once one can construct different graph protos which draw on the same window, one can fairly easily wire things up in an appropriate fashion. -- Dr. Harvey J. Stein Berger Financial Research abel@netvision.net.il From majordom@umnstat.stat.umn.edu Wed Mar 5 09:06:49 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05264; Wed, 5 Mar 1997 09:06:49 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05259; Wed, 5 Mar 1997 09:06:43 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa02141; 5 Mar 97 10:06 EST Received: from slip-1-12.acc.Virginia.EDU by elvis.med.Virginia.EDU (5.65c/1.34) id AA18071; Wed, 5 Mar 1997 10:06:14 -0500 Date: 5 Mar 97 10:08:48 -0500 Subject: Re: graph overlays [was: :CLEAR-CONTENT ] From: Majewski To: "Harvey J. Stein" Cc: stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >For example, I've wanted boxes around 3d plots instead of axes. Under >the current implementation, what I did was turn off the axes and add >lines to the plot. However, I think this is an ugly solution, and can >interfere with plot manipulation (because of the extra elements in the >plot). It's also a pain because one needs to keep track of the lines >added so that one can turn off the axes again, (if they haven't >already been erased by a :clear-lines message). I suppose an >alternative would have been to use the overlay facilities, but then >I'd have to go and do the 3d -> screen mappings myself, which is a >pain. > >Add to this all the additional things one would want from a box around >one's data, such as perhaps a grid on the bottom & two back sides of >the box, or tick marks, or labels with values, and both of the above >seem like poor solutions. I agree. I haven't looked at the spin-proto implementation yet. I suppose a 3d version of my 2-d graphic overlays could be done. The 2d reals only have to call :real-to-canvas. I suppose spin-proto has to call :apply-transformation and perhaps some other methods. For the 3d boxes, you could just redefine :draw-axes to do a box. ( You might want to add an axes-color slot and setter/getter method.) Doing the 2d real-to-canvas was not too much of a pain -- and the advantage of OO programming is that once it's done the first time, you don't have to do it again. ( That's why I have an abstract real-graf- proto that all of my real-space 2d objects inherit from. ) >Another thing that I've found to be awkward is working with colored >plots. For example, if I want to plot 3 functions in 3 different >colors, I have to, for each function, record the number of lines, >plot the function, get the number of lines, and change the color of >the new lines. I'd be nice if there was an easier way to do this - >perhaps default colors that wouldn't affect the entire graph, but >would just be used for all new elements added to a graph. This is exactly what my classes do. ( Take a look at the URL I posted for a screen dump of an example. ) I am plotting an unknown spectrum, with plot-points, and also adding an overlay that colors the area under the points. After doing a fit, I draw reference peaks scaled by the fit-results coefficients in different colors, and add a linked namelist containing the reference names. Clicking on either the name or clicking under the curve selects that reference in both linked windows. I also draw the residuals as another colored region, and I draw the continuums ( background regions that are used for measuring the total mass and are needed to scale the coeffecients into concentration. ) in another fill color. [ These plots are quite "busy" which is why I needed to use filled areas as an alternative to line plots. ] All of these overlays are individually selectable and their line widths, line-color, fill-color, state, etc. can me changed, as well as their order in the overlay list. The jitterplot examples do something different with overlays: they do jittered ( randomized slightly in the x-dimension ) distribution plot of grouped data, and they add a standard boxplot in an overlay, as well as a cloud-plot -- 2 filled ovals of 1 and 2 standard deviations. It also adds a grouping to the data points, so you can easily change the color of all the points in group-2, for example. -- Steve Majewski From majordom@umnstat.stat.umn.edu Wed Mar 5 13:10:07 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06938; Wed, 5 Mar 1997 13:10:07 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06934; Wed, 5 Mar 1997 13:10:00 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa28081; 5 Mar 97 14:03 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA20574; Wed, 5 Mar 1997 14:03:00 -0500 Date: Wed, 5 Mar 1997 14:02:59 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Cc: Luke Tierney Subject: spin-proto bugs in 3.49 ? Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk After messages about trying overlays in 3-d plots, I was trying out some of the spin-proto functions ( without any of my mods loaded ) and I find: spin-function, without adding a optional :color keyword arg gives the error: Error: unknown color - NIL Happened in: # If you append ":color 'red" or some other color to the arg list you don't get an error. I think I got a similar error with :add-function method. With :abcplane method, I get the same error, however, if you try to add a :color keyword arg to the method call, it complains about too many args. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Wed Mar 5 13:19:55 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA07026; Wed, 5 Mar 1997 13:19:55 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA07022; Wed, 5 Mar 1997 13:19:49 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa03278; 5 Mar 97 14:12 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA15540; Wed, 5 Mar 1997 14:12:34 -0500 Date: Wed, 5 Mar 1997 14:12:33 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Cc: Luke Tierney Subject: spin-proto color bug... Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk It's looks like an old bug I uncovered. It looks like the problem is that :add-surface has a color keyword argument with no default value. ( defaults to Nil ) in graph2.lsp: (defmeth spin-proto :add-surface (x y z &key (draw t) (type 'solid) (spline 3) color) :add-function will pass a color arg on via apply: (apply #'send self :add-surface x y z args))) but :abcplane does calls :add-function with no extra args, so color defaults to Nil. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Wed Mar 5 14:10:21 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA07512; Wed, 5 Mar 1997 14:10:21 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA07507; Wed, 5 Mar 1997 14:10:09 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa02007; 5 Mar 97 15:07 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA17777; Wed, 5 Mar 1997 15:07:22 -0500 Date: Wed, 5 Mar 1997 15:07:21 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Cc: Luke Tierney Subject: other 3.49 problems ? In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The xlispstat-3-49-files.sea distribution seems to be missing some lisp sources -- loading from the workspace is OK, but if I try to load init, I get errors from missing common2.lsp and others. ( These missing files are in the source code distribution though. ) If I try to load graph2.lsp, I get the error: Error: not a valid graph address - try reallocating the object Happened in: # which appears to be from the line: (send spin-proto :scale-type 'variable) [ Found while trying to make a fix for the previously posted bug. ] ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@nokomis.stat.umn.edu Wed Mar 5 14:43:18 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA07901; Wed, 5 Mar 1997 14:43:18 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA07896; Wed, 5 Mar 1997 14:43:09 -0600 From: Luke Tierney Message-Id: <9703052043.AA07896@nokomis.stat.umn.edu> Subject: Re: other 3.49 problems ? To: sdm7g@Virginia.EDU (Steven D. Majewski) Date: Wed, 5 Mar 1997 14:43:09 -0600 (CST) Cc: stat-lisp-news@nokomis.stat.umn.edu, luke@nokomis.stat.umn.edu In-Reply-To: from "Steven D. Majewski" at Mar 5, 97 03:07:21 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 817 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > > The xlispstat-3-49-files.sea distribution seems to be missing some > lisp sources -- loading from the workspace is OK, but if I try to > load init, I get errors from missing common2.lsp and others. > ( These missing files are in the source code distribution though. ) > I'll look into that > If I try to load graph2.lsp, I get the error: > > Error: not a valid graph address - try reallocating the object > Happened in: # > > > which appears to be from the line: > > (send spin-proto :scale-type 'variable) > Change this to (send spin-proto :scale-type 'variable :draw nil) The color problem in spin-function is new -- I added :color to :add-surface but never checked it without color. This is best handled by a little internal change in the C code, which I have made. luke From majordom@umnstat.stat.umn.edu Thu Mar 6 17:33:05 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27479; Thu, 6 Mar 1997 17:33:05 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27473; Thu, 6 Mar 1997 17:32:56 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa14462; 6 Mar 97 18:32 EST Received: from bootp-17-17.bootp.Virginia.EDU by elvis.med.Virginia.EDU (5.65c/1.34) id AA14292; Thu, 6 Mar 1997 18:32:54 -0500 Date: 6 Mar 97 18:38:12 -0500 Subject: (get-decoded-time) to ascii time-stamp object From: sdm7g@Virginia.EDU To: stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Here's a first hack at an object to format the (get-decoded-time) values into an ascii string like that output by posix ctime(). -- Steve M. ;; ( sec min hr date month year day-of-week DST TZ ) ( defproto time-stamp '( times ) '( weekdays months )) ( send time-stamp :slot-value 'weekdays (list "Sun" "Mon" "Tue" "Wed" "Thu" "Fri" "Sat" )) ( send time-stamp :slot-value 'months (list "Jan" "Feb" "Mar" "Apr" "May" "Jun" "Jul" "Aug" "Sep" "Oct" "Nov" "Dec" )) ( defmeth time-stamp :now () ( slot-value 'times ( multiple-value-list (get-decoded-time)))) ( defmeth time-stamp :isnew () ( send self :now )) ( defmeth time-stamp :weekday () ( elt ( slot-value 'weekdays ) ( elt (slot-value 'times) 6 ))) ( defmeth time-stamp :month () ( elt ( slot-value 'months ) ( elt (slot-value 'times) 4 ))) ( defmeth time-stamp :yr () ( elt ( slot-value 'times ) 5 )) ( defmeth time-stamp :asc-time () ( apply #'format Nil "~2,0d:~2,0d:~2,0d" ( reverse ( subseq ( slot-value 'times ) 0 3 )))) ( defmeth time-stamp :day () ( elt ( slot-value 'times ) 3 )) ( defmeth time-stamp :ctime () ( format nil "~A ~A ~D ~A ~D" ( send self :weekday ) ( send self :month ) ( send self :day ) ( send self :asc-time ) ( send self :yr ))) ( defmeth time-stamp :print ( &optional (str nil) ) ( format str "~A" ( send self :ctime ))) From majordom@umnstat.stat.umn.edu Thu Mar 6 17:47:17 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27609; Thu, 6 Mar 1997 17:47:17 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27605; Thu, 6 Mar 1997 17:47:08 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa25225; 6 Mar 97 18:47 EST Received: from bootp-17-17.bootp.Virginia.EDU by elvis.med.Virginia.EDU (5.65c/1.34) id AA22282; Thu, 6 Mar 1997 18:47:06 -0500 Date: 6 Mar 97 18:52:24 -0500 Subject: modal dialog focus From: sdm7g@Virginia.EDU To: stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk OK -- I'm doing the following to create a dialog with several text fields. [ using Mac XlispStat 3.49 ] When it's created, the cursor/focus is in the last editable-text item. Is there a way to force it to the first element in the dialog ? [ Any other tips on how to handle the formatting for this sort of thing ? ] -- Steve M. ( defun ledit ( label &optional ( def "" ) ( tlen 40 )) ( list ( send text-item-proto :new label :size '( 100 16 )) ( send edit-text-item-proto :new def :size '( 200 16 )))) ( defun get-many-strings-dialog ( &rest labels ) ( let* (( items ( mapcar #'ledit labels )) ( items ( append items ( list ( list (send text-item-proto :new "Comments:" :size '( 100 16) ) (send edit-text-item-proto :new "" :size '( 200 100 )))))) ( d ( send ok-or-cancel-dialog-proto :new items :ok-action #'(lambda () ( split-list ( sendtoall (combine items) :text ) 2 ))))) ( send d :modal-dialog ))) ( get-many-strings-dialog "Experimenter:" "Specimin-ID:" "Structure:" "Treatment" ) From majordom@umnstat.stat.umn.edu Thu Mar 6 18:13:05 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27757; Thu, 6 Mar 1997 18:13:05 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27749; Thu, 6 Mar 1997 18:12:54 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa13579; 6 Mar 97 19:12 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA18539; Thu, 6 Mar 1997 19:12:48 -0500 Date: Thu, 6 Mar 1997 19:12:47 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: modal dialog focus ... and don't show on creation ? In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I changed the previous function to return the dialog rather than to call it's :modal-dialog method. This works more precisely as I wanted: the editable-text dialog items keep their :text values once set, so that they remain the defaults the next time I call :mocal-dialog. But -- how do I keep the dialog box from appearing initially when I first create the dialog. I just want to create it and keep it in reserve for when it's needed. ( which will be on every file save. ) ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Thu Mar 6 20:05:38 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28348; Thu, 6 Mar 1997 20:05:38 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28344; Thu, 6 Mar 1997 20:05:30 -0600 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.7.5/10.1) id VAA24970; Thu, 6 Mar 1997 21:05:27 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 6 Mar 1997 21:05:53 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: Forrest Young Subject: Bug? Feature? Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi folks: I've got a problem I'm wondering about with the functions sv-decomp and eigen. I'm using XLispStat 3.44 on a Mac. If I have a square symmetric (indefinite) matrix B and I do (setf svd (sv-decomp B)) (setf singvals (second svd)) Then I get singular values which, as advertised, are in decreasing order. If, instead, I do (setf eig (eigen B)) (setf evals (first eig)) Then I get eigenvalues that are the same as the singular values, except that some differ in sign. Since the eigenvalues are also arranged in order, they are ordered differently than the singular values. If I do (setf evals2 (reverse (sort-data (abs evals)))) Then evals2 is the same as singvals, within acceptable tolerance. (The functions eigenvalues and eigenvectors work the same as eigen.) Is this a bug? Is this as it should be? If not a bug, why? ----- Forrest Young Professor, Psychometrics CB# 3270 Davie Hall email: mailto:forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software and Documentation site: http://forrest.psych.unc.edu/research/ViSta.html ViSta Mirrors: http://ssda.anu.edu.au/mirrors/ViSta.html Australia http://sunsite.univie.ac.at/ViSta Austria http://micros.hensa.ac.uk/mirrors/vista UK (ac.uk only 8AM-8PM GMT) http://www.psych.unc.edu/fwy/research USA From majordom@umnstat.stat.umn.edu Fri Mar 7 12:02:20 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10885; Fri, 7 Mar 1997 12:02:20 -0600 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10877; Fri, 7 Mar 1997 12:02:11 -0600 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA08862 (5.67b/IDA-1.5 for ); Fri, 7 Mar 1997 13:00:55 -0500 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA3089 for ; Fri, 7 Mar 1997 13:00:08 -0400 Message-Id: <33208248.727E@kodak.com> Date: Fri, 07 Mar 1997 13:02:00 -0800 From: Jeff Morris Reply-To: morris@Kodak.COM Organization: Eastman Kodak Company X-Mailer: Mozilla 3.0 (Win16; I) Mime-Version: 1.0 To: xlispstatnews Subject: GLIM Question Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I am trying to fit a loglinear model with a linear trend and am having trouble. The example comes from Haberman 1978 and is a 2x2 table of attitude rating by year. The attitudes are "Too Harshly", "Not Harshly Enough", "About Right", "Don't Know", "No Answer", for the years 1972-1975. I've defined the variables as listed below and fit a log-linear model of independence which works fine. But how do I introduce into the model attitude*(linear year)? As always, thanks in advance for the help! Jeff > (def attitude (repeat '(th nh ar dk na) '(4 4 4 4 4))) ATTITUDE > (def year (repeat '(72 73 74 75) 5)) YEAR > (def yearl '(72 73 74 75)) YEARL > (def number '(105 68 42 61 1066 1092 580 1174 265 196 72 144 173 138 51 104 4 10 8 7)) (poissonreg-model (append (indicators attitude) (indicators year)) number) . . . output . . . -- Jeffrey D. Morris, Ph.D. Health & Environment Labs Eastman Kodak Company morris@kodak.com From majordom@umnstat.stat.umn.edu Fri Mar 7 13:19:04 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11585; Fri, 7 Mar 1997 13:19:04 -0600 Return-Path: Received: from sulu.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11581; Fri, 7 Mar 1997 13:18:56 -0600 Received: from zen by sulu.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id LAA00819; Fri, 7 Mar 1997 11:18:28 -0800 (PST) Date: Fri, 7 Mar 1997 11:18:26 -0800 (PST) From: Thomas Lumley X-Sender: thomas@zen Reply-To: Thomas Lumley To: Jeff Morris Cc: xlispstatnews Subject: Re: GLIM Question In-Reply-To: <33208248.727E@kodak.com> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Fri, 7 Mar 1997, Jeff Morris wrote: >> the variables as listed below and fit a log-linear model of >> independence which works fine. But how do I introduce into the >> model attitude*(linear year)? Try this: > (def year-by-attitude (mapcar #'(lambda (x) (* x year)) (indicators attitude)) ) > year-by-attitude ((0 0 0 0 72 73 74 75 0 0 0 0 0 0 0 0 0 0 0 0) (0 0 0 0 0 0 0 0 72 73 74 75 0 0 0 0 0 0 0 0) (0 0 0 0 0 0 0 0 0 0 0 0 72 73 74 75 0 0 0 0) (0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 72 73 74 75)) I have also written a model formula package (available from the UCLA archive or http://www.biostat.washington.edu/~thomas/gee.html) which allows slightly simpler specification of these sorts of models: eg (def your-model (as-formula '((factor year) (factor atts) (interaction (list year atts) :is-factor (list nil t))))) returns a model formula object (here atts is a string version of attitudes using eg "th" instead of th). This object contains a design matrix, variable names, and grouping information that can be used to construct eg a Wald test for the set of four interaction variables). There are also modifications for regression-model so that it accepts model formula objects as predictors and displays these Wald tests. The following function extends this modification to poisson regression (defun poissonreg-formula (formula-object y &rest args &key (predictor-names nil has-names) (print-summary t) (intercept t) (print t)) "Like poissonreg-model but uses a formula instead of the x argument" (let* ( (name-list (if has-names predictor-names (send formula-object :name-list))) (result (apply #'poissonreg-model (list (send formula-object ':design-matrix) y ':predictor-names name-list ':print (null print-summary) args))) ) (send result :add-slot 'formula formula-object) (if print-summary (send result :display-with-formula t)) result )) so that > (poissonreg-formula your-model number) Iteration 1: deviance = 14.1668 Iteration 2: deviance = 13.8720 Block Wald Chisq p-value YEAR 359.79 0.0000 ATTS 56.481 0.0000 (INTERACTION YEAR ATTS) 71.252 0.0000 and > (send * :display) Weighted Least Squares Estimates: Constant 5.55276 (5.565291E-2) (YEAR 73) -0.249581 (4.693474E-2) (YEAR 74) -1.13147 (7.637781E-2) (YEAR 75) -0.648430 (0.102355) (ATTS DK) -2.87334 (3.86849) (ATTS NA) -28.8277 (11.7598) (ATTS NH) -16.3790 (2.68014) (ATTS TH) -5.02354 (4.56693) (YEAR (ATTS DK)) 3.416852E-2 (5.286612E-2) (YEAR (ATTS NA)) 0.349938 (0.159762) (YEAR (ATTS NH)) 0.247364 (3.661617E-2) (YEAR (ATTS TH)) 5.638767E-2 (6.238886E-2) Scale taken as: 1.00000 Deviance: 13.8714 Number of cases: 20 Degrees of freedom: 8 NIL Unfortunately the coding for the dummy variables is not the same as given by (indicators). I code the alphanumerically lowest category by zeroes, (indicators) codes the first observation by zeroes. Thomas Lumley ------------------------------------------------------+------ Biostatistics : "Never attribute to malice what : Uni of Washington : can be adequately explained by : Box 357232 : incompetence" - Hanlon's Razor : Seattle WA 98195-7232 : : ------------------------------------------------------------ From majordom@umnstat.stat.umn.edu Fri Mar 7 15:20:04 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13060; Fri, 7 Mar 1997 15:20:04 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13054; Fri, 7 Mar 1997 15:19:50 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id ab08474; 7 Mar 97 16:19 EST Received: from slip-5-7.acc.Virginia.EDU by elvis.med.Virginia.EDU (5.65c/1.34) id AA21785; Fri, 7 Mar 1997 16:19:19 -0500 Date: 7 Mar 97 16:20:59 -0500 Subject: Re: XLISP-STAT: Can I change standard font? From: Steve Majewski To: "Dom. Early" Cc: stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Cyberdog/2.0b1 Newsgroups: comp.lang.lisp,comp.lang.lisp.x X-News-Servers: news.virginia.edu X-Newsgroups-To: nntp://news.virginia.edu/comp.lang.lisp, nntp://news.virginia.edu/comp.lang.lisp.x Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Dominic Early wrote on news:comp.lang.lisp, news:comp.lang.lisp.x : ( This reply CCed to -- you might have better luck with questions there! ) >I am using XLISP-STAT and am building a system which uses an embedded >high level language, as part of my PhD research. I would like to be able to >use non-standard characters and am wondering if it is possible to easily >change the default font used by the system? I am using a PowerMac with >XLISP STAT 3.48b, and had thought about hacking the system file to modify >the font at a system level but would much rather something less hacky. >If you can think of a cross platform solution all the better! If you create or edit a "XLS Preferences" file in the same directory as the XLISP-STAT application, you can put in lines like: ListenerFontSize 12 ListenerFontName Monaco EditFontName Courier EditFontSize 12 GraphFontName Helvetica GraphFontSize 10 Color yes to change some of the defaults. However, this does *not* change the System font (Chicago) that is used to draw the window title bars, menus, etc. For that, you either need to either hack the system resourecs to do it globally, or modify XlispStat to do it locally. [ "XLS Preferences" is documented in the doc/changes.* file ] >I am also trying to construct some complex dialog boxes under XLISP-STAT, >if >anyone has any examples I would be very greatful for some sample code. A >particular problem I am having at the moment is with list-item-proto's. I >override their :do-action method to do something but this method is not >called when I single click on the list-item-proto, only when I double click >on it. Is this (dare I say it!) a bug or a deliberate feature? I just >think it strange when the parameter list for the standard :do-action >method >is as follows do action(&optional dbl-click) which has a boolean parameter >indicating whether it is a double-click or not, yet the method is only >called on a double click, making the parameter redundant! > >What can I say about XLISP-STAT, its a great system, cheers Luke T. & co! > ------ Example 1: non-modal-dialog -- the :action methods for each item modify the target when clicked. This makes a non-modal dialog window that allows setting of the graphs x & y max and turns on full view or partial scrolling. ( I note that my list-item also requires a double click -- I guessing this is a feature. ) This is one of my earliest attempts. I'ld probably do it differently now: ------ ( defmeth scatterplot-proto :make-range-widget () ( let* ((xmax ( cadr (send self :range 0))) (ymax ( cadr (send self :range 1))) (hmax-label ( send text-item-proto :new "Horiz. max:" )) (hmax-value ( send text-item-proto :new "" :text-length 4 )) (hmax-scroll ( send interval-scroll-item-proto :new ( list 1 xmax ) :text-item hmax-value :action #'(lambda (x) ( send self :range 0 0 x )))) (vmax-label ( send text-item-proto :new "Vert. max:" )) (vmax-value ( send text-item-proto :new "" :text-length 6 )) (vmax-values (nice-dseq ymax)) (vmax-scroll ( send sequence-scroll-item-proto :new vmax-values :text-item vmax-value :action #'(lambda (y) (send self :range 1 0 y )))) (hs-top ( send text-item-proto :new "Horiz. Scroll range:" )) (hscroll ( send list-item-proto :new ( list "None" "1/2" "1/3" "1/4" "1/5" ) :selection 0 :action #'(lambda (n) ( when n ( let (( n ( send hscroll :selection ))) (send self :has-h-scroll ( if (zerop n) nil (* (car (send self :size)) (1+ n))))))))) (vscroll (send toggle-item-proto :new "use vertical scroll" ))) ------ example 2: modal dialog. For a modal dialog, you usually don't need an :action for the dialog items except for the final OK that dismisses the dialog. That needs to gather up all of the :text or :value values from the items. When called with something like: ( setf dlg ( many-strings-dialog "Experimenter:" "Specimen-ID:" "Structure:" "Treatment" )) ( send dlg :modal-dialog ) It will return something like: (("Experimenter:" "Steve") ("Specimen-ID:" "10111") ("Structure:" "Cyto") ("Treatment" "control") ("Comments:" "")) [ i.e. a list of labels and their edit-text entries. ] and when you call :modal-dialog again, it will appear with those values as the new defaults. ------ ( defun ledit ( label &optional ( def "" ) ( tlen 40 )) ( list ( send text-item-proto :new label :size '( 100 16 )) ( send edit-text-item-proto :new def :size '( 200 16 )))) ( defun many-strings-dialog ( &rest labels ) ( let* (( items ( mapcar #'ledit labels )) ( items ( append items ( list ( list (send text-item-proto :new "Comments:" :size '( 100 16) ) (send edit-text-item-proto :new "" :size '( 200 100 )))))) ( d ( send ok-or-cancel-dialog-proto :new items :show nil :ok-action #'(lambda () ( split-list ( sendtoall (combine items) :text ) 2 ))))) d )) ---- Steve Majewski From majordom@nokomis.stat.umn.edu Fri Mar 7 16:31:28 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA06485; Fri, 7 Mar 1997 16:31:28 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA06480; Fri, 7 Mar 1997 16:31:19 -0600 From: Luke Tierney Message-Id: <9703072231.AA06480@nokomis.stat.umn.edu> Subject: Re: XLISP-STAT: Can I change standard font? To: sdm7g@Virginia.EDU (Steve Majewski) Date: Fri, 7 Mar 1997 16:31:19 -0600 (CST) Cc: Dom.Early@tcd.ie, stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Steve Majewski" at Mar 7, 97 04:20:59 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 584 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > ------ > Example 1: non-modal-dialog -- the :action methods for each item > modify the target when clicked. This makes a non-modal dialog window > that allows setting of the graphs x & y max and turns on full view or > partial scrolling. ( I note that my list-item also requires a double > click -- > I guessing this is a feature. ) This is one of my earliest attempts. I'ld > It is a bug in the mac version -- on Windows and X11 single clicks call the method also and they should on the mac too. I'll fix that (and break someone's mac program I'm sure :-)) luke From majordom@umnstat.stat.umn.edu Mon Mar 10 11:31:59 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23078; Mon, 10 Mar 1997 11:31:59 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23074; Mon, 10 Mar 1997 11:31:48 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa19344; 10 Mar 97 12:31 EST Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA17158; Mon, 10 Mar 1997 12:31:38 -0500 Date: Mon, 10 Mar 1997 12:31:38 -0500 (EST) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Cc: Luke Tierney Subject: an extra level of indirection/abstraction in plot functions Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk A discussion about graphics API's on another list reminded me of this issue -- I'ld like to float it by and see if it is a concern to anyone else: I have sometimes been forced to use modifications to scatterplot-proto's behaviour, rather than creating a subtype, because the standard ploting function have a hardwired prototype. This applies for both major changes like my overlay graphics, and minor setting of defaults, like size or background color. If an extra level of indirection/abstraction is inserted into those default functions, for example, if plot-lines, instead of calling scatterplot-proto's :new method, tries to create an instance of "user-lineplot-proto" - which would by default be bound to scatterplot-proto -- this would leave room for insertion of user plot defaults without having to modify the base classes. My "lisp logic" tells me not to expect any difference in performance in the default case where "user-lineplot-proto" is just an alias to scatterplot-proto, but I haven't tested this. [ BTW: *Is* there an overhead associated with the depth of inheritance in classes ? Is there any performance reason to prefer flatter class hierarchies ? ] ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Mon Mar 10 14:46:36 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24423; Mon, 10 Mar 1997 14:46:36 -0600 Return-Path: Received: from gateway.ets.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24419; Mon, 10 Mar 1997 14:46:26 -0600 Received: from research.ets.org by gateway.ets.org with SMTP id AA22898 (5.65c/IDA-1.4.4); Mon, 10 Mar 1997 15:45:07 -0500 Received: from comet.research.ets.org by research.ets.org with SMTP id AA19181 (5.67b/IDA-1.4.3 for sdm7g@Virginia.EDU); Mon, 10 Mar 1997 15:45:06 -0500 Received: by comet.research.ets.org (SMI-8.6/SMI-SVR4) id PAA17796; Mon, 10 Mar 1997 15:45:06 -0500 Date: Mon, 10 Mar 1997 15:45:06 -0500 Message-Id: <199703102045.PAA17796@comet.research.ets.org> From: Russell Almond To: sdm7g@Virginia.EDU Cc: stat-lisp-news@umnstat.stat.umn.edu, luke@umnstat.stat.umn.edu In-Reply-To: (sdm7g@Virginia.EDU) Subject: Re: an extra level of indirection/abstraction in plot functions Reply-To: ralmond@ets.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk If an extra level of indirection/abstraction is inserted into those default functions, for example, if plot-lines, instead of calling scatterplot-proto's :new method, tries to create an instance of "user-lineplot-proto" - which would by default be bound to scatterplot-proto -- this would leave room for insertion of user plot defaults without having to modify the base classes. This is an example of a "Meta-Object Protocol" or what I was calling an Open Implementation Protocol in several "generic" functions which can be specialized. Therefore if you were using Common Lisp (with CLOS as opposed to Xlisp) you might write: (defgeniric fit-model (model-type data-set) (:documentation "Fits a model and prints a table of summary statistics.")) (defgeneric build-data-matrix (model-type data-set) (:documentation "Builds the data matrix for fitting a particular model.") (:returns y x)) (defgeneric solve-data-matrix (model-type y x) (:documentation "Finds the parameters of the model.")) (defgeneric print-model-summary (fit-model) (:documentation "Prints the results of the model fit.")) (defmethod fit-model ((model-type Model) (data-set T)) (multiple-value-bind (y x) (build-data-matrix model-type data-set)) (let ((fit-model (solve-data-matrix model-type y x))) (print-model-summary fit-model) (values fit-model model-type y x))) The idea is that if you were just introducing a new type of model (say GLM as opposed to LM) you might only need to rewrite the solve-data-matrix method for your new model type (as long as your interface for the fit model type was sufficiently well defined). But you might need to rewrite the generic function to handle something more complex (like multiple imputation) even though you could still call some of the same component functions. Probably the best place to read about this kind of thinking is in the book "The Art of the Metaobject Protocol" by Gregor Kiczales, Jime Des Rivieres and Daniel Bobrow. My "lisp logic" tells me not to expect any difference in performance in the default case where "user-lineplot-proto" is just an alias to scatterplot-proto, but I haven't tested this. [ BTW: *Is* there an overhead associated with the depth of inheritance in classes ? Is there any performance reason to prefer flatter class hierarchies ? ] My feeling for xlisp-stat was that most of the method dispatch was handled at run-time, unlike CLOS where quite a bit was done at compile time, especially if you carefully declared all of the variables. (John Macdonald wrote some code in CLOS which was as fast or faster than similar code in Fortran: he just needed to make all of the declarations which were required in Fortran and optional in Lisp.) On the other hand, the performance overhead is just an extra method call (or in the case of a deeply nested hierarchy a linear search through the class heirarchy). You need to weigh performance versus the need for code re-use. On modern machines programmer time is usually more expensive than CPU cycles. --Russell Almond Educational Testing Service Research Statistics Group, 15-T Princeton, NJ 08541 Phone: 609-734-1557 FAX: 609-734-5420 Email: almond@acm.org, ralmond@ets.org http://bayes.stat.washington.edu/almond/almond.html From majordom@umnstat.stat.umn.edu Wed Mar 12 14:53:45 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27024; Wed, 12 Mar 1997 14:53:45 -0600 Return-Path: Received: from stats2.asu.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27016; Wed, 12 Mar 1997 14:53:37 -0600 Received: from stats2 (localhost [127.0.0.1]) by stats2.asu.edu with SMTP (8.7.1/8.7.1) id NAA23996 for ; Wed, 12 Mar 1997 13:53:35 -0700 (MST) From: daddy@imap2.asu.edu Date: Wed, 12 Mar 1997 13:53:34 -0700 (MST) X-Sender: daddy@stats2 To: stat-lisp-news@umnstat.stat.umn.edu Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello Folks- I have two questions. First, is the list archived, and if so, how can I view it? Secondly, how can I view the original code for graph-proto? I want to make a special graph and need to alter some things. I have search the .lsp files that came with my distribution, but have not found the code. One thing I would like to do for example would be to have a histogram with frequncies along the ticks of the y-axis rather than the proportion. Thanks- Dan Dan Huston Phone (602) 965-2420 Measurement, Statistics and Methodological Studies Fax (602) 965-0300 Psychology in Education dan.huston@asu.edu 325 Payne Hall http://seamonkey.ed.asu.edu/~huston Arizona State University Tempe, AZ 85287-0611 From majordom@umnstat.stat.umn.edu Wed Mar 12 15:46:57 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27516; Wed, 12 Mar 1997 15:46:57 -0600 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27512; Wed, 12 Mar 1997 15:46:47 -0600 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA16969 (5.67b/IDA-1.5 for ); Wed, 12 Mar 1997 16:45:35 -0500 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA28222 for ; Wed, 12 Mar 1997 16:44:47 -0400 Message-Id: <33274E83.6ECE@kodak.com> Date: Wed, 12 Mar 1997 16:46:59 -0800 From: Jeff Morris Reply-To: morris@Kodak.COM Organization: Eastman Kodak Company X-Mailer: Mozilla 3.0 (Win16; I) Mime-Version: 1.0 To: xlispstatnews Subject: regression with categories and x's Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I have run into a curious lispish thing in fitting a regression model (also GLIMs) when I include both a quantitative variable, say dose, and a category, say prep. In order to include the category I can use the (indicators prep) command to create the indicator variables for prep. What you get is a list where each element is a list. For example: dose (3.4 5.2 7.0 8.5 10.5 13 18 21 28 6.5 10 14 21.5 29) > prep (STD STD STD STD STD STD STD STD STD TEST TEST TEST TEST TEST) > (indicators prep) ((0 0 0 0 0 0 0 0 0 1 1 1 1 1)) If I want now to fit a logistic model to this data, I would naively do the following: (binomialreg-model (append dose (indicators prep)) r n) for > r (0 5 11 14 18 21 23 30 27 2 10 18 21 27) and > n (33 32 38 37 40 37 31 37 30 40 30 40 35 37) But when I do I get the error: Error: dimensions do not match Happened in: # But, of course, if I define dose so that it is a list with its one element a list, everything works out fine, i.e. (def dose-2 '((3.4 5.2 7.0 8.5 10.5 13 18 21 28 6.5 10 14 21.5 29))) (binomialreg-model (append dose-2 (indicators prep)) r n) then everything works fine. Can anyone tell me a better way to define the dose so that I don't have to create dose-2 as I did above or a better way to define the indicator variables for prep? (I realize that prep, with only 2 levels, can be coded easily without using the indicators command, but I would like be able to use it for a variable with an arbitrary number of levels) Thanks as always! Jeff -- Jeffrey D. Morris, Ph.D. Health & Environment Labs Eastman Kodak Company morris@kodak.com From majordom@umnstat.stat.umn.edu Wed Mar 12 16:10:40 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27825; Wed, 12 Mar 1997 16:10:40 -0600 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27821; Wed, 12 Mar 1997 16:10:29 -0600 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA18709 (5.67b/IDA-1.5 for ); Wed, 12 Mar 1997 17:09:22 -0500 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA28556 for ; Wed, 12 Mar 1997 17:08:34 -0400 Message-Id: <33275416.986@kodak.com> Date: Wed, 12 Mar 1997 17:10:46 -0800 From: Jeff Morris Reply-To: morris@Kodak.COM Organization: Eastman Kodak Company X-Mailer: Mozilla 3.0 (Win16; I) Mime-Version: 1.0 To: xlispstatnews Subject: Re: regression with categories and x's References: <33274E83.6ECE@kodak.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Let me be the first to comment on my question by realizing that I left off the link structure, so that the model statements should read: (binomialreg-model (append dose (indicators prep)) r n :link logit-link) Wouldn't want to confuse anyone :>) Sorry! Jeff Morris wrote: > > I have run into a curious lispish thing in fitting a regression > model (also GLIMs) when I include both a quantitative variable, say > dose, and a category, say prep. In order to include the category > I can use the (indicators prep) command to create the indicator > variables for prep. What you get is a list where each element is > a list. For example: > > dose > (3.4 5.2 7.0 8.5 10.5 13 18 21 28 6.5 10 14 21.5 29) > > prep > (STD STD STD STD STD STD STD STD STD TEST TEST TEST TEST TEST) > > (indicators prep) > ((0 0 0 0 0 0 0 0 0 1 1 1 1 1)) > > If I want now to fit a logistic model to this data, I would > naively do the following: > > (binomialreg-model (append dose (indicators prep)) r n) > > for > > r > (0 5 11 14 18 21 23 30 27 2 10 18 21 27) > and > > n > (33 32 38 37 40 37 31 37 30 40 30 40 35 37) > > But when I do I get the error: > Error: dimensions do not match > Happened in: # > > But, of course, if I define dose so that it is a list with its one > element a list, everything works out fine, i.e. > > (def dose-2 '((3.4 5.2 7.0 8.5 10.5 13 18 21 28 6.5 10 14 21.5 29))) > > (binomialreg-model (append dose-2 (indicators prep)) r n) > > then everything works fine. Can anyone tell me a better way to define > the dose so that I don't have to create dose-2 as I did above or a > better way to define the indicator variables for prep? > > (I realize that prep, with only 2 levels, can be coded easily without > using the indicators command, but I would like be able to use it for > a variable with an arbitrary number of levels) > > Thanks as always! > > Jeff > > -- > Jeffrey D. Morris, Ph.D. > Health & Environment Labs > Eastman Kodak Company > morris@kodak.com -- Jeffrey D. Morris, Ph.D. Health & Environment Labs Eastman Kodak Company morris@kodak.com From majordom@umnstat.stat.umn.edu Wed Mar 12 18:15:16 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28664; Wed, 12 Mar 1997 18:15:16 -0600 Return-Path: Received: from salmon.maths.tcd.ie by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28660; Wed, 12 Mar 1997 18:15:03 -0600 Received: from orange.stats.tcd.ie by salmon.maths.tcd.ie with SMTP id aa11412; 13 Mar 97 0:14 +0000 Received: from [194.125.148.193] by stats.tcd.ie (4.1/SMI-4.1) id AA27968; Thu, 13 Mar 97 00:02:25 GMT X-Sender: dbearly@orange.stats.tcd.ie (Unverified) Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 13 Mar 1997 00:14:17 +0000 To: stat-lisp-news@umnstat.stat.umn.edu From: "Dom. Early" Subject: HOW-TO? Problem relating to spin-plots Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, I am trying to build some way of interpreting a mouse-click on a plot that returns the location (in real coordinates, i.e. w.r.t. the data of the plot, not canvas coordinates) of the click. For scatter-plots this is a pretty simple matter, I override the do-click method and use the graph-proto :canvas-to-real method to work out the scaling from canvas coordinates to real coordinates. However the problem is considerably more difficult with spin-plots because of their 3-D nature, and the fact that we view them as 3-d objects projected onto a 2-d screen. In the simplest case consider a spin-plot where the view point is parallel to the z axis through the origin (exactly the view you see when you create a spin-plot inititally), I will call this view point V. A mouse-click can be interpreted as the projection of a line parallel to the z-axis, where the line is of the form z = ax + by. Specifically if a click occurs at (10, 20), say, in canvas coordiantes, we can transform these to say (5, 10) in real coordinates using :canvas-to-real method. Now we can correctly state that the point clicked represents a line from the point (5, 10, -infinity) to the point (5, 10, +infinity). Things get hairy when we allow (as XLispStat so nicely does) the plot to be rotated on each axis. Our line now takes the form z = ax + by + c. I want to be able to calculate the intersection of the line with the each axis (allowing for the possibility that the line is parallel to any axis). How should I approach this problem can anyone tell me? I do notice that spin-plots only alow a fixed and equal aspect ratio so this makes the canvas to real coordinates conversion a lot simpler (i.e. a canvas ordinate will scale to the same real ordinate be it on the x, y or z plane). I guess we can consider our line as of the form z = ax + by, then calculate some sort of transformation T which will move the view point from V to V', where T is the transformation that encapsulates all rotations that have occured to the plot from its initial state. Then apply this transformation to our line, somehow. How can I work out what rotations have occured so far to our spin-plot? How can I apply a transformation to a line? I guess a lot of this stuff is handled in the internals of spin-proto and the solution lies within (or is this wishful thinking??). I hope I've stated my problem in an understandable fashion, Any views, comments (or code, even better!!) would be greatly appreciated!! Cheers, Dom. +--------------------------------------------------------------------------+ | Dominic Early e-mail: Dom.Early@tcd.ie | | A.31, Department of Statistics, Phone: +353-1-608 1917 | | Trinity College, Fax: +353-1-661 5046 | | Dublin 2, Ireland. | | | | http://www2.tcd.ie/Statistics/staff/dbearly.html | +--------------------------------------------------------------------------+ _/_/_/_/ _/_/_/ _/ _/ _/_/_/_/_/ _/_/_/ _/_/_/_/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/_/_/_/ _/_/_/ _/ _/ _/ _/ _/_/_/ _/_/_/_/ From majordom@umnstat.stat.umn.edu Wed Mar 12 19:05:46 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28898; Wed, 12 Mar 1997 19:05:46 -0600 Return-Path: Received: from nessie.mcc.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28894; Wed, 12 Mar 1997 19:05:38 -0600 Received: from Tedistan.annex.umist.ac.uk by nessie.mcc.ac.uk with SMTP (PP); Thu, 13 Mar 1997 01:05:28 +0000 Message-Id: From: Ted.Harding@nessie.mcc.ac.uk (Ted Harding) Subject: Re: HOW-TO? Problem relating to spin-plots To: Dom.Early@tcd.ie (Dom. Early) Date: Thu, 13 Mar 1997 01:04:55 +0000 (GMT) Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: from "Dom. Early" at Mar 13, 97 00:14:17 am X-Mailer: ELM [version 2.4 PL23] Content-Type: text Content-Length: 3533 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk ( Re Message From: Dom. Early ) > > Status: > > Hi, > > I am trying to build some way of interpreting a mouse-click on a plot that > returns the location (in real coordinates, i.e. w.r.t. the data of the > plot, not canvas coordinates) of the click. > > For scatter-plots this is a pretty simple matter, I override the do-click > method and use the graph-proto :canvas-to-real method to work out the > scaling from canvas coordinates to real coordinates. However the problem > is considerably more difficult with spin-plots because of their 3-D nature, > and the fact that we view them as 3-d objects projected onto a 2-d screen. > > In the simplest case consider a spin-plot where the view point is parallel > to the z axis through the origin (exactly the view you see when you create > a spin-plot inititally), I will call this view point V. A mouse-click can > be interpreted as the projection of a line parallel to the z-axis, where > the line is of the form z = ax + by. Specifically if a click occurs at > (10, 20), say, in canvas coordiantes, we can transform these to say (5, 10) > in real coordinates using :canvas-to-real method. Now we can correctly > state that the point clicked represents a line from the point (5, 10, > -infinity) to the point (5, 10, +infinity). > > Things get hairy when we allow (as XLispStat so nicely does) the plot to be > rotated on each axis. Our line now takes the form z = ax + by + c. I want > to be able to calculate the intersection of the line with the each axis > (allowing for the possibility that the line is parallel to any axis). How > should I approach this problem can anyone tell me? > > I do notice that spin-plots only alow a fixed and equal aspect ratio so > this makes the canvas to real coordinates conversion a lot simpler (i.e. a > canvas ordinate will scale to the same real ordinate be it on the x, y or z > plane). I guess we can consider our line as of the form z = ax + by, then > calculate some sort of transformation T which will move the view point from > V to V', where T is the transformation that encapsulates all rotations that > have occured to the plot from its initial state. Then apply this > transformation to our line, somehow. How can I work out what rotations > have occured so far to our spin-plot? How can I apply a transformation to > a line? I guess a lot of this stuff is handled in the internals of > spin-proto and the solution lies within (or is this wishful thinking??). > > I hope I've stated my problem in an understandable fashion, Any views, > comments (or code, even better!!) would be greatly appreciated!! > > Cheers, > Dom. Obviously you can't click on an arbitrary point of the canvas and hope for its 3-D coordinates because a whole line corresponds to that point. What you can hope for (says he without getting his hands dirty ... ) is to be able to click AT or NEAR a data point, and get the 3-D coordinates (or some other label) of the data-point which is plotted nearest to the mouse-click. For that, the required info is in there somewhere: in fact, of course, you can already select data-points in a spin-plot (e.g. start the demo "xlispstat tourdemo" and, when the iris data come up, draw a box round some of the plotted points on the canvas; these then get highlighted and if you then rotate the plot the highlighted points rotate). So xlispstat KNOWS. I don't know how to ask it for the info, though. Yet. Best wishes, Ted. (Ted.Harding@nessie.mcc.ac.uk) From majordom@umnstat.stat.umn.edu Wed Mar 12 19:39:41 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29210; Wed, 12 Mar 1997 19:39:41 -0600 Return-Path: Received: from salmon.maths.tcd.ie by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29206; Wed, 12 Mar 1997 19:39:31 -0600 Received: from orange.stats.tcd.ie by salmon.maths.tcd.ie with SMTP id aa15041; 13 Mar 97 1:39 +0000 Received: from [194.125.148.193] ([194.125.145.142]) by stats.tcd.ie (4.1/SMI-4.1) id AA28067; Thu, 13 Mar 97 01:26:53 GMT X-Sender: dbearly@orange.stats.tcd.ie Message-Id: In-Reply-To: References: from "Dom. Early" at Mar 13, 97 00:14:17 am Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 13 Mar 1997 01:39:19 +0000 To: Ted Harding From: "Dom. Early" Subject: Re: HOW-TO? Problem relating to spin-plots Cc: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk ***I wrote : >> >> >> ... ***Ted replied: > >Obviously you can't click on an arbitrary point of the canvas and hope for >its 3-D coordinates because a whole line corresponds to that point. > >What you can hope for (says he without getting his hands dirty ... ) is to >be able to click AT or NEAR a data point, and get the 3-D coordinates >(or some other label) of the data-point which is plotted nearest to the >mouse-click. > Yup, I already realise this, but this is beyond what I am looking for, what I want is the parameters to the line equation z = ax + by + c which describes the infinitely long line on which the point where I clicked must exist. What I hope to do is to calculate the range boundries of a rectangle drawn upon a spin-plot, i.e. "selection range = ((c1 < x < c2) & (c3 < y < c4) & (c5 < z < c6))" where c1..cn are constants in range -infinity to +infinity, and before anyone says it (just let all constants equal -/+ infinity, problem solved!) I want to be as precise as possible. I have thought of, and implemented, a method which uses the values of the selected points (i.e. the points within the rectangle drawn on the spin-plot) to calculate the minimum bounding cube of the selection, but this does not describe exactly what is drawn, nor does it work for selections containing no points. Thanks Dom. +--------------------------------------------------------------------------+ | Dominic Early e-mail: Dom.Early@tcd.ie | | A.31, Department of Statistics, Phone: +353-1-608 1917 | | Trinity College, Fax: +353-1-661 5046 | | Dublin 2, Ireland. | | | | http://www2.tcd.ie/Statistics/staff/dbearly.html | +--------------------------------------------------------------------------+ _/_/_/_/ _/_/_/ _/ _/ _/_/_/_/_/ _/_/_/ _/_/_/_/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/_/_/_/ _/_/_/ _/ _/ _/ _/ _/_/_/ _/_/_/_/ From majordom@umnstat.stat.umn.edu Thu Mar 13 05:19:17 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10323; Thu, 13 Mar 1997 05:19:17 -0600 Return-Path: Received: from melia.qut.edu.au by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10317; Thu, 13 Mar 1997 05:18:58 -0600 Received: from pigeon.qut.edu.au (lunney@pigeon.qut.edu.au) by melia.qut.edu.au (PMDF V5.0-7 #20429) id <01IGGV84YSFK00BRER@melia.qut.edu.au>; Thu, 13 Mar 1997 21:21:31 +1000 Received: from localhost (lunney@localhost) by pigeon.qut.edu.au (8.8.5/8.8.5) with SMTP id VAA13889; Thu, 13 Mar 1997 21:17:50 +1000 (EST) Date: Thu, 13 Mar 1997 21:17:48 +1000 (EST) From: Paul Lunney Subject: Summary: S-plus v xlisp-stat To: s-news@utstat.toronto.edu, stat-lisp-news@umnstat.stat.umn.edu Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Content-Transfer-Encoding: 7BIT Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I got several requests for a summary of responses to a recent request for information re S-plus v xlisp-stat, so here it is. One of the posts mentions a program called R as having something of the flavour of S, but not being available in Windows. I'm not familiar with R, but along these lines I would suggest the program 'MacAnova' (http://www.stat.umn.edu/~gary/macanova/macanova.home.html). Despite the name, it's also available for other platforms. It's a great program. Which reminds me... I'd like to publicly thank those who develop and extend fine freely-available software such as xlisp-stat and MacAnova. May lots of good things come your way. My original post follows, and then the replies. Thanks again to all who replied. Paul Lunney > Could anyone who is familiar with S-plus (Win) and xlisp-stat (Win) give > me some idea of what they might use each for? ie what are the strengths > and weaknesses for different types of tasks. > > My impression is that xlisp-stat is superior graphically and may be > ultimately more flexible, but that S-plus has got oodles of built-in > functions to reduce some of the analytic programing effort. Please tell me > if this couldn't be more wrong... *---------------------------------------------------------------------. * From deleeuw@math.ucla.edu Thu Feb 27 18:34:43 1997 Date: Wed, 26 Feb 1997 13:31:28 -0800 (PST) From: Jan Deleeuw To: ralmond@ets.org Cc: p.lunney@qut.edu.au, s-news@utstat.toronto.edu, stat-lisp-news@umnstat.stat.umn.edu Subject: Re: S-plus and xlisp-stat Generally, I think the comparisons of S-plus and XLS have been fair. Also, there is one in Statistical Science some time ago. Although we have both, the five major and overriding reasons why I would not even consider using S-plus are perhaps worth repeating. (a) XLS comes with complete source code -- in ANSI C. Generally it is a breeze to port it to a new flavor of UNIX. It is quite easy to extend it at compile time, and by now it is also easy to extend it at runtime with object code. One can always go back to the source in case of doubt. One can look at the source and marvel. (b) XLS runs on the Mac, in fact that is where both XLS and I prefer to be most of the time. The other two are just necessary evils. (c) I can give it away to my students and my family, in unlimited quantities, and for their own home computers, no matter what hardware and OS they have. I can provide WWW interfaces to the program which run for free from the web servers -- I dont think the S-plus people would appreciate a similar service. (d) Porting from Common Lisp is now easy. Porting from FORTRAN to CL, and then to XL is now easy. Lisp is a general purpose language. There are lots of books on Lisp. Lisp is a beautiful language. Because of XLS I now read books such as "On Lisp" or Norvig or Steele. (e) Dynamic graphics for demos, instructional modules, and research tools cannot really be done in S. Computing and making static plots is perhaps marginally easier in S, but this is not what I do. -- Jan de Leeuw; UCLA Department of Statistics; UCLA Statistical Consulting US mail: 8118 Math Sciences, 405 Hilgard Ave, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw *---------------------------------------------------------------------. * From rga6711@comet.research.ets.org Thu Feb 27 18:34:43 1997 Date: Wed, 26 Feb 1997 16:08:04 -0500 From: Russell Almond Reply-To: ralmond@ets.org To: p.lunney@qut.edu.au Cc: s-news@utstat.toronto.edu, stat-lisp-news@umnstat.stat.umn.edu Subject: Re: S-plus and xlisp-stat Here are a couple of the major differences: 1) Graphics Engine. One of the biggest drawbacks of S is that it still has a plotter based graphics package. I find S easier to use for fancy static graphics with lots of overlays. Xlisp-stat's big advantage comes from its ability to support dynamic graphics. 2) Memory model S stores data objects on disc while XLS stores them in memory. One consequence is that S storage is persistent. XLS should also be faster with some operations as a consequence, but S will handle larger data sets (although with the price of memory, thats not as much of a problem as it used to be). 3) Programming Ease I'd say this one is a toss-up. S has a more C-like syntax for those who have a mental block about parentheses. S also has a truely bizarre object model (class identity and method dispatch are completely divorced from object encapsulation.) Also, I find the S lazy evaluation harder to understand than the Lisp eager evaluation rules. Unfortunately, xlisp lacks many of the debugging tools which makes Lisp so easy to use. (Which reminds me, did I ever post my debugging code to the list?) 4) Functionality S-Plus definetly has more, and more thorough testing. Both collections of functions reflect the research interestsof their respective authors: S tends to favor robust procedures and XLS Bayesian. On the other hand, if you really need a specific function, try SAS. 5) Support Luke does a great job for a one man show, but part of the price of S-Plus is the ability to call somebody on the phone to hold your hand while you go through the mysteries of configuring S-Plus to work on your PC with your printer. Also, there are training couses available. 6) Price XLS costs fewer dollars, but morein sweat equity. Thus, it tends to be favored by researchers and teachers who are willing to put in the effort to get it to do what they want. --Russell Almond *---------------------------------------------------------------------. * From lynd.bacon@lba.com Thu Feb 27 18:34:43 1997 Date: Wed, 26 Feb 1997 12:40:42 -0600 From: Lynd Bacon To: Paul Lunney , stat-lisp-news@umnstat.stat.umn.edu Subject: Re: S-plus and xlisp-stat At 09:23 PM 2/26/97 +1000, Paul Lunney wrote: > >Could anyone who is familiar with S-plus (Win) and xlisp-stat (Win) give >me some idea of what they might use each for? ie what are the strengths >and weaknesses for different types of tasks. > I've been an SPlus user for several years, but have only started working with xlisp-stat recently. What attracted me to xlisp-stat was: (1) the ease with which you can use it to develop and prototype dynamic graphics. (2) Its open and extensible architecture; (3) The price is right; (4) I once took a course in LISP (you don't need to know LISP to use xlisp-stat. I'm just fond of OOP languages!) >My impression is that xlisp-stat is superior graphically and may be >ultimately more flexible, but that S-plus has got oodles of built-in >functions to reduce some of the analytic programing effort. Please tell me >if this couldn't be more wrong... > Splus has a _huge_ number of objects, and the pile seems to grow daily. xlisp-stat is not exactly lacking in the area of built-in fuctions: check out Luke Tierney's 1990 book (Wiley), and you'll see what I mean! I'd say the learning curve for Splus is steeper than for xlisp-stat, but some would disagree. I'm not sure what you mean by "more flexible." Since Splus has so many functions, it may reduce programming effort if the functions are what you need. A 32-bit MS Windows version of SPlus is still being beta-tested. It should be available later this year (June, maybe?). A 32 bit version of xlisp-stat has been available for a bit. Hope the above is some help, -lynd bacon *---------------------------------------------------------------------. * From sdm7g@Virginia.EDU Thu Feb 27 18:34:43 1997 Date: Wed, 26 Feb 1997 11:59:56 +0000 From: sdm7g@Virginia.EDU To: s-news@utstat.toronto.edu, stat-lisp-news@umnstat.stat.umn.edu Cc: Paul Lunney Subject: Re: S-plus and xlisp-stat On Wed, Feb 26, 1997 11:23 AM, Paul Lunney wrote: > >Could anyone who is familiar with S-plus (Win) and xlisp-stat (Win) give >me some idea of what they might use each for? ie what are the strengths >and weaknesses for different types of tasks. Well, I'm not very familiar with them on Windows, but I've used both. >My impression is that xlisp-stat is superior graphically and may be >ultimately more flexible, but that S-plus has got oodles of built-in >functions to reduce some of the analytic programing effort. Please tell me >if this couldn't be more wrong... I think your impression is basically correct. S-plus is a higher-level, user-oriented language, with a lot of statistical procedures built in -- it's an interactive statictics application. LispStat is more flexible, but not as complete or polished. It's a language for writing interactive statictical applications or for experimenting with new types of interactive statictical graphics. If you want to do something "standard", it's going to be easier in S-plus. If you want to try to do something new or not supported in S-plus, it's probably going to be easier in LispStat. S-plus has several good books about using it for doing statistics. But the documentation on how to really customize it is a bit thin. LispStat documentation is thinner in general, but it's more programmer oriented. Plus, it's Lisp - for which there is plenty of books, tutorials, & documentation. The above is a wild and subjective generalization, but I don't think it's too far off the mark. I did quite a bit of evaluation of them both before launching on a project -- I can't swear that what I wanted to do was impossible to do in S-plus, but if it was possible, there were no examples or documentation to guide me in that sort of low level programming. Some examples of my XlispStat graphics are on: http://galen.med.virginia.edu/~sdm7g/XES/Docs/xesapp.html http://galen.med.virginia.edu/~sdm7g/XES/Docs/jitterplot-examples.html ( I've now got interface code to our spectrometer, so I can graph and analyze data real-time.) ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. *---------------------------------------------------------------------. * From kilroy@pele.pulmcc.washington.edu Thu Feb 27 18:34:43 1997 Date: Wed, 26 Feb 1997 09:36:52 -0800 (PST) From: Steve McKinney To: Paul Lunney Subject: Re: S-plus and xlisp-stat Couldn't be more right, actually. I use both on a Unix platform. Lisp-Stat's ease of programming and graphics capabilities allow me to experiment and devise new graphing techniques, but its lack of stat functions is a weak spot. Conversely, S-plus has routines that are hard to match in terms of numerical accuracy, speed and range of applications. S-plus has great graphics, but linking information in graphs and coloring are too difficult given S-plus's copy-everything style. Fortunately in Unix I can pass commands off to a background S-plus process, and pipe the results back into LispStat for graphing. Best of both worlds. I'm not sure if this is doable in Windows95 yet, but the day is approaching. I do use S-plus in the Windows environment, it is still a great package running under Windows95. I tried LispStat under windows3.1 but the 20,000 byte stack was easily overwhelmed by any reasonable recursive lisp function. I haven't tried LispStat 3.48 under Windows95, but supposedly Windows95 has a bigger, or unlimited stack. You can also make calls to Windows API functions to talk between applications, (I do this in Visual Basic but haven't tried it under LispStat). Soon our day will come... \`|'/ (. .) ---uuu-U-uuu--- kilroy was here *---------------------------------------------------------------------. * From thomas@biostat.washington.edu Thu Feb 27 18:34:43 1997 Date: Wed, 26 Feb 1997 09:14:44 -0800 (PST) From: Thomas Lumley To: Paul Lunney Subject: Re: S-plus and xlisp-stat Xlisp-Stat is more flexible graphically (particularly for producing dynamic graphics where it is easy to link plots, add slider controls and so on). It is also much much faster. The quality of the printed graphics is substantially inferior to S-Plus but there is an interface to the gnuplot program that produces high-quality printed output. The fact that it is a dialect of LISP makes it harder to learn initially. Another advantage of lispstat, not relevant to you, is that it runs on Macs and on Linux, the free Unix clone. S-PLUS will not run on Macs or on any Unix or Unix clone for Intel chips. S-PLUS has, as you observed, oodles of built-in functions. It is probably easier to learn. More people program their favorite new ideas in S-PLUS than any other language. I use S-PLUS for most data analysis. People *are* starting to write built-in functions for xlispstat. For example, I have written code for Generalised Estimating Equations and for Cox regression. Still, I would say that the main markets for lispstat are teaching (because of the dynamic graphics and because students can run it at home for no cost) and quick-and-dirty simulations (for which S is prohitively slow but xlispstat is fast enough that the saving in programmer time over C/FORTRAN is worthwhile). Thomas Lumley -----------------------------------------------------+------ Biostatistics : "Never attribute to malice what : Uni of Washington : can be adequately explained by : Box 357232 : incompetence" - Hanlon's Razor : Seattle WA 98195-7232 : : ------------------------------------------------------------ *---------------------------------------------------------------------. * From Chris.Brunsdon@newcastle.ac.uk Thu Feb 27 18:34:43 1997 Date: Wed, 26 Feb 1997 11:51:52 +0000 From: Chris Brunsdon To: Paul Lunney Subject: Re: S-plus and xlisp-stat My impression is pretty much the same as yours. Another distinction is, of course, that xlisp-stat is public domain, whereas S-plus is not. Oh - and I don't think there is a Mac version of S-plus. Although you specifically mention Windows versions the last point could be important if any code you create is to be shared among others using non-PC hardware. ============================================================================== + Phone +44 091 222 7570 Dr. Chris Brunsdon, + Mobile 0374 840556 Lecturer in Computer Based Methods + Fax +44 091 222 8811 Dept. Of Town and Country Planning + Telex 53654 (UNINEWG) University of Newcastle Upon Tyne UK + NE1 7RU + chris.brunsdon@newcastle.ac.uk ============================================================================ == *---------------------------------------------------------------------. * From Osman.Buyukisik@ae.ge.com Fri Feb 28 10:49:29 1997 Date: Thu, 27 Feb 1997 07:57:44 -0500 From: U-E59264-Osman Buyukisik To: deleeuw@math.ucla.edu Cc: ralmond@ets.org, p.lunney@qut.edu.au, s-news@utstat.toronto.edu, stat-lisp-news@umnstat.stat.umn.edu Subject: Re: S-plus and xlisp-stat >>>>> "Jan" == Jan Deleeuw writes: Jan> (a) XLS comes with complete source code -- in ANSI Jan> C. Generally it is a breeze to port it to a new flavor of Jan> UNIX. It is quite easy to extend it at compile time, and by Jan> now it is also easy to extend it at runtime with object Jan> code. One can always go back to the source in case of Jan> doubt. One can look at the source and marvel. Even though S and S-plus are commercial products, there is GPL'd free S like program called "R." Authors suggest the S "Blue book" for a manual. Except for complex numbers, it is mostly compatible. INcludes free source code. Jan> (b) XLS runs on the Mac, in fact that is where both XLS and I Jan> prefer to be most of the time. The other two are just Jan> necessary evils. I know there is a windoze version. Mac may not be supported yet :( Jan> (c) I can give it away to my students and my family, in Jan> unlimited quantities, and for their own home computers, no Jan> matter what hardware and OS they have. I can provide WWW Jan> interfaces to the program which run for free from the web Jan> servers -- I dont think the S-plus people would appreciate a Jan> similar service. You can do the same with R. Jan> (d) Porting from Common Lisp is now easy. Porting from Jan> FORTRAN to CL, and then to XL is now easy. Lisp is a general Jan> purpose language. There are lots of books on Lisp. Lisp is a Jan> beautiful language. Because of XLS I now read books such as Jan> "On Lisp" or Norvig or Steele. I think the next version may be written in some functional language. Could be better than Lisp :) Jan> (e) Dynamic graphics for demos, instructional modules, and Jan> research tools cannot really be done in S. Computing and Jan> making static plots is perhaps marginally easier in S, but Jan> this is not what I do. Here xlisp-stat shines. Just my $0.02 Osman From Osman.Buyukisik@ae.ge.com Fri Feb 28 10:49:29 1997 Date: Thu, 27 Feb 1997 08:57:30 -0500 From: U-E59264-Osman Buyukisik To: Osman.Buyukisik@ae.ge.com Cc: deleeuw@math.ucla.edu, ralmond@ets.org, p.lunney@qut.edu.au, s-news@utstat.toronto.edu, stat-lisp-news@umnstat.stat.umn.edu Subject: Re: S-plus and xlisp-stat OOPS, I just found out that my reply was in error (about R): Jan> (b) XLS runs on the Mac, in fact that is where both XLS and I Jan> prefer to be most of the time. The other two are just Jan> necessary evils. I know there is a windoze version (of R). Mac may not be supported yet :( ^ |_______________ this is wrong. Mac versions of R are complete, but not windoze ! Sorry ! Osman From majordom@umnstat.stat.umn.edu Thu Mar 13 10:34:52 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11792; Thu, 13 Mar 1997 10:34:52 -0600 Return-Path: Received: from gateway.ets.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11788; Thu, 13 Mar 1997 10:34:41 -0600 Received: from research.ets.org by gateway.ets.org with SMTP id AA13898 (5.65c/IDA-1.4.4 for ); Thu, 13 Mar 1997 11:34:40 -0500 Received: from comet.research.ets.org by research.ets.org with SMTP id AA21094 (5.67b/IDA-1.4.3 for stat-lisp-news@nokomis.stat.umn.edu); Thu, 13 Mar 1997 11:34:38 -0500 Received: by comet.research.ets.org (SMI-8.6/SMI-SVR4) id LAA20208; Thu, 13 Mar 1997 11:34:37 -0500 Date: Thu, 13 Mar 1997 11:34:37 -0500 Message-Id: <199703131634.LAA20208@comet.research.ets.org> From: Russell Almond To: stat-lisp-news@umnstat.stat.umn.edu Subject: Building data for binomial Regression. Reply-To: ralmond@ets.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk (Sorry I accidentally deleted the original posting so I can't quite site it.) But the thrust of the question was how to take two variable data structures, dose and (indicator response), and build them into a structure that could be passed to binomial-reg (or some similar function). The answer is the list function (binomial-reg (list dose (indicator response)) n r :link binomial-link) [or something like that, I deleted the original post and haven't looked up the functions] should do what you want. In any case, (list dose (indicator response)) will yeild the data structure you need a list of two elements the first being the list dose and the second being the list returned from (indicator response). It become slightly trickier if you need to build up a data frame (to use the S terminology) type structure from a combination of variable vectors and other data frames. In that case, you would need to write your own analog of the S cbind function. Something like: (defun cbind (first &rest data) "This function merges a number of items which could be variables (lists of values) or data matrixes (lists of lists of values)." (unless (listp first) (error "cbind: Expected a list, got ~S" first)) (let (result) (if (listp (car first)) ;; First is a data matrix (setf result (copy-list first)) ; We are going to ; destructively modify this ; list so better copy it ;; First is just a data vector (setf result (list first))) (apply #'cbind-aux result (last result) data))) (defun cbind-aux (result result-last first &rest data) ;; Properly tail recursive version of cbind (unless (listp first) (error "cbind: Expected a list, got ~S" first)) (let (new-result) (if (listp (car first)) ;; First is a data matrix (setf result (copy-list first)) ; copy so we can destroy it ;; First is just a data vector (setf result (list first))) ;; result-last is a pointer to the last cons cell of result so we ;; don't need to do implicit looping to append. (setf (cdr result-last) new-result) (if (null data) result ; End condition (apply #'cbind-aux result (last new-result) data)))) I think you might be able to get away with something less obscure in xlisp because I don't think the compiler does tail recursion optimization. Hope this helps. --Russell Almond From majordom@umnstat.stat.umn.edu Thu Mar 13 10:58:18 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11938; Thu, 13 Mar 1997 10:58:18 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11934; Thu, 13 Mar 1997 10:58:07 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa16175; 13 Mar 97 11:57 EST Received: from slip-1-10.acc.Virginia.EDU by elvis.med.Virginia.EDU (5.65c/1.34) id AA22419; Thu, 13 Mar 1997 11:56:03 -0500 Date: 13 Mar 97 11:57:52 -0500 Subject: Re: HOW-TO? Problem relating to spin-plots From: "Steven D. Majewski" To: stat-lisp-news@umnstat.stat.umn.edu, "Dom. Early" X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk You can get/set the current transformation matrix from a spin plot with the :transformation method. The following ALMOST draws a line into the plane. Sometimes I get a point visible and sometimes I get a short line segment. Perhaps this is due to round off or some other truncation. But when you rotate or spin the plot, it looks approximately correct. ( I just hacked this out and minimally tested it manually -- you can hook it into do-click to get it to work automatically. I just made :do-click report the canvas (x,y) and I typed them in manually. ) Change the -1, 1 Z values to make it longer. I was testing it on a range ( -3 3 ) spin plot, so that was a convenient test value. [ Does a really large z value get you into greater roundoff problems? ] ( defmeth s :zline ( cx cy &key (color 'red)) ( let* (( real-xy ( send self :canvas-to-real cx cy )) ( real-x ( car real-xy )) ( real-y ( cadr real-xy )) ( transmat ( send self :transformation )) ( v1 ( matmult ( vector real-x real-y -1 ) transmat )) ( v2 ( matmult ( vector real-x real-y 1 ) transmat ))) ( send s :add-lines ( map 'cons #'list v1 v2 ) :color color ))) The more general question about how to pick or input 3-d points or lines is an interesting one. Gestural input would be nice to experiment with, but this may be difficult with the current XLispStat graphics implementation. - Steve Majewski - From majordom@umnstat.stat.umn.edu Thu Mar 13 11:12:57 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12129; Thu, 13 Mar 1997 11:12:57 -0600 Return-Path: Received: from sulu.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12125; Thu, 13 Mar 1997 11:12:47 -0600 Received: from zen by sulu.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id JAA15030; Thu, 13 Mar 1997 09:12:43 -0800 (PST) Date: Thu, 13 Mar 1997 09:12:41 -0800 (PST) From: Thomas Lumley X-Sender: thomas@zen To: ralmond@ets.org Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Building data for binomial Regression. In-Reply-To: <199703131634.LAA20208@comet.research.ets.org> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Thu, 13 Mar 1997, Russell Almond wrote: > > (Sorry I accidentally deleted the original posting so I can't quite > site it.) > > But the thrust of the question was how to take two variable data > structures, dose and (indicator response), and build them into a > structure that could be passed to binomial-reg (or some similar > function). > > The answer is the list function > > (binomial-reg (list dose (indicator response)) n r :link > binomial-link) > This doesn't quite work because indicator returns a list of lists (indicators for k categories require k-1 lists of 1s and 0s). Two things which do work are: (cons dose (indicator response)) (append (list dose) (indicator response)) Thomas Lumley ------------------------------------------------------+------ Biostatistics : "Never attribute to malice what : Uni of Washington : can be adequately explained by : Box 357232 : incompetence" - Hanlon's Razor : Seattle WA 98195-7232 : : ------------------------------------------------------------ From majordom@umnstat.stat.umn.edu Thu Mar 13 11:18:04 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12193; Thu, 13 Mar 1997 11:18:04 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12187; Thu, 13 Mar 1997 11:17:53 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa00174; 13 Mar 97 12:17 EST Received: from slip-1-10.acc.Virginia.EDU by elvis.med.Virginia.EDU (5.65c/1.34) id AA01554; Thu, 13 Mar 1997 12:17:20 -0500 Date: 13 Mar 97 12:19:56 -0500 Subject: Re: 2 questions... From: "Steven D. Majewski" To: stat-lisp-news@umnstat.stat.umn.edu, daddy@imap2.asu.edu X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >Hello Folks- > >I have two questions. First, is the list archived, and if so, how can I >view it? I'm not sure, but take a look on either http://www.stat.ucla.edu/, or in the umn archives. >Secondly, how can I view the original code for graph-proto? I want to >make a special graph and need to alter some things. I have search the >.lsp files that came with my distribution, but have not found the code. > >One thing I would like to do for example would be to have a histogram >with frequncies along the ticks of the y-axis rather than the proportion. Some of the graphic code is in the xlispstat C sources -- you can look at the various iview* files. Some of the code is in xlisp, and is mostly in graphics.lsp, graph2.lsp,. graph3.lsp. There are a lot of contributed modules that will give you some hints of how to do extensions. ( They are probably a better place to start than the sources) I know there are a couple of alternate histogram implementations -- maybe one of them already does what you want. [ And again, I have lots of code for doing real-space, colored overlay graphics objects and some other things on http://galen.med.virginia.edu/~sdm7g/ among the XES application sources. I have updated the files there recently, and will do so again today, but the documentation now lags behind the sources. ] - Steve Majewski - From majordom@umnstat.stat.umn.edu Thu Mar 13 11:37:11 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12379; Thu, 13 Mar 1997 11:37:11 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12375; Thu, 13 Mar 1997 11:37:01 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa12934; 13 Mar 97 12:36 EST Received: from slip-1-10.acc.virginia.edu (slip-2-14.acc.Virginia.EDU) by elvis.med.Virginia.EDU (5.65c/1.34) id AA21178; Thu, 13 Mar 1997 12:36:17 -0500 Date: 13 Mar 97 12:28:37 -0500 Subject: Re: HOW-TO? Problem relating to spin-plots From: "Steven D. Majewski" To: stat-lisp-news@umnstat.stat.umn.edu, "Dom. Early" X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk > The following ALMOST draws a line into the plane. Please note that the previous message contains code that I experimentally hacked out over a long coffee this morning. I'm not sure it's quite exactly right, but it'll get you started in the right direction. ( It looks to me like there is more than a roundoff problem. You might have to look at the C code to see what's really expected by the transformations. ) - Steve Majewski - From majordom@umnstat.stat.umn.edu Thu Mar 13 22:32:03 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16436; Thu, 13 Mar 1997 22:32:03 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16432; Thu, 13 Mar 1997 22:31:54 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa21085; 13 Mar 97 23:31 EST Received: from slip-2-15.acc.Virginia.EDU by elvis.med.Virginia.EDU (5.65c/1.34) id AA26544; Thu, 13 Mar 1997 23:31:08 -0500 Date: 13 Mar 97 23:33:14 -0500 Subject: Re: HOW-TO? Problem relating to spin-plots From: "Steven D. Majewski" To: stat-lisp-news@umnstat.stat.umn.edu, "Dom. Early" X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk [ I wrote in previous posts: ] >> The following ALMOST draws a line into the plane. [ ... ] > > ( It looks to me like there is more >than a roundoff problem. You might have to look at the C code >to see what's really expected by the transformations. ) It appears that if you check NO SCALING in the Spinner Menu Options dialog, then that procedure works perfectly - it will draw a line into the screen Z-axis at the canvas x,y coordinates clicked. It will be only be visible, on end as a dot, until you rotate the plot. With Scaling ON ( either the Fixed or Variable option on ), there is clearly another step to the transform besides the canvas-to-real and matmult by the transformation matrix. Anyone know how to reverse that effect ? - Steve Majewski - From majordom@umnstat.stat.umn.edu Fri Mar 14 02:23:55 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25962; Fri, 14 Mar 1997 02:23:55 -0600 Return-Path: Received: from kasuga.cc.chuo-u.ac.jp by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25958; Fri, 14 Mar 1997 02:23:38 -0600 Received: from isg.indsys.chuo-u.ac.jp (isg.indsys.chuo-u.ac.jp [133.91.120.136]) by kasuga.cc.chuo-u.ac.jp (8.7.6+2.6Wbeta7/3.4W4) with SMTP id RAA20423 for ; Fri, 14 Mar 1997 17:24:55 +0900 (JST) Received: from [133.91.120.87] (mactk) by isg.indsys.chuo-u.ac.jp (4.1/6.4J.6-fps0) id AA28517; Fri, 14 Mar 97 17:27:54 JST Message-Id: <9703140827.AA28517@isg.indsys.chuo-u.ac.jp> X-Mailer: Macintosh Eudora Pro Version 2.1.4-Jr2 Mime-Version: 1.0 Content-Type: text/plain; charset="ISO-2022-JP" Date: Fri, 14 Mar 1997 17:23:39 +0900 To: stat-lisp-news@umnstat.stat.umn.edu From: kamakura@indsys.chuo-u.ac.jp (Toshinari Kamakura) Subject: spin-function Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, Th function "spin-function" does not work properly on the version 3.48 and 3.49. For example, in the Tierney's textboook we can find the function f(x)=x^2+y^2. But we cannot obtain the picture as appeared in p. 60. ++++++++++++++++++++++++++ Toshinari Kamakura Department of Industrial and Systems Engineering Chuo University 1-13-27 Kasuga email: kamakura@indsys.chuo-u.ac.jp Bunkyo-ku Tokyo 112, Japan From majordom@umnstat.stat.umn.edu Sun Mar 16 14:32:49 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26209; Sun, 16 Mar 1997 14:32:49 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26205; Sun, 16 Mar 1997 14:32:41 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.7.5/10.1) id PAA22212; Sun, 16 Mar 1997 15:32:34 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sun, 16 Mar 1997 15:29:33 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: Forrest Young Subject: What is xlisp.hlp file? Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi ... In the 3.44 distribution (I havn't checked more recent ones) there is a file named xlisp.hlp, which I had always assumed was what contained the data for the (help ...) function. Now I discover (help ...) works without the xlisp.hlp file. So...what is this file for? --- Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: mailto:forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Voice: 1-919-962-5038 ViSta Software and Documentation site: http://forrest.psych.unc.edu/research/ USA Home Site ViSta Mirrors: http://ssda.anu.edu.au/mirrors/ViSta.html Australia http://sunsite.univie.ac.at/ViSta/ Austria http://micros.hensa.ac.uk/mirrors/vista/ UK (ac.uk only 8AM-8PM GMT) http://www.psych.unc.edu/fwy/research/ USA ___ (@ @) ------------------------------------------------ooO--(_)--Ooo--------- ||| From majordom@umnstat.stat.umn.edu Sun Mar 16 21:16:07 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28100; Sun, 16 Mar 1997 21:16:07 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28095; Sun, 16 Mar 1997 21:15:59 -0600 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.7.5/10.1) id WAA15604; Sun, 16 Mar 1997 22:15:40 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sun, 16 Mar 1997 22:15:52 -0500 To: vista@unc.edu, edstat-l@jse.stat.ncsu.edu, stat-lisp-news@umnstat.stat.umn.edu From: Forrest Young Subject: New Release of ViSta Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Announcing a major new release of ViSta - The Visual Statistics System For Unix, MSWindows and MacOS Version 4.27, March 14th, 1997 Copyright (c) 1991-97 by Forrest W. Young This new release brings major new (and improved) data analysis and visualization methods to ViSta for Unix, updating it to the capabilities of ViSta for MSWindows and MacOS. It also adds Data Simulation, with an optional Central Limit Theorem visualization. This release is available for all three platforms, unifying the code for the first time. ViSta is available for free for Microsoft Windows, MacOS, and for Unix with X11. ViSta and its documentation are available from our WWW and FTP servers at: WWW: http://forrest.psych.unc.edu/research/vista.html FTP: ftp://www.psych.unc.edu/pub/forrest/vista/ ViSta is also mirrored in Australia, Austria, United Kingdom, and USA at http://ssda.anu.edu.au/mirrors/ViSta.html (Australia) http://sunsite.univie.ac.at/ViSta (Austria) http://micros.hensa.ac.uk/mirrors/vista (UK - ac.uk only 8AM-8PM GMT) http://www.psych.unc.edu/fwy/research (USA) The new release will not be available on the Australian mirror for a few days. GENERAL DESCRIPTION ViSta features visual tools for understanding data structure, for seeing the results of statistical analyses; for guiding the analyst during an analysis session; and for structuring the analyst's data analysis process. ViSta is designed for a wide range of users. It is particularly useful for those learning statistics and for their teachers. ViSta is being used to teach introductory statistics; introductory and advanced multivariate statistics; statistical graphics; and computational statistics. ViSta is also being used for research and development in statistical graphics and computational statistics. ViSta is designed for an audience of users having a very wide range of data analysis sophistication, ranging from novice to expert. ViSta provides seamlessly integrated data analysis environments specifically tailored to the user's level of expertise. Guidance is available for students and novices, and tools are available for teachers and other experts to create guidance for these novices. A structured graphical user interface is available for competent users, and a command line interface is available for sophisticated users. The complete Lisp-Stat (Tierney, 1990) programming environment is available to researchers, programmers and graduate students who wish to extend ViSta's capabilities. ViSta performs univariate and multivariate statistical data analysis and visualization. ViSta's data analysis and visualization capabilities include: exploratory graphics and descriptive statistics; univariate tests and visualizations; analysis of variance; regression analysis; principal components analysis; multidimensional scaling and correspondence analysis. All analysis methods have extensive visualization features. NEW FEATURES FOR ALL PLATFORMS: 1) Data can be simulated by sampling from any of 10 hypothetical population distributions. 2) The data simulation feature includes an optional visualization which interactively demonstrates the central limit theorem. A click of the mouse generates one (or several) new samples whose means and standard deviations are added to histograms of their sampling distributions. 3) Part of the code is provided as compiled byte-code for MSWindows and MacOS to improve efficiency. NEW FEATURES IN ViSta FOR UNIX: ViSta for Unix now includes all of the features previously available only in the MSWindows and MacOS versions. These include: 1) A new OLS, robust and monotonic regression module. 2) Redundancy analysis (new option of the multivariate regression module). 3) Extensive visual regression diagnostics. 4) A visualization method for Univariate Analysis. 5) A completed visualization method for Analysis of Variance. 6) An improved visualization method for Multivariate Regression. 7) A multivariate regression test. Minor changes made in ViSta for Unix which were already made in ViSta for MacOS and MSWindows, include: 1) Import Data works with strings as well as symbols and numbers. 2) Create-Data can remove rows of multivariate data with missing values. 3) Datasheet menu items now allow adding several obs/vars simultaneously. 4) Datasheet shows matrix names for matrix data. 5) Datasheet window scrolling and sizing improved. In addition, window and dialog layout, input and output file dialog boxes, graphic fonts, and the exit process have been improved in the Unix version. BUG FIXES IN ALL VERSIONS: Several bugs have been fixed in all versions of ViSta. These include: 1) Data with missing values are imported properly 2) Data with missing values propagate properly during data management. 3) There is a new option for row-wise removal of data with missing values. 4) Summary Statistics are reported for the selected subset of variables. 5) Univariate Analysis performed on the selected numeric variable. 6) Univariate Analysis reports correct confidence intervals. 7) Regression Analysis added variable plot now has an X-axis button. 8) Regression Analysis module does robust regression when requested in the dialog box. ----- Forrest Young Professor, Psychometrics CB# 3270 Davie Hall email: mailto:forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software and Documentation site: http://forrest.psych.unc.edu/research/ViSta.html ViSta Mirrors: http://ssda.anu.edu.au/mirrors/ViSta.html Australia http://sunsite.univie.ac.at/ViSta Austria http://micros.hensa.ac.uk/mirrors/vista UK (ac.uk only 8AM-8PM GMT) http://www.psych.unc.edu/fwy/research USA From majordom@umnstat.stat.umn.edu Tue Mar 18 12:39:53 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26554; Tue, 18 Mar 1997 12:39:53 -0600 Return-Path: Received: from salmon.maths.tcd.ie by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26550; Tue, 18 Mar 1997 12:39:42 -0600 Received: from orange.stats.tcd.ie by salmon.maths.tcd.ie with SMTP id aa18961; 18 Mar 97 18:39 +0000 Received: from [194.125.145.148] by stats.tcd.ie (4.1/SMI-4.1) id AA08668; Tue, 18 Mar 97 18:26:47 GMT X-Sender: dbearly@orange.stats.tcd.ie Message-Id: In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sat, 15 Mar 1997 19:30:49 +0000 To: "Steven D. Majewski" From: "Dom. Early" Subject: Re: HOW-TO? Problem relating to spin-plots Cc: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk > >It appears that if you check NO SCALING in the Spinner Menu >Options dialog, then that procedure works perfectly - it will >draw a line into the screen Z-axis at the canvas x,y coordinates >clicked. It will be only be visible, on end as a dot, until you >rotate the plot. > > With Scaling ON ( either the Fixed or Variable option on ), >there is clearly another step to the transform besides the >canvas-to-real and matmult by the transformation matrix. > > Anyone know how to reverse that effect ? > >- Steve Majewski >- Sorry I haven't got back earlier, thanks to Steve Majewski and Ted Harding for their input. I have been playing around with Steve Majewski's code which works nicely for spin-plots with 'no scaling' selected in the options menu item. I've rewritten Steve's code slightly to make investigation a little easier (below). I've noticed that when you switch from 'variable scaling' to 'No scaling', the code is prone to some sort of errors, i.e. the lines are placed exactly at the required point, and somtimes are not completely orthogonal to the viewing plane. However if you switch from 'fixed scaling' to 'no scaling', the code seems to work fine. In fact if you switch from 'Variable scaling' to 'fixed scaling' and then to 'no scaling' it works fine also, STRANGE! Maybe this might shine some light on what is going on internally? Thanks to Steve, I will continue to tinker with this problem on my 'long coffee breaks' and let you know if I come up with anything! I hope you will do likewise? Thanks for the help, Dom. ------------------------------------------------------------------------------ ( defmeth spin-proto :zline ( cx cy &key (color 'red)) (let*( (real-xy (send self :canvas-to-real cx cy )) (real-x (car real-xy)) (real-y (cadr real-xy)) (transmat (send self :transformation)) ;; When spin plots are first created the transformation is nil ;; which causes problems below with matmult, so we simply ;; initialise the transformation to the identity matrix (transmat (if transmat transmat #2a((1 0 0) (0 1 0) (0 0 1)))) (v1 ( matmult (vector real-x real-y -1) transmat)) (v2 ( matmult (vector real-x real-y 1) transmat))) (send self :add-lines (map 'cons #'list v1 v2) :color color)) ) ;; Here we extend do-select-click functionality to add a ZLINE, to make ;; testing a bit easier (defmeth spin-proto :do-select-click(pX pY pM1 pM2) (send self :zline pX pY) (call-next-method pX pY pM1 pM2) ) (def s (spin-plot '((1 2 3) (3 4 5) (4 5 6)))) +--------------------------------------------------------------------------+ | Dominic Early e-mail: Dom.Early@tcd.ie | | A.31, Department of Statistics, Phone: +353-1-608 1917 | | Trinity College, Fax: +353-1-661 5046 | | Dublin 2, Ireland. | | | | http://www2.tcd.ie/Statistics/staff/dbearly.html | +--------------------------------------------------------------------------+ _/_/_/_/ _/_/_/ _/ _/ _/_/_/_/_/ _/_/_/ _/_/_/_/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/_/_/_/ _/_/_/ _/ _/ _/ _/ _/_/_/ _/_/_/_/ From majordom@nokomis.stat.umn.edu Wed Mar 19 09:49:47 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA16550; Wed, 19 Mar 1997 09:49:47 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA16545; Wed, 19 Mar 1997 09:49:42 -0600 From: Luke Tierney Message-Id: <9703191549.AA16545@nokomis.stat.umn.edu> Subject: xlispstat 3.50 released as standard distribution To: stat-lisp-news@nokomis.stat.umn.edu Date: Wed, 19 Mar 1997 09:49:41 -0600 (CST) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 490 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk xlispstat 3.50 is now the standard distribution. It is available in ftp://ftp.stat.umn.edu/pub/xlispstat This release fixes bugs found in 3.49. There is a good chance that this will be the final release to include a Win16 version. I may also cut back the Macintosh distribution to a PPC version and a single 68K version for 020 processors and a coprocessor (SoftFPU will allow this to run on systems without coprocessors). Please let me know if this would cause serious problems. luke From majordom@umnstat.stat.umn.edu Wed Mar 19 12:31:55 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13100; Wed, 19 Mar 1997 12:31:55 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13092; Wed, 19 Mar 1997 12:31:43 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa10776; 19 Mar 97 13:31 EST Received: from slip-1-12.acc.Virginia.EDU by elvis.med.Virginia.EDU (5.65c/1.34) id AA18110; Wed, 19 Mar 1997 13:30:58 -0500 Date: 19 Mar 97 13:18:27 -0500 Subject: Re: HOW-TO? Problem relating to spin-plots From: "Steven D. Majewski" To: "Dom. Early" Cc: stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk > >However if you switch from 'fixed scaling' to 'no scaling', the code seems >to work fine. In fact if you switch from 'Variable scaling' to 'fixed >scaling' and then to 'no scaling' it works fine also, STRANGE! Maybe this >might shine some light on what is going on internally? > The thing to check is: ( send plot :scale ( send plot :current-variables )) When these are all approximately equal to 1.0, then the code works, no mater which option is set. It looks like no-scaling forces it to (1.0 1.0 1.0). What the scaling is equal to for the other options depends on some other factors, especially the range of the values in the plot. [ One of my first test spin-plots happened to be unit-ranged in all dimensions, so the scales factors were all close to 1.0 for all scale options - this kept me from noticing the scale effect at first. ] It should be possible to put the scale factor into that code and have something that works for all cases. I just haven't yet been able to figure out what the inverse scale function is -- is it applied to the real or the canvas values ? - Steve Majewski From majordom@umnstat.stat.umn.edu Thu Mar 20 12:11:50 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28740; Thu, 20 Mar 1997 12:11:50 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28735; Thu, 20 Mar 1997 12:11:33 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa15279; 20 Mar 97 13:11 EST Received: from slip-2-11.acc.Virginia.EDU by elvis.med.Virginia.EDU (5.65c/1.34) id AA14020; Thu, 20 Mar 1997 13:10:34 -0500 Date: 20 Mar 97 13:13:07 -0500 Subject: Re: HOW-TO? Problem relating to spin-plots From: "Steven D. Majewski" To: "Steven D. Majewski" , "Dom. Early" Cc: stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk OK -- Try the following code. In addition to a :scale, there is also a :shift that has to be corrected for. I found this by hunt and peck experimenting, with a bit of peeking at the sources, however, I haven't figured out the whole flow of control for these modules from the sources, so it's mostly what works experimentally. I can't for the life of me imagine why I have to *both* divide and multiply by the scales in 2 different places. Shift option clicking will draw a red line segment orthogonal to the viewing plane. - Steve Majewski ;;; -------------------------- ;;; test 3-d line drawing ... ;;; -------------------------- ;; scale factor for test -- set b to any number ( setf b 2 ) ( setf S ( spin-function #'(lambda (x y) (+ -1 (* x x) (* y y))) (- b) b (- b) b :color 'green )) ( defmeth s :do-click ( x y a b ) ( let (( scaled ( send self :canvas-to-scaled x y ))) ( format T "CANVAS: ~A SCALED: ~A C->Real: ~A\n" ( list x y ) scaled ( send self :canvas-to-real x y ))) ( when ( and a b ) ( send self :zline x y )) ( call-next-method x y a b )) ( defmeth s :zline ( cx cy &key (color 'red)) ( let* (( real-xy ( send self :canvas-to-real cx cy )) ( real-x ( car real-xy )) ( real-y ( cadr real-xy )) ( transmat ( send self :transformation )) ( shifts ( send self :shift ( send self :current-variables ))) ( scales ( send self :scale ( send self :current-variables )))) ( if ( null transmat ) ( setf transmat ( send self :transformation #2A((1. 0. 0.) (0. 1. 0.) (0. 0. 1.))))) ( let (( v1 ( matmult (/ ( vector real-x real-y (- b)) scales) transmat )) ( v2 ( matmult (/ ( vector real-x real-y b ) scales) transmat ))) ( send s :add-lines ( map 'cons #'list (+ ( * v1 scales ) shifts ) (+ ( * v2 scales ) shifts )) :color color )))) ( let* (( n 8 )) ( send s :add-surface (rseq (- b) b n) (rseq (- b) b n) ( make-array ( list n n ) :initial-contents ( repeat 0.0 (* n n))) :color 'blue )) From majordom@umnstat.stat.umn.edu Thu Mar 20 17:05:58 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA01984; Thu, 20 Mar 1997 17:05:58 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA01979; Thu, 20 Mar 1997 17:05:45 -0600 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.7.5/10.1) id SAA16821; Thu, 20 Mar 1997 18:05:40 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 20 Mar 1997 18:06:03 -0500 To: vista@unc.edu, edstat-l@jse.stat.ncsu.edu, stat-lisp-news@umnstat.stat.umn.edu From: Forrest Young Subject: ViSta for Windows Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk To users of ViSta for Windows: Regarding the recently released new version of ViSta for Windows: A user has pointed out that the program group icon that is created for ViSta for Windows does not properly point to the ViSta executable. The problem is that the installation process creates a file called VISTA.EXE, but the program group is expecting a file named WXLS.EXE. I will fix this in the near future. In the meantime, you can either change the name of the VISTA.EXE program to be WXLS.EXE, or you can change the properties of the program group icon to call VISTA.EXE instead of WXLS.EXE when the icon is double-clicked. Sorry about the problem. ----- Forrest Young Professor, Psychometrics CB# 3270 Davie Hall email: mailto:forrest@unc.edu UNC Psychometric Lab www: http://forrest.psych.unc.edu Chapel Hill, NC 27599-3270 USA voice: 1-919-962-5038 ViSta Software and Documentation site: http://forrest.psych.unc.edu/research/ViSta.html ViSta Mirrors: http://ssda.anu.edu.au/mirrors/ViSta.html Australia http://sunsite.univie.ac.at/ViSta Austria http://micros.hensa.ac.uk/mirrors/vista UK (ac.uk only 8AM-8PM GMT) http://www.psych.unc.edu/fwy/research USA From majordom@umnstat.stat.umn.edu Thu Mar 20 23:32:08 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03885; Thu, 20 Mar 1997 23:32:08 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03881; Thu, 20 Mar 1997 23:31:55 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa25088; 21 Mar 97 0:31 EST Received: from slip-1-8.acc.Virginia.EDU by elvis.med.Virginia.EDU (5.65c/1.34) id AA17156; Fri, 21 Mar 1997 00:31:40 -0500 Date: 21 Mar 97 00:34:27 -0500 Subject: One more time -- this makes more sense [was: HOW-TO? Problem relating to spin-plots ] From: "Steven D. Majewski" To: "Dom. Early" Cc: stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk OK - This one works the same, but it makes slightly more sense. If you change to :CANVAS-TO-REAL to :CANVAS-TO-SCALED, then the extra division by scale factors is not required. ( And I'm starting to figure out the transformation steps that must be used in the plots. ) - Steve Majewski ( defmeth s :zline ( cx cy &key (color 'red)) ( let* (( real-xy ( send self :canvas-to-scaled cx cy )) ( real-x ( car real-xy )) ( real-y ( cadr real-xy )) ( transmat ( send self :transformation )) ( cv ( send self :current-variables )) ( shifts ( send self :shift cv )) ( scales ( send self :scale cv ))) ( if ( null transmat ) ( setf transmat ( send self :transformation #2A((1. 0. 0.) (0. 1. 0.) (0. 0. 1.))))) ( let (( v1 ( matmult ( vector real-x real-y (- b)) transmat )) ( v2 ( matmult ( vector real-x real-y b ) transmat ))) ( send s :add-lines ( map 'cons #'list (+ ( * v1 scales ) shifts ) (+ ( * v2 scales ) shifts )) :color color )))) From majordom@umnstat.stat.umn.edu Fri Mar 21 15:28:31 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17420; Fri, 21 Mar 1997 15:28:31 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17416; Fri, 21 Mar 1997 15:28:21 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.7.5/10.1) id QAA24195; Fri, 21 Mar 1997 16:28:15 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 21 Mar 1997 16:28:00 -0500 To: vista@unc.edu, edstat-l@jse.stat.ncsu.edu, stat-lisp-news@umnstat.stat.umn.edu From: Forrest Young Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk To users of the new release of ViSta for Windows: Yesterday I reported that a user had noticed that the ViSta program group icon that is created during the configuration process didn't work properly. I am pleased to say that new ViSta for Windows zip files are now available in which this problem is taken care of. These files can be downloaded via the web from http://forrest.psych.unc.edu/research/ViSta/winvista.html or by ftp from ftp://www.psych.unc.edu/pub/forrest/vista/mswindows/vista32.zip (Win95 or Win3.1+32bit extension) or ftp://www.psych.unc.edu/pub/forrest/vista/mswindows/vista16.zip (Win3.1 only). If you wish, you can use the files you have already downloaded and either 1) ignore the program group (especially under Win95, where it is superfluous); 2) change the name of the VISTA.EXE program to be WXLS32.EXE (or WXLS16.EXE); or 3) change the properties of the program group icon to point to VISTA.EXE instead of WXLS32.EXE (or WXLS16.EXE). --- Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: mailto:forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Voice: 1-919-962-5038 ViSta Software and Documentation site: http://forrest.psych.unc.edu/research/ USA Home Site ViSta Mirrors: http://ssda.anu.edu.au/mirrors/ViSta.html Australia http://sunsite.univie.ac.at/ViSta/ Austria http://micros.hensa.ac.uk/mirrors/vista/ UK (ac.uk only 8AM-8PM GMT) http://www.psych.unc.edu/fwy/research/ USA ___ (@ @) ------------------------------------------------ooO--(_)--Ooo--------- ||| From majordom@umnstat.stat.umn.edu Fri Mar 21 15:29:53 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17430; Fri, 21 Mar 1997 15:29:53 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17426; Fri, 21 Mar 1997 15:29:45 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.7.5/10.1) id QAA24344; Fri, 21 Mar 1997 16:29:40 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 21 Mar 1997 16:29:27 -0500 To: vista@unc.edu, edstat-l@jse.stat.ncsu.edu, stat-lisp-news@umnstat.stat.umn.edu From: Forrest Young Subject: Re: New ViSta for Windows Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk To users of the new release of ViSta for Windows: Yesterday I reported that a user had noticed that the ViSta program group icon that is created during the configuration process didn't work properly. I am pleased to say that new ViSta for Windows zip files are now available in which this problem is taken care of. These files can be downloaded via the web from http://forrest.psych.unc.edu/research/ViSta/winvista.html or by ftp from ftp://www.psych.unc.edu/pub/forrest/vista/mswindows/vista32.zip (Win95 or Win3.1+32bit extension) or ftp://www.psych.unc.edu/pub/forrest/vista/mswindows/vista16.zip (Win3.1 only). If you wish, you can use the files you have already downloaded and either 1) ignore the program group (especially under Win95, where it is superfluous); 2) change the name of the VISTA.EXE program to be WXLS32.EXE (or WXLS16.EXE); or 3) change the properties of the program group icon to point to VISTA.EXE instead of WXLS32.EXE (or WXLS16.EXE). --- Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: mailto:forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Voice: 1-919-962-5038 ViSta Software and Documentation site: http://forrest.psych.unc.edu/research/ USA Home Site ViSta Mirrors: http://ssda.anu.edu.au/mirrors/ViSta.html Australia http://sunsite.univie.ac.at/ViSta/ Austria http://micros.hensa.ac.uk/mirrors/vista/ UK (ac.uk only 8AM-8PM GMT) http://www.psych.unc.edu/fwy/research/ USA ___ (@ @) ------------------------------------------------ooO--(_)--Ooo--------- ||| From majordom@umnstat.stat.umn.edu Mon Mar 31 11:03:22 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA08400; Mon, 31 Mar 1997 11:03:22 -0600 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA08395; Mon, 31 Mar 1997 11:03:12 -0600 Received: from elvis.med.virginia.edu by mail.virginia.edu id ab08730; 31 Mar 97 12:02 EST Received: from slip-1-6.acc.Virginia.EDU by elvis.med.Virginia.EDU (5.65c/1.34) id AA17243; Mon, 31 Mar 1997 12:02:25 -0500 Date: 31 Mar 97 12:05:31 -0500 Subject: :element-type 'c-char & other c-* types & native/dynload routines. From: "Steven D. Majewski" To: stat-lisp-news@umnstat.stat.umn.edu Cc: Luke Tierney X-Mailer: Cyberdog/2.0b1 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk What is the status of the implementation of the c-char, c-float, c-(etc), data types. Has anyone actually tried using these for interfacing to native C routines ? Luke: Do you have any plans in the works to adapt the dynload routines to use these datatypes ? I would like to pass large arrays of floats, ints and chars to dynamically loaded code modules. Looking at the current interface, I see that that is some copying and translation from Lisp <> C-types. This is what I would like to get rid of, as well as provide a more flexible interface. Anyone have any ideas on what a better Lisp->C API should look like? - Steve Majewski From majordom@umnstat.stat.umn.edu Mon Mar 31 14:36:30 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10030; Mon, 31 Mar 1997 14:36:30 -0600 Return-Path: Received: from ts037p7.pop3b.netvision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10000; Mon, 31 Mar 1997 14:35:58 -0600 Received: (from hjstein@localhost) by bacall.nnt.com (8.7.4/8.7.3) id XAA00451; Mon, 31 Mar 1997 23:34:17 +0300 Date: Mon, 31 Mar 1997 23:34:17 +0300 Message-Id: <199703312034.XAA00451@bacall.nnt.com> From: "Harvey J. Stein" To: "Steven D. Majewski" Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: :element-type 'c-char & other c-* types & native/dynload routines. In-Reply-To: References: Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Steven D. Majewski writes: > > > What is the status of the implementation of the c-char, c-float, c-(etc), > data types. Has anyone actually tried using these for interfacing to > native C routines ? > > Luke: Do you have any plans in the works to adapt the dynload routines > to use these datatypes ? > > I would like to pass large arrays of floats, ints and chars to dynamically > loaded code modules. Looking at the current interface, I see that that > is some copying and translation from Lisp <> C-types. This is what I would > like to get rid of, as well as provide a more flexible interface. > > Anyone have any ideas on what a better Lisp->C API should look like? I have a version of the foreign fcn interface which passes data directly when it can (as opposed to copying it first). This would be for typed vectors, which are basically C arrays internally. For example, (make-array 100 :element-type 'c-double) makes a typed array. The only tricky part is to keep things in this form. If you use vectorized operations to add two typed vectors, the result is no longer typed. I've been using it to speed up nonlinear optimizations - I have the objective function coded in fortran. I link it in & call it with large arrays as arguments. My ffi version avoids all the copying & garbage collection that the native version does, and thus greatly speeds up the optimizations. I can email it to you if you're interested. -- Harvey J. Stein Berger Financial Research abel@netvision.net.il From majordom@nokomis.stat.umn.edu Mon Mar 31 15:01:48 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA04484; Mon, 31 Mar 1997 15:01:48 -0600 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA04476; Mon, 31 Mar 1997 15:01:33 -0600 From: Luke Tierney Message-Id: <9703312101.AA04476@nokomis.stat.umn.edu> Subject: Re: :element-type 'c-char & other c-* types & native/dynload To: sdm7g@Virginia.EDU (Steven D. Majewski) Date: Mon, 31 Mar 1997 15:01:33 -0600 (CST) Cc: stat-lisp-news@nokomis.stat.umn.edu, luke@nokomis.stat.umn.edu In-Reply-To: from "Steven D. Majewski" at Mar 31, 97 12:05:31 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 346 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > Luke: Do you have any plans in the works to adapt the dynload routines > to use these datatypes ? > The basic plan is to build a low level FFI that passes data directly to C routines and to reimplement the current facilities using the new low level interface. But it is likely to be a number of months before I get to that point. luke From majordom@umnstat.stat.umn.edu Tue Apr 1 14:40:41 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26629; Tue, 1 Apr 1997 14:40:41 -0600 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26616; Tue, 1 Apr 1997 14:40:23 -0600 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.7.5/10.1) id PAA29750; Tue, 1 Apr 1997 15:39:49 -0500 (GMT+5:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Tue, 1 Apr 1997 15:39:01 -0500 To: vista@unc.edu, edstat-l@jse.stat.ncsu.edu, stat-l@vm1.mcgill.ca, stat-lisp-news@umnstat.stat.umn.edu, sug-l@listserv.unc.edu, allstat@mailbase.ac.uk From: Forrest Young Subject: ViSta: Two Notes Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Two notes for ViSta users: 1) A user discovered that the help files for ViSta for Unix were formatted improperly. If you already have ViSta for Unix, you can download only the reformatted help files from the ViSta home page. The entire download module has also been fixed for those downloading ViSta for Unix in the future. 2) The Australian mirror has still not updated its files to include the new release. I will let you know when they do. --- Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: mailto:forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Voice: 1-919-962-5038 ViSta Software and Documentation site: http://forrest.psych.unc.edu/research/ USA Home Site ViSta Mirrors: http://sunsite.univie.ac.at/ViSta/ Austria http://micros.hensa.ac.uk/mirrors/vista/ UK (ac.uk only 8AM-8PM GMT) http://www.psych.unc.edu/fwy/research/ USA ___ (@ @) ------------------------------------------------ooO--(_)--Ooo--------- ||| From majordom@umnstat.stat.umn.edu Tue Apr 1 15:53:52 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27917; Tue, 1 Apr 1997 15:53:52 -0600 Return-Path: Received: from bremmer.Mines.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27899; Tue, 1 Apr 1997 15:53:40 -0600 Received: (from kasper@localhost) by bremmer.Mines.EDU (8.6.12/8.6.9) id OAA31222; Tue, 1 Apr 1997 14:53:35 -0700 Date: Tue, 1 Apr 1997 14:53:35 -0700 Message-Id: <199704012153.OAA31222@bremmer.Mines.EDU> From: Kasper van Wyk To: stat-lisp-news@umnstat.stat.umn.edu Subject: static variables Reply-To: kasper@dix.mines.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, In a version of matrix multiplication for sparse matrices I need to define an array of certain length multiple times. To speed up the process, I would like to define such an array once, after which I "point to this "reference" array: (let* (x refarray)) However, doing so, causes refarray to take on the value of x. "why" and " how do I prevent this" are my questions. -- Kasper van Wijk +(303) 273 3464 (work) +(303) 384 9602 (home) -- From majordom@umnstat.stat.umn.edu Tue Apr 1 15:57:21 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28010; Tue, 1 Apr 1997 15:57:21 -0600 Return-Path: Received: from sulu.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27998; Tue, 1 Apr 1997 15:57:11 -0600 Received: from chekov by sulu.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id NAA25589; Tue, 1 Apr 1997 13:57:09 -0800 (PST) Date: Tue, 1 Apr 1997 13:57:07 -0800 (PST) From: Thomas Lumley X-Sender: thomas@chekov To: stat-lisp-news@umnstat.stat.umn.edu Subject: cox regression In-Reply-To: <9612092108.AA01690@nokomis.stat.umn.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I have a new (alpha) version of code for cox regression models. It now handles arbitrary time-dependent covariates (ie anything you can write a function to specify). Also new: tests of proportional hazards assumption based on scaled Schoenfeld residuals (as in survival4 for S). Available from http://www.biostat.washington.edu/~thomas/cox.html Thomas Lumley -----------------------------------------------------+------ Biostatistics : "Never attribute to malice what : Uni of Washington : can be adequately explained by : Box 357232 : incompetence" - Hanlon's Razor : Seattle WA 98195-7232 : : ------------------------------------------------------------ From majordom@umnstat.stat.umn.edu Tue Apr 1 16:06:39 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28159; Tue, 1 Apr 1997 16:06:39 -0600 Return-Path: Received: from sulu.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28154; Tue, 1 Apr 1997 16:06:27 -0600 Received: from chekov by sulu.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id OAA25809; Tue, 1 Apr 1997 14:06:23 -0800 (PST) Date: Tue, 1 Apr 1997 14:06:21 -0800 (PST) From: Thomas Lumley X-Sender: thomas@chekov To: kasper@dix.mines.edu Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: static variables In-Reply-To: <199704012153.OAA31222@bremmer.Mines.EDU> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Tue, 1 Apr 1997, Kasper van Wyk wrote: > To speed up the process, I would like to define such an > array once, after which I "point to this "reference" array: > > (let* (x refarray)) > > However, doing so, causes refarray to take on the value of > x. "why" and " how do I prevent this" are my questions. What you want is (let* ((x refarray)) more code ) ie, more parentheses. Thomas Lumley ------------------------------------------------------+------ Biostatistics : "Never attribute to malice what : Uni of Washington : can be adequately explained by : Box 357232 : incompetence" - Hanlon's Razor : Seattle WA 98195-7232 : : ------------------------------------------------------------ From majordom@umnstat.stat.umn.edu Tue Apr 1 19:47:21 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29530; Tue, 1 Apr 1997 19:47:21 -0600 Return-Path: Received: from salmon.maths.tcd.ie by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29526; Tue, 1 Apr 1997 19:47:07 -0600 Received: from orange.stats.tcd.ie by salmon.maths.tcd.ie with SMTP id aa09791; 2 Apr 97 2:47 +0100 Received: from [194.125.134.55] by stats.tcd.ie (4.1/SMI-4.1) id AA07786; Wed, 2 Apr 97 02:34:16 BST X-Sender: dbearly@orange.stats.tcd.ie Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 2 Apr 1997 01:46:58 +0000 To: stat-lisp-news@umnstat.stat.umn.edu From: "Dom. Early" Subject: barchart class (part 1 of 3) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, while working on some code that will form part of my Ph.D. thesis I had to build a BarChart class that provided all the standard facilities of the built in graph types such as histogram. There are several implementations of barcharts available for XLISP-STAT but most provided a limited subset of the functionality of the built in plots. I came up with a histogram subclass that kinda hacks a histogram so it looks like a barchart, and then draw labels on the bins. You can do most everything with the barchart that you can do with a histogram such as brushing and selecting, hiding and focusing on selections and so on. The interface is practically the same as with the histogram, execpt you can use non-numeric data, e.g. (barchart '(foo bar foo foo bar bar foobar foo foobar bar 1 1 2 3 3 3)) You may notice that the standard slicer menu option is not provided under the barchart menu, as using a slicer doesn't really make sense with categorical data anyway. With this implementation you are limited to at most 24 different categories of data (as we are limited to 50 bins in a histogram). This should still prove useful for many datasets though. One difficulty arose to do with scaling and rounding which could leave two bins contiguous - a considerable amount of the code will automatically recognise and rectify this problem by resizing the plot until the problem 'goes away'. You may notice this happening when you resize plots, i.e. they appear to grow again! A nice touch is the ability to use vertical labels which reduces screen clutter and label overlapping. The source is divided into two files, the first is my standard set of utilities and the second is the barchart class which uses the utility functions heavily. If my utility functions cause too many identifier clashes with other peoples code then please let me know and I will see about building them into the class itself or using more standard lisp identifiers. enjoy, Dom. +--------------------------------------------------------------------------+ | Dominic Early e-mail: Dom.Early@tcd.ie | | A.31, Department of Statistics, Phone: +353-1-608 1917 | | Trinity College, Fax: +353-1-661 5046 | | Dublin 2, Ireland. | | | | http://www2.tcd.ie/Statistics/staff/dbearly.html | +--------------------------------------------------------------------------+ _/_/_/_/ _/_/_/ _/ _/ _/_/_/_/_/ _/_/_/ _/_/_/_/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/_/_/_/ _/_/_/ _/ _/ _/ _/ _/_/_/ _/_/_/_/ From majordom@umnstat.stat.umn.edu Tue Apr 1 19:47:04 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29523; Tue, 1 Apr 1997 19:47:04 -0600 Return-Path: Received: from salmon.maths.tcd.ie by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29519; Tue, 1 Apr 1997 19:46:53 -0600 Received: from orange.stats.tcd.ie by salmon.maths.tcd.ie with SMTP id aa09784; 2 Apr 97 2:46 +0100 Received: from [194.125.134.55] by stats.tcd.ie (4.1/SMI-4.1) id AA07783; Wed, 2 Apr 97 02:33:58 BST X-Sender: dbearly@orange.stats.tcd.ie Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 2 Apr 1997 01:46:39 +0000 To: stat-lisp-news@umnstat.stat.umn.edu From: "Dom. Early" Subject: barchart class (part 3 of 3) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Save this file as barchart.lsp ------8X------8X------8X------8X------8X------8X------8X------8X------8X------8X (require "DOMUTILS") (require "GRAPHS") ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; BARCHART.lsp : Ver 1.0 beta (2 April 1997) ;; ;; A prototype class for a bar chart ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; ;; (c) Dom. Early 1997 ;; ;; Dom. Early ;; Dept. of Statistics ;; Trinity College Dublin ;; Ireland ;; ;; ;;DEVELOPMENT ENVIRONMENT: ;; ;; Machine: PowerMac 7200/90 ;; Operating system: MacOS 7.5.5 ;; XLISP-STAT: 3.50 Beta (PowerMac Binary) ;; ;; ;;SOURCE CODE ;; Source code is optimised for a 4 space tab width and indented according ;; to my own personal preferences (sorry, I know hardened lisp'ers will hate ;; it, so pretty print it! 8) ;; ;; ;;COMMENTS: ;; Users are free to use this code, UNMODIFIED, in their packages, provided ;; they clearly acknowledge it's use and credit the author. ;; ;; If you have any bug reports, suggestions or comments about this code, ;; or if you have suggestions on how to improve it, please let me know ;; via email. Also, if you find this code useful, I'd appreciate your ;; letting me know that as well. ;; ;; This code has only been tested on non Power Mac platforms so I am unsure ;; as to how well or badly it performs on Windows/Unix boxes. ;; Comments from "beta testers" for these platforms are especially welcome. ;; ;; ;; Enjoy, ;; Dom. ;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;;OVERVIEW: ;; The primary function in the code below is to produce a well-integrated ;; barchart type plot to XLisp-Stat. Other implementations of barcharts ;; are designed to provide basic barchart facilities only, and lack many ;; of the features of the built-in plots. ;; ;; My barchart-proto is implemented as a subclass of histogram-proto ;; and I have tried to implement the package as closely to the standard ;; histogram interface as possible. ;; ;; Essentially a barchart of say, 5 categories is built using a histogram ;; of 9 bins, with every second bin containing zero elements. My code then ;; adds labels to each bin, the label being the value of the category. ;; ;; ;;USAGE: ;; Use in a similar manner to histogram except barchart accepts non-numeric ;; data. e.g. ;; ;; >(barchart '(foo bar foobar foo bar foo 1 1 2 3 "foo" "FOOBAR")) ;; ;; This produces a barchart with the following categories: ;; '(1 2 3 BAR FOO FOOBAR foo) ;; ;; (note that it converts all elements to strings so foobar is equivalent to ;; "FOOBAR"). ;; ;; In physically small barcharts or barcharts with a lot of categories labels ;; can obscure each other so I have provided an option to switch to vertically ;; oriented labels. Select Options... from the Barchart menu when the ;; barchart is the active window and click on the Vertical Labels checkbox. ;; ;; ;;ADVANTAGES: ;; Has all features of standard plots including selecting and brushing, ;; linking, hiding/focusing selections, selection colors, and plot options. ;; ;; ;;DISADVANTAGES: ;; Limited to 25 categories ;; ;; ;;KNOWN PROBLEMS ;; A problem with to do with scaling/rounding errors that occurs when the ;; the barchart is resized to a very small size, or when the barchart ;; contains a lot of categories could leave two contiguous non-empty bars. ;; A fix simply tests for this problem and then resizes the barchart until ;; the problem 'goes away'. Certain other functions which call the plot ;; method :adjust-to-data may also cause the graph size to be adjusted ;; slightly under certain circumstances. ;; ;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; PROTOTYPE DECLARATION ;; ;; sBarLabelList - Used to store the labels for each bar ;; sVerticalLabels - Boolean flags whether to display labels vertically ;; sScaleFix - Boolean flags whether in scale fix cycle ;; sDoScaleFix - Boolean flags whether scale fix is enabled ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defproto Barchart-Proto '(sBarLabelList sVerticalLabels sScaleFix sDoScaleFix) () histogram-proto) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-Proto :BarLabel(pBarIndex) "Args: (pBarIndex:NATURAL) Return nth bar label from our list of labels" (GetElement (send self :BarLabelList) pBarIndex) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-Proto :VerticalLabels(&optional (pVerticalLabels nil pSet)) "Args: (pVerticalLabels:BOOLEAN) Gettor/Settor for sVerticalLabels flag, which flags whether the labels should be printed horizontally or vertically" (when pSet (send self :slot-value 'sVerticalLabels pVerticalLabels) (send self :AdjustMargins) ) (send self :slot-value 'sVerticalLabels) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-Proto :ScaleFix(&optional (pScaleFix nil pSet)) "Args: (pScaleFix:BOOLEAN) Gettor/Settor for sScaleFix flag, which flags whether are in the scale fix cycle" (if pSet (send self :slot-value 'sScaleFix pScaleFix) ) (send self :slot-value 'sScaleFix) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-Proto :DoScaleFix(&optional (pDoScaleFix nil pSet)) "Args: (pDoScaleFix:BOOLEAN) Gettor/Settor for sDoScaleFix flag, which flags whether scale fixing is enabled" (if pSet (send self :slot-value 'sDoScaleFix pDoScaleFix) ) (send self :slot-value 'sDoScaleFix) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :BarLabelList(&optional (pList nil pSet)) "Args: (pList:LIST OF STRINGS) Gettor/Settor for sBarLabelList slot, which contains the list of labels for the barchart bars" (if pSet (send self :slot-value 'sBarLabelList pList) ) (send self :slot-value 'sBarLabelList) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :AdjustMargins() "Args: () Sets the margins of the plot to accomadate the labels of the bars, according to the label orientation" (when (send self :VerticalLabels) ;; Adjust the margins to allow for vertical label strings ;; Algorithm : ;; Find the biggest label string length ;; Set self margin to allow for that string length ;; Resize ourself to include the new margin (let( (maxtw 0) ) (dolist(label (send self :BarLabelList) nil) (if(< maxtw (send self :text-width label)) (setf maxtw (send self :text-width label)) ) ) (send self :margin 0 0 0 (- maxtw 20)) (send self :size (+ (car (send self :size)) maxtw) (+ (cadr (send self :size)) maxtw) ) ) ) (when (not (send self :VerticalLabels)) (send self :margin 0 0 0 0) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :CountElements(pList) "Args: (pList:LIST OF VALUES) Inputs a list of values Returns a list of unique values from the input list, together with a list of the count of each unique element in the input list" (let( (elementlist (sort (UniqueList pList) #'string<)) (elementcount '()) (curindex 0) ) (dotimes (i (length elementlist)) (setf elementcount (AddElement elementcount 0)) ) (dolist(element pList nil) (setf curindex (GetElementIndex element elementlist)) (setf elementcount (SetElement elementcount curindex (+ 1 (GetElement elementcount curindex)) ) ) ) (list elementlist elementcount) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :CheckBinCounts() "Args: () Ensure that every second bar is empty so histogram appears as a barchart. Returns true if any of the odd bars is non-empty. Used to fix the scaling problem" (let( (count 0) (warning false) (element nil) (bincounts (send self :bin-counts)) ) (loop (if (or warning (Nil? bincounts)) (return) ) (setf element (car bincounts)) (setf bincounts (cdr bincounts)) (if (oddp count) (if (Neq 0 element) (setf warning true) ) ) (setf count (+ count 1)) ) warning ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :AdjustFrameSize() "Args: () Increments the horizontal aspect of the fame-size by one. Used to fix the scaling problem" (send self :frame-size (1+ (car (send self :frame-size))) (cadr (send self :frame-size)) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :Redraw() "Args: () Overloads the histogram-proto method to redraw the barchart. Draws the histogram and then adds the labels" (when (not (send self :DoScaleFix)) (send self :start-buffering) (send self :erase-window) (call-next-method) (send self :LabelBars) (send self :buffer-to-screen) ) (when (send self :DoScaleFix) (when (not (send self :ScaleFix)) (send self :start-buffering) ) ;; (send self :erase-window) (call-next-method) (when (not (send self :ScaleFix)) (send self :ScaleFix true) (if (send self :CheckBinCounts) (loop (send self :AdjustFrameSize) (send self :adjust-to-data :draw false) (send self :resize) (send self :erase-window) (send self :redraw) (if (not (send self :CheckBinCounts)) (return) ) ) ) (send self :ScaleFix false) ) (when (not (send self :ScaleFix)) (send self :LabelBars) (send self :buffer-to-screen) ) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :LabelBars() "Args: () Draws the labels for each bar either vertically or horizontally" (let*( (crect (send self :content-rect)) (left (car crect)) (top (car (cdr crect))) (width (car (cdr (cdr crect)))) (height (car (cdr (cdr (cdr crect))))) (bottom (+ top height)) (nbars (send self :num-bins)) (bwidth (/ (third crect) nbars)) (ta (send self :text-ascent)) (tw 0) (maxtw 0) (curlabel "") ) (dotimes(i nbars) (when (evenp i) (setf curlabel (send self :BarLabel (/ i 2))) (if(< maxtw (send self :text-width curlabel)) (setf maxtw (send self :text-width curlabel)) ) ) ) (dotimes(i nbars) (when (evenp i) (setf curlabel (send self :BarLabel (/ i 2))) (setf tw (send self :text-width curlabel)) (if (send self :VerticalLabels) (send self :draw-string-up curlabel (round (+ (* i bwidth) left ta)) (+ bottom maxtw 5) ) (send self :draw-string curlabel (round (+ (* i bwidth) left)) (+ bottom (+ ta 5)) ) ) ) ) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :adjust-to-data(&key (draw true)) "Args: (draw:BOOLEAN) Overloads the histogram-proto method, redrawing if draw is true. Resets the number of bins and sets the scaled range" (call-next-method :draw draw) (send self :num-bins (- (* (length (send self :BarLabelList)) 2) 1)) (send self :x-axis true nil 0) (send self :scaled-range 0 0 (- (length (send self :BarLabelList)) 1)) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun barchart(pList &key (title "Barchart")) "Args: (pList:LIST title:STRING) Creates a barchart from the data in pList" (let*( (bchart (send Barchart-proto :new 1 :show false)) (strlist (mapcar #'ToString pList)) (barcounts (send bchart :CountElements strlist)) (labels (car barcounts)) (counts (cadr barcounts)) (histlist '()) ) (when (< (length labels) 25) (dolist(element strlist nil) (setf histlist (AddElement histlist (* 1 (GetElementIndex element labels))) ) ) (send bchart :title title) (send bchart :add-points histlist) (send bchart :scaled-range 0) (send bchart :BarLabelList labels) (send bchart :adjust-to-data :draw false) (send bchart :VerticalLabels false) (send bchart :DoScaleFix true) (send bchart :show-window) ) (when (not (< (length labels) 25)) (Message-Dialog "cannot plot barcharts of more than 24 categories") (setf bchart nil) ) bchart ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;; (defmeth Barchart-proto :make-options-dialog-items() "Args: () Overload the histogram-proto method to add an option to draw labels vertically" (remove nil (AddElement (call-next-method) (send graph-toggle-item-proto :new "Vertical Labels" self :VerticalLabels) ) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (send Barchart-proto :title "Barchart") (send Barchart-proto :menu-title "Barchart") (send Barchart-proto :fixed-aspect nil) (send Barchart-proto :size 250 125) (send Barchart-proto :menu-template '( link ;; showing-labels mouse resize-brush ;; --------------- dash erase-selection focus-on-selection show-all ;; symbol color selection ;; --------------- dash ;; slicer rescale options #+unix save-image ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (provide "BARCHART") +--------------------------------------------------------------------------+ | Dominic Early e-mail: Dom.Early@tcd.ie | | A.31, Department of Statistics, Phone: +353-1-608 1917 | | Trinity College, Fax: +353-1-661 5046 | | Dublin 2, Ireland. | | | | http://www2.tcd.ie/Statistics/staff/dbearly.html | +--------------------------------------------------------------------------+ _/_/_/_/ _/_/_/ _/ _/ _/_/_/_/_/ _/_/_/ _/_/_/_/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/_/_/_/ _/_/_/ _/ _/ _/ _/ _/_/_/ _/_/_/_/ From majordom@umnstat.stat.umn.edu Tue Apr 1 19:47:44 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29540; Tue, 1 Apr 1997 19:47:44 -0600 Return-Path: Received: from salmon.maths.tcd.ie by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29533; Tue, 1 Apr 1997 19:47:28 -0600 Received: from orange.stats.tcd.ie by salmon.maths.tcd.ie with SMTP id aa09801; 2 Apr 97 2:47 +0100 Received: from [194.125.134.55] by stats.tcd.ie (4.1/SMI-4.1) id AA07790; Wed, 2 Apr 97 02:34:39 BST X-Sender: dbearly@orange.stats.tcd.ie Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 2 Apr 1997 01:46:58 +0000 To: stat-lisp-news@umnstat.stat.umn.edu From: "Dom. Early" Subject: barchart class (part 1 of 3) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, while working on some code that will form part of my Ph.D. thesis I had to build a BarChart class that provided all the standard facilities of the built in graph types such as histogram. There are several implementations of barcharts available for XLISP-STAT but most provided a limited subset of the functionality of the built in plots. I came up with a histogram subclass that kinda hacks a histogram so it looks like a barchart, and then draw labels on the bins. You can do most everything with the barchart that you can do with a histogram such as brushing and selecting, hiding and focusing on selections and so on. The interface is practically the same as with the histogram, execpt you can use non-numeric data, e.g. (barchart '(foo bar foo foo bar bar foobar foo foobar bar 1 1 2 3 3 3)) You may notice that the standard slicer menu option is not provided under the barchart menu, as using a slicer doesn't really make sense with categorical data anyway. With this implementation you are limited to at most 24 different categories of data (as we are limited to 50 bins in a histogram). This should still prove useful for many datasets though. One difficulty arose to do with scaling and rounding which could leave two bins contiguous - a considerable amount of the code will automatically recognise and rectify this problem by resizing the plot until the problem 'goes away'. You may notice this happening when you resize plots, i.e. they appear to grow again! A nice touch is the ability to use vertical labels which reduces screen clutter and label overlapping. The source is divided into two files, the first is my standard set of utilities and the second is the barchart class which uses the utility functions heavily. If my utility functions cause too many identifier clashes with other peoples code then please let me know and I will see about building them into the class itself or using more standard lisp identifiers. enjoy, Dom. +--------------------------------------------------------------------------+ | Dominic Early e-mail: Dom.Early@tcd.ie | | A.31, Department of Statistics, Phone: +353-1-608 1917 | | Trinity College, Fax: +353-1-661 5046 | | Dublin 2, Ireland. | | | | http://www2.tcd.ie/Statistics/staff/dbearly.html | +--------------------------------------------------------------------------+ _/_/_/_/ _/_/_/ _/ _/ _/_/_/_/_/ _/_/_/ _/_/_/_/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/_/_/_/ _/_/_/ _/ _/ _/ _/ _/_/_/ _/_/_/_/ From majordom@umnstat.stat.umn.edu Tue Apr 1 20:15:07 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29798; Tue, 1 Apr 1997 20:15:07 -0600 Return-Path: Received: from salmon.maths.tcd.ie by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29542; Tue, 1 Apr 1997 19:47:44 -0600 Received: from orange.stats.tcd.ie by salmon.maths.tcd.ie with SMTP id aa09813; 2 Apr 97 2:47 +0100 Received: from by stats.tcd.ie (4.1/SMI-4.1) id AB07790; Wed, 2 Apr 97 02:34:50 BST X-Sender: dbearly@orange.stats.tcd.ie Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 2 Apr 1997 01:46:39 +0000 To: stat-lisp-news@umnstat.stat.umn.edu From: "Dom. Early" Subject: barchart class (part 3 of 3) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Save this file as barchart.lsp ------8X------8X------8X------8X------8X------8X------8X------8X------8X------8X (require "DOMUTILS") (require "GRAPHS") ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; BARCHART.lsp : Ver 1.0 beta (2 April 1997) ;; ;; A prototype class for a bar chart ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; ;; (c) Dom. Early 1997 ;; ;; Dom. Early ;; Dept. of Statistics ;; Trinity College Dublin ;; Ireland ;; ;; ;;DEVELOPMENT ENVIRONMENT: ;; ;; Machine: PowerMac 7200/90 ;; Operating system: MacOS 7.5.5 ;; XLISP-STAT: 3.50 Beta (PowerMac Binary) ;; ;; ;;SOURCE CODE ;; Source code is optimised for a 4 space tab width and indented according ;; to my own personal preferences (sorry, I know hardened lisp'ers will hate ;; it, so pretty print it! 8) ;; ;; ;;COMMENTS: ;; Users are free to use this code, UNMODIFIED, in their packages, provided ;; they clearly acknowledge it's use and credit the author. ;; ;; If you have any bug reports, suggestions or comments about this code, ;; or if you have suggestions on how to improve it, please let me know ;; via email. Also, if you find this code useful, I'd appreciate your ;; letting me know that as well. ;; ;; This code has only been tested on non Power Mac platforms so I am unsure ;; as to how well or badly it performs on Windows/Unix boxes. ;; Comments from "beta testers" for these platforms are especially welcome. ;; ;; ;; Enjoy, ;; Dom. ;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;;OVERVIEW: ;; The primary function in the code below is to produce a well-integrated ;; barchart type plot to XLisp-Stat. Other implementations of barcharts ;; are designed to provide basic barchart facilities only, and lack many ;; of the features of the built-in plots. ;; ;; My barchart-proto is implemented as a subclass of histogram-proto ;; and I have tried to implement the package as closely to the standard ;; histogram interface as possible. ;; ;; Essentially a barchart of say, 5 categories is built using a histogram ;; of 9 bins, with every second bin containing zero elements. My code then ;; adds labels to each bin, the label being the value of the category. ;; ;; ;;USAGE: ;; Use in a similar manner to histogram except barchart accepts non-numeric ;; data. e.g. ;; ;; >(barchart '(foo bar foobar foo bar foo 1 1 2 3 "foo" "FOOBAR")) ;; ;; This produces a barchart with the following categories: ;; '(1 2 3 BAR FOO FOOBAR foo) ;; ;; (note that it converts all elements to strings so foobar is equivalent to ;; "FOOBAR"). ;; ;; In physically small barcharts or barcharts with a lot of categories labels ;; can obscure each other so I have provided an option to switch to vertically ;; oriented labels. Select Options... from the Barchart menu when the ;; barchart is the active window and click on the Vertical Labels checkbox. ;; ;; ;;ADVANTAGES: ;; Has all features of standard plots including selecting and brushing, ;; linking, hiding/focusing selections, selection colors, and plot options. ;; ;; ;;DISADVANTAGES: ;; Limited to 25 categories ;; ;; ;;KNOWN PROBLEMS ;; A problem with to do with scaling/rounding errors that occurs when the ;; the barchart is resized to a very small size, or when the barchart ;; contains a lot of categories could leave two contiguous non-empty bars. ;; A fix simply tests for this problem and then resizes the barchart until ;; the problem 'goes away'. Certain other functions which call the plot ;; method :adjust-to-data may also cause the graph size to be adjusted ;; slightly under certain circumstances. ;; ;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; PROTOTYPE DECLARATION ;; ;; sBarLabelList - Used to store the labels for each bar ;; sVerticalLabels - Boolean flags whether to display labels vertically ;; sScaleFix - Boolean flags whether in scale fix cycle ;; sDoScaleFix - Boolean flags whether scale fix is enabled ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defproto Barchart-Proto '(sBarLabelList sVerticalLabels sScaleFix sDoScaleFix) () histogram-proto) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-Proto :BarLabel(pBarIndex) "Args: (pBarIndex:NATURAL) Return nth bar label from our list of labels" (GetElement (send self :BarLabelList) pBarIndex) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-Proto :VerticalLabels(&optional (pVerticalLabels nil pSet)) "Args: (pVerticalLabels:BOOLEAN) Gettor/Settor for sVerticalLabels flag, which flags whether the labels should be printed horizontally or vertically" (when pSet (send self :slot-value 'sVerticalLabels pVerticalLabels) (send self :AdjustMargins) ) (send self :slot-value 'sVerticalLabels) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-Proto :ScaleFix(&optional (pScaleFix nil pSet)) "Args: (pScaleFix:BOOLEAN) Gettor/Settor for sScaleFix flag, which flags whether are in the scale fix cycle" (if pSet (send self :slot-value 'sScaleFix pScaleFix) ) (send self :slot-value 'sScaleFix) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-Proto :DoScaleFix(&optional (pDoScaleFix nil pSet)) "Args: (pDoScaleFix:BOOLEAN) Gettor/Settor for sDoScaleFix flag, which flags whether scale fixing is enabled" (if pSet (send self :slot-value 'sDoScaleFix pDoScaleFix) ) (send self :slot-value 'sDoScaleFix) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :BarLabelList(&optional (pList nil pSet)) "Args: (pList:LIST OF STRINGS) Gettor/Settor for sBarLabelList slot, which contains the list of labels for the barchart bars" (if pSet (send self :slot-value 'sBarLabelList pList) ) (send self :slot-value 'sBarLabelList) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :AdjustMargins() "Args: () Sets the margins of the plot to accomadate the labels of the bars, according to the label orientation" (when (send self :VerticalLabels) ;; Adjust the margins to allow for vertical label strings ;; Algorithm : ;; Find the biggest label string length ;; Set self margin to allow for that string length ;; Resize ourself to include the new margin (let( (maxtw 0) ) (dolist(label (send self :BarLabelList) nil) (if(< maxtw (send self :text-width label)) (setf maxtw (send self :text-width label)) ) ) (send self :margin 0 0 0 (- maxtw 20)) (send self :size (+ (car (send self :size)) maxtw) (+ (cadr (send self :size)) maxtw) ) ) ) (when (not (send self :VerticalLabels)) (send self :margin 0 0 0 0) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :CountElements(pList) "Args: (pList:LIST OF VALUES) Inputs a list of values Returns a list of unique values from the input list, together with a list of the count of each unique element in the input list" (let( (elementlist (sort (UniqueList pList) #'string<)) (elementcount '()) (curindex 0) ) (dotimes (i (length elementlist)) (setf elementcount (AddElement elementcount 0)) ) (dolist(element pList nil) (setf curindex (GetElementIndex element elementlist)) (setf elementcount (SetElement elementcount curindex (+ 1 (GetElement elementcount curindex)) ) ) ) (list elementlist elementcount) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :CheckBinCounts() "Args: () Ensure that every second bar is empty so histogram appears as a barchart. Returns true if any of the odd bars is non-empty. Used to fix the scaling problem" (let( (count 0) (warning false) (element nil) (bincounts (send self :bin-counts)) ) (loop (if (or warning (Nil? bincounts)) (return) ) (setf element (car bincounts)) (setf bincounts (cdr bincounts)) (if (oddp count) (if (Neq 0 element) (setf warning true) ) ) (setf count (+ count 1)) ) warning ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :AdjustFrameSize() "Args: () Increments the horizontal aspect of the fame-size by one. Used to fix the scaling problem" (send self :frame-size (1+ (car (send self :frame-size))) (cadr (send self :frame-size)) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :Redraw() "Args: () Overloads the histogram-proto method to redraw the barchart. Draws the histogram and then adds the labels" (when (not (send self :DoScaleFix)) (send self :start-buffering) (send self :erase-window) (call-next-method) (send self :LabelBars) (send self :buffer-to-screen) ) (when (send self :DoScaleFix) (when (not (send self :ScaleFix)) (send self :start-buffering) ) ;; (send self :erase-window) (call-next-method) (when (not (send self :ScaleFix)) (send self :ScaleFix true) (if (send self :CheckBinCounts) (loop (send self :AdjustFrameSize) (send self :adjust-to-data :draw false) (send self :resize) (send self :erase-window) (send self :redraw) (if (not (send self :CheckBinCounts)) (return) ) ) ) (send self :ScaleFix false) ) (when (not (send self :ScaleFix)) (send self :LabelBars) (send self :buffer-to-screen) ) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :LabelBars() "Args: () Draws the labels for each bar either vertically or horizontally" (let*( (crect (send self :content-rect)) (left (car crect)) (top (car (cdr crect))) (width (car (cdr (cdr crect)))) (height (car (cdr (cdr (cdr crect))))) (bottom (+ top height)) (nbars (send self :num-bins)) (bwidth (/ (third crect) nbars)) (ta (send self :text-ascent)) (tw 0) (maxtw 0) (curlabel "") ) (dotimes(i nbars) (when (evenp i) (setf curlabel (send self :BarLabel (/ i 2))) (if(< maxtw (send self :text-width curlabel)) (setf maxtw (send self :text-width curlabel)) ) ) ) (dotimes(i nbars) (when (evenp i) (setf curlabel (send self :BarLabel (/ i 2))) (setf tw (send self :text-width curlabel)) (if (send self :VerticalLabels) (send self :draw-string-up curlabel (round (+ (* i bwidth) left ta)) (+ bottom maxtw 5) ) (send self :draw-string curlabel (round (+ (* i bwidth) left)) (+ bottom (+ ta 5)) ) ) ) ) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defmeth Barchart-proto :adjust-to-data(&key (draw true)) "Args: (draw:BOOLEAN) Overloads the histogram-proto method, redrawing if draw is true. Resets the number of bins and sets the scaled range" (call-next-method :draw draw) (send self :num-bins (- (* (length (send self :BarLabelList)) 2) 1)) (send self :x-axis true nil 0) (send self :scaled-range 0 0 (- (length (send self :BarLabelList)) 1)) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun barchart(pList &key (title "Barchart")) "Args: (pList:LIST title:STRING) Creates a barchart from the data in pList" (let*( (bchart (send Barchart-proto :new 1 :show false)) (strlist (mapcar #'ToString pList)) (barcounts (send bchart :CountElements strlist)) (labels (car barcounts)) (counts (cadr barcounts)) (histlist '()) ) (when (< (length labels) 25) (dolist(element strlist nil) (setf histlist (AddElement histlist (* 1 (GetElementIndex element labels))) ) ) (send bchart :title title) (send bchart :add-points histlist) (send bchart :scaled-range 0) (send bchart :BarLabelList labels) (send bchart :adjust-to-data :draw false) (send bchart :VerticalLabels false) (send bchart :DoScaleFix true) (send bchart :show-window) ) (when (not (< (length labels) 25)) (Message-Dialog "cannot plot barcharts of more than 24 categories") (setf bchart nil) ) bchart ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;; (defmeth Barchart-proto :make-options-dialog-items() "Args: () Overload the histogram-proto method to add an option to draw labels vertically" (remove nil (AddElement (call-next-method) (send graph-toggle-item-proto :new "Vertical Labels" self :VerticalLabels) ) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (send Barchart-proto :title "Barchart") (send Barchart-proto :menu-title "Barchart") (send Barchart-proto :fixed-aspect nil) (send Barchart-proto :size 250 125) (send Barchart-proto :menu-template '( link ;; showing-labels mouse resize-brush ;; --------------- dash erase-selection focus-on-selection show-all ;; symbol color selection ;; --------------- dash ;; slicer rescale options #+unix save-image ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (provide "BARCHART") +--------------------------------------------------------------------------+ | Dominic Early e-mail: Dom.Early@tcd.ie | | A.31, Department of Statistics, Phone: +353-1-608 1917 | | Trinity College, Fax: +353-1-661 5046 | | Dublin 2, Ireland. | | | | http://www2.tcd.ie/Statistics/staff/dbearly.html | +--------------------------------------------------------------------------+ _/_/_/_/ _/_/_/ _/ _/ _/_/_/_/_/ _/_/_/ _/_/_/_/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/_/_/_/ _/_/_/ _/ _/ _/ _/ _/_/_/ _/_/_/_/ From majordom@umnstat.stat.umn.edu Tue Apr 1 20:15:33 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29805; Tue, 1 Apr 1997 20:15:33 -0600 Return-Path: Received: from salmon.maths.tcd.ie by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29538; Tue, 1 Apr 1997 19:47:36 -0600 Received: from orange.stats.tcd.ie by salmon.maths.tcd.ie with SMTP id aa09807; 2 Apr 97 2:47 +0100 Received: from by stats.tcd.ie (4.1/SMI-4.1) id AB07790; Wed, 2 Apr 97 02:34:43 BST X-Sender: dbearly@orange.stats.tcd.ie Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 2 Apr 1997 01:47:16 +0000 To: stat-lisp-news@umnstat.stat.umn.edu From: "Dom. Early" Subject: barchart class (part 2 of 3) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Save this file as domutils.lsp ------8X------8X------8X------8X------8X------8X------8X------8X------8X------8X ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; DOMUTILS.lsp : Ver 1.0 beta (2 April 1997) ;; ;; A set of utility functions and constants ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; ;; (c) Dom. Early 1997 ;; ;; Dom. Early ;; Dept. of Statistics ;; Trinity College Dublin ;; Ireland ;; ;; ;;DEVELOPMENT ENVIRONMENT: ;; ;; Machine: PowerMac 7200/90 ;; Operating system: MacOS 7.5.5 ;; XLISP-STAT: 3.50 Beta (PowerMac Binary) ;; ;; ;;SOURCE CODE ;; Source code is optimised for a 4 space tab width and indented according ;; to my own personal preferences (sorry, I know hardened lisp'ers will hate ;; it, so pretty print it! 8) ;; ;; ;;COMMENTS: ;; Users are free to use this code, UNMODIFIED, in their packages, provided ;; they clearly acknowledge it's use and credit the author. ;; ;; If you have any bug reports, suggestions or comments about this code, ;; or if you have suggestions on how to improve it, please let me know ;; via email. Also, if you find this code useful, I'd appreciate your ;; letting me know that as well. ;; ;; This code has only been tested on non Power Mac platforms so I am unsure ;; as to how well or badly it performs on Windows/Unix boxes. ;; Comments from "beta testers" for these platforms are especially welcome. ;; ;; ;; Enjoy, ;; Dom. ;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; ;;OVERVIEW: ;; The primary aim of the code below is to provide clearer constants, ;; and utility functions to make coding simpler and code more readable. ;; These functions are general in nature and can be used by a wide range ;; of packages. I hope the identifiers used don't clash with other ;; people's code. ;; ;; Note: In speed-critical areas some functions may impede performance ;; where used ;; ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Define a global true for readability ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (setf true t) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; define a global false for readability ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (setf false nil) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Define a nil predicate for readability ;; PARAMETERS ;; Value - the value to be tested ;; ;; RETURNS ;; boolean ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun Nil?(pValue) "Args: (pValue) Returns true if pValue is nil, otherwise false" (eq nil pValue) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Define a list type predicate for readability ;; PARAMETERS ;; Value - the value to be tested ;; ;; RETURNS ;; boolean ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun List?(pValue) "Args: (pValue) Returns true if pValue is a list, otherwise false" (listp pValue) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Define a number type predicate for readability ;; PARAMETERS ;; Value - the value to be tested ;; ;; RETURNS ;; boolean ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun Number?(pValue) "Args: (pValue) Returns true if pValue is a number, otherwise false" (numberp pValue) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Define a real number type predicate for readability ;; PARAMETERS ;; Value - the value to be tested ;; ;; RETURNS ;; boolean ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun Real?(pValue) "Args: (pValue) Returns true if pValue is a real, otherwise false" (realp pValue) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Define an integer number type predicate for readability ;; PARAMETERS ;; Value - the value to be tested ;; ;; RETURNS ;; boolean ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun Integer?(pValue) "Args: (pValue) Returns true if pValue is an integer, otherwise false" (integerp pValue) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Define a natural number type predicate for readability ;; PARAMETERS ;; Value - the value to be tested ;; ;; RETURNS ;; boolean ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun Natural?(pValue) "Args: (pValue) Returns true if pValue is a natural number, otherwise false" (and (integerp pValue) (>= pValue 0)) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Define a string type predicate for readability ;; PARAMETERS ;; Value - the value to be tested ;; ;; RETURNS ;; boolean ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun String?(pValue) "Args: (pValue) Returns true if pValue is a string, otherwise false" (stringp pValue) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Define a character type predicate for readability ;; PARAMETERS ;; Value - the value to be tested ;; ;; RETURNS ;; boolean ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun Character?(pValue) "Args: (pValue) Returns true if pValue is a character, otherwise false" (if (symbolp pValue) (= 1 (length (symbol-name pValue))) (characterp pValue) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Define an empty list predicate for readability ;; PARAMETERS ;; Value - the value to be tested ;; ;; RETURNS ;; boolean ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun Empty?(pValue) "Args: (pValue) Returns true if pValue is an empty list (nil), otherwise false" (nil? pValue) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Define a Not Equal operator for readability ;; PARAMETERS ;; Value - the value to be tested ;; ;; RETURNS ;; boolean ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun Neq(pValue1 pValue2) "Args: (pValue1 pValue2) Returns true if pValue1 is not equal to pValue2, otherwise false" (not (eq pValue1 pValue2)) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Define an Error Handling Routine ;; PARAMETERS ;; String - File containing routine ;; String - Routine ;; String - Error message ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun DomsErr(pFileName pRoutine pErrorMsg) "Args: (pFilename pRoutine pErrorMsg) Diagnostic display of the error catching routine" (princ (format nil " File : ~a" pFileName)) (terpri) (princ (format nil "Routine : ~a" pRoutine)) (terpri) (princ (format nil " Error : ~a" pErrorMsg)) (terpri) nil ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Function to generate a list of random numbers ;; PARAMETERS ;; N - the number of elements in output list, must be greater than 0 ;; Range - random numbers will be between 0 and Range, if integer entered ;; then a list of integers is generated, if real then a list of ;; reals ;; ;; RETURNS ;; list ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun RandomSequence(pN pRange) "Args: (pN pRange) Returns a list of pN random numbers in the range 0 - pRange" (if (and (natural? pN) (number? pRange) (< 0 pN)) (let ( (output '()) ) (dotimes(i pN output) (setf output (cons (random pRange) output)) ) ) (DomsErr "Generals.lsp" "RandomSequence" "Bad parameters") ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Function to test if a value is in a list ;; PARAMETERS ;; Value - the value to be searched for in the list ;; List - the list to be searched, must be a list ;; ;; RETURNS ;; boolean ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Member test required to catch strings (defun InList?(pValue pList) "Args: (pValue pList) Returns true if an item of value pValue is in pList, otherwise false" (if (list? pList) (not (empty? (member pValue pList :test #'equal))) false ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Function to return the index of an element if it is in a list, ;; indexes begin at 0 ;; PARAMETERS ;; Value - the value to be searched for ;; List - the list to be searched, must be a list ;; ;; RETURNS ;; number ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun GetElementIndex(pValue pList &optional(pIndex 0)) "Args: (pValue pList) Returns the index of the first item of value pValue in pList, or nil if the item is not in the list. Note: Indexing begins at index 0" (if (list? pList) (let( (tail (member pValue pList :test #'equal)) ) (if tail (- (length pList) (length tail)) nil ) ) (DomsErr "Generals.lsp" "GetElementIndex" "Parameter must be a list") ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Function to get a value from a list given its index, indexes begin at 0 ;; PARAMETERS ;; List - the list to be searched, must be a list ;; Index - the index of the element ;; ;; RETURNS ;; element ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun GetElement(pList pIndex) "Args: (pList pIndex) Returns the item at index pIndex in pList, or nil if the item is not in the list. Note: Indexing begins at index 0" (if (and (list? pList) (natural? pIndex)) (nth pIndex pList) (DomsErr "Generals.lsp" "GetElement" "Bad parameters") ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Destructive Function to set a value in a list given its index, ;; indexes begin at 0 ;; PARAMETERS ;; List - the list to be searched, must be a list ;; Index - the index of the element ;; Value - the new value of the element ;; ;; RETURNS ;; List ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun SetElement(pInList pIndex pValue) "Args: (pList pIndex pValue) Returns a list similar to pList but with item at index pIndex set to pValue. Note: Indexing begins at index 0" (cond ((and (list? pInList) (natural? pIndex) (> (length pInList) pIndex)) (setf (nth pIndex pInList) pValue) pInList ) (true (DomsErr "Generals.lsp" "SetElement" "Bad parameters") ) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Function to add a value to a list ;; PARAMETERS ;; List - the list into which the value is stored, must be a list ;; any - the value to be added ;; ;; RETURNS ;; List ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun AddElement(pList pValue) "Args: (pList pValue) Returns a list with pValue appended to pList" (if (list? pList) (append pList (list pValue)) (DomsErr "Generals.lsp" "AddElement" "Bad parameters") ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Function to get last character from a string ;; PARAMETERS ;; String - the list to be searched, must be a string ;; ;; RETURNS ;; char ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun GetStringTerminator(pString) "Args: (pString) Returns the last character of pString" (if (and (string? pString) (not (eq (length pString) 0))) (char pString (- (length pString) 1)) (DomsErr "Generals.lsp" "GetStringTerminator" "Parameter must be a string") ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Function to partially copy a string ;; PARAMETERS ;; String - the string to be copied, must be a string ;; N - the number of characters to copy, if greater than ;; the length of the string, the string only is returned ;; ;; RETURNS ;; string ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun StrnCpy(pSrc pN &optional (pDest "") (pCount 0)) "Args: (pString pN) Returns a string containing up to n characters of pString" (if (and (string? pSrc) (natural? pN)) (if (or (eq pCount pN) (eq pCount (length pSrc))) pDest (StrnCpy pSrc pN (concatenate 'String pDest (string (char pSrc pCount))) (+ pCount 1) ) ) (DomsErr "Generals.lsp" "StrnCpy" "Bad parameters") ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Function to concanenate n strings ;; PARAMETERS ;; String...String ;; strings to be concatenated together ;; ;; RETURNS ;; string ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun StrCat(&rest pArgs) "Args: (&rest pStrings) Returns a string containing the contcatenation of all argument strings" (let( (result "") ) (if (some #'(lambda(x) (not (string? x))) pArgs) (DomsErr "Generals.lsp" "StrnCpy" "Bad parameters") (dolist(astr pArgs result) (setf result (concatenate 'String result astr)) ) ) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Function to subtract 2 strings ;; PARAMETERS ;; String1 String2 ;; string2 to be subtracted from string1 ;; ;; RETURNS ;; string ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun StrDiff(pStr1 pStr2) "Args: (pStr1 pStr2) Returns a pStr1 less pStr2 if pStr1 is of form <*>" (let( (result pStr1) (lstr2 (length pStr2)) ) (if (or (not (string? pStr1)) (not (string? pStr2))) (DomsErr "Generals.lsp" "StrnCpy" "Bad parameters") (dotimes(i lstr2 result) (setf result (string-right-trim (tostring (char pStr2 (- lstr2 (+ i 1)))) result ) ) ) ) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Function to remove the last character from a string ;; PARAMETERS ;; String - the string to be modified, must be a string. If ;; the string is empty it returns an empty string ;; ;; RETURNS ;; string ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun RemoveStringTerminator(pString) "Args: (pString) Returns a string less the last character of the string" (if (string? pString) (if (eq pString "") "" (StrnCpy pString (- (length pString) 1)) ) (DomsErr "Generals.lsp" "RemoveStringTerminator" "Parameter must be a string") ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Function to create a menu 'divider' ;; PARAMETERS ;; none ;; ;; RETURNS ;; menu-item-proto ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun NewDashedMenuItem() "Args: () returns a new menu dividing menu-item-proto object" (send menu-item-proto :new "-") ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Function to remove dupliacates from a list of elements ;; eg (1 2 3 3 4 4 5) -> (1 2 3 4 5) ;; PARAMETERS ;; List - A list of elements, possibly containing duplicates ;; ;; RETURNS ;; List - List of unique elements in same order as input ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun UniqueList(pList) "Args: (pList) Returns a list containing each unique element in pList" (let( (outlist '()) ) (dolist(element pList nil) (setf outlist (if (not (inlist? element outlist)) (AddElement outlist element) outlist ) ) ) outlist ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun GetType(pX) "Args: (pItem) Returns the type of pItem" (if (String? pX) 'STRING (if (Integer? pX) 'INTEGER (if (Real? pX) 'REAL nil ) ) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun ToString(pValue &optional pForce false) "Args: (pValue &optional pForce) Converts pValue to a string, if pForce is TRUE then it converts even strings to nested strings" (if (or pForce (not (String? pValue))) (with-output-to-string (s) (prin1 pValue s)) pValue ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun String2Symbol(pString) "Args (pString) Converts a string to a symbol" (if (string? pString) (read (make-string-input-stream pString)) (DomsErr"Generals.lsp" "String2Symbol" "Parameter must be a string") ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun Available?(pString) "Args: (pString) Returns true if the Module specified in pString is loaded" (if (string? pString) (dolist(element *modules* nil) (if (string= element pString) (return true) ) ) false ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun FlattenTree(pExp &optional(pResult nil)) (if (nil? pExp) pResult (if (List? (car pExp)) (append (append pResult (FlattenTree (car pExp))) (FlattenTree (cdr pExp))) (append (append pResult (list (car pExp))) (Flattentree (cdr pExp))) ) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (defun IsInstance?(pProto pObject) (if (objectp pobject) (eq pProto (send pObject :slot-value 'PROTO-NAME)) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; ;; Note - this does not handle multiple inheritance yet (defun IsDescendant?(pProto pObject) (if (objectp pObject) (if (eq pProto (send pObject :slot-value 'PROTO-NAME)) true (IsDescendant? pProto (car (send pObject :parents))) ) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ; (provide "DOMUTILS") From majordom@umnstat.stat.umn.edu Fri Apr 4 15:53:27 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16989; Fri, 4 Apr 1997 15:53:27 -0600 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16981; Fri, 4 Apr 1997 15:53:15 -0600 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA13173 (5.67b/IDA-1.5 for ); Fri, 4 Apr 1997 16:52:00 -0500 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA11719 for ; Fri, 4 Apr 1997 16:51:03 -0400 Message-Id: <3345A261.DD5@kodak.com> Date: Fri, 04 Apr 1997 16:52:49 -0800 From: Jeff Morris Reply-To: morris@Kodak.COM Organization: Eastman Kodak Company X-Mailer: Mozilla 3.0 (Win16; I) Mime-Version: 1.0 To: xlispstatnews Subject: Successive divisions Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Here's my weekend lisp exercise for all you die-hards out there who can't find anything to do outside (it's finally going to be warm here in Upstate so I'll be out fishing)!! I have a number, call it "initial" and a list call it "counts". What I would like to do is create a new list where the first element is "initial" and each sucessive element is divided by 2 say. This is continued until the new list is the same length as "counts". That is, I am creating a list with the serial dilutions of a starting value the same length as another list. As always, appreciate any help -- Jeffrey D. Morris, Ph.D. Health & Environment Labs Eastman Kodak Company morris@kodak.com From majordom@umnstat.stat.umn.edu Fri Apr 4 16:50:36 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17391; Fri, 4 Apr 1997 16:50:36 -0600 Return-Path: Received: from neopath.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17387; Fri, 4 Apr 1997 16:50:26 -0600 Received: from denali.neopath.com by neopath.com (SMI-8.6/SMI-SVR4) id OAA06406; Fri, 4 Apr 1997 14:45:29 -0800 Received: from pilchuck.neopath.com by denali.neopath.com (5.x/SMI-SVR4) id AA04835; Fri, 4 Apr 1997 14:45:48 -0800 Received: by pilchuck.neopath.com (5.x/SMI-SVR4) id AA11464; Fri, 4 Apr 1997 14:45:48 -0800 Date: Fri, 4 Apr 1997 14:45:48 -0800 From: keith@neopath.com (Keith Frost) Message-Id: <9704042245.AA11464@pilchuck.neopath.com> To: morris@Kodak.COM Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <3345A261.DD5@kodak.com> (message from Jeff Morris on Fri, 04 Apr 1997 16:52:49 -0800) Subject: Re: Successive divisions Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk ; Okay, while you're fishing, I'll bite. ;) ; In the hopes that someone will show me a much better way: (defun dilute (len &optional (start 1.0) (divisor 2.0) (accum nil)) (if (<= len 0) (reverse accum) (dilute (- len 1) (/ start divisor) divisor (cons start accum)))) From majordom@umnstat.stat.umn.edu Fri Apr 4 17:04:15 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17493; Fri, 4 Apr 1997 17:04:15 -0600 Return-Path: Received: from mathfs.math.montana.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17483; Fri, 4 Apr 1997 17:04:03 -0600 Received: (from jimrc@localhost) by mathfs.math.montana.edu (8.8.3/8.8.3) id QAA20544; Fri, 4 Apr 1997 16:03:56 -0700 Date: Fri, 4 Apr 1997 16:03:56 -0700 (MST) From: Jim Robison-Cox To: Jeff Morris Cc: xlispstatnews Subject: Re: Successive divisions In-Reply-To: <3345A261.DD5@kodak.com> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Fri, 4 Apr 1997, Jeff Morris wrote: > I have a number, call it "initial" and a list call it "counts". > What I would like to do is create a new list where the first element > is "initial" and each sucessive element is divided by 2 say. This is > continued until the new list is the same length as "counts". That is, > I am creating a list with the serial dilutions of a starting value the > same length as another list. That's a nice little problem, I couldn't resist. You could do this: (def serial-dil (* initial (^ 2 (- 0 (iseq (length counts) ))))) But it would make sense to write it as a function: (defun serial-diltn ( init ct) (* init (^ 2 (- 0 (iseq (length ct)))))) > (serial-diltn 120 (iseq 1 10)) (120 60.0 30.0 15.0 7.5 3.75 1.875 0.9375 0.46875 0.234375) Jim Robison-Cox ____________ Department of Math Sciences | | phone: (406)994-5340 2-214 Wilson Hall \ BZM, MT | FAX: (406)994-1789 Montana State University | *_______| Bozeman, MT 59717 \_| e-mail: jimrc@math.montana.edu From majordom@umnstat.stat.umn.edu Fri Apr 4 18:52:50 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA18036; Fri, 4 Apr 1997 18:52:50 -0600 Return-Path: Received: from kasuga.cc.chuo-u.ac.jp by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA18032; Fri, 4 Apr 1997 18:52:41 -0600 Received: from isg.indsys.chuo-u.ac.jp (isg.indsys.chuo-u.ac.jp [133.91.120.136]) by kasuga.cc.chuo-u.ac.jp (8.7.6+2.6Wbeta7/3.4W4) with SMTP id JAA17845 for ; Sat, 5 Apr 1997 09:53:51 +0900 (JST) Received: from [133.91.120.87] (mactk) by isg.indsys.chuo-u.ac.jp (4.1/6.4J.6-fps0) id AA04736; Sat, 5 Apr 97 09:57:15 JST Message-Id: <9704050057.AA04736@isg.indsys.chuo-u.ac.jp> X-Mailer: Macintosh Eudora Pro Version 2.1.4-Jr2 Mime-Version: 1.0 Content-Type: text/plain; charset="ISO-2022-JP" Date: Sat, 5 Apr 1997 09:53:03 +0900 To: xlispstatnews From: kamakura@indsys.chuo-u.ac.jp (Toshinari Kamakura) Subject: Re: Successive divisions Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello. I would like to change a number to the corresponding string. For example, 1 <=> "1" , 2<=>"2", ... Some times I need the string such as "c1", "c2", ... How can I create the string "c*"? ++++++++++++++++++++++++++ Toshinari Kamakura Department of Industrial and Systems Engineering Chuo University 1-13-27 Kasuga email: kamakura@indsys.chuo-u.ac.jp Bunkyo-ku Tokyo 112, Japan From majordom@umnstat.stat.umn.edu Fri Apr 4 23:09:09 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA18752; Fri, 4 Apr 1997 23:09:09 -0600 Return-Path: Received: from sputnik.shlrc.mq.edu.au by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA18748; Fri, 4 Apr 1997 23:08:59 -0600 Received: from srsuna.shlrc.mq.edu.au (localhost [127.0.0.1]) by sputnik.shlrc.mq.edu.au (8.7.6/8.7.3) with ESMTP id PAA00502 for ; Sat, 5 Apr 1997 15:06:08 +1000 Message-Id: <199704050506.PAA00502@sputnik.shlrc.mq.edu.au> To: stat-lisp-news@umnstat.stat.umn.edu Subject: (system ...) on windows Date: Sat, 05 Apr 1997 15:06:08 +1000 From: Steve Cassidy Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'm having troubles with calling an external program on the Windows version of XLS, I wonder if anyone has a better solution that the one I came up with. THe problem is that I call an external program to extract data from a speech database, the data is then written to a text file and read in to xls. On unix this is fine as (system ...) waits for the command to complete before returning but on windows (system ...) returns immediately. Hence I have had to hack code to sit waiting on the file which is being written to see when it first of all exists and then when it has stopped growing -- then I assume the process has finished and I can read the data: (defun wait-for-file (file) "ARGS: FILE wait until file exists and has stopped growing. Used to wait until another process has finished writing to the file" ;; first wait until the file exists or we get bored (let ((failed nil)) (do ((its 0 (1+ its))) ((or (> its 200)(probe-file file)) (if (> its 200) (setq failed t))) (format t "its = ~a\n" its)) (format t "failed is ~a\n" failed) (if (not failed) ;; then wait for it to stop growing (with-open-file (stream file) (do ((length 0 (file-length stream)) (lastlength 0 length) (sameits 0) (zeroits 0)) ((or (> zeroits 200) (> sameits 50)) (if (> zeroits 200) (setq failed t))) (format t "zeroits = ~a, sameits = ~a, length = ~a\n" zeroits sameits length) ;; if the file is stuck at zero we will wait for 200 ;; iterations before quiting (if (= length 0) (incf zeroits)) ;; increment count if length is not zero (ie some data has ;; been written) and the length is the same as last time ;; we will then wait until the length doesn't change for 50 ;; iterations (if (and (> length 0) (= length lastlength)) (incf sameits) (setf sameits 0))))) (format t "failed is ~a\n" failed) (not failed))) This is of course an inherently falible thing to do -- I have no way of knowing if the external call has really completed or just crashed halfway through, or it may just take longer than my wait time to begin writing data. This seems to be a problem inherent in WinExec, the fn call used to implement (system ...), a quick scan revealed a family of functions (spawn*) in the microsoft api which allow the caller to wait on the child process -- however they are only available for win32 applications. Apart from hacking the source can anyone see a more reliable way of synchronising the two processes under the current xls system. Many thanks, Steve Cassidy SHLRC, Macquarie University Sydney From majordom@umnstat.stat.umn.edu Sat Apr 5 08:10:54 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28402; Sat, 5 Apr 1997 08:10:54 -0600 Return-Path: Received: from mcmail.CIS.McMaster.CA by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28398; Sat, 5 Apr 1997 08:10:45 -0600 Received: from fox (quark16.slip.yorku.ca [130.63.190.173]) by mcmail.CIS.McMaster.CA (8.7.1/8.7.1) with SMTP id JAA15739; Sat, 5 Apr 1997 09:10:33 -0500 (EST) Message-Id: <3.0.1.32.19970405090522.006c3d84@mcmail.cis.mcmaster.ca> X-Sender: jfox@mcmail.cis.mcmaster.ca X-Mailer: Windows Eudora Light Version 3.0.1 (32) Date: Sat, 05 Apr 1997 09:05:22 -0500 To: morris@Kodak.COM From: John Fox Subject: Re: Successive divisions Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <3345A261.DD5@kodak.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk At 04:52 PM 04/04/97 -0800, you wrote: >I have a number, call it "initial" and a list call it "counts". >What I would like to do is create a new list where the first element >is "initial" and each sucessive element is divided by 2 say. This is >continued until the new list is the same length as "counts". That is, >I am creating a list with the serial dilutions of a starting value the >same length as another list. Maybe I misunderstand, but if not, here's a simple solution: (defun dilutions (initial counts &optional (factor 2)) (let ((factors (^ factor (iseq (length counts))))) (/ initial factors))) > (def counts (poisson-rand 10 1)) COUNTS > (dilutions 10 counts) (10 5 2.5 1.25 0.625 0.3125 0.15625 0.078125 0.0390625 0.0195312) > (dilutions 10 counts 3) (10 3.33333 1.11111 0.37037 0.123457 0.0411523 0.0137174 0.00457247 0.00152416 0.000508053) John |----------------------------------------------------| | John Fox jfox@McMaster.ca | | Department of Sociology McMaster University | | World Wide Web: http://davinci.socsci.mcmaster.ca/ | |----------------------------------------------------| From majordom@umnstat.stat.umn.edu Sat Apr 5 14:27:19 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29630; Sat, 5 Apr 1997 14:27:19 -0600 Return-Path: Received: from sulu.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29626; Sat, 5 Apr 1997 14:27:09 -0600 Received: from chekov by sulu.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id MAA03728; Sat, 5 Apr 1997 12:27:06 -0800 (PST) Date: Sat, 5 Apr 1997 12:27:05 -0800 (PST) From: Thomas Lumley X-Sender: thomas@chekov To: xlispstatnews Subject: Numbers to strings In-Reply-To: <9704050057.AA04736@isg.indsys.chuo-u.ac.jp> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Sat, 5 Apr 1997, Toshinari Kamakura wrote: > I would like to change a number to the corresponding string. > For example, 1 <=> "1" , 2<=>"2", ... > Some times I need the string such as "c1", "c2", ... > The (format) function can be made to return, rather than print, a string. So: (def x 1) (format nil "~a" x) returns "1" and (def x (iseq 10)) (mapcar #'(lambda (a) (format nil "c~a" a)) x) returns the list ("c0" "c2" "c3" ... "c9") Along the same lines you can turn symbols (like a-symbol) into strings (like "a-symbol") with the (symbol-name) function. This may be important if you read a data file that has unquoted character variables (which will end up as symbols) and you want them as strings. Thomas Lumley ------------------------------------------------------+------ Biostatistics : "Never attribute to malice what : Uni of Washington : can be adequately explained by : Box 357232 : incompetence" - Hanlon's Razor : Seattle WA 98195-7232 : : ------------------------------------------------------------ From majordom@umnstat.stat.umn.edu Mon Apr 7 03:17:08 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23133; Mon, 7 Apr 1997 03:17:08 -0500 Return-Path: Received: from marraco.udl.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23129; Mon, 7 Apr 1997 03:16:49 -0500 Received: from caragol.udl.es (caragol.udl.net [10.69.1.2]) by marraco.udl.es (8.8.5/8.8.5) with ESMTP id KAA03766 for ; Mon, 7 Apr 1997 10:18:58 -0100 (GMT) Received: from caragol-1878 (caragol-1878 [10.100.7.86]) by caragol.udl.es (8.7.5/8.7.3) with SMTP id KAA03717 for ; Mon, 7 Apr 1997 10:15:02 -0100 (GMT) Date: Mon, 7 Apr 1997 10:15:02 -0100 (GMT) From: Jaume March Message-Id: <199704071115.KAA03717@caragol.udl.es> To: stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Pronto E-Mail [version 2.01] Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello. John M Neuhaus (Statistical Methods in Medical Research 1992;1: page 254) wrote: "Mixed-effects analysis using this model [that involves numerical integration and can be computationally very intensive] are becoming popular since the method is included in the EGRET software package." where is this EGRET package? (My Internet browser is still looking for...) and the real question is: Anyone knows a way to do the same in wxls? I belive wxls has a solution. Any views, comments (or code, even better!!) would be greatly appreciated!! :):):) Jaume March :):):) School of Medicine :):):) University of Lleida :):):) Lleida, 25198 :):):) Jaume.March@cmb.udl.es As always, appreciate any help From majordom@umnstat.stat.umn.edu Mon Apr 7 14:54:21 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA26679; Mon, 7 Apr 1997 14:54:21 -0500 Return-Path: Received: from sulu.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA26672; Mon, 7 Apr 1997 14:54:09 -0500 Received: from chekov by sulu.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id MAA13814; Mon, 7 Apr 1997 12:54:06 -0700 (PDT) Date: Mon, 7 Apr 1997 12:54:04 -0700 (PDT) From: Thomas Lumley X-Sender: thomas@chekov To: stat-lisp-news@umnstat.stat.umn.edu Subject: mixed effects models In-Reply-To: <199704071115.KAA03717@caragol.udl.es> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Mon, 7 Apr 1997, Jaume March wrote: > John M Neuhaus (Statistical Methods in Medical Research 1992;1: page 254) > wrote: > > "Mixed-effects analysis using this model [that involves numerical > integration and can be computationally very intensive] are becoming popular > since the method is included in the EGRET software package." > > where is this EGRET package? (My Internet browser is still looking for...) Until recently it was produced by SERC, a Seattle company. I understand it has just been sold. > > and the real question is: > > Anyone knows a way to do the same in wxls? There is code to fit generalised linear mixed models by two different algorithms at the UCLA archive. One method is Schall's (equivalent to the PQL method of Breslow & Clayton) which is known to perform badly for binary data. The other method is due to Jan de Leeuw, but I don't know how it works (or how well). There is no reason in principle why you couldn't do this in lispstat, but I don't think anyone has actually implemented the calculations. Programming the numerical integration efficiently would be tedious. Thomas Lumley ------------------------------------------------------+------ Biostatistics : "Never attribute to malice what : Uni of Washington : can be adequately explained by : Box 357232 : incompetence" - Hanlon's Razor : Seattle WA 98195-7232 : : ------------------------------------------------------------ From majordom@umnstat.stat.umn.edu Mon Apr 7 21:28:10 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29796; Mon, 7 Apr 1997 21:28:10 -0500 Return-Path: Received: from kasuga.cc.chuo-u.ac.jp by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29786; Mon, 7 Apr 1997 21:27:54 -0500 Received: from isg.indsys.chuo-u.ac.jp (isg.indsys.chuo-u.ac.jp [133.91.120.136]) by kasuga.cc.chuo-u.ac.jp (8.7.6+2.6Wbeta7/3.4W4) with SMTP id LAA02967 for ; Tue, 8 Apr 1997 11:29:13 +0900 (JST) Received: from [133.91.120.87] (mactk) by isg.indsys.chuo-u.ac.jp (4.1/6.4J.6-fps0) id AA02881; Tue, 8 Apr 97 11:32:15 JST Message-Id: <9704080232.AA02881@isg.indsys.chuo-u.ac.jp> X-Mailer: Macintosh Eudora Pro Version 2.1.4-Jr2 Mime-Version: 1.0 Content-Type: text/plain; charset="ISO-2022-JP" Date: Tue, 8 Apr 1997 11:28:01 +0900 To: stat-lisp-news@umnstat.stat.umn.edu From: kamakura@indsys.chuo-u.ac.jp (Toshinari Kamakura) Subject: generation of variables Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, I have a simular type of problem like generation of string: How I create the variable 'c1 'c2 ... ? When we use the function "make-color", we assign the color table to variables. For example, (make-color 'c1 1 1 0) If we can change the variable in the program depending on the other variable, color change is easy. a>2 => (make-color 'c1 1 1 0) 1 (make-color 'c2 0 1 1) 0 (make-color 'c3 1 0 1) . . . On the macintosh the function "free-color" does not work properly. Once we assign a color to the variable, we cannot change the color. This is a bug. Therefore, it is very convenient to generate the variable. ++++++++++++++++++++++++++ Toshinari Kamakura Department of Industrial and Systems Engineering Chuo University 1-13-27 Kasuga email: kamakura@indsys.chuo-u.ac.jp Bunkyo-ku Tokyo 112, Japan From majordom@umnstat.stat.umn.edu Tue Apr 8 08:56:27 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11086; Tue, 8 Apr 1997 08:56:27 -0500 Return-Path: Received: from gateway.ets.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11082; Tue, 8 Apr 1997 08:56:16 -0500 Received: from research.ets.org by gateway.ets.org with SMTP id AA27194 (5.65c/IDA-1.4.4 for ); Tue, 8 Apr 1997 09:56:05 -0400 Received: from comet.research.ets.org by research.ets.org with SMTP id AA05884 (5.67b/IDA-1.4.3 for kamakura@indsys.chuo-u.ac.jp); Tue, 8 Apr 1997 09:56:05 -0400 Received: by comet.research.ets.org (SMI-8.6/SMI-SVR4) id JAA08713; Tue, 8 Apr 1997 09:56:04 -0400 Date: Tue, 8 Apr 1997 09:56:04 -0400 Message-Id: <199704081356.JAA08713@comet.research.ets.org> From: Russell Almond To: kamakura@indsys.chuo-u.ac.jp Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <9704080232.AA02881@isg.indsys.chuo-u.ac.jp> (kamakura@indsys.chuo-u.ac.jp) Subject: Re: generation of variables Reply-To: ralmond@ets.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk The magic function is intern which takes a string and makes a symbol with that print-name. (NB! Lisp stores symbol names interally in upper case.) Therefore you could use something like this: (let ((Color-symbol-number 0)) ;; Defun inside let form gives persistant variable associated ;; with function (defun make-new-color (r b g) "Makes a new color with a unique name and returns the name" (let ((new-name (intern (format nil "COLOR-%d" (incf color-symbol-number))))) (make-color new-name r b g") new-name)) ) Actually, I'm forgetting an easier way to do this with the gensym function. (defun make-new-color (r b g) "Makes a new color with a unique name and returns the name" (let ((new-name (gensym "COLOR-"))) (make-color new-name r b g") new-name)) (There is a subtle difference having to do with packages between gensym and intern in ANSI Common Lisp, but I don't think Xlisp makes that distinction.) I hope this helps. --Russell Almond almond@acm.org From majordom@umnstat.stat.umn.edu Tue Apr 8 09:37:06 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11340; Tue, 8 Apr 1997 09:37:06 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11334; Tue, 8 Apr 1997 09:36:52 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa23880; 8 Apr 97 10:36 EDT Received: from slip-5-8.acc.Virginia.EDU by elvis.med.Virginia.EDU (5.65c/1.34) id AA22969; Tue, 8 Apr 1997 10:36:07 -0400 Date: 8 Apr 97 10:38:56 -0400 Subject: Re: generation of variables From: "Steven D. Majewski" To: Toshinari Kamakura , stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Cyberdog/2.0 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >When we use the function "make-color", we assign the >color table to variables. > >For example, > >(make-color 'c1 1 1 0) > >If we can change the variable in the program >depending on the other variable, color change >is easy. > > a>2 => (make-color 'c1 1 1 0) > 1 (make-color 'c2 0 1 1) > 0 (make-color 'c3 1 0 1) > . > . > . >On the macintosh the function "free-color" does not work >properly. Once we assign a color to the variable, we >cannot change the color. This is a bug. Therefore, >it is very convenient to generate the variable. > In what way ( in using what version ) does this not work ? It appears to work fine for me. ( v 3.50 ): ( make-color 'mygreen 0.0 1.0 0.0 ) [ plot something using :color 'mygreen ] ( free-color 'mygreen ) ( make-color 'mygreen 0.2 1.0 0.5 ) [ plot something else using 'mygreen ] - and if I leave the first plot showing, it evens changes color when I change the color bindings of 'mygreen. What I know doesn't work on the Mac is #'parse-color . I have some additional color functions for the mac that translate an ascii RGB.txt file from and X11 distribution and lets you use all of those colortable names. It also modifies make-color to add new colors to the table, so parse color will work on colors not in the original table and added dynamically. ( I needed this so I could add functions to return a lighter or darker shade of a named color, and I should use them in some of my plot code to make sure that colors chose for lines are not close to the background colors. ) - Steve Majewski From majordom@umnstat.stat.umn.edu Tue Apr 8 09:58:17 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11563; Tue, 8 Apr 1997 09:58:17 -0500 Return-Path: Received: from dmsna.dms.unina.it by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11538; Tue, 8 Apr 1997 09:57:22 -0500 Received: from ppp-2.inopera.it ([194.184.82.101]) by dmsna.dms.unina.it (8.6.12/8.6.9) with SMTP id QAA03328; Tue, 8 Apr 1997 16:56:09 +0200 Message-Id: <199704081456.QAA03328@dmsna.dms.unina.it> X-Sender: pagnotta@143.225.253.1 X-Mailer: Windows Eudora Light Version 1.5.2 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sun, 30 Mar 1997 21:51:18 +0000 To: stat-lisp-news@umnstat.stat.umn.edu From: "Stefano M. Pagnotta" Subject: fortran and lisp Cc: "Jan Deleeuw" Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi all, I retrived the 'clusfind' fortran programs by Rousseeuw from Statlib. It would be interesting having such a programs running under Xlisp. Jan Deleeuw wrote some functions for agglomerative nesting, but it's cumbersome to write again in Xlisp the all set of programs. Do you know where is, if any, an automatic translator from fortran to lisp? Thank you, Stefano. ------------------------------------------ "Stefano M. Pagnotta" Dipartimento di Matematica e Statistica Universita' di Napoli "Federico II" tel/fax +39.(0)825.38101 (home) http://www.dms.unina.it/~pagnotta From majordom@umnstat.stat.umn.edu Tue Apr 8 10:37:37 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11973; Tue, 8 Apr 1997 10:37:37 -0500 Return-Path: Received: from maui.medctr.luc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11967; Tue, 8 Apr 1997 10:37:23 -0500 Received: (from ael@localhost) by maui.medctr.luc.edu (8.7.3/8.6.9) id KAA03998 for stat-lisp-news@umnstat.stat.umn.edu; Tue, 8 Apr 1997 10:36:29 -0500 (CDT) Date: Tue, 8 Apr 1997 10:36:29 -0500 (CDT) From: Andy Long Message-Id: <199704081536.KAA03998@maui.medctr.luc.edu> Content-Type: text Apparently-To: Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk stat-lisp-news@umnstat.stat.umn.edu Hi folks, I just started using 3.50 (rather than 3.47) from within emacs, and I'm having trouble because control-d is now interpreted as "exit from xlispstat" (and I can't seem to remap control-d to something innocuous). Anyone have a fix? Andy From majordom@umnstat.stat.umn.edu Tue Apr 8 10:48:14 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12083; Tue, 8 Apr 1997 10:48:14 -0500 Return-Path: Received: from ts037p7.pop3b.netvision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12077; Tue, 8 Apr 1997 10:48:01 -0500 Received: (from hjstein@localhost) by blinky.bfr.co.il (8.7.6/8.6.9) id SAA01840; Tue, 8 Apr 1997 18:39:15 +0300 Date: Tue, 8 Apr 1997 18:39:15 +0300 Message-Id: <199704081539.SAA01840@blinky.bfr.co.il> From: "Harvey J. Stein" To: "Stefano M. Pagnotta" Cc: stat-lisp-news@umnstat.stat.umn.edu, "Jan Deleeuw" Subject: fortran and lisp In-Reply-To: <199704081456.QAA03328@dmsna.dms.unina.it> References: <199704081456.QAA03328@dmsna.dms.unina.it> Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Stefano M. Pagnotta writes: > Hi all, > I retrived the 'clusfind' fortran programs by Rousseeuw from Statlib. > It would be interesting having such a programs running under Xlisp. > Jan Deleeuw wrote some functions for agglomerative nesting, > but it's cumbersome to write again in Xlisp the all set of programs. > Do you know where is, if any, an automatic translator from fortran to lisp? > Thank you, > Stefano. I think you'd probably be better off compiling the fortran code & calling it from xlispstat. You can compile it to a .o or a shared library, dynamically link it in, and just write wrappers for the few fcns you want to use to convert the data from the native lisp formats to the formats needed by the fortran code. -- Harvey J. Stein Berger Financial Research abel@netvision.net.il From majordom@umnstat.stat.umn.edu Tue Apr 8 12:47:42 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02073; Tue, 8 Apr 1997 12:47:42 -0500 Return-Path: Received: from post3.inre.asu.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02069; Tue, 8 Apr 1997 12:47:34 -0500 Received: from stats1.asu.edu (stats1.asu.edu) by asu.edu (PMDF V5.0-6 #7723) id <01IHGKO33QZ498PZAJ@asu.edu> for stat-lisp-news@umnstat.stat.umn.edu; Tue, 08 Apr 1997 10:47:27 -0700 (MST) Received: from stats1 (LOCALHOST [127.0.0.1]) by stats1.asu.edu with SMTP (8.7.1/8.7.1) id KAA01467 for ; Tue, 08 Apr 1997 10:47:25 -0700 (MST) Date: Tue, 08 Apr 1997 10:47:25 -0700 (MST) From: DAN.HUSTON@asu.edu Subject: putting a graph on its side X-Sender: daddy@stats1 To: stat-lisp-news@umnstat.stat.umn.edu Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Content-Transfer-Encoding: 7BIT Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello Folks, I am looking for a way to make a histogram (or other univariate plot) on its side, like this: -------------| | | ---------------------| | | ---------------------| | | -------------| Any idea how I would do that? Thanks, Dan Dan Huston Phone (602) 965-2420 Measurement, Statistics and Methodological Studies Fax (602) 965-0300 Psychology in Education dan.huston@asu.edu 325 Payne Hall http://seamonkey.ed.asu.edu/~huston Arizona State University Tempe, AZ 85287-0611 From majordom@umnstat.stat.umn.edu Tue Apr 8 13:39:35 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11340; Tue, 8 Apr 1997 09:37:06 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11334; Tue, 8 Apr 1997 09:36:52 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa23880; 8 Apr 97 10:36 EDT Received: from slip-5-8.acc.Virginia.EDU by elvis.med.Virginia.EDU (5.65c/1.34) id AA22969; Tue, 8 Apr 1997 10:36:07 -0400 Date: 8 Apr 97 10:38:56 -0400 Subject: Re: generation of variables From: "Steven D. Majewski" To: Toshinari Kamakura , stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Cyberdog/2.0 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >When we use the function "make-color", we assign the >color table to variables. > >For example, > >(make-color 'c1 1 1 0) > >If we can change the variable in the program >depending on the other variable, color change >is easy. > > a>2 => (make-color 'c1 1 1 0) > 1 (make-color 'c2 0 1 1) > 0 (make-color 'c3 1 0 1) > . > . > . >On the macintosh the function "free-color" does not work >properly. Once we assign a color to the variable, we >cannot change the color. This is a bug. Therefore, >it is very convenient to generate the variable. > In what way ( in using what version ) does this not work ? It appears to work fine for me. ( v 3.50 ): ( make-color 'mygreen 0.0 1.0 0.0 ) [ plot something using :color 'mygreen ] ( free-color 'mygreen ) ( make-color 'mygreen 0.2 1.0 0.5 ) [ plot something else using 'mygreen ] - and if I leave the first plot showing, it evens changes color when I change the color bindings of 'mygreen. What I know doesn't work on the Mac is #'parse-color . I have some additional color functions for the mac that translate an ascii RGB.txt file from and X11 distribution and lets you use all of those colortable names. It also modifies make-color to add new colors to the table, so parse color will work on colors not in the original table and added dynamically. ( I needed this so I could add functions to return a lighter or darker shade of a named color, and I should use them in some of my plot code to make sure that colors chose for lines are not close to the background colors. ) - Steve Majewski From majordom@umnstat.stat.umn.edu Tue Apr 8 14:05:32 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11563; Tue, 8 Apr 1997 09:58:17 -0500 Return-Path: Received: from dmsna.dms.unina.it by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11538; Tue, 8 Apr 1997 09:57:22 -0500 Received: from ppp-2.inopera.it ([194.184.82.101]) by dmsna.dms.unina.it (8.6.12/8.6.9) with SMTP id QAA03328; Tue, 8 Apr 1997 16:56:09 +0200 Message-Id: <199704081456.QAA03328@dmsna.dms.unina.it> X-Sender: pagnotta@143.225.253.1 X-Mailer: Windows Eudora Light Version 1.5.2 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sun, 30 Mar 1997 21:51:18 +0000 To: stat-lisp-news@umnstat.stat.umn.edu From: "Stefano M. Pagnotta" Subject: fortran and lisp Cc: "Jan Deleeuw" Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi all, I retrived the 'clusfind' fortran programs by Rousseeuw from Statlib. It would be interesting having such a programs running under Xlisp. Jan Deleeuw wrote some functions for agglomerative nesting, but it's cumbersome to write again in Xlisp the all set of programs. Do you know where is, if any, an automatic translator from fortran to lisp? Thank you, Stefano. ------------------------------------------ "Stefano M. Pagnotta" Dipartimento di Matematica e Statistica Universita' di Napoli "Federico II" tel/fax +39.(0)825.38101 (home) http://www.dms.unina.it/~pagnotta From majordom@umnstat.stat.umn.edu Tue Apr 8 14:29:23 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11973; Tue, 8 Apr 1997 10:37:37 -0500 Return-Path: Received: from maui.medctr.luc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11967; Tue, 8 Apr 1997 10:37:23 -0500 Received: (from ael@localhost) by maui.medctr.luc.edu (8.7.3/8.6.9) id KAA03998 for stat-lisp-news@umnstat.stat.umn.edu; Tue, 8 Apr 1997 10:36:29 -0500 (CDT) Date: Tue, 8 Apr 1997 10:36:29 -0500 (CDT) From: Andy Long Message-Id: <199704081536.KAA03998@maui.medctr.luc.edu> Content-Type: text Apparently-To: Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk stat-lisp-news@umnstat.stat.umn.edu Hi folks, I just started using 3.50 (rather than 3.47) from within emacs, and I'm having trouble because control-d is now interpreted as "exit from xlispstat" (and I can't seem to remap control-d to something innocuous). Anyone have a fix? Andy From majordom@umnstat.stat.umn.edu Tue Apr 8 14:35:32 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12083; Tue, 8 Apr 1997 10:48:14 -0500 Return-Path: Received: from ts037p7.pop3b.netvision.net.il by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12077; Tue, 8 Apr 1997 10:48:01 -0500 Received: (from hjstein@localhost) by blinky.bfr.co.il (8.7.6/8.6.9) id SAA01840; Tue, 8 Apr 1997 18:39:15 +0300 Date: Tue, 8 Apr 1997 18:39:15 +0300 Message-Id: <199704081539.SAA01840@blinky.bfr.co.il> From: "Harvey J. Stein" To: "Stefano M. Pagnotta" Cc: stat-lisp-news@umnstat.stat.umn.edu, "Jan Deleeuw" Subject: fortran and lisp In-Reply-To: <199704081456.QAA03328@dmsna.dms.unina.it> References: <199704081456.QAA03328@dmsna.dms.unina.it> Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Stefano M. Pagnotta writes: > Hi all, > I retrived the 'clusfind' fortran programs by Rousseeuw from Statlib. > It would be interesting having such a programs running under Xlisp. > Jan Deleeuw wrote some functions for agglomerative nesting, > but it's cumbersome to write again in Xlisp the all set of programs. > Do you know where is, if any, an automatic translator from fortran to lisp? > Thank you, > Stefano. I think you'd probably be better off compiling the fortran code & calling it from xlispstat. You can compile it to a .o or a shared library, dynamically link it in, and just write wrappers for the few fcns you want to use to convert the data from the native lisp formats to the formats needed by the fortran code. -- Harvey J. Stein Berger Financial Research abel@netvision.net.il From majordom@umnstat.stat.umn.edu Tue Apr 8 21:30:55 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03388; Tue, 8 Apr 1997 21:30:55 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03374; Tue, 8 Apr 1997 21:30:46 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id ab18927; 8 Apr 97 22:30 EDT Received: from slip-21-1.itc.Virginia.EDU by elvis.med.Virginia.EDU (5.65c/1.34) id AA26099; Tue, 8 Apr 1997 22:29:16 -0400 Date: 8 Apr 97 22:32:28 -0400 Subject: Re: generation of variables From: "Steven D. Majewski" To: Toshinari Kamakura Cc: stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Cyberdog/2.0 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk > >Please try the following example. > >;;; case 1 > >(setf w (send graph-window-proto :new)) >(send w :use-color t) > >(make-color 'my-color 1 1 0) >(send w :draw-color 'my-color) >(send w :draw-line 50 50 100 100) > >(free-color 'my-color) >(make-color 'my-color 1 0 1) >(send w :draw-color 'my-color) >(send w :draw-line 100 50 50 100) > > In the case above, 'my-color is STILL the :draw-color for the window. I suspect it can't be totally freed. ( It's hard to test what's going on because these lines aren't permanent, aren't redrawn and they tend to disappear if you try to do anything that causes a redraw. ) Try inserting an assignment of :draw-color to another color before trying to free-it. This change below works. ;;; case 1B (setf w (send graph-window-proto :new)) (send w :use-color t) (make-color 'my-color 1 1 0) (send w :draw-color 'my-color) (send w :draw-line 50 50 100 100) ( send w :draw-color 'black ) ;;; ADD THIS LINE (free-color 'my-color) (make-color 'my-color 1 0 1) (send w :draw-color 'my-color) (send w :draw-line 100 50 50 100) >;;; case 2 > >(make-color 'my-color1 1 1 0) >(send w :draw-color 'my-color1) >(send w :draw-line 50 50 100 100) > >(make-color 'my-color2 1 0 1) >(send w :draw-color 'my-color2) >(send w :draw-line 100 50 50 100) > >This is a case that I posted several months ago. >I believe that Prof. Tierney note it. > I'm not exactly sure what you are trying to do -- if you use :draw-line, the lines aren't permanent, so it's hard to see what the effect is. Try this line by line and watch the effect. This works as I would expect: The lines are in two different colors until the :redraw, and then they are both in the new color. ( make-color 'this-color 1 1 0 ) ( setf p ( plot-lines ( iseq 10 ) (iseq 10) :color 'this-color )) ( free-color 'this-color ) ( make-color 'this-color 1 0 1 ) ( send p :add-lines ( iseq 10 ) ( - 9 ( iseq 10 )) :color 'this-color ) ( send p :redraw ) - Steve Majewski From majordom@umnstat.stat.umn.edu Wed Apr 16 12:41:26 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02963; Wed, 16 Apr 1997 12:41:26 -0500 Return-Path: Received: from maui.medctr.luc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02959; Wed, 16 Apr 1997 12:41:20 -0500 Received: (from ael@localhost) by maui.medctr.luc.edu (8.7.3/8.6.9) id MAA10391 for stat-lisp-news@stat.umn.edu; Wed, 16 Apr 1997 12:40:23 -0500 (CDT) Date: Wed, 16 Apr 1997 12:40:23 -0500 (CDT) From: Andy Long Message-Id: <199704161740.MAA10391@maui.medctr.luc.edu> To: stat-lisp-news@umnstat.stat.umn.edu Subject: turtle graphics Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi All, Has anyone implemented turtle graphics in xlispstat? If so, is the code available? Thanks, Andy From majordom@umnstat.stat.umn.edu Thu Apr 17 09:29:18 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA18145; Thu, 17 Apr 1997 09:29:18 -0500 Return-Path: Received: from sungod.ccs.yorku.ca by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA18141; Thu, 17 Apr 1997 09:29:12 -0500 Received: from nexus.yorku.ca (nexus.yorku.ca [130.63.236.20]) by sungod.ccs.yorku.ca (8.8.5/8.6.11) with SMTP id KAA11847; Thu, 17 Apr 1997 10:29:09 -0400 (EDT) Received: from hotspur.psych.yorku.ca (hotspur.psych.yorku.ca [130.63.134.26]) by nexus.yorku.ca (8.6.11/8.6.11) with SMTP id KAA21092; Thu, 17 Apr 1997 10:29:08 -0400 Received: by hotspur.psych.yorku.ca (NX5.67c/NX3.0S) id AA00381; Thu, 17 Apr 97 10:29:08 -0400 Date: Thu, 17 Apr 97 10:29:08 -0400 From: Michael Friendly Message-Id: <9704171429.AA00381@hotspur.psych.yorku.ca> Received: by NeXT.Mailer (1.87.1) Received: by NeXT Mailer (1.87.1) To: Andy Long , stat-lisp-news@umnstat.stat.umn.edu Subject: Re: turtle graphics Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Not xlispstat, but I've done it in SAS. See: http://www.math.yorku.ca/SCS/sugi/saslogo.html Should be relatively easy to transscribe to lisp. -Michael From majordom@umnstat.stat.umn.edu Fri Apr 18 13:27:07 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05038; Fri, 18 Apr 1997 13:27:07 -0500 Return-Path: Received: from wara.bolnet.bo by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05030; Fri, 18 Apr 1997 13:26:56 -0500 Received: from kjetil (dial02.bolnet.bo [166.114.1.12]) by wara.bolnet.bo (8.7.4/8.7.3) with SMTP id OAA12495 for ; Fri, 18 Apr 1997 14:30:25 +0400 Message-Id: <3355C470.76B6@wara.bolnet.bo> Date: Thu, 17 Apr 1997 02:34:24 -0400 From: kjetil halvorsen X-Mailer: Mozilla 2.01 (Win95; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: downloading mac version via pc-windows References: <9704080232.AA02881@isg.indsys.chuo-u.ac.jp> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Sorry to bother the list with this, but now I have tried to long without success: I want to put XLisp-stat on my old se30 in addition to the pc, I downloaded the files from the net, but how can I get them from the pc to mac? I copied them easily to the mac, as macos recognizes pc diskettes, but my stuffit (unstuffit) wont recognize them. Does the files get corrupted in the transfer, or is there something else which is wrong? Thanks, Kjetil. From majordom@nokomis.stat.umn.edu Fri Apr 18 14:46:04 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA07998; Fri, 18 Apr 1997 14:46:04 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA07993; Fri, 18 Apr 1997 14:45:55 -0500 From: Luke Tierney Message-Id: <9704181945.AA07993@nokomis.stat.umn.edu> Subject: 68K Mac 881 version of 3-50 is bad To: stat-lisp-news@nokomis.stat.umn.edu Date: Fri, 18 Apr 1997 14:45:55 -0500 (CDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 346 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk The version of XLISP-STAT 3.50 for 68K Macintoshes with a coporcessor crashes on calls to math functions because it was accidentally linked with the wrong library. I have removed the executable file from the distrtibution. I hope to replace it with a working one in a week or two. Until then, please use the 020 version. Sorry about that. luke From majordom@umnstat.stat.umn.edu Sat Apr 19 15:13:34 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19480; Sat, 19 Apr 1997 15:13:34 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19476; Sat, 19 Apr 1997 15:13:29 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.7.5/10.1) id QAA00845; Sat, 19 Apr 1997 16:12:17 -0400 (GMT+4:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: In-Reply-To: <199704191854.VAA10588@bacall.nnt.com> References: <33568682.19B07517@nervm.nerdc.ufl.edu> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sat, 19 Apr 1997 16:10:27 -0500 To: "Harvey J. Stein" From: Forrest Young Subject: Re: screen shots? Cc: vista@unc.edu, stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk > Paul W. Box wrote: > > >I was wondering if ViSta has any utilities for graphic output? I have > > >been grabbing screen shots of some of the windows and converting them to > > >tiff or postscript format, but is there another way? > > Forrest Young replied: > > No, there is no direct way of printing with ViSta. If you want to print > > graphics you have to do screen shots. If you want to print text you have to > > copy the text and paste it into a text processor and use it to print. > > > > If only XLS provided printing ... > > Harvey Stein asked: >Why can't you just send the window the :save-image message? > >> (send graph-proto :help :save-image) >SAVE-IMAGE >Method args: (&optional (file "image.ps")) >Saves a postscript image of the window to file named FILE by buffering, >sending the window the :REDRAW message, and dumping the buffer to a file. Forrest Young replies: Because I didn't know about it? ;-( I see that it is documented in Luke's book, but it looks like a Unix-only feature. I tried it with Mac XLS 4.26 and it's not found. Does it work under MS-Windows? Or with more recent versions of XLS? If not, could it be made to work more generally in the future? (After all, postscript is useful under non-unix environments). Another thought: Is the work that Jan deLeeuw/Frederic Udina did on printing relevant here? I'll add a PRINT menu item to the FILE menu in the next release for whatever operating systems this works under. --- Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: mailto:forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Voice: 1-919-962-5038 ViSta Software and Documentation site: http://forrest.psych.unc.edu/research/ USA Home Site ViSta Mirrors: http://ssda.anu.edu.au/mirrors/ Australia http://www.sci.usq.edu.au/general/vista/ Australia http://sunsite.univie.ac.at/ViSta/ Austria http://cons-dev.univ-lyon1.fr/vista/vista.html France http://micros.hensa.ac.uk/mirrors/vista/ UK (ac.uk only 8AM-8PM GMT) http://www.psych.unc.edu/fwy/ USA ___ (@ @) ------------------------------------------------ooO--(_)--Ooo--------- ||| From majordom@umnstat.stat.umn.edu Sat Apr 19 15:52:18 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19658; Sat, 19 Apr 1997 15:52:18 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19654; Sat, 19 Apr 1997 15:52:11 -0500 Received: from [164.67.20.83] (ts16-6.wla.ts.ucla.edu) by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA22769; Sat, 19 Apr 97 13:52:02 PDT X-Sender: deleeuw@laplace.stat.ucla.edu Message-Id: In-Reply-To: References: <199704191854.VAA10588@bacall.nnt.com> <33568682.19B07517@nervm.nerdc.ufl.edu> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sat, 19 Apr 1997 13:55:42 -0700 To: forrest@unc.edu From: Jan de Leeuw Subject: Re: screen shots? Cc: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >I see that it is documented in Luke's book, but it looks like a Unix-only >feature. I tried it with Mac XLS 4.26 and it's not found. Does it work >under MS-Windows? Or with more recent versions of XLS? If not, could it be >made to work more generally in the future? (After all, postscript is useful >under non-unix environments). Another thought: Is the work that Jan >deLeeuw/Frederic Udina did on printing relevant here? > >I'll add a PRINT menu item to the FILE menu in the next release for >whatever operating systems this works under. > The De Leeuw/Udina stuff, which can certainly be improved, gets the contents of the slots of a graph-proto and writes a file with instructions for gnuplot. Running gnuplot then approximates the original plot (and there is a whole range of output devices such as gif, fig, latex, postscript etc). Bernard Walter has written similar code that outputs fig, which can be edited in xfig, but is less portable than gnuplot. Ideally, someone should write yet another module of this sort which outputs postscript (and not just a bitmap encapsulated in postscript headers and footers). This should also come in handy in display postscript systems such as Rhapsody. Of course that module should be part of the standard distribution, not just an add-on in Lisp. --- Jan --- Jan de Leeuw; Professor and Director, Interdivisional Program in Statistics; US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Sun Apr 20 08:36:28 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA01066; Sun, 20 Apr 1997 08:36:28 -0500 Return-Path: Received: from sungod.ccs.yorku.ca by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA01062; Sun, 20 Apr 1997 08:36:23 -0500 Received: from nexus.yorku.ca (nexus.yorku.ca [130.63.236.20]) by sungod.ccs.yorku.ca (8.8.5/8.6.11) with SMTP id JAA09028 for ; Sun, 20 Apr 1997 09:36:07 -0400 (EDT) Received: from hotspur.psych.yorku.ca (hotspur.psych.yorku.ca [130.63.134.26]) by nexus.yorku.ca (8.6.11/8.6.11) with SMTP id JAA09991 for ; Sun, 20 Apr 1997 09:36:01 -0400 Received: by hotspur.psych.yorku.ca (NX5.67c/NX3.0S) id AA01238; Sun, 20 Apr 97 09:36:01 -0400 Date: Sun, 20 Apr 97 09:36:01 -0400 From: Michael Friendly Message-Id: <9704201336.AA01238@hotspur.psych.yorku.ca> Received: by NeXT.Mailer (1.87.1) Received: by NeXT Mailer (1.87.1) To: stat-lisp-news@umnstat.stat.umn.edu Subject: Announce: Regression Graphics Workshop (Toronto) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Statistical Consulting Service Playfair Society York University University of Toronto +=================================+ | Workshop on Regression Graphics | +=================================+ Wednesday, April 30 8:30 - 5:00pm Registration and Location Information ------------------------------------- http://www.math.yorku.ca/SCS/Courses/S97.html Voice: 416 736-5061 (Institute for Social Research) Instructors ----------- R. Dennis Cook Sanford Weisberg University of Minnesota Dept. of Applied Statistics Length and format ----------------- One-day course: 8:30-9:30 Lecture 1 10:00-11:00 Lecture 2 11:30-12:30 Lecture 3 < lunch break > 1:45-2:30 Lecture 4 3:00-3:45 Lecture 5 4:15-5:00 Lecture 6 Abstract -------- This course will provide an overview of the ideas of regression graphics, which is a collection of methodological tools that are used to discover and understand dependence of a response on predictors primarily through the use of simple graphs in one, two and three dimensions. It will mostly follow the outline of the book "An Introduction to Regression Graphics", published by Wiley in 1994. Prerequisite for this course is familiarity with standard regression methodology at the level of one of the major textbooks in this area. The course will also introduce the R-code, which is Xlisp-Stat computer code that accompanies the book and allows the user to use all the methods described in the book and in the course. The program is very easy to use, and requires no knowledge of lisp. Objectives ---------- The objectives of this course is to help university-level instructors gain the confidence to use regression graphics in their own teaching, and to provide training in the methodology for practicing statisticians for use in their own work. Particpants will also be given access to the R-code2, the lastest version of the R-code. Outline of the course --------------------- 1. Introduction illustrating graphical issues. 2. Foundations: Structural dimension, sufficient summary plots, linear 3. Finding summary plots and exploring the importance of linear predic- tors. 4. Illustrations, 3D plots, residual plots 5. Graphics for regressions with a binary response 6. Graphics for model assessment. Throughout the course, we will include hints for teaching, and demonstrate the use of the R-code for doing the analyses suggested. Target population ----------------- The target population is graduate students, university-level instructors who teach regression analysis, and practicing statisticians who wish to learn the latest ideas in using graphs to understand regression problems. Learning outcomes ----------------- Regression analysis is one of the fundamental tools of the practicing statistician. The traditional role of graphics in regression, at least with many pre- dictors, has really been peripheral, dealing mostly with questions of model adequacy. Regression graphics moves graphs to the center of analysis. This requires some new theory, but this theory can be presented at a very general and intuitive level. Our hope is to encourage the participants to use this approach in their own work, and particularly in their own teaching. About the Instructors --------------------- Both instructors are professors of applied statistics at the University of Minnesota. Much of their research over the last two decades, and much of their teaching, has been in the areas of regression and graphics. They have jointly written over 25 works, including the textbook An Introduction to Regression Graphics that is the basis of this course and the 1982 monograph "Residuals and Influence in Regression". They are both fellows of the American Statistical Association. From majordom@umnstat.stat.umn.edu Tue Apr 22 10:11:13 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA01268; Tue, 22 Apr 1997 10:11:13 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA01264; Tue, 22 Apr 1997 10:11:05 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa13902; 22 Apr 97 11:10 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA25545; Tue, 22 Apr 1997 15:10:51 GMT Date: Tue, 22 Apr 1997 11:10:50 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: vista@unc.edu Cc: "Paul W. Box" , Xlisp List Subject: Re: file import utility In-Reply-To: <335C5564.2A005640@nervm.nerdc.ufl.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk [ question from vista list, answer cc-ed to xlisp-stat list ] On Tue, 22 Apr 1997, Paul W. Box wrote: > I don't know if this is of interest to this group or not, but I have > just written a perl script to import freelist data into ViSta. The file > takes a list, written as: > > Observ1: ducks; oranges; peaches > Observ2: ducks; elephants; oranges; gum > Observ3: peaches; ducks; gum > Observ4: banjos > > and creates a data structure that, once imported into ViSta, gives a > matrix like > [ ... ] > > I have no idea if I am duplicating something that is already included in > ViSta somewhere else, or if I have created something that is entirely > specific to my needs and noone else's. If there is interest I am > willing to make this code available (with the proper caveats) or place > it on a user contribution site. > It's not too much more difficult to write those sorts of filters in Lisp -- the advantage being that you don't need an extra step external to the program to read in your data. I don't recall what Vista expects a table to look like. There were a lot of other miscellaneous table code in the xlispstat archive. I have some functions in and ../Sources/genutil.lsp> that built tables and read in some non-lisp data files. ( Somewhere else in that directory is some code to read in EMMFF ascii files that have a sequence of "#KEYWORD : value" header lines that give some parameters for the following spectral data. ) When I have the time, I'll try to write up a brief guide on how to do this sort of thing so you can write you own custom data importers, but one brief tip is that you can modify the lisp *readtable* temporarily, and can change the interpretation of chars. You can change "," and ";" for example to be classified as "whitespace" so the normal lisp reader can read in non-lisp lists of numbers. ( I also change "#" to be a comment indicator, like ";" -- lots of unix produced data files use "#" as a comment indicator ( gnuplot for one example ) and for the EMMFF files above, I read in the numeric data and the identifying header with two different routines, so that using those header lines to carry information is optional. ) This technique might work on Vista's reader. Just patch the readtable to interpret ";" as whitespace instead of a comma -- I'm not sure if you'll get an error on the trailing ":" , but you can patch that char also. The only 'trick' is that these patches have to be temporary -- only for reading data files, not for reading lisp sources or command lines -- or else you'll get all sorts of errors from lisp. You have to wrap the changes so that if there is an input error or other abnormal termination, the proper default settings get restored. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Wed Apr 23 07:56:16 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16470; Wed, 23 Apr 1997 07:56:16 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16466; Wed, 23 Apr 1997 07:56:09 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.7.5/10.1) id IAA07015; Wed, 23 Apr 1997 08:56:07 -0400 (GMT+4:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 23 Apr 1997 08:54:07 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: Forrest Young Subject: Re: printing graphics Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi Lispers: I'm taking the liberty of forwarding this reply from Luke to my message to him about priting graphics. Does anyone out there have the time and expertise to do what Luke suggests? (I don't have the expertise). Perhaps this would require two separate people, one for Windows and one for Mac. Forrest >From: Luke Tierney >Subject: Re: printing graphics >To: forrest@unc.edu (Forrest Young) >Date: Wed, 23 Apr 1997 05:41:47 -0500 (CDT) >Mime-Version: 1.0 >Status: > >> >> Luke: >> >> I assume you saw the recent set of messages about printing graphics. >> >> Is there any chance you can include the Unix version's >> (send graph-proto :save-image) >> in the distributions for Mac and Windows? It would add a feature that I >> think many people would appreciate. >> >> Forrest >> > >It could be done. Someone would just have to write the Mac and Windows >version of reading the bitmap pixel by pixel. But that isn't the right >thing to do, since on both platforms the save to clipboard function >already produces something internaly that you should be able to just >write to a file (PICT on Mac, Metafile on Windows). So an easier and >better solution in principle is to just write these out to a file. I >tinkered with this on the Mac once and couldn't get it to work but was >probably doing something stupid. In principle, it should only take a >couple of hours for each of the platforms to get the file writing to >work. The problem is I don't have those couple of hours -- I'm working >on other aspects that are higher priority to me. If someone else >wanted to take this on I'd be happy to put in their code. > >luke > --- Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: mailto:forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Voice: 1-919-962-5038 ViSta Software and Documentation site: http://forrest.psych.unc.edu/research/ USA Home Site ViSta Mirrors: http://ssda.anu.edu.au/mirrors/ Australia http://www.sci.usq.edu.au/general/vista/ Australia http://sunsite.univie.ac.at/ViSta/ Austria http://cons-dev.univ-lyon1.fr/vista/vista.html France http://micros.hensa.ac.uk/mirrors/vista/ UK (ac.uk only 8AM-8PM GMT) http://www.psych.unc.edu/fwy/ USA ___ (@ @) ------------------------------------------------ooO--(_)--Ooo--------- ||| From majordom@umnstat.stat.umn.edu Wed Apr 23 09:47:18 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17111; Wed, 23 Apr 1997 09:47:18 -0500 Return-Path: Received: from maui.medctr.luc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17107; Wed, 23 Apr 1997 09:47:11 -0500 Received: (from ael@localhost) by maui.medctr.luc.edu (8.7.3/8.6.9) id JAA15935 for stat-lisp-news@stat.umn.edu; Wed, 23 Apr 1997 09:46:13 -0500 (CDT) Date: Wed, 23 Apr 1997 09:46:13 -0500 (CDT) From: Andy Long Message-Id: <199704231446.JAA15935@maui.medctr.luc.edu> To: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Forest, et al: I've written a postscript routine, but the hitch is that I use all my own methods for the plots (and maybe lots of other personal stuff). That project is documented in http://sparc20.medctr.luc.edu/~ael/classes/classes.html I have a gzipped tarfile that you can download, and which you can try out relatively simply (I've never used this stuff in Windows, but I use it on my mac and unix). You might be able to get by with my routine without everything else I've done - all you need to do is add a slot to your plot called plot-item, which is a list of lists of things in the plot. That includes stings, pairs of lists, etc. I'll include the code below, to give you the flavor. Good luck; write to me directly if you have any questions/comments. Andy ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;; ;; postscript ;; (defun postscript(filename &key plot (bulsize 1) width height no-axes no-box axes-box (fmt "~,3,3g") ) " ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;; ;; postscript Doc: Creation Date:1/27/97 Author: Andy Long ;; ;; Description/Arguments: (postscript filename &key plot (bulsize 1) width height no-axes no-box axes-box (fmt \"~,3,3g\") ) Requires: filename (a string). Options: plot is a plot-name, if you don't want to use the most recent; bulsize is the size of points; width is the width of the postscript window; height is the height of the postscript window; no-axes is true if you want to can the axes; no-box is true if you want to can the box; axes-box is true if you want the axes part of the box; fmt is a format for the axis labels (defaults to \"~,3,3g\"). ;; Notes: ;; See Also: ;; Examples: (new-plot) (plot sin) (postscript \"postscript.example0.ps\") (plot (sample (iseq 100) 10) (sample (iseq 100) 10)) (postscript \"postscript.example1.ps\") (plot (rand.u 10) (rand.u 10)) (postscript \"postscript.example2.ps\") " (let* ( (plot (if plot plot (first (agw)))) (f (open filename :direction :output)) (stuff (send plot :plot-item)) (variable-labels (send plot :variable-label '(0 1))) (width (if width width 400)) (height (if height height 400)) (domain (send plot :range 0)) (dmin (first domain)) (dmax (second domain)) (ddiff (- dmax dmin)) (range (send plot :range 1)) (rmin (first range)) (rmax (second range)) (rdiff (- rmax rmin)) started bullet-size lineson linewidth linetype point-labels plot-symbol text (tinyx (/ ddiff 100)) (tinyy (/ rdiff 100)) (labels-showing (send plot :showing-labels)) ) (format f "%!PS %%BoundingBox: 0 0 ~d ~d /m {moveto} def /l {lineto} def /r {rlineto} def /s {stroke} def /n {newpath} def /c {closepath} def 2 setlinejoin /rtext{dup stringwidth pop -1 div 0 rmoveto show} def /ltext{show} def /ctext{dup stringwidth pop -2 div 0 rmoveto show} def /1b{1 0 360 arc c fill} def /2b{2 0 360 arc c fill} def /3b{3 0 360 arc c fill} def /4b{4 0 360 arc c fill} def /5b{5 0 360 arc c fill} def /6b{6 0 360 arc c fill} def 1.000000 1.000000 scale /Courier-Bold findfont 14 scalefont setfont " (+ 100 width) (+ 100 height) ) (if no-box (if axes-box (format f " s n 100 100 m 100 ~d l 100 100 m s n " (+ 100 width) (+ 100 height) ) ) (format f " s n 100 100 m 100 ~d l 100 ~d m 100 100 l s n " (+ 100 width) (+ 100 width) (+ 100 width) (+ 100 height) (+ 100 width) (+ 100 height) (+ 100 height) (+ 100 height) ) ) (if no-axes () (let () ;; Plot the y-axis if visible: (if (and (< 0 dmax) (> 0 dmin)) (setq stuff (append stuff (list (list 'lines '(0 0) (list rmin rmax)) )))) ;; Plot the x-axis if visible: (if (and (< 0 rmax) (> 0 rmin)) (setq stuff (append stuff (list (list 'lines (list dmin dmax) '(0 0)) )))) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;; axis tic marks: ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; (setq stuff (append stuff (mapcar #'(lambda(x) (list 'lines (list dmin (- dmin tinyx)) (list x x)) ) (rseq rmin rmax *y-tics*) ))) (setq stuff (append stuff (mapcar #'(lambda(x) (list 'lines (list x x) (list rmin (- rmin tinyy))) ) (rseq dmin dmax *x-tics*) ))) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;; axis tic labels: ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; (setq stuff (append stuff (mapcar #'(lambda(x) (list (format nil fmt x) (- dmin (* 5 tinyx)) (- x (* tinyy (size (format nil fmt x)))) 90 )) (rseq rmin rmax *y-tics*) ))) (setq stuff (append stuff (mapcar #'(lambda(x) (list (format nil fmt x) (- x (* tinyx (size (format nil fmt x)))) (- rmin (* 5 tinyy)) )) (rseq dmin dmax *x-tics*) ))) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;; variable labels: ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; (if variable-labels (setq stuff (append stuff (list (list (first variable-labels) (+ dmin (half ddiff) (- (* tinyx (size (first variable-labels))))) (- rmin (* 10 tinyy)) ) (list (second variable-labels) (- dmin (* 10 tinyx)) (+ rmin (half rdiff) (- (* tinyy (size (second variable-labels)))) ) 90 ) ) ) ) ) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; ;; Usual plot-item stuff, now: ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; (dolist (i stuff) (setq bullet-size bulsize) (setq lineson *lines*) (setq point-labels nil) (if lineson (let () (setq linewidth *line-width*) (setq linetype *line-type*) )) (setq text nil) (cond ((or (functionp i) (listp i)) (cond ((functionp i) (let () (setq x (rseq dmin dmax *points*)) (setq y (mapcar i x)) ) ) ((listp (first i)) (let () (setq x (first i)) (setq y (second i)) ) (if (and *point-labels* (not lineson) labels-showing (= (length *point-labels*) (length x)) ) (let () (setq point-labels *point-labels*) (setq angle nil) ) ) ) ((equal (first i) 'lines) (let () (setq x (second i)) (setq y (third i)) (setq lineson t) (setq linewidth (if (fifth i) (fifth i) linewidth)) (setq linetype (if (sixth i) (sixth i) linetype)) ) ) ((equal (first i) 'points) (let () (setq x (second i)) (setq y (third i)) (setq lineson nil) (setq plot-symbol (if (fifth i) (fifth i))) (setq bullet-size (if (sixth i) (sixth i) bulsize)) ) ) ((equal (first i) 'function) (let () (setq x (rseq dmin dmax *points*)) (setq y (mapcar (second i) x)) ;; third is color, can't use it yet (if (fourth i) (setq lineson (fourth i))) (setq linewidth (if (fifth i) (fifth i) linewidth)) (setq linetype (if (sixth i) (sixth i) linetype)) ) ) ((stringp (first i)) (let () (setq x (list (second i))) (setq y (list (third i))) (setq text (first i)) (setq lineson t) (setq angle (if (fourth i) (fourth i) nil)) ) ) (t (print "Search me!")) ) (setq started nil) (dolist (j (weave (mapcar #'(lambda(x) (+ 100 (floor (* width (/ (- x dmin) ddiff))))) x ) (mapcar #'(lambda(x) (+ 100 (floor (* height (/ (- x rmin) rdiff))))) y ) )) #| ;; What follows is my attempt at trimming the plot. It also trimmed the ;; axis labels, so I decided to get back to it when I get some time.... (setq j (list (if (< (first j) 0) 0 (if (> (first j) (+ 100 width)) (+ 100 width) (first j))) (if (< (second j) 0) 0 (if (> (second j) (+ 100 height)) (+ 100 height) (second j))) ) ) |# (setq j (list (if (> (first j) (+ 100 width)) (+ 100 width) (first j)) (if (> (second j) (+ 100 height)) (+ 100 height) (second j)) ) ) (if (and lineson (not started) linewidth) (if (or (equal linetype 'solid) (equal linetype 'normal)) (format f "~%~d setlinewidth~%0 setgray" linewidth) (format f "~%~d setlinewidth~%.5 setgray" linewidth) ) ) (if started (if lineson (format f "~%~d ~d l~%s n ~d ~d m" (first j) (second j) (first j) (second j)) (let () (format f "~%s n ~d ~d ~db" (first j) (second j) bullet-size) (if point-labels (let( (text (first point-labels)) ) (format f "~%s n ~d ~d m" (first j) (second j)) (setq point-labels (circulate point-labels)) (if angle (format f "~%~g rotate~%(~a) ltext~%-~g rotate~%" angle text angle) (format f "~%(~a) ltext" text) ) ) ) ) ) (let () (if lineson (format f "~%s n ~d ~d m" (first j) (second j)) (format f "~%s n ~d ~d ~db" (first j) (second j) bullet-size) ) (let() (if (or text point-labels) (let() (if point-labels (let() (setq text (first point-labels)) (setq point-labels (circulate point-labels)) )) (format f "~%s n ~d ~d m" (first j) (second j)) (if angle (format f "~%~g rotate~%(~a) ltext~%-~g rotate~%" angle text angle) (format f "~%(~a) ltext" text) ) ) ) ) ) ) (setq started t) ) ) (t (print "can't handle it yet!")) ) ) (format f "~%showpage~%") (close f) ) ) From majordom@umnstat.stat.umn.edu Wed Apr 23 10:51:17 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17665; Wed, 23 Apr 1997 10:51:17 -0500 Return-Path: Received: from hickory.tds-eagan.lmco.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17659; Wed, 23 Apr 1997 10:51:11 -0500 Received: from eaggate1.tds-eagan.lmco.com by hickory (5.x/SMI-SVR4.109) id AA13429; Wed, 23 Apr 1997 10:52:14 -0500 Received: by eaggate1.tds-eagan.lmco.com with SMTP (Microsoft Exchange Server Internet Mail Connector Version 4.0.994.63) id <01BC4FD4.7DC736C0@eaggate1.tds-eagan.lmco.com>; Wed, 23 Apr 1997 10:52:52 -0500 Message-Id: From: "Artz, Ray E @EAG" To: 'stat-lisp-news' Subject: Re: Printing graphics Date: Wed, 23 Apr 1997 10:51:00 -0500 X-Mailer: Microsoft Exchange Server Internet Mail Connector Version 4.0.994.63 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello Lispers, Here is something which is by no means a complete solution to the problem, but which may be of some help to Windows users: There is a freeware utility Clipmeta.exe available in the the dviwin distribution (available from CTAN in /dviware/dviwin/dviwin29.zip; CTAN sites include ftp://ftp.dante.de/tex-archive/ and a mirror at http://www.cdrom.com/pub/tex/ctan/). This utility allows the contents of the windows clipboard to be saved as a windows metafont (.wmf) file which can later be pasted into another application such a word-processor. So one copies the graph via edit-copy, then calls clipmeta which asks for a filename. This is, for me, often more convenient than having to paste the clipboard directly to another application "on-the-fly". Also, the dviwin previewer provides a "special" for the inclusion of .wmf files into tex documents. Clearly, this is far from a complete solution: 1) It is obviously not portable to non-windows systems. 2) I would rather have postscript than *.wmf . 3) The graphs are still bitmaps instead of vector graphics with editable contents. I would personally like to see something like Jan DeLeeuw's system (which apparently does give genuine vector-graphic plots, though I haven't tried it yet), but which lets one get a postscript file directly without having to go through gnuplot. In the meantime, I'll use clipmeta when I'm in a hurry and try to find time to try out Jan's approach. Ray E. Artz, Ph.D. Staff Engineer Precision Agriculture Segment Lockheed Martin P.O. Box 64525 MS U2F27 St. Paul, MN 55164-0525 TEL: (612) 456-2328 FAX: (612) 456-2362 Ray.E.Artz@lmco.com Street Address (Fed. Ex., etc.): MS U2F27 3333 Pilot Knob Road Eagan, MN 55121 Also, Ph.D. Candidate School of Statistics University of Minnesota From majordom@umnstat.stat.umn.edu Thu Apr 24 09:31:36 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04104; Thu, 24 Apr 1997 09:31:36 -0500 Return-Path: Received: from mailtest.ccs.yorku.ca by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04100; Thu, 24 Apr 1997 09:31:29 -0500 Received: from nexus.yorku.ca (nexus.yorku.ca [130.63.236.20]) by sungod.ccs.yorku.ca (8.8.5/8.6.11) with SMTP id KAA21039 for ; Thu, 24 Apr 1997 10:31:19 -0400 (EDT) Received: from hotspur.psych.yorku.ca (hotspur.psych.yorku.ca [130.63.134.26]) by nexus.yorku.ca (8.6.11/8.6.11) with SMTP id KAA23296 for ; Thu, 24 Apr 1997 10:31:13 -0400 Received: by hotspur.psych.yorku.ca (NX5.67c/NX3.0S) id AA07180; Thu, 24 Apr 97 10:31:12 -0400 Date: Thu, 24 Apr 97 10:31:12 -0400 From: Michael Friendly Message-Id: <9704241431.AA07180@hotspur.psych.yorku.ca> Received: by NeXT.Mailer (1.87.1) Received: by NeXT Mailer (1.87.1) To: "'stat-lisp-news'" Subject: Announce: Mosaic displays weblet Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Not a lisp-stat item, but thought folks here might be interested. I have a new Javascript-enabled SAS-powered weblet available. It does my Mosaic Displays of data in an n-way contingency table, using data entered in a form, uploaded from a file, or sample data on my server. It generates dynamic client-side image maps, so you can click on the graphs and see cell values displayed (if your browser supports JavaScript.) It's located at: http://www.math.yorku.ca/SCS/Online/ Comments welcome. -Michael --- Michael Friendly Internet: friendly@hotspur.psych.yorku.ca (NeXTmail OK) Psychology Dept York University Voice: 416 736-5118 Fax: 416 736-5814 4700 Keele Street http://www.math.yorku.ca/SCS/friendly.html Toronto, ONT M3J 1P3 CANADA From majordom@umnstat.stat.umn.edu Tue May 6 04:05:48 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29950; Tue, 6 May 1997 04:05:48 -0500 Return-Path: Received: from hermes.fundp.ac.be by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29946; Tue, 6 May 1997 04:05:40 -0500 Received: from grt-kfps.math.fundp.ac.be by hermes.fundp.ac.be; (5.65v3.2/1.1.8.2/14Dec95-0443PM) id AA31648; Tue, 6 May 1997 11:05:53 +0200 Message-Id: <318DC0E4.30F72698@fundp.ac.be> Date: Mon, 06 May 1996 11:05:40 +0200 From: Isabelle REGINSTER Organization: FUBNDP - MATH - GRT X-Mailer: Mozilla 3.01Gold (X11; I; Linux 2.0.18 i686) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: cross-tables Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, I'm a beginner in XLispStat. I must make cross-tables from two columns of a file. For instance, the first colums is the sex (the value F or M) and the second the age. I want a table computing the number of sex F and M for all ages : age | M | F ------------ 20 | 3 | 5 25 | 6 | 3 How can I make this ? Thanks in advance. Isabelle -- -------------------------------------- Isabelle REGINSTER Facultes Universitaires ND de la Paix Departement de Mathematique Groupe de Recherche sur les Transports (Pr. Ph.Toint) Rempart de la Vierge, 8 B.5000 Namur (Belgium) tel : 32-81-72.49.23 fax : 32-81-72.49.14 isabelle.reginster@fundp.ac.be -------------------------------------- From majordom@umnstat.stat.umn.edu Wed May 7 08:28:44 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16363; Wed, 7 May 1997 08:28:44 -0500 Return-Path: Received: from comtch.iea.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16359; Wed, 7 May 1997 08:28:37 -0500 Received: from localhost (andersr@localhost) by comtch.iea.com (8.8.5/8.8.5) with SMTP id NAA25664 for ; Wed, 7 May 1997 13:28:35 GMT Date: Wed, 7 May 1997 06:28:35 -0700 (PDT) From: "Roderick A. Anderson" X-Sender: andersr@comtch To: Lisp Stat News Subject: [Q] Getting unsubscribed Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Sorry to hit the whole list with this but I've lost the instructions for getting on and off this list. Seems the rm command in a mail directory wipes the slate clean ... ;-) Oh yeah, what are the instructions? TIA, Rod From majordom@umnstat.stat.umn.edu Thu May 8 09:04:46 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA01972; Thu, 8 May 1997 09:04:46 -0500 Return-Path: Received: from zam149.zam.kfa-juelich.de by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA01968; Thu, 8 May 1997 09:04:39 -0500 Received: from hlrz9 by fz-juelich.de (5.65v3.2/DEC-AXP) id AA21839; Thu, 8 May 1997 16:04:37 +0200 Received: from hlrz4.hlrz.kfa-juelich.de by hlrz9.hlrz.kfa-juelich.de (SMI-8.6/SMI-4.1) id QAA07310; Thu, 8 May 1997 16:04:33 +0200 Reply-To: jmu@hlrz9.hlrz.kfa-juelich.de Received: by hlrz4.hlrz.kfa-juelich.de (SMI-8.6/SMI-4.1) id QAA06487; Thu, 8 May 1997 16:04:34 +0200 Date: Thu, 8 May 1997 16:04:34 +0200 Message-Id: <199705081404.QAA06487@hlrz4.hlrz.kfa-juelich.de> From: Johannes Mueller-Gerking To: stat-lisp-news@umnstat.stat.umn.edu Subject: XLS and ilisp Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Dear all, Yesterday I tried to install ilisp-5.8 for use with xlispstat, but couldn't get very far. I was able to start XLS in an emacs buffer, but it was not possible to send a defun from some file to the lisp process (eval-defun-lisp). With the settings as they were, I got the error "ilisp-eval-command not defined" (or similar). I then copied the corresponding definition from the ilisp-cl.el into ilisp-xls.el, and recompiled. Now I got another "... undefined" error. I then realized that the xlisp dialect as defined in ilisp-xls.el inherits from the ilisp dialect, which is difficult to understand for me because ilisp doesn't appear as dialect in the docs. Ok, scheme also inherits from ilisp. Hm. When changing the parent mode to clisp, there were no more error-messages, but after 5 min or so at 80% of an Ultra processor, I thought that there would be no further results either, and stopped. My brief question (after that lengthy intro, sorry): before I dig deeper into it, is there anybody who could provide me with a more complete version of (defdialect xlisp "Xlisp" ilisp (setq ilisp-load-command "(load \"%s\")" ilisp-last-command "*") ) than this version from ilisp-xls.el? Thank you very much! Johannes From majordom@umnstat.stat.umn.edu Thu May 8 11:07:20 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02984; Thu, 8 May 1997 11:07:20 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02972; Thu, 8 May 1997 11:07:10 -0500 Received: from [164.67.20.144] (ts20-3.wla.ts.ucla.edu) by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA16754; Thu, 8 May 97 09:06:59 PDT X-Sender: deleeuw@laplace.stat.ucla.edu Message-Id: In-Reply-To: <199705081404.QAA06487@hlrz4.hlrz.kfa-juelich.de> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 8 May 1997 09:06:51 -0700 To: jmu@hlrz9.hlrz.kfa-juelich.de, stat-lisp-news@umnstat.stat.umn.edu From: Jan de Leeuw Subject: Re: XLS and ilisp Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk At 4:04 PM +0200 5/8/97, Johannes Mueller-Gerking wrote: >Dear all, > >Yesterday I tried to install ilisp-5.8 for use with xlispstat, but >couldn't get very far. I was able to start XLS in an emacs buffer, but >it was not possible to send a defun from some file to the lisp process >(eval-defun-lisp). With the settings as they were, I got the error >"ilisp-eval-command not defined" (or similar). I then copied the >corresponding definition from the ilisp-cl.el into ilisp-xls.el, and >recompiled. Now I got another "... undefined" error. > >I then realized that the xlisp dialect as defined in ilisp-xls.el >inherits from the ilisp dialect, which is difficult to understand for >me because ilisp doesn't appear as dialect in the docs. Ok, scheme >also inherits from ilisp. Hm. > >When changing the parent mode to clisp, there were no more >error-messages, but after 5 min or so at 80% of an Ultra processor, I >thought that there would be no further results either, and stopped. > >My brief question (after that lengthy intro, sorry): before I dig >deeper into it, is there anybody who could provide me with a more >complete version of >(defdialect xlisp "Xlisp" ilisp > (setq ilisp-load-command "(load \"%s\")" > ilisp-last-command "*") > ) >than this version from ilisp-xls.el? > >Thank you very much! > >Johannes The ilisp stuff is for ilisp 5.6 -- ilisp is quite an unstable system, with maintainers and basic interfaces changing a lot, so probably 5.8 is quite different from 5.6. Basically what I did is copy the scheme interface and do some minor tweaks -- I am sure the same route is still open for 5.8. Actually, it's on my list of things to do, which is however (5/8/97) 3.5 miles long. By the way, I think it is on Luke's list too. --- Jan --- Jan de Leeuw; Professor and Director, Interdivisional Program in Statistics; US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Thu May 8 14:53:44 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04389; Thu, 8 May 1997 14:53:44 -0500 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04385; Thu, 8 May 1997 14:53:36 -0500 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA08683 (5.67b/IDA-1.5 for ); Thu, 8 May 1997 15:52:09 -0400 Received: from [150.103.229.15] by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA15206 for ; Thu, 8 May 1997 15:50:56 -0400 Message-Id: <33725927.1FE7@kodak.com> Date: Thu, 08 May 1997 15:52:23 -0700 From: Jeff Morris Reply-To: morris@Kodak.COM Organization: Eastman Kodak Company X-Mailer: Mozilla 3.0 (Win16; I) Mime-Version: 1.0 To: xlispstatnews Subject: Software Verification Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk This is a bit off the subject, but since the issue concerns programs I have written in XLS I thought someone might have some input. Good Laboratory Practices, the bane of industrial labs, requires that software be "verified". Does anyone have any experience with verifying software or suggestions of a procedure that would sufficiently appease regulators. Any thoughts would be appreciated! -- Jeffrey D. Morris, Ph.D. Health & Environment Labs Eastman Kodak Company morris@kodak.com From majordom@umnstat.stat.umn.edu Thu May 8 15:17:22 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04590; Thu, 8 May 1997 15:17:22 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04586; Thu, 8 May 1997 15:17:16 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa26523; 8 May 97 15:59 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA25786; Thu, 8 May 1997 15:59:05 -0400 Date: Thu, 8 May 1997 15:59:04 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Jan de Leeuw Cc: jmu@hlrz9.hlrz.kfa-juelich.de, stat-lisp-news@umnstat.stat.umn.edu Subject: Re: XLS and ilisp In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk What is ilisp ? ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Thu May 8 15:30:38 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04739; Thu, 8 May 1997 15:30:38 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04735; Thu, 8 May 1997 15:30:32 -0500 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id NAA17898; Thu, 8 May 1997 13:30:10 -0700 (PDT) From: Jan Deleeuw Message-Id: <199705082030.NAA17898@galton.math.ucla.edu> Subject: Re: XLS and ilisp To: sdm7g@Virginia.EDU (Steven D. Majewski) Date: Thu, 8 May 1997 13:30:10 -0700 (PDT) Cc: jmu@hlrz9.hlrz.kfa-juelich.de, stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: from "Steven D. Majewski" at May 8, 97 03:59:04 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk > > > What is ilisp ? > > ---| Steven D. Majewski (804-982-0831) |--- > ---| Department of Molecular Physiology and Biological Physics |--- > ---| University of Virginia Health Sciences Center |--- > ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- > By doing just a little every day, you can gradually > let the task completely overwhelm you. > > ilisp is a bunch of emacs lisp files that make it easy to edit and run various lisps from emacs -- you edit your lisp in one window and submit it directly to the interpreter that is running in another window. It can be configured for different lisp systems. If you set it up properly you approximate the ease of use of the mac version, but you have a lot of additional power because all the other emacs functions are available as well. -- Jan de Leeuw; Professor and Director, Interdivisional Program in Statistics; US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Thu May 8 15:54:09 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04955; Thu, 8 May 1997 15:54:09 -0500 Return-Path: Received: from fhat.stat.sc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04951; Thu, 8 May 1997 15:54:02 -0500 Received: (from rossini@localhost) by fhat.stat.sc.edu (8.6.12/8.6.12) id QAA11059; Thu, 8 May 1997 16:52:22 -0400 To: "Steven D. Majewski" Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: XLS and ilisp References: From: Anthony Rossini Date: 08 May 1997 16:52:22 -0400 In-Reply-To: "Steven D. Majewski"'s message of Thu, 8 May 1997 15:59:04 -0400 (EDT) Message-Id: Lines: 15 X-Mailer: Gnus v5.4.50/XEmacs 20.2(beta4) X-Face: -JZ8Fs2;tGK&eaV#. U#%d52]n51?][w92.tu+>13)STQ>>>> "Steven" == Steven D Majewski writes: Steven> What is ilisp ? a reasonably nice inferior-lisp mode in Emacs/XEmacs. (an improvement on running xlispstat, scheme, or common lisp (many variants) from within Emacs, from M-x shell or M-x terminal). -- -tony (Anthony Rossini) Statistics Dept., U South Carolina rossini@stat.sc.edu Columbia, SC 29208 http://www.stat.sc.edu/~rossini/ 803-777-3578(O) 803-777-4048 (fax) "'Bother,' said Pooh, as Cthulu rose up and ate him." From majordom@umnstat.stat.umn.edu Thu May 8 16:14:55 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05244; Thu, 8 May 1997 16:14:55 -0500 Return-Path: Received: from gateway.ets.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05240; Thu, 8 May 1997 16:14:48 -0500 Received: from research.ets.org by gateway.ets.org with SMTP id AA19966 (5.65c/IDA-1.4.4 for ); Thu, 8 May 1997 17:14:46 -0400 Received: by research.ets.org id AA22962 (5.67b/IDA-1.4.3 for stat-lisp-news@umnstat.stat.umn.edu); Thu, 8 May 1997 17:14:44 -0400 Date: Thu, 8 May 1997 17:14:44 -0400 Message-Id: <199705082114.AA22962@research.ets.org> From: Russell Almond To: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <199705082030.NAA17898@galton.math.ucla.edu> (message from Jan Deleeuw on Thu, 8 May 1997 13:30:10 -0700 (PDT)) Subject: Re: XLS and ilisp Reply-To: ralmond@ets.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk ilisp is a bunch of emacs lisp files that make it easy to edit and run various lisps from emacs -- you edit your lisp in one window and submit it directly to the interpreter that is running in another window. It can be configured for different lisp systems. If you set it up properly you approximate the ease of use of the mac version, but you have a lot of additional power because all the other emacs functions are available as well. What is the matter with inferior-lisp-mode (which is built into lisp). I thought that this was a standard package in emacs and it might work better than ilisp, or at least be easier to set up. I thought was just a matter of setting one variable in emacs. --Russell Almond From majordom@umnstat.stat.umn.edu Thu May 8 16:36:55 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05540; Thu, 8 May 1997 16:36:55 -0500 Return-Path: Received: from fhat.stat.sc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05536; Thu, 8 May 1997 16:36:48 -0500 Received: (from rossini@localhost) by fhat.stat.sc.edu (8.6.12/8.6.12) id RAA11176; Thu, 8 May 1997 17:35:56 -0400 To: ralmond@ets.org Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: XLS and ilisp References: <199705082114.AA22962@research.ets.org> From: Anthony Rossini Date: 08 May 1997 17:35:56 -0400 In-Reply-To: Russell Almond's message of Thu, 8 May 1997 17:14:44 -0400 Message-Id: Lines: 19 X-Mailer: Gnus v5.4.50/XEmacs 20.2(beta4) X-Face: -JZ8Fs2;tGK&eaV#. U#%d52]n51?][w92.tu+>13)STQ>>>> "Russell" == Russell Almond writes: Russell> What is the matter with inferior-lisp-mode (which is Russell> built into lisp). I thought that this was a standard Russell> package in emacs and it might work better than ilisp, or Russell> at least be easier to set up. I thought was just a Russell> matter of setting one variable in emacs. Nothing is wrong with it -- but ilisp has many more features (similar to comparing standard TeX mode with AUC-TeX -- there are good reasons for both... except that ilisp used to not be as stable; I've not used it recently so can't compare...). -- -tony (Anthony Rossini) Statistics Dept., U South Carolina rossini@stat.sc.edu Columbia, SC 29208 http://www.stat.sc.edu/~rossini/ 803-777-3578(O) 803-777-4048 (fax) "'Bother,' said Pooh, as Cthulu rose up and ate him." From majordom@umnstat.stat.umn.edu Thu May 8 16:59:30 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05744; Thu, 8 May 1997 16:59:30 -0500 Return-Path: Received: from fhat.stat.sc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05738; Thu, 8 May 1997 16:59:23 -0500 Received: (from rossini@localhost) by fhat.stat.sc.edu (8.6.12/8.6.12) id RAA11189; Thu, 8 May 1997 17:58:34 -0400 To: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: XLS and ilisp From: Anthony Rossini Date: 08 May 1997 17:58:34 -0400 Message-Id: Lines: 34 X-Mailer: Gnus v5.4.50/XEmacs 20.2(beta4) X-Face: -JZ8Fs2;tGK&eaV#. U#%d52]n51?][w92.tu+>13)STQ>>>> "Jan" == Jan de Leeuw writes: Jan> At 4:04 PM +0200 5/8/97, Johannes Mueller-Gerking wrote: >> Yesterday I tried to install ilisp-5.8 for use with xlispstat, >> but couldn't get very far. I was able to start XLS in an emacs Jan> The ilisp stuff is for ilisp 5.6 -- ilisp is quite an Jan> unstable system, with maintainers and basic interfaces Actually, the recent (beta/alpha) version of S-mode will handle XLispStat, though not to the same functionality as Splus or R. If interested, retrieve: http://www.stat.sc.edu/~rossini/projects/S-mode-4.9-b5.tar.gz or ftp://ftp.math.sc.edu/rossini/S-mode-4.9-b5.tar.gz The command for XLispStat is M-x XLS, in this scenario. The current plan is to have all S dialects as well as XLispStat and a nice mechanism for ViSta supported from within Emacs or XEmacs. S-mode 5.0 is currently code-named istat-5.0 (implying interactive, interpretative statistics in Emacs). best, -- -tony (Anthony Rossini) Statistics Dept., U South Carolina rossini@stat.sc.edu Columbia, SC 29208 http://www.stat.sc.edu/~rossini/ 803-777-3578(O) 803-777-4048 (fax) "'Bother,' said Pooh, as Cthulu rose up and ate him." ------- End of forwarded message ------- From majordom@umnstat.stat.umn.edu Thu May 8 17:03:40 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05796; Thu, 8 May 1997 17:03:40 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05792; Thu, 8 May 1997 17:03:33 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa01412; 8 May 97 16:14 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA27136; Thu, 8 May 1997 16:14:06 -0400 Date: Thu, 8 May 1997 16:14:05 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Xlisp List Subject: error bands for line plots. Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I recently had a need for plotting the calculated standard errors around a line plot of my data, so instead of using error-bars, I used a colored region around the line, + or - the error. This was quite easy using my color graph overlay code: ( defmeth scatterplot-proto :add-error-band ( x y yerror &key fill-color ) ( let (( xband ( combine x ( reverse x ))) ( yband ( combine ( + y yerror ) (reverse ( - y yerror ))))) ( send self :add-overlay ( poly ( transpose ( list xband yband )) :fill-color fill-color )))) It creates a polygon region by folding y+error and y-error. ( i.e. one of them is reversed so it folds back and closes ) [ This will eventually get documented and added to the overlay methods at http://galen.med.virginia.edu/~sdm7g/XES/Sources/ ] ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Thu May 8 20:19:28 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06388; Thu, 8 May 1997 20:19:28 -0500 Return-Path: Received: from laplace.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06384; Thu, 8 May 1997 20:19:22 -0500 Received: (from jbond@localhost) by laplace.math.ucla.edu (8.8.3/8.8.3) id SAA10009 for stat-lisp-news@umnstat.stat.umn.edu; Thu, 8 May 1997 18:19:21 -0700 (PDT) From: Jason Bond Message-Id: <199705090119.SAA10009@laplace.math.ucla.edu> Subject: Saving Color Plots To: stat-lisp-news@umnstat.stat.umn.edu Date: Thu, 8 May 1997 18:19:21 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Does anyone out there know how to save plots in color postscript form? The save-to-file in the plots menu saves black and white only. Jason Bond From majordom@umnstat.stat.umn.edu Thu May 8 20:39:07 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06508; Thu, 8 May 1997 20:39:07 -0500 Return-Path: Received: from hotelling.stat.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06504; Thu, 8 May 1997 20:39:02 -0500 Received: from localhost (udina@localhost) by hotelling.stat.unc.edu (8.8.4/8.7.3) with SMTP id VAA04591; Thu, 8 May 1997 21:37:59 -0400 (EDT) X-Authentication-Warning: hotelling: udina owned process doing -bs Date: Thu, 8 May 1997 21:37:58 -0400 (EDT) From: Frederic Udina X-Sender: udina@hotelling To: Jason Bond Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Saving Color Plots In-Reply-To: <199705090119.SAA10009@laplace.math.ucla.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, On Thu, 8 May 1997, Jason Bond wrote: > Does anyone out there know how to save plots in color postscript > form? The save-to-file in the plots menu saves black and white only. > I need that too. If it's not available, I would suggest establishing some standard about how to store colored polygons in a XLS graph-proto object. That way, is someone can find a way to translate plot to postscript including filled polygons, it will work for weverybody. The same call for standars is for storing text. For example, it could be, slot 'text Every entry is a list '(text x y h v &optional font size color) slot must contain a list s such entries. x y h v follow the help of :draw-text method slot 'polygons contain a list of polygons, each one being a list (coordinates color ... ???) where coordinates is ... any suggestion? Frederic Udina (from UNC, Chapel Hill, USA) _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Ramon Trias Fargas, 25-27 | Compuserve: 100555.603 08005 Barcelona | WWWeb: http://libiya.upf.es/ SPAIN | or http://www.econ.upf.es/~udina ___________________________________/_\_________________________________ From majordom@umnstat.stat.umn.edu Fri May 16 12:37:42 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24010; Fri, 16 May 1997 12:37:42 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24006; Fri, 16 May 1997 12:37:34 -0500 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.7.5/10.1) id NAA28564; Fri, 16 May 1997 13:37:26 -0400 (GMT+4:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 16 May 1997 13:38:19 -0500 To: vista@unc.edu, edstat-l@jse.stat.ncsu.edu, stat-l@vm1.mcgill.ca, stat-lisp-news@umnstat.stat.umn.edu From: Forrest Young Subject: French Translation Cc: leguelte@biomserv.univ-lyon1.fr Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi folks: Louis Le Guelte, at the University of Lyon, is translating the ViSta software into French, and he asks me: >I am still trying to find French equivalent for LEVERAGES, DFFITS, and >VARIANCE INFLATION FACTOR that are used in the regression routine. > >Have you any english-french-speaking statistician who could give >me the equivalent? You can either respond to me or to Louis at mailto:leguelte@biomserv.univ-lyon1.fr ----- Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: mailto:forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Voice: 1-919-962-5038 ViSta Software and Documentation site: http://forrest.psych.unc.edu/research/ USA Home Site ViSta Mirrors: http://ssda.anu.edu.au/mirrors Australia http://www.sci.usq.edu.au/general/vista Australia http://sunsite.univie.ac.at/ViSta Austria http://cons-dev.univ-lyon1.fr/vista France http://micros.hensa.ac.uk/mirrors/vista UK (ac.uk only 8AM-8PM GMT) http://www.psych.unc.edu/fwy USA ___ (@ @) ------------------------------------------------ooO--(_)--Ooo--------- ||| From majordom@umnstat.stat.umn.edu Tue May 20 16:06:09 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15552; Tue, 20 May 1997 16:06:09 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15540; Tue, 20 May 1997 16:06:01 -0500 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id OAA11880 for stat-lisp-news@umnstat.stat.umn.edu; Tue, 20 May 1997 14:05:58 -0700 (PDT) From: Jan Deleeuw Message-Id: <199705202105.OAA11880@galton.math.ucla.edu> Subject: Archive To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Tue, 20 May 1997 14:05:58 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I am putting the xlispstat archive partly in an mSQL database, to conform to the bookmarks database on our main server, and the bookmarks databases in the IMS homepage and the Journal of Statistical Software homepage. In fact, these are all (cleverly) disguised subtrees of a single large bookmarks database. See http://www.stat.ucla.edu/bookmarks/bookmarks.phtml?book=xls&parent=/ and then go to artificial intelligence or mathematics/libraries which are already partially filled. The leaves of the tree are still unix directories, which get an interface provided by Apache, but the branches of the tree are mSQL records. It will take me a while to move the stuff, in the meantime the basic interface from the UCLA Statistics homepage will remain the same. ---- Jan -- Jan de Leeuw; Professor and Director, Interdivisional Program in Statistics; US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon May 26 18:41:24 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06173; Mon, 26 May 1997 18:41:24 -0500 Return-Path: Received: from out1.ibm.net by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06169; Mon, 26 May 1997 18:41:16 -0500 Received: (from uucp@localhost) by out1.ibm.net (8.6.9/8.6.9) id XAA233250 for ; Mon, 26 May 1997 23:41:15 GMT Message-Id: <199705262341.XAA233250@out1.ibm.net> Received: from slip166-72-239-241.sc.us.ibm.net(166.72.239.241) by out1.ibm.net via smap (V1.3mjr) id smaXQEDmb; Mon May 26 23:41:03 1997 Comments: Authenticated sender is From: "Brent Hutto" To: stat-lisp-news@umnstat.stat.umn.edu Date: Mon, 26 May 1997 19:39:58 +0000 Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7BIT Subject: Getting graphics in an X window (Linux) X-Mailer: Pegasus Mail for Win32 (v2.53/R1) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I've just set up a Debian GNU/Linux system on my computer and one of the first things I wanted to get running was xlispstat. Release 3-50 wouldn't compile (I'll post another message about that later) so I'm using 3-47 instead. I want to have access to interactive graphics in pop-up windows like I've seen in the MS-Windows version of xlispstat. If I look at my environment variables, I have DISPLAY defined as :0.0 which I believe is correct given that I'm not on a network, just running the X server on the same machine that is running xlispstat. Yet when I try something like (plot-lines (list 1 2 3) (list 1 2 3)) I get some escape codes or whatnot from the GNUPLOT Tektronix routines. For some reason, xlispstat doesn't seem to notice that I'm running under X windows. My setup is a PC running Debian GNU/Linux 1.2.8 and the XFree86 3.2 software. My window manager is one called "wm2" and I'm just opening a generic xterm and typing xlispstat at the shell prompt. Is there perhaps a command-line switch that I should invoke? Or maybe I need to define DISPLAY as something else. Any help is appreciated. From majordom@umnstat.stat.umn.edu Tue Jun 3 02:30:33 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27488; Tue, 3 Jun 1997 02:30:33 -0500 Return-Path: Received: from hermes.fundp.ac.be by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27484; Tue, 3 Jun 1997 02:30:26 -0500 Received: from sturm.math.fundp.ac.be by hermes.fundp.ac.be; (5.65v3.2/1.1.8.2/14Dec95-0443PM) id AA23037; Tue, 3 Jun 1997 09:31:15 +0200 Message-Id: <3393C818.3D0B6296@fundp.ac.be> Date: Tue, 03 Jun 1997 09:30:32 +0200 From: Isabelle REGINSTER Organization: FUNDP - MATH - GRT X-Mailer: Mozilla 3.01Gold (X11; I; Linux 2.0.18 i686) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: write in file Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, I'm a beginner in xlispstat. I would like to write the result of calculations in a file. Per instance : >(def toto (+1 2)) ; I would like to write "toto 3" in a file. How can I do this ? Thanks in advance Isabelle -- -------------------------------------- Isabelle REGINSTER Facultes Universitaires ND de la Paix Departement de Mathematique Groupe de Recherche sur les Transports (Pr. Ph.Toint) Rempart de la Vierge, 8 B.5000 Namur (Belgium) tel : 32-81-72.49.23 fax : 32-81-72.49.14 isabelle.reginster@fundp.ac.be -------------------------------------- From majordom@umnstat.stat.umn.edu Tue Jun 10 07:31:55 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14829; Tue, 10 Jun 1997 07:31:55 -0500 Return-Path: Received: from hermes.fundp.ac.be by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14825; Tue, 10 Jun 1997 07:31:39 -0500 Received: from sturm.math.fundp.ac.be by hermes.fundp.ac.be; (5.65v3.2/1.1.8.2/14Dec95-0443PM) id AA16966; Tue, 10 Jun 1997 14:32:07 +0200 Message-Id: <339D490A.6A123D77@fundp.ac.be> Date: Tue, 10 Jun 1997 14:31:06 +0200 From: Isabelle REGINSTER Organization: FUNDP - MATH - GRT X-Mailer: Mozilla 3.01Gold (X11; I; Linux 2.0.18 i686) Mime-Version: 1.0 To: xlispstat Subject: format Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, Who can me explain the different options of this format function ? The help give me only this : loading in help file information - this will take a minute ...done FORMAT [function-doc] Args: (destination control &rest args) Very basic implementation of Common Lisp format function. Only A, S, D, F, E, G, %, and ~ directives are supported. D, % and ~ can take one argument, R, E and G can take two. What is A, S, D, etc... ? Thank you in advance. Isabelle -- -------------------------------------- Isabelle REGINSTER Facultes Universitaires ND de la Paix Departement de Mathematique Groupe de Recherche sur les Transports (Pr. Ph.Toint) Rempart de la Vierge, 8 B.5000 Namur (Belgium) tel : 32-81-72.49.23 fax : 32-81-72.49.14 isabelle.reginster@fundp.ac.be -------------------------------------- From majordom@umnstat.stat.umn.edu Tue Jun 10 09:26:00 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15346; Tue, 10 Jun 1997 09:26:00 -0500 Return-Path: Received: from odin.mdacc.tmc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15342; Tue, 10 Jun 1997 09:25:49 -0500 Received: by odin.mdacc.tmc.edu; (5.65/1.1.8.2/15Aug95-0158PM) id AA07316; Tue, 10 Jun 1997 09:25:55 -0500 Date: Tue, 10 Jun 1997 09:25:55 -0500 (CDT) From: Ren Yu To: stat-lisp-news@umnstat.stat.umn.edu Subject: save 'draw-text' Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi! I am new to xlisp-stat. I have a simple question that how can I keep the 'text' on the graph which generated by 'draw-text' or 'draw-string'? Right now once I click the graph or other window, the printed text is dispeared. Thanks! Ren From majordom@nokomis.stat.umn.edu Tue Jun 10 10:25:21 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA25414; Tue, 10 Jun 1997 10:25:21 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA25407; Tue, 10 Jun 1997 10:25:02 -0500 From: Luke Tierney Message-Id: <9706101525.AA25407@nokomis.stat.umn.edu> Subject: Re: format To: isabelle.reginster@fundp.ac.be (Isabelle REGINSTER) Date: Tue, 10 Jun 1997 10:25:01 -0500 (CDT) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: <339D490A.6A123D77@fundp.ac.be> from "Isabelle REGINSTER" at Jun 10, 97 02:31:06 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1288 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk Isabelle REGINSTER wrote: > > Hello, > > Who can me explain the different options of this format function ? The > help give me only this : > > loading in help file information - this will take a minute ...done > FORMAT > [function-doc] > Args: (destination control &rest args) > Very basic implementation of Common Lisp format function. Only A, S, D, > F, E, G, %, and ~ directives are supported. D, % and ~ can take one > argument, R, E and G can take two. > > What is A, S, D, etc... ? > > Thank you in advance. > > Isabelle > Two possible references for the CL format function are Common Lisp, the Language, 2nd edition, Chapter 22:Output Functions:Formatted Output to Character Streams http://www.cs.cmu.edu/Groups/AI/html/cltl/cltl2.html and Common Lisp HyperSpec http://www.cs.cmu.edu/Groups/AI/html/cltl/cltl2.html Keep in mind that only a subset of format directives is supported. luke -- Luke Tierney University of Minnesota Phone: 612-625-7843 School of Statistics Fax: 612-624-8868 206 Church Street email: luke@stat.umn.edu Minneapolis, MN 55455 USA WWW: http://www.stat.umn.edu From majordom@umnstat.stat.umn.edu Tue Jun 10 14:29:13 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17849; Tue, 10 Jun 1997 14:29:13 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17845; Tue, 10 Jun 1997 14:29:03 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa26654; 10 Jun 97 15:28 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA19044; Tue, 10 Jun 1997 15:28:58 -0400 Date: Tue, 10 Jun 1997 15:28:57 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Ren Yu Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: save 'draw-text' In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Tue, 10 Jun 1997, Ren Yu wrote: > > Hi! > I am new to xlisp-stat. I have a simple question that how can I keep > the 'text' on the graph which generated by 'draw-text' or 'draw-string'? > Right now once I click the graph or other window, the printed text is > dispeared. > :draw-text, :draw-string, and any of the :draw- methods are immediate mode drawing. They do not add points, lines or text to the graph in the way that :add-lines, :add-points do. You need you :draw-text method to be called by the graph's :redraw method if you want it to be a persistant graphics object. ( Besides the redraw on uncovering the graph, you will see if you try to copy and paste the graphics window into another program, none of the non redrawn graphics gets copied -- copy does and implicit :redraw. ) [1] You can put the :draw-text into the graph's :redraw method. [2] You can use a more general solution, as in: ( and some other supporting files in that directory -- see ) which creates graph overlays for rectangles, lines, points, labels, etc. as well as composit graph objects ( a grouping of any of the primitives ) which are automatically called on the graph's :redraw. [3] If you want to do something similar, but different, the comments in the source code explain some of what's going on. Some of it is explained in the XlispStat book. Some of it has been changed slightly in 3.50. The is lots of discussion of this in the mailing list archive. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- By doing just a little every day, you can gradually let the task completely overwhelm you. From majordom@umnstat.stat.umn.edu Wed Jun 11 00:54:43 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22125; Wed, 11 Jun 1997 00:54:43 -0500 Return-Path: Received: from laplace.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22112; Wed, 11 Jun 1997 00:54:35 -0500 Received: (from jbond@localhost) by laplace.math.ucla.edu (8.8.3/8.8.3) id WAA25699; Tue, 10 Jun 1997 22:54:17 -0700 (PDT) From: Jason Bond Message-Id: <199706110554.WAA25699@laplace.math.ucla.edu> Subject: Re: save 'draw-text' To: ryu@odin.mdacc.tmc.edu (Ren Yu) Date: Tue, 10 Jun 1997 22:54:16 -0700 (PDT) Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: from "Ren Yu" at Jun 10, 97 09:25:55 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Try the following .lsp file. There are instructions at the top of the file. http://www.stat.ucla.edu/develop/lisp/xlisp/xlisp-stat/code/statistics /plotting/strays/labels.lsp It provides a graphical interface to putting point labels on plots (and keeping them there..through modifying the graph-protos redraw method... option #1 on Dr. Majewski's suggestion list). Labels can be placed, aligned, erased, or moved. Jason Bond UCLA Statistics > > > Hi! > I am new to xlisp-stat. I have a simple question that how can I keep > the 'text' on the graph which generated by 'draw-text' or 'draw-string'? > Right now once I click the graph or other window, the printed text is > dispeared. > > Thanks! > > Ren > > > > From majordom@umnstat.stat.umn.edu Wed Jun 11 10:55:17 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03719; Wed, 11 Jun 1997 10:55:17 -0500 Return-Path: Received: from anorien.mcs.surrey.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03715; Wed, 11 Jun 1997 10:55:08 -0500 Received: from godel.ee.surrey.ac.uk by anorien.mcs.surrey.ac.uk with smtp (Smail3.1.29.1-ident) id m0wbpjk-000SBqC; Wed, 11 Jun 97 16:55 BST Received: by godel.ee.surrey.ac.uk (Smail3.1.29.1 #8) id m0wbpji-0002neC; Wed, 11 Jun 97 16:55 BST Message-Id: From: map1rs@ee.surrey.ac.uk (Richard Stevens) Subject: Maximisation Problem To: stat-lisp-news@umnstat.stat.umn.edu Date: Wed, 11 Jun 1997 16:55:01 +0100 (BST) Reply-To: map1rs@surrey.ac.uk In-Reply-To: <318DC0E4.30F72698@fundp.ac.be> from "Isabelle REGINSTER" at May 6, 96 11:05:40 am X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 892 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Forgive me if this question is as dumb as I first thought, but I've had a lot more trouble than I expected. I'm trying to maximise a function lik using (newtonmax #'lik ... ) in the usual way; however, this is a function whose values are usually in the region 10^-320. Newtonmax invariably finds the gradient is less than gradient tolerance immediately - is this because the tolerance is a constant, and so a function with such small return values will always have a gradient less than the tolerance? Or does it measure the tolerance relative to the function value? I did indeed try the obvious, ie (defun newlik (x) (* 1e320 (lik x))) in order to maximise newlik instead. However, (newlik x) invariably evaluates to #.POSITIVE-INFINITY even for x values such that (lik x) is about 1e-318. I can't find a way to prevent the overflow. Thanks in advance for any help, Richard Stevens From majordom@umnstat.stat.umn.edu Wed Jun 11 11:09:48 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03978; Wed, 11 Jun 1997 11:09:48 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03973; Wed, 11 Jun 1997 11:09:40 -0500 Received: from [164.67.20.171] (ts21-14.wla.ts.ucla.edu) by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA03279; Wed, 11 Jun 97 09:08:02 PDT X-Sender: deleeuw@laplace.stat.ucla.edu Message-Id: In-Reply-To: References: <318DC0E4.30F72698@fundp.ac.be> from "Isabelle REGINSTER" at May 6, 96 11:05:40 am Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 11 Jun 1997 09:07:58 -0700 To: map1rs@surrey.ac.uk, stat-lisp-news@umnstat.stat.umn.edu From: Jan de Leeuw Subject: Re: Maximisation Problem Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk How about maximizing loglik ? At 4:55 PM +0100 6/11/97, Richard Stevens wrote: >Forgive me if this question is as dumb as I first thought, but I've had a lot >more trouble than I expected. > >I'm trying to maximise a function lik using (newtonmax #'lik ... ) in the >usual way; however, this is a function whose values are usually in the region >10^-320. Newtonmax invariably finds the gradient is less than gradient >tolerance immediately - is this because the tolerance is a constant, and so a >function with such small return values will always have a gradient less than >the tolerance? Or does it measure the tolerance relative to the function >value? > >I did indeed try the obvious, ie (defun newlik (x) (* 1e320 (lik x))) in order >to maximise newlik instead. However, (newlik x) invariably evaluates to >#.POSITIVE-INFINITY even for x values such that (lik x) is about 1e-318. I >can't find a way to prevent the overflow. > >Thanks in advance for any help, > >Richard Stevens --- Jan de Leeuw; Professor and Director, Interdivisional Program in Statistics; US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw ============================================================================ No matter where you go, there you are. --- Buckaroo Banzai ============================================================================ From majordom@umnstat.stat.umn.edu Wed Jun 11 17:14:51 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA08357; Wed, 11 Jun 1997 17:14:51 -0500 Return-Path: Received: from newton.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA08349; Wed, 11 Jun 1997 17:14:22 -0500 Received: from libiya.upf.es (libiya.upf.es [193.145.60.105]) by newton.upf.es (8.8.5/8.8.5) with SMTP id AAA18777 for ; Thu, 12 Jun 1997 00:13:15 +0100 (WET DST) Received: by libiya.upf.es (1.38.193.4/15.6) id AA28865; Thu, 12 Jun 1997 00:11:11 +0100 Date: Thu, 12 Jun 1997 00:11:11 +0100 (WETDST) From: Frederic Udina To: lisp stat list Subject: invisible cursor in windows 3.1x Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi Has anybody noticed that when installing a cross coursor in a white over black window in Xlisp-Stat running over Windoze 3.1 the cursor is not visible? Is it an known bug or is my own problem? I have no problems running the code on Mac, Win95 or Unix, but a user reports that problem using Win3.1 with XLS-3.50. I insist, the window has black background, with white background there is no problem at all. Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Ramon Trias Fargas, 25-27 | Compuserve: 100555.603 08005 Barcelona | WWWeb: http://libiya.upf.es/ SPAIN | or http://www.econ.upf.es/~udina ___________________________________/_\_________________________________ From majordom@umnstat.stat.umn.edu Thu Jun 12 05:27:13 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA20193; Thu, 12 Jun 1997 05:27:13 -0500 Return-Path: Received: from anorien.mcs.surrey.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA20189; Thu, 12 Jun 1997 05:26:22 -0500 Received: from godel.ee.surrey.ac.uk by anorien.mcs.surrey.ac.uk with smtp (Smail3.1.29.1-ident) id m0wc751-000S8RC; Thu, 12 Jun 97 11:26 BST Received: by godel.ee.surrey.ac.uk (Smail3.1.29.1 #8) id m0wc750-0002ngC; Thu, 12 Jun 97 11:26 BST Message-Id: From: map1rs@ee.surrey.ac.uk (Richard Stevens) Subject: Re: Maximisation Problem To: stat-lisp-news@umnstat.stat.umn.edu Date: Thu, 12 Jun 1997 11:26:10 +0100 (BST) In-Reply-To: <199706111710.AA06404@research.ets.org> from "Russell Almond" at Jun 11, 97 01:10:53 pm X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 268 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Thanks to Russell and Jan for suggesting I work with (log lik). At first this gave underflow, but now I have a routine that directly calculates log-lik (as opposed to (defun log-lik (x) (log (lik x)))) it seems to work. Cheers, Richard Stevens University of Surrey From majordom@umnstat.stat.umn.edu Thu Jun 12 16:27:12 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24778; Thu, 12 Jun 1997 16:27:12 -0500 Return-Path: Received: from odin.mdacc.tmc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24772; Thu, 12 Jun 1997 16:27:04 -0500 Received: by odin.mdacc.tmc.edu; (5.65/1.1.8.2/15Aug95-0158PM) id AA06625; Thu, 12 Jun 1997 16:27:01 -0500 Date: Thu, 12 Jun 1997 16:27:01 -0500 (CDT) From: Ren Yu To: Jason Bond Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: save 'draw-text' In-Reply-To: <199706110554.WAA25699@laplace.math.ucla.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Many thanks to everyone (Frederic Udina, Steven D. Majewski and Jason Bond) for the response to the question I posted. Your suggestions are very helpful. Thanks again! Ren Yu U T M. D. Anderson Cancer Center Houston, TX 77030 U. S. A. ryu@odin.mdacc.tmc.edu From majordom@nokomis.stat.umn.edu Fri Jun 13 10:16:49 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA06264; Fri, 13 Jun 1997 10:16:49 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA06231; Fri, 13 Jun 1997 10:15:19 -0500 From: Luke Tierney Message-Id: <9706131515.AA06231@nokomis.stat.umn.edu> Subject: Re: invisible cursor in windows 3.1x To: udina@upf.es (Frederic Udina) Date: Fri, 13 Jun 1997 10:15:19 -0500 (CDT) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Frederic Udina" at Jun 12, 97 00:11:11 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 942 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk Frederic Udina wrote: > > > Hi > > Has anybody noticed that when installing a cross coursor in a white > over black window in Xlisp-Stat running over Windoze 3.1 the cursor is > not visible? > Is it an known bug or is my own problem? I have no problems running the > code on Mac, Win95 or Unix, but a user reports that problem using > Win3.1 with XLS-3.50. > > I insist, the window has black background, with white background there is > no problem at all. > The cross is one of MS's standard system cursors, so any problems with it are Redmond's -- it looks like they arranged for it to use xor drawing on Win95. -- Luke Tierney University of Minnesota Phone: 612-625-7843 School of Statistics Fax: 612-624-8868 206 Church Street email: luke@stat.umn.edu Minneapolis, MN 55455 USA WWW: http://www.stat.umn.edu From majordom@umnstat.stat.umn.edu Fri Jun 13 11:50:02 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA08555; Fri, 13 Jun 1997 11:50:02 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA08551; Fri, 13 Jun 1997 11:49:55 -0500 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id JAA13989; Fri, 13 Jun 1997 09:49:54 -0700 (PDT) From: Jan Deleeuw Message-Id: <199706131649.JAA13989@galton.math.ucla.edu> Subject: Jason's Demos To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Fri, 13 Jun 1997 09:49:53 -0700 (PDT) Cc: jbond@laplace.math.ucla.edu (Jason Bond), jose@laplace.math.ucla.edu (Jose Garcia) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I have updated the directory of Xlisp-Stat demos on http://www.stat.ucla.edu/textbook/demos/ It now has foo.xli for the files that need to be downloaded and run and foo.lsp for the files that need to be downloaded and displayed in the browser (as text files). Our server is set up appropriately, you have to configure your client accordingly. -- Jan de Leeuw; Professor and Director, Interdivisional Program in Statistics; US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Fri Jun 13 15:50:33 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10538; Fri, 13 Jun 1997 15:50:33 -0500 Return-Path: Received: from odin.mdacc.tmc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10534; Fri, 13 Jun 1997 15:50:25 -0500 Received: by odin.mdacc.tmc.edu; (5.65/1.1.8.2/15Aug95-0158PM) id AA17979; Fri, 13 Jun 1997 15:50:33 -0500 Date: Fri, 13 Jun 1997 15:50:33 -0500 (CDT) From: Ren Yu To: stat-lisp-news@umnstat.stat.umn.edu Subject: paste text Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi! everyone, I need to link 2 graphs together and when select part of first plot (plot-points), I want to paste the result based on selection as text on the second graph (for example, I want to show the correlation between selected x,y, detail listed below). Is there a way to do this? Thank you very much! (def plot1 (plot-points x y :title "scatterplot 1" :variable-labels (list "x" "y"))) (setf w (send graph-proto :new 2)) (defun corr (x y) (/ (sum (* (- x (mean x)) (- y (mean y)))) (* (- (length x) 1) (standard-deviation x) (standard-deviation y)))) (def tmp (send plot1 :selection)) (def result (corr (select x tmp) (select y tmp))) I want to show result (for example, 0.512) on 'w' as text. Ren ~ From majordom@umnstat.stat.umn.edu Tue Jun 17 04:52:39 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03368; Tue, 17 Jun 1997 04:52:39 -0500 Return-Path: Received: from hermes.fundp.ac.be by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03364; Tue, 17 Jun 1997 04:52:29 -0500 Received: from sturm.math.fundp.ac.be by hermes.fundp.ac.be; (5.65v3.2/1.1.8.2/14Dec95-0443PM) id AA27680; Tue, 17 Jun 1997 11:53:17 +0200 Message-Id: <33A65E50.294B47BF@fundp.ac.be> Date: Tue, 17 Jun 1997 11:52:16 +0200 From: Isabelle REGINSTER Organization: FUNDP - MATH - GRT X-Mailer: Mozilla 3.01Gold (X11; I; Linux 2.0.18 i686) Mime-Version: 1.0 To: xlispstat Subject: invalid dotted pair ??? Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, I would like to read a file with 30 columns. This my command : (def mydata (read-data-columns "/usr/users/testbeta/data/yreisen.sas")) and when I press "return", there is an error message : Error: invalid dotted pair Happened in: # What is it ? Is there an invalid charater in my file ? I have already delete the ":". Thanks in advance for your answers and Have a nice day. Isabelle -- -------------------------------------- Isabelle REGINSTER Facultes Universitaires ND de la Paix Departement de Mathematique Groupe de Recherche sur les Transports (Pr. Ph.Toint) Rempart de la Vierge, 8 B.5000 Namur (Belgium) tel : 32-81-72.49.23 fax : 32-81-72.49.14 isabelle.reginster@fundp.ac.be -------------------------------------- From majordom@umnstat.stat.umn.edu Tue Jun 17 09:18:31 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04271; Tue, 17 Jun 1997 09:18:31 -0500 Return-Path: Received: from hermes.fundp.ac.be by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04267; Tue, 17 Jun 1997 09:18:20 -0500 Received: from sturm.math.fundp.ac.be by hermes.fundp.ac.be; (5.65v3.2/1.1.8.2/14Dec95-0443PM) id AA26254; Tue, 17 Jun 1997 16:19:17 +0200 Message-Id: <33A69CA7.232616BD@fundp.ac.be> Date: Tue, 17 Jun 1997 16:18:15 +0200 From: Isabelle REGINSTER Organization: FUNDP - MATH - GRT X-Mailer: Mozilla 3.01Gold (X11; I; Linux 2.0.18 i686) Mime-Version: 1.0 To: xlispstat Subject: variable with characters Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, I have another question. My variable content any 0, 1 and "." and I would like to count the number of 0, 1 and "." This is my script : (def var (list 1 0 0 1 "." 1)) (def nbr0 (length (which (= var 0)))) and the error message : Error: not a number - "." Happened in: # How can I count ? What must I put in place of the = ? Thanks in advance. Isabelle -- -------------------------------------- Isabelle REGINSTER Facultes Universitaires ND de la Paix Departement de Mathematique Groupe de Recherche sur les Transports (Pr. Ph.Toint) Rempart de la Vierge, 8 B.5000 Namur (Belgium) tel : 32-81-72.49.23 fax : 32-81-72.49.14 isabelle.reginster@fundp.ac.be -------------------------------------- From majordom@umnstat.stat.umn.edu Wed Jun 18 08:39:02 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19808; Wed, 18 Jun 1997 08:39:02 -0500 Return-Path: Received: from hermes.fundp.ac.be by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19800; Wed, 18 Jun 1997 08:38:37 -0500 Received: from sturm.math.fundp.ac.be by hermes.fundp.ac.be; (5.65v3.2/1.1.8.2/14Dec95-0443PM) id AA11653; Wed, 18 Jun 1997 15:39:20 +0200 Message-Id: <33A7E4C7.7F3A4BD0@fundp.ac.be> Date: Wed, 18 Jun 1997 15:38:15 +0200 From: Isabelle REGINSTER Organization: FUNDP - MATH - GRT X-Mailer: Mozilla 3.01Gold (X11; I; Linux 2.0.18 i686) Mime-Version: 1.0 To: xlispstat Subject: name of variable Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello everybody, I would like to define the name of a variable with the value of an other variable : (def var (list 1 2 3)) (dolist ( i var ) ( def newvar-i toto)) where newvar-i = newvar-1 the first time, newvar-2 the second time, ... I don't know how say to my script it's the value of i and no the character i. How make this ? Thanks in advance. Isabelle ------------------------------------- Isabelle REGINSTER Facultes Universitaires ND de la Paix Departement de Mathematique Groupe de Recherche sur les Transports (Pr. Ph.Toint) Rempart de la Vierge, 8 B.5000 Namur (Belgium) tel : 32-81-72.49.23 fax : 32-81-72.49.14 isabelle.reginster@fundp.ac.be -------------------------------------- From majordom@umnstat.stat.umn.edu Thu Jun 19 10:35:52 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06626; Thu, 19 Jun 1997 10:35:52 -0500 Return-Path: Received: from hickory.tds-eagan.lmco.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06622; Thu, 19 Jun 1997 10:35:42 -0500 Received: from eaggate1.tds-eagan.lmco.com by hickory (5.x/SMI-SVR4.109) id AB06490; Thu, 19 Jun 1997 10:37:03 -0500 Received: by eaggate1.tds-eagan.lmco.com with SMTP (Microsoft Exchange Server Internet Mail Connector Version 4.0.994.63) id <01BC7C9C.D74D21B0@eaggate1.tds-eagan.lmco.com>; Thu, 19 Jun 1997 10:37:52 -0500 Message-Id: From: "Artz, Ray E @EAG" To: Isabelle REGINSTER , 'stat-lisp-news' Subject: RE: name of variable Date: Thu, 19 Jun 1997 10:38:00 -0500 X-Mailer: Microsoft Exchange Server Internet Mail Connector Version 4.0.994.63 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Isabella wrote: __________________________ I would like to define the name of a variable with the value of an other variable : (def var (list 1 2 3)) (dolist ( i var ) ( def newvar-i toto)) where newvar-i = newvar-1 the first time, newvar-2 the second time, ... I don't know how say to my script it's the value of i and no the character i. How make this ? ___________________________________ _____________________________________ Try this: (def var (list 1 2 3)) (def values (list 6 8 10)) (dolist ( i var ) ( set (intern (string-upcase (format nil "newvar-~d" i))) (select values (- i 1)))) The expression (string-upcase (format nil "newvar-~d" i)) is used here to emphasize upper case, in the present instance the expression (format nil "NEWVAR-~d" i) should work just as well. Note the this will not cause the variables newvar-1,...,newvar-3 to appear on the lsip-stat *variables* list (accessible by typing *variables* or (variables) ). If you want that to occur, you can try the following code. (I am not completely comfortable with this direct access of *variables* , but I don't know a more elegant method to accomplish the objective -- the macro "def" won't work here because it doesn't evaluate its first argurment.) (def var (list 1 2 3)) (def values (list 6 8 10)) (dolist ( i var ) ( set (intern (string-upcase (format nil "newvar-~d" i))) (select values (- i 1))) ( pushnew (find-symbol (string-upcase (format nil "newvar-~d" i))) *variables*)) --Ray Ray E. Artz, Ph.D. Staff Engineer Precision Agriculture Segment Lockheed Martin P.O. Box 64525 MS U2F27 St. Paul, MN 55164-0525 TEL: (612) 456-2328 FAX: (612) 456-2362 Ray.E.Artz@lmco.com Street Address (Fed. Ex., etc.): MS U2F27 3333 Pilot Knob Road Eagan, MN 55121 Also (alter ego) ... Ph.D. Candidate School of Statistics University of Minnesota Ray.E.Artz-1@tc.umn.edu Home: 13140 Doyle's Court Apple Valley, MN 55124 (612) 432-0330 From majordom@umnstat.stat.umn.edu Sat Jun 21 15:23:39 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06949; Sat, 21 Jun 1997 15:23:39 -0500 Return-Path: Received: from wara.bolnet.bo by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06945; Sat, 21 Jun 1997 15:23:31 -0500 Received: from default ([166.114.10.94]) by wara.bolnet.bo (8.8.5/8.8.5) with SMTP id QAA01748 for ; Sat, 21 Jun 1997 16:27:10 +0400 Message-Id: <33AAE470.68CD@wara.bolnet.bo> Date: Fri, 20 Jun 1997 16:13:37 -0400 From: kjetil halvorsen X-Mailer: Mozilla 3.0Gold (Win95; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: loglikelihood function References: <199706110554.WAA25699@laplace.math.ucla.edu> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I am trying to write a log-likelihood function for a simple ordered regression problem, and sincerely cannot understand why the following doesn't work: (defun log-lik-lat (n) (let ((n1 (first n)) (n2 (rest n))) (flet ((ft (theta m &rest u) (let* ((res (mapcar #'normal-cdf u)) (res1 (combine 0.0 res 1.0)) (dif (difference res1))) (sum (break) (* m (log dif) )) ))) #'(lambda (theta &rest u) (+ (ft 0.0 n1 u) (ft theta n2 u)))))) (def n (list(list 73 45 28 29 25) (list 25 27 34 42 72)) ) (def p (log-lik-lat n)) (funcall p 1 -1.0 -0.5 0.5 1) I get the following error: Error: bad argument type - NIL Happened in: # Also: How can I identify the subroutine causing the problem from the above cryptic error? (I am using xlsp 3.50 on windows 95) Thank you, Kjetil Halvorsen. From majordom@umnstat.stat.umn.edu Mon Jun 23 12:33:02 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02553; Mon, 23 Jun 1997 12:33:02 -0500 Return-Path: Received: from sulu.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02545; Mon, 23 Jun 1997 12:32:52 -0500 Received: from zen by sulu.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id KAA00556; Mon, 23 Jun 1997 10:32:49 -0700 (PDT) Date: Mon, 23 Jun 1997 10:32:31 -0700 (PDT) From: Thomas Lumley X-Sender: thomas@zen To: stat-lisp-news@umnstat.stat.umn.edu Subject: survival analysis In-Reply-To: <33AAE470.68CD@wara.bolnet.bo> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Another revision of my survival analysis package is available from http://www.biostat.washington.edu/~thomas/cox.html Apart from various minor fixes the main new thing is the ability to calculate and plot the survival functions for arbitrary covariate patterns. There is a dynamic graphics demonstration of the effect of covariates on the fitted survival curve for a Cox model, with sliders to vary each of the covariates. The demonstration is also set up as a single file for use over the net. The data set I have provided for this is not great, as two of the covariates are discrete, but you can run it on any data of your choice. - Thomas Lumley Summary of package tcoxreg.lsp ------------------------------ This fits the stratified Cox regression model to data with right censoring and optional left truncation. Both the Efron and Breslow (Peto) adjustments for ties are available. In addition to time-dependence in step-function form which can be done with truncation/censoring there is a facility for arbitrary continuous time-dependence (which is very slow). Plots of the baseline survival and hazard functions and of the scaled Schoenfeld residuals are available, as are tests of the proportional hazards assumption based on (scaled) Schoenfeld residuals (Grambsch and Therneau, Biometrika, 1994) From majordom@umnstat.stat.umn.edu Tue Jun 24 12:30:11 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA18400; Tue, 24 Jun 1997 12:30:11 -0500 Return-Path: Received: from newton.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA18396; Tue, 24 Jun 1997 12:30:00 -0500 Received: from libiya.upf.es (libiya.upf.es [193.145.60.105]) by newton.upf.es (8.8.5/8.8.5) with SMTP id TAA26828 for ; Tue, 24 Jun 1997 19:28:57 +0100 (WET DST) Received: by libiya.upf.es (1.38.193.4/15.6) id AA04210; Tue, 24 Jun 1997 19:30:50 +0100 Date: Tue, 24 Jun 1997 19:30:50 +0100 (WETDST) From: Frederic Udina To: lisp stat list Subject: workspaces Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi there, I'm using (or trying to!) xls 3.50 on a Win95 machine. When I start a fresh xlisp-stat and issue (save-workspace "filename") I got a filename.wks file as expected. If then I drag the icon onto the lisp-stat icon, it doesn't start properly. If I rename the filename.wks to xlisp.wks, it works. Is there any other way to start xlisp-stat using this new workspace? Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Ramon Trias Fargas, 25-27 | Compuserve: 100555.603 08005 Barcelona | WWWeb: http://libiya.upf.es/ SPAIN | or http://www.econ.upf.es/~udina ___________________________________/_\_________________________________ From majordom@umnstat.stat.umn.edu Wed Jun 25 09:39:34 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03244; Wed, 25 Jun 1997 09:39:34 -0500 Return-Path: Received: from odin.mdacc.tmc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03240; Wed, 25 Jun 1997 09:39:26 -0500 Received: by odin.mdacc.tmc.edu; (5.65/1.1.8.2/15Aug95-0158PM) id AA15069; Wed, 25 Jun 1997 09:39:28 -0500 Date: Wed, 25 Jun 1997 09:39:28 -0500 (CDT) From: Ren Yu To: stat-lisp-news@umnstat.stat.umn.edu Subject: xlisp-s Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, Does anyone use xlisp-s to get S and xlispstat interface recently ? I downloaed the 'xlisp-s' file from STATLIB. When I followed the instruction of the files I down-loaded I got some error message at xlisp-stat prmopt: for example, 1. file does not exist - "userhome.fsl", 2. file does not exist - "/home/odin/ryu/tmp/S.to.CL" where "/home/odin/ryu' is my home directory. Does this mean I didn't get all files? How can I get those files ? Thank you for your help! Ren Yu UT MDACC Houston, TX 77036 ryu@odin.mdacc.tmc.edu From majordom@umnstat.stat.umn.edu Thu Jun 26 08:27:47 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17443; Thu, 26 Jun 1997 08:27:47 -0500 Return-Path: Received: from cagw1.att.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17439; Thu, 26 Jun 1997 08:27:42 -0500 Received: from toontown.colmd.att.com by caig1.att.att.com (SMI-8.6/EMS-1.2 sol2) id JAA15372; Thu, 26 Jun 1997 09:20:31 -0400 Received: from hobbes.colmd.att.com by toontown.colmd.att.com (SMI-8.6/SMI-SOL2.5.1/COLMD-MH1.0) id JAA18921; Thu, 26 Jun 1997 09:24:32 -0400 Received: by hobbes.colmd.att.com (SMI-8.6/SMI-SVR4/SOL2.5.1/COLMD-C1.0) id JAA25123; Thu, 26 Jun 1997 09:21:17 -0400 Date: Thu, 26 Jun 1997 09:21:17 -0400 From: jvl@colmd.att.com (Juniata V. Lees) Message-Id: <199706261321.JAA25123@hobbes.colmd.att.com> To: stat-lisp-news@umnstat.stat.umn.edu Subject: compiling for Solaris with gcc Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, I have searched the archive (the one I could find!) but can't find pointers on compiling for Solaris 2.5.1 with gcc. I am getting errors on memmove redefined (among others) and I can see that the configure program prints a line that says something like 'looking for memmove ... (cached) no'. If anyone can point me to a resource or tell me what I should do to compile I would appreciated the help. If there is an archive or FAQ page please send that info too. I have Solaris 2.5.1 for X86. When I ran the configure program I used the option 'with-gcc' and also set the X11 libs to point to /usr/openwin. Thank you, Juniata Lees j.v.lees@att.com From majordom@umnstat.stat.umn.edu Mon Jun 30 21:41:13 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04239; Mon, 30 Jun 1997 21:41:13 -0500 Return-Path: Received: from atlas.socsci.umn.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04235; Mon, 30 Jun 1997 21:41:07 -0500 Received: from kukoonest (tc1-3.globalnet.com.tw [202.39.153.119]) by atlas.socsci.umn.edu (8.8.5/8.8.5) with SMTP id VAA06735 for ; Mon, 30 Jun 1997 21:41:00 -0500 (CDT) Message-Id: <3368D891.7EC3@atlas.socsci.umn.edu> Date: Thu, 01 May 1997 10:53:21 -0700 From: Aging but still tough Harry Reply-To: harry@atlas.socsci.umn.edu Organization: Thinking GIS X-Mailer: Mozilla 3.0 (WinNT; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: sources for spatial statistics (or spatial analysis) references Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello netters, I am looking for refercences for spatial statistics (or spatial analysis.) Could someone show me where is the best starting point? Thanks! Best, harry -- :( harry@atlas.socsci.umn.edu SSRFC, UMN, MPLS,(612)625-8556 :| kchang@fruit.cc-sun.fcu.edu.tw Voice/FAX:(04)381-5514 :) Homepage http://www.globalnet.com.tw/~kchang <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> < I am back, tough as usual, and ready to go from here .. > < ~ ~ My heart is beating as hard as it could .. > < V V My mind is as clear as crystal .. > < ^ Thinking GIS @259-0680 .. > <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<>>>>>>>>>>>>>>>>>>>>>>>>>>>>> From majordom@umnstat.stat.umn.edu Tue Jul 1 07:59:41 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA14604; Tue, 1 Jul 1997 07:59:41 -0500 Return-Path: Received: from bsfiles.nerc-bas.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA14597; Tue, 1 Jul 1997 07:59:34 -0500 Received: from pcmail.nerc-bas.ac.uk by bsfiles.nerc-bas.ac.uk (8.8.3/NERC-1.4(Solaris 2.x, SMTP)) id NAA09330; Tue, 1 Jul 1997 13:59:27 +0100 (BST) Received: from Cambridge-Message_Server by pcmail.nerc-bas.ac.uk with Novell_GroupWise; Tue, 01 Jul 1997 14:01:43 +0000 Message-Id: X-Mailer: Novell GroupWise 4.1 Date: Tue, 01 Jul 1997 13:59:01 +0000 From: "Alistair W.A. Murray" To: stat-lisp-news@umnstat.stat.umn.edu Subject: Cluster analysis Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I have a largish data set (179 units) on which I need to do some hierarchical cluster analysis. I found the agnes (agglomerative nesting) function written by Jan de Leeuw and a mst (minimum spanning) function written by David Betz. Unfortunately neither of these are documented and I can't work out just what they do. The agnes function returns a very long list of numbers - but what are they? Is there any graphical display available? I haven't tried the mst function yet but it doesn't look like it does any plotting. Has anyone written code to produce graphics from this function? Anyone got a dendrogram function? What I really want to be able to do is to link the mst or dendrogram to various namelists containing information about the units and to an x,y plot of their location to see if the clusters have common properties and any spatio-temporal continuity. Any help would be much appreciated. Many thanks, Alistair Murray Biometrician - British Antarctic Survey. This email address may not be used for unsolicited email. I do not give permission for it to be stored in any data base for the purpose of sending mailshots nor do I grant permission for it to be passed to any third party for this purpose. Thank you. From majordom@umnstat.stat.umn.edu Tue Jul 1 11:32:23 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15861; Tue, 1 Jul 1997 11:32:23 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15857; Tue, 1 Jul 1997 11:32:17 -0500 Received: from [164.67.20.190] (ts23-1.wla.ts.ucla.edu) by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA17447; Tue, 1 Jul 97 09:32:02 PDT X-Sender: deleeuw@laplace.stat.ucla.edu Message-Id: In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Tue, 1 Jul 1997 09:31:58 -0700 To: "Alistair W.A. Murray" From: Jan de Leeuw Subject: Re: Cluster analysis Cc: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >I have a largish data set (179 units) on which I need to do some >hierarchical cluster analysis. I found the agnes (agglomerative nesting) >function written by Jan de Leeuw and a mst (minimum spanning) function >written by David Betz. Unfortunately neither of these are documented and I >can't work out just what they do. The agnes function returns a very long >list of numbers - but what are they? Is there any graphical display >available? Agnes does agnes -- agglomerative nesting. (defun agnes (diss &key (type "average") (ultrametric nil)) It prints two lists: the list of successive clusterings, and the list of cluster values. Input is a dissimilarity matrix, type can be ward, gower, weighted, single, average, and complete. There is no graphics. minimum-spanning-tree computes the minimum spanning tree, i.e. it gives a list of the edges of the tree given a dissimilarity matrix. --- Jan de Leeuw; Professor and Director, Interdivisional Program in Statistics; US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw ============================================================================ No matter where you go, there you are. --- Buckaroo Banzai ============================================================================ From majordom@umnstat.stat.umn.edu Tue Jul 1 12:28:11 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16198; Tue, 1 Jul 1997 12:28:11 -0500 Return-Path: Received: from wara.bolnet.bo by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16194; Tue, 1 Jul 1997 12:28:04 -0500 Received: from default (cl10.entelnet.bo [166.114.10.40]) by wara.bolnet.bo (8.8.5/8.8.5) with SMTP id NAA15434; Tue, 1 Jul 1997 13:31:56 +0400 Message-Id: <33B7E687.5730@wara.bolnet.bo> Date: Mon, 30 Jun 1997 13:02:00 -0400 From: kjetil halvorsen X-Mailer: Mozilla 3.0Gold (Win95; I) Mime-Version: 1.0 To: "Alistair W.A. Murray" Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Cluster analysis References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I have also tried to find out about agnes, my incomplete understanding up to now is: Agnes takes as input a distance matrix, so in the case of 179 units it will be 179*179 symmetric. Agnes does not compute the distance matrix from a variables*cases matrix, so this the user has to supply. The output virtually is the dendrogram in list form, so with 5 cases one might get something like: ( ((1) (2) (3) (4) (5))( (1 2) (3) (4) (5))( (1 2) (3) (4 5)) ( ( 1 2 3) (4 5) ) ( (1 2 3 4 5) ) ) (with some extra spaces added, and hopefully parenthesis correctly counted!) This would correspond to a dendrogram as: 1-\ >-\ The output is a level 3 list, first level a list of 2-/ \ succesive agglomerative solutions, each of this solutions >-----\ is a list of groupings, at start al solutions are singletons, / \ and at each new step in the agglomerative process, two 3----/ \ groups are united. >-------- / 4-\ / >---------/ 5-/ I would also be interested in code for producing graphics from this function. Hope this helps, Kjetil Halvorsen. Alistair W.A. Murray wrote: > > I have a largish data set (179 units) on which I need to do some > hierarchical cluster analysis. I found the agnes (agglomerative nesting) > function written by Jan de Leeuw and a mst (minimum spanning) function > written by David Betz. Unfortunately neither of these are documented and I > can't work out just what they do. The agnes function returns a very long > list of numbers - but what are they? Is there any graphical display > available? I haven't tried the mst function yet but it doesn't look like > it does any plotting. Has anyone written code to produce graphics from > this function? Anyone got a dendrogram function? What I really want to be > able to do is to link the mst or dendrogram to various namelists containing > information about the units and to an x,y plot of their location to see if > the clusters have common properties and any spatio-temporal continuity. > Any help would be much appreciated. > Many thanks, > > Alistair Murray > Biometrician - British Antarctic Survey. > > This email address may not be used for unsolicited email. > I do not give permission for it to be stored in any data > base for the purpose of sending mailshots nor do I grant > permission for it to be passed to any third party for > this purpose. Thank you. From majordom@umnstat.stat.umn.edu Sun Jul 6 16:09:00 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA07083; Sun, 6 Jul 1997 16:09:00 -0500 Return-Path: Received: from galton.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA07076; Sun, 6 Jul 1997 16:08:54 -0500 Received: (from deleeuw@localhost) by galton.math.ucla.edu (8.8.3/8.8.3) id OAA18109 for stat-lisp-news@umnstat.stat.umn.edu; Sun, 6 Jul 1997 14:08:51 -0700 (PDT) From: Jan Deleeuw Message-Id: <199707062108.OAA18109@galton.math.ucla.edu> Subject: WWW Interface To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Sun, 6 Jul 1997 14:08:51 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk At http://www.stat.ucla.edu/calculators/xlispstat there is a new Xlisp-Stat calculator that allows anybody to execute Xlisp-Stat expressions or programs on our server. It is written in PHP/FI, and it updates a CGI/Perl program by Jose Garcia. It is not possible to do graphics at the moment, and the system and popen functions are disabled, for obvious reasons. The page comes with an Xlisp-Stat QuickRef, and with the Xlisp-Plus manual. -- Jan de Leeuw; Professor and Director, Interdivisional Program in Statistics; US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw From majordom@umnstat.stat.umn.edu Mon Jul 7 08:57:13 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA18202; Mon, 7 Jul 1997 08:57:13 -0500 Return-Path: Received: from hickory.tds-eagan.lmco.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA18198; Mon, 7 Jul 1997 08:57:08 -0500 Received: from eaggate1.tds-eagan.lmco.com by hickory (5.x/SMI-SVR4.109) id AA26233; Mon, 7 Jul 1997 08:58:37 -0500 Received: by eaggate1.tds-eagan.lmco.com with SMTP (Microsoft Exchange Server Internet Mail Connector Version 4.0.994.63) id <01BC8AB4.1572C240@eaggate1.tds-eagan.lmco.com>; Mon, 7 Jul 1997 08:59:31 -0500 Message-Id: From: "Artz, Ray E @EAG" To: 'stat-lisp-news' Subject: Re: sources for spatial statistics Date: Mon, 7 Jul 1997 09:01:00 -0500 X-Mailer: Microsoft Exchange Server Internet Mail Connector Version 4.0.994.63 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Harry wrote: Hello netters, I am looking for refercences for spatial statistics (or spatial analysis.) Could someone show me where is the best starting point? Thanks! ____________________________________ Hello Harry: The following sites may be of interest: http://java.ei.jrc.it/rem/gregoire -- Spatial Statistics site (A very good site when I looked at it several months ago, but my be offline: I couldn't get through when I tried this morning.) http://www.stat.ucla.edu , especially sections on Xlisp (with some spatial statistics functions), and the online statistical computing journal (see vol 01 for Splus code for variogram diagnostics) Also, you can find some spatial stat and GIS (Geographical Information System) links at Lance Waller's home page: http://muskie.biostat.umn.edu/~lance/ Ray Artz Ray E. Artz, Ph.D. Staff Engineer Precision Agriculture Segment Lockheed Martin P.O. Box 64525 MS U2F27 St. Paul, MN 55164-0525 TEL: (612) 456-2328 FAX: (612) 456-2362 Ray.E.Artz@lmco.com Street Address (Fed. Ex., etc.): MS U2F27 3333 Pilot Knob Road Eagan, MN 55121 From majordom@umnstat.stat.umn.edu Mon Jul 7 10:11:04 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA18604; Mon, 7 Jul 1997 10:11:04 -0500 Return-Path: Received: from rgborafsa.er.usgs.gov by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA18592; Mon, 7 Jul 1997 10:10:58 -0500 Received: from [130.11.54.131] (mcmac.er.usgs.gov [130.11.54.131]) by rgborafsa.er.usgs.gov (Geomail 1.2.3) with SMTP id LAA09274 for ; Mon, 7 Jul 1997 11:11:17 -0400 Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 7 Jul 1997 11:11:42 -0400 To: stat-lisp-news@umnstat.stat.umn.edu From: "Richard B McCammon Reston, VA " (Richard McCammon (USGS)) Subject: Re: sources for spatial statistics Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk You may be interested to look on the Internet at the special issue "Exploratory Cartographic Visualization", guest edited by Alan MacEachren and Menno-Jan Kraak, that appeared as C&G vol 23, number 4. The same material is on a CD-ROM in a back pocket of the printed version. The web version is free. The URL is http://www.elsevier.nl/inca/homepage/sad/cageo/home.htm __________________________________________________________ snail mail: Richard B. McCammon Eastern Minerals Team U.S. Geological Survey MS-954 12201 Sunrise Valley Drive Reston, VA 20192 USA e-mail: mccammon@rgborafsa.er.usgs.gov or mccammon@130.11.54.100 Phone: 703-648-6150 Fax: 703-648-6383 From majordom@umnstat.stat.umn.edu Tue Jul 8 13:27:24 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03751; Tue, 8 Jul 1997 13:27:24 -0500 Return-Path: Received: from pele.pulmcc.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03747; Tue, 8 Jul 1997 13:27:14 -0500 Received: by pele.pulmcc.washington.edu (5.65/UW-NDC Revision: 2.22 ) id AA16555; Tue, 8 Jul 97 11:41:56 -0700 From: Steve McKinney Message-Id: <9707081841.AA16555@pele.pulmcc.washington.edu> Subject: Re: Cluster analysis To: kjetil@wara.bolnet.bo (kjetil halvorsen) Date: Tue, 8 Jul 97 11:41:56 PDT Cc: awm@pcmail.nbs.ac.uk, stat-lisp-news@umnstat.stat.umn.edu, kilroy@pele.pulmcc.washington.edu (kilroy) In-Reply-To: <33B7E687.5730@wara.bolnet.bo>; from "kjetil halvorsen" at Jun 30, 97 1:02 pm X-Mailer: ELM [version 2.3 PL11] Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I developed and continue to use a suite of lispstat functions which I call Autopaint, that allow me to examine multiple low-dimensional plots of a high dimensional data set and link patterns across graphs using color. Much of my work involves cluster analysis. Using Autopaint, I can put up a dendrogram, paint portions of the dendrogram and Autopaint maps those colors to the corresponding data objects' data point representations in other graphs such as a scatterplot matrix, grand tour, 3d spin etc. I use Autopaint/LispStat in conjunction with Splus, doing much of my number crunching in Splus (so for example I use the Splus hclust routine for clustering). I adapted a set of common-lisp to S routines written by Wayne Olford at U of Waterloo, that allows the lisp-stat user to invoke a background Splus session, pass off data to that process, then get results back. Sometimes it's a bit clunky, but it beats rewriting all statistical software into yet another computer language. I incorporated the object-oriented data representation discussed in Tierney's "LispStat" book, Chapter 10 Section 6 'An Alternative Linking Strategy", which works very well and allows the user to select subsets and color them and have the selections and colors correctly updated in all plots displaying any amount of the total data set. All of this still operates under old faithful and reliable LispStat version 2.1 release 2. I do not yet have a manual for the suite of programs, but I would be happy to share the code which others could adapt or just use as examples when writing their own code. Let me know if you'd like a copy. \`|'/ (. .) ---uuu-U-uuu--- kilroy was here From majordom@umnstat.stat.umn.edu Wed Jul 9 05:11:47 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15104; Wed, 9 Jul 1997 05:11:47 -0500 Return-Path: Received: from yurim.skku.ac.kr by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15100; Wed, 9 Jul 1997 05:11:38 -0500 Received: from mhuh.skku.ac.kr (ppp235-seoul.interpia.net [203.236.151.110]) by yurim.skku.ac.kr (8.8.5/8.8.5) with SMTP id TAA04778 for ; Wed, 9 Jul 1997 19:12:06 +1000 (KDT) Message-Id: <33C363E5.7305@yurim.skku.ac.kr> Date: Wed, 09 Jul 1997 19:11:49 +0900 From: Moon Yul Huh Reply-To: myhuh@yurim.skku.ac.kr Organization: Sung Kyun Kwan University X-Mailer: Mozilla 3.0Gold (Win95; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: Calling Visual C++ from XLISP-STAT Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Dear Colleagues, Anybody was successful calling Visual C++ from within XLS? Will appreciate your suggestion. -- Moon Yul Huh Department of Statistics, Sung Kyun Kwan University Myungruyn-Dong, Chongro-Ku, Seoul, Korea Tel: 82 2 7600463 FAX: 82 2 7640622 From majordom@umnstat.stat.umn.edu Wed Jul 9 05:19:37 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15145; Wed, 9 Jul 1997 05:19:37 -0500 Return-Path: Received: from yurim.skku.ac.kr by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15141; Wed, 9 Jul 1997 05:19:28 -0500 Received: from mhuh.skku.ac.kr (ppp235-seoul.interpia.net [203.236.151.110]) by yurim.skku.ac.kr (8.8.5/8.8.5) with SMTP id TAA04855 for ; Wed, 9 Jul 1997 19:19:57 +1000 (KDT) Message-Id: <33C365B6.4B59@yurim.skku.ac.kr> Date: Wed, 09 Jul 1997 19:19:34 +0900 From: Moon Yul Huh Reply-To: myhuh@yurim.skku.ac.kr Organization: Sung Kyun Kwan University X-Mailer: Mozilla 3.0Gold (Win95; I) Mime-Version: 1.0 To: XLS user Subject: Calling Visual C++ from XLISP-STAT Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Dear Colleagues Anyone was successful calling functions written in Visual C++ from within XKISP-STAT? Moon Yul Huh Department of Statistics, Sung Kyun Kwan University Myungruyn-Dong, Chongro-Ku, Seoul, Korea Tel: 82 2 7600463 FAX: 82 2 7640622 From majordom@umnstat.stat.umn.edu Wed Jul 9 05:20:18 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15155; Wed, 9 Jul 1997 05:20:18 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15151; Wed, 9 Jul 1997 05:20:14 -0500 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.7.5/10.1) id GAA01364; Wed, 9 Jul 1997 06:20:12 -0400 (GMT+4:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 9 Jul 1997 06:20:38 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: Forrest Young Subject: Mac Text windows Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi Lisp-Staters: With the Mac version of Lisp-Stat you can send text to a text window. This can be done by: (setf w (send display-window-proto :new )) (send w :paste-string string) However, if STRING has more text than will show in the window, then what you see is the LAST portion of the text rather than the FIRST. Does anyone know how to change this so the first portion is viewed? ----- Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: mailto:forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Voice: 1-919-962-5038 ViSta Software and Documentation site: http://forrest.psych.unc.edu/research/ USA Home Site ViSta Mirrors: http://ssda.anu.edu.au/mirrors Australia http://www.sci.usq.edu.au/general/vista Australia http://sunsite.univie.ac.at/ViSta Austria http://cons-dev.univ-lyon1.fr/vista France http://micros.hensa.ac.uk/mirrors/vista UK (ac.uk only 8AM-8PM GMT) http://www.psych.unc.edu/fwy USA ___ (@ @) ------------------------------------------------ooO--(_)--Ooo--------- ||| From majordom@umnstat.stat.umn.edu Sun Jul 13 05:55:34 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11348; Sun, 13 Jul 1997 05:55:34 -0500 Return-Path: Received: from lersews1.okayama-u.ac.jp by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11344; Sun, 13 Jul 1997 05:55:10 -0500 Received: from [150.46.86.54] ([150.46.86.54]) by lersews1.okayama-u.ac.jp (8.7.5+2.6Wbeta6/3.3W9) with SMTP id UAA14688 for ; Sun, 13 Jul 1997 20:56:54 +0900 (JST) Date: Sun, 13 Jul 1997 20:56:54 +0900 (JST) Message-Id: <199707131156.UAA14688@lersews1.okayama-u.ac.jp> To: stat-lisp-news@umnstat.stat.umn.edu From: le06621@lersews1.okayama-u.ac.jp (Yoshinori Shibasaki) Subject: code of Lag sequential Mime-Version: 1.0 Content-Type: text/plain; charset=iso-2022-jp X-Mailer: Eudora-J(1.3.8.5-J13) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hajimemashite I study how preshool children play in free play. I want to use Lag sequential method to analyze it. I had written the following program in xlisp-stat (Win-16; version 2.1R3 Alpha 6). But I give the following message from interpreter: error: index out of range - 11 (when lag is 2) I think that this program can check if index is out of range in Line 12. Variable 'len' in function 'conditional-probability' is set in Line 6: If 'data' contains 10 element and 'lag' is 2, 'len' will be 7. Then when 'cnt' is 8, line 12 '(if (< cnt len)' will be nil and interpreter will not evaluate line 13, 14. But interpreter does that. I have no idea. This program is my first program in xLisp-stat. I lose my way. Please help me. Sayonara... -------- 1: (defun conditional-probability (data focal contigent lag) 2: (let 3: ((frequency-of-focal-element 0) (frequency-of-contigent-element 0) 4: (len 0) (cnt 0)) 5: 6: (setf len (- (length data) 1 lag)) 7: (print "Starting Conditional...") 8: 9: 10: (dolist (x data) 11: 12: (if (< cnt len) 13: (if (and (= x contigent) 14: (= (select data (+ cnt lag)) focal)) 15: 16: (setf frequency-of-focal-element 17: (+ frequency-of-focal-element 1)))) 18: 19: (setf cnt (+ cnt 1))) 20: (print "Exit Loop") (setf frequency-of-contigent-element (count contigent data)) (if (or (= frequency-of-focal-element 0) (= frequency-of-contigent-element 0)) nil (/ frequency-of-focal-element frequency-of-contigent-element)))) (defun correct-z-statistics (focal contigent) (let ((lag 0) (z-statistics 0) (conditional-probability-of-focal-element 0) (unconditional-probability-of-focal-element 0) (unconditional-probability-of-contigent-element 0) (frequency-of-all-element 0)) (setf lag (get-value-dialog "lag:")) ;(setf data (read-data-columns "a:/data.txt")) (setf data (list 0 1 1 1 0 1 0 1 0 1)) (setf unconditional-probability-of-focal-element (count focal data)) (setf unconditional-probability-of-contigent-element (count contigent data)) (setf frequency-of-all-element (length data)) (setf conditional-probability-of-focal-element (conditional-probability data focal contigent lag)) (print "Computing z statistics") (setf z-statistics (/ (- conditional-probability-of-focal-element unconditional-probability-of-focal-element) (sqrt (/ (* unconditional-probability-of-focal-element (- 1.0 unconditional-probability-of-focal-element) (- 1.0 unconditional-probability-of-contigent-element)) (* (- frequency-of-all-element lag) unconditional-probability-of-contigent-element))))) (format t "Correct-z-statistics: ~,3g" z-statistics))) (correct-z-statistics 1 0) ------------- Yoshinori Shibasaki Email: le06621@lersews1.okayama-u.ac.jp URL: http://www.okayama-u.ac.jp/user/le/psycho/member/94shiba/shiba.html From majordom@umnstat.stat.umn.edu Tue Jul 15 03:18:13 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA08552; Tue, 15 Jul 1997 03:18:13 -0500 Return-Path: Received: from yurim.skku.ac.kr by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA08548; Tue, 15 Jul 1997 03:16:51 -0500 Received: from mhuh.skku.ac.kr ([203.252.32.152]) by yurim.skku.ac.kr (8.8.5/8.8.5) with SMTP id RAA04455; Tue, 15 Jul 1997 17:17:08 +1000 (KDT) Message-Id: <33CB31F3.F39@yurim.skku.ac.kr> Date: Tue, 15 Jul 1997 17:16:52 +0900 From: Moon Yul Huh Reply-To: myhuh@yurim.skku.ac.kr Organization: Sung Kyun Kwan University X-Mailer: Mozilla 3.0Gold (Win95; I) Mime-Version: 1.0 To: myhuh@yurim.skku.ac.kr Cc: XLS user Subject: Re: Cluster analysis References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Alistair W.A. Murray wrote: > > I have a largish data set (179 units) on which I need to do some > hierarchical cluster analysis. I found the agnes (agglomerative nesting) > function written by Jan de Leeuw and a mst (minimum spanning) function > written by David Betz. Unfortunately neither of these are documented and I > can't work out just what they do. The agnes function returns a very long > list of numbers - but what are they? Is there any graphical display > available? I haven't tried the mst function yet but it doesn't look like > it does any plotting. Has anyone written code to produce graphics from > this function? Anyone got a dendrogram function? What I really want to be > able to do is to link the mst or dendrogram to various namelists containing > information about the units and to an x,y plot of their location to see if > the clusters have common properties and any spatio-temporal continuity. > Any help would be much appreciated. > Many thanks, > > Alistair Murray > Biometrician - British Antarctic Survey. > > This email address may not be used for unsolicited email. > I do not give permission for it to be stored in any data > base for the purpose of sending mailshots nor do I grant > permission for it to be passed to any third party for > this purpose. Thank you. We developed menu-oriented XLISP-STAT programs for drawing dendrogram cluster analysis. You can download the programs by connecting the following addres: http://myhuh.skku.ac.kr/software/software.html Hope you enjoy the program. Moon Yul Huh Department of Statistics, Sung Kyun Kwan University Myungruyn-Dong, Chongro-Ku, Seoul, Korea Tel: 82 2 7600463 FAX: 82 2 7640622 From majordom@umnstat.stat.umn.edu Tue Jul 15 03:26:40 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA08587; Tue, 15 Jul 1997 03:26:40 -0500 Return-Path: Received: from yurim.skku.ac.kr by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA08583; Tue, 15 Jul 1997 03:26:07 -0500 Received: from mhuh.skku.ac.kr ([203.252.32.152]) by yurim.skku.ac.kr (8.8.5/8.8.5) with SMTP id RAA04646; Tue, 15 Jul 1997 17:21:34 +1000 (KDT) Message-Id: <33CB32FE.71E2@yurim.skku.ac.kr> Date: Tue, 15 Jul 1997 17:21:18 +0900 From: Moon Yul Huh Reply-To: myhuh@yurim.skku.ac.kr Organization: Sung Kyun Kwan University X-Mailer: Mozilla 3.0Gold (Win95; I) Mime-Version: 1.0 To: "Alistair W.A. Murray" Cc: XLS user , Jan de Leeuw Subject: Drawing Dendrogram for Cluster analysis wriiten in XLISP-STAT References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Alistair W.A. Murray wrote: > > I have a largish data set (179 units) on which I need to do some > hierarchical cluster analysis. I found the agnes (agglomerative nesting) > function written by Jan de Leeuw and a mst (minimum spanning) function > written by David Betz. Unfortunately neither of these are documented and I > can't work out just what they do. The agnes function returns a very long > list of numbers - but what are they? Is there any graphical display > available? I haven't tried the mst function yet but it doesn't look like > it does any plotting. Has anyone written code to produce graphics from > this function? Anyone got a dendrogram function? What I really want to be > able to do is to link the mst or dendrogram to various namelists containing > information about the units and to an x,y plot of their location to see if > the clusters have common properties and any spatio-temporal continuity. > Any help would be much appreciated. > Many thanks, > > Alistair Murray > Biometrician - British Antarctic Survey. > > This email address may not be used for unsolicited email. > I do not give permission for it to be stored in any data > base for the purpose of sending mailshots nor do I grant > permission for it to be passed to any third party for > this purpose. Thank you. We developed menu-oriented XLISP-STAT programs for drawing dendrogram cluster analysis. You can download the programs by connecting the following addres: http://myhuh.skku.ac.kr/software/software.html The program may run very slow with large number of observations. This is intrinsic problem for XLISP-STAT. Hope you enjoy the program. _____ Moon Yul Huh Department of Statistics, Sung Kyun Kwan University Myungruyn-Dong, Chongro-Ku, Seoul, Korea Tel: 82 2 7600463 FAX: 82 2 7640622 From majordom@umnstat.stat.umn.edu Tue Jul 15 09:44:47 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10062; Tue, 15 Jul 1997 09:44:47 -0500 Return-Path: Received: from anorien.mcs.surrey.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10058; Tue, 15 Jul 1997 09:44:41 -0500 Received: from godel.ee.surrey.ac.uk by anorien.mcs.surrey.ac.uk with smtp (Smail3.1.29.1-ident) id m0wo8qD-000SBrC; Tue, 15 Jul 97 15:44 BST Received: by godel.ee.surrey.ac.uk (Smail3.1.29.1 #8) id m0wo8qC-0002nkC; Tue, 15 Jul 97 15:44 BST Message-Id: From: map1rs@ee.surrey.ac.uk (Richard Stevens) Subject: Labelling plots To: stat-lisp-news@umnstat.stat.umn.edu Date: Tue, 15 Jul 1997 15:44:35 +0100 (BST) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 586 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello world. I'd like to plot a graph - using plot-function or perhaps plot-lines - with labelled axes. I'm looking at page 123 of the book by Luke Tierney (Section 4.1, "Input/Output", at the end of subsection 4.1.3) and it tells me "the :labels keyword argument to plot-function takes a list of strings to use as the axis labels" However, I'm executing (plot-function #'p6 0 1 :labels (list 'x 'y) ) and I get a plot with no axis-labels. What am I doing wrong? Please help - I'd like to include these graphs in my thesis! AdThanksvance, Richard Stevens University of Surrey From majordom@umnstat.stat.umn.edu Tue Jul 15 10:56:29 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10424; Tue, 15 Jul 1997 10:56:29 -0500 Return-Path: Received: from anorien.mcs.surrey.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10414; Tue, 15 Jul 1997 10:56:23 -0500 Received: from aragorn.ee.surrey.ac.uk by anorien.mcs.surrey.ac.uk with smtp (Smail3.1.29.1-ident) id m0wo9xe-000SAhC; Tue, 15 Jul 97 16:56 BST Received: by aragorn.ee.surrey.ac.uk (Smail3.1.29.1 #8) id m0wo9xd-0003zqC; Tue, 15 Jul 97 16:56 BST Message-Id: From: map1rs@ee.surrey.ac.uk (Richard Stevens) Subject: Re: Labelling plots To: stat-lisp-news@umnstat.stat.umn.edu Date: Tue, 15 Jul 1997 16:56:21 +0100 (BST) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 51 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Thanks folks - problem now fixed. Richard Stevens From majordom@umnstat.stat.umn.edu Wed Jul 16 14:37:12 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25012; Wed, 16 Jul 1997 14:37:12 -0500 Return-Path: Received: from salmon.maths.tcd.ie by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25008; Wed, 16 Jul 1997 14:37:05 -0500 Received: from orange.stats.tcd.ie by salmon.maths.tcd.ie with SMTP id aa17442; 16 Jul 97 20:37 +0100 Received: from [194.125.134.161] ([194.125.148.33]) by stats.tcd.ie (4.1/SMI-4.1) id AA01580; Wed, 16 Jul 97 20:22:53 BST X-Sender: dbearly@orange.stats.tcd.ie Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 16 Jul 1997 20:36:31 +0100 To: stat-lisp-news@umnstat.stat.umn.edu From: "Dom. Early" Subject: Histogram bin width calculation question Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, I've been playing around with histograms in XLISPSTAT recently and have come up with the following question. Suppose I have a set x (def x '(1 2 2 3 3 3 4 4 4 4 5 5 5 5 5)) I create a histogram of x (def hx (histogram x)) First I call scaled-range to see what the default range of the histogram is (send hx :scaled-range 0) and I get (1.0 5.0) i.e. the bins of hx range from 1.0 to 5.0 inclusive so as to capture all the data. Now I set the number of bins in hx to be 5 (send hx :num-bins 5) I can clearly see the staircase efect of the histogram However if I set the number of bins in hx to be 4 (send hx :num-bins 4) I get a rather weird effect. The bin ranges seem to be as follows 1 <= r1 <= 2 2 < r2 < 3 3 <= r3 <= 4 4 < r3 <= 5 Here it seems that bin widths are not fixed. What I need to know are the ranges of each bin. Can anyone give me some clues as to how to calculate this? I was trying something like: (def bwidth (/ (- (cadr (send hx :scaled-range 0)) (car (send hx :scaled-range 0))) (send hx :num-bins))) (def minhx (car (send hx :scaled-range 0))) (def ranges '()) (dotimes(i (send hx :num-bins) ranges) (setf ranges (append ranges (list (list (+ minhx (* i bwidth)) (+ minhx (* (+ i 1) bwidth))))))) which yields ((1.0 2.0) (2.0 3.0) (3.0 4.0) (4.0 5.0)) in other words 1.0 <= r1 < 2.0 2.0 <= r2 < 3.0 3.0 <= r3 < 4.0 4.0 <= r4 < 5.0 This is certainly flawed in that the upper limit of the last bin must be <= 5.0, it also assumes all bins are the same width, which is clearly not the case. Can anyone give me an idea of how I can accurately predict the lower and upper limits of a histogram in xlisp-stat? Thanks for any help, Dom. +--------------------------------------------------------------------------+ | Dominic Early e-mail: Dom.Early@tcd.ie | | A.31, Department of Statistics, Phone: +353-1-608 1917 | | Trinity College, Fax: +353-1-661 5046 | | Dublin 2, Ireland. | | | | http://www2.tcd.ie/Statistics/staff/dbearly.html | +--------------------------------------------------------------------------+ _/_/_/_/ _/_/_/ _/ _/ _/_/_/_/_/ _/_/_/ _/_/_/_/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/_/_/_/ _/_/_/ _/ _/ _/ _/ _/_/_/ _/_/_/_/ From majordom@umnstat.stat.umn.edu Fri Jul 18 04:57:32 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22385; Fri, 18 Jul 1997 04:57:32 -0500 Return-Path: Received: from yurim.skku.ac.kr by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22381; Fri, 18 Jul 1997 04:57:25 -0500 Received: from mhuh.skku.ac.kr ([203.252.32.152]) by yurim.skku.ac.kr (8.8.5/8.8.5) with SMTP id SAA25730; Fri, 18 Jul 1997 18:57:15 +1000 (KDT) Message-Id: <33CF3DEA.4F01@yurim.skku.ac.kr> Date: Fri, 18 Jul 1997 18:56:58 +0900 From: Moon Yul Huh Reply-To: myhuh@yurim.skku.ac.kr Organization: Sung Kyun Kwan University X-Mailer: Mozilla 3.0Gold (Win95; I) Mime-Version: 1.0 To: Jan de Leeuw Cc: XLS user Subject: Programs for Data Exploration Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Dear Colleagues, I have several programs for data exploration. The programs are wriiten in XLS,and utilize many properties of dynamic graphics programming techniques. I will be much obliged to have any comments. You can download the programs by connecting to the following address: http://myhuh.skku.ac.kr/software/dynamic.html Hope you enjoy the program. -- Moon Yul Huh Department of Statistics, Sung Kyun Kwan University Myungruyn-Dong, Chongro-Ku, Seoul, Korea Tel: 82 2 7600463 FAX: 82 2 7640622 From majordom@umnstat.stat.umn.edu Fri Jul 18 08:51:22 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22969; Fri, 18 Jul 1997 08:51:22 -0500 Return-Path: Received: from newton.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22965; Fri, 18 Jul 1997 08:51:14 -0500 Received: from libiya.upf.es (libiya.upf.es [193.145.60.105]) by newton.upf.es (8.8.5/8.8.5) with SMTP id PAA13581; Fri, 18 Jul 1997 15:50:08 +0100 (WET DST) Received: by libiya.upf.es (1.38.193.4/15.6) id AA11783; Fri, 18 Jul 1997 15:51:47 +0100 Date: Fri, 18 Jul 1997 15:51:47 +0100 (WETDST") From: Frederic Udina To: "Dom. Early" Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Histogram bin width calculation question In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Well, the xlisp-stat histogram function simply do not work well. Just try playing around with the window width (I mean the window that display your histogram) and you will see nice things. There are several replacement histogram packages. I actually have one in ftp://ftp.upf.es/pub/stat/anchor-position/ that was specialized for the work on the paper you will find there. On Wed, 16 Jul 1997, Dom. Early wrote: > Hi, > I've been playing around with histograms in XLISPSTAT recently and > have come up with the following question. > >......... > > Can anyone give me an idea of how I can accurately predict the lower and > upper limits of a histogram in xlisp-stat? > Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Ramon Trias Fargas, 25-27 | WWWeb: http://libiya.upf.es/ 08005 Barcelona SPAIN | or http://www.econ.upf.es/~udina ____________________________________-__________________________________ >> Please >> DON'T include this e-mail address in any commercial mailing list << From majordom@umnstat.stat.umn.edu Fri Jul 18 09:25:41 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23165; Fri, 18 Jul 1997 09:25:41 -0500 Return-Path: Received: from seralph6.essex.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23161; Fri, 18 Jul 1997 09:25:35 -0500 Received: from ircsun10.essex.ac.uk by seralph6.essex.ac.uk; (5.65v3.2/1.1.8.2/14Feb95-1154AM) id AA14625; Fri, 18 Jul 1997 15:25:33 +0100 Date: Fri, 18 Jul 97 15:25:32 BST Message-Id: <10488.9707181425@ircsun10.essex.ac.uk> From: Brendan Halpin To: stat-lisp-news%umnstat.stat.umn.edu@seralph6.essex.ac.uk Subject: l i s t s e r v Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk set digest -- Brendan Halpin |E-mail: brendan@essex.ac.uk ESRC Research Centre on Micro-Social Change |PGP key on servers, finger University of Essex, Colchester CO4 3SQ, UK |Work: +44-1206-873790 http://www.irc.essex.ac.uk/~brendan |Home: +44-1206-825250 From majordom@umnstat.stat.umn.edu Thu Jul 24 12:51:21 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16231; Thu, 24 Jul 1997 12:51:21 -0500 Return-Path: Received: from newton.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16227; Thu, 24 Jul 1997 12:51:13 -0500 Received: from libiya.upf.es (libiya.upf.es [193.145.60.105]) by newton.upf.es (8.8.5/8.8.5) with SMTP id TAA18125 for ; Thu, 24 Jul 1997 19:50:05 +0100 (WET DST) Received: by libiya.upf.es (1.38.193.4/15.6) id AA14576; Thu, 24 Jul 1997 19:51:36 +0100 Date: Thu, 24 Jul 1997 19:51:36 +0100 (WETDST") From: Frederic Udina To: lisp stat list Subject: compiling in SunOS Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi folks, trying to compile 3-50 I got: ...everything OK till here... cc -DUNIX -DNEWGC -DBYTECODE -DGNUPLOT -O -DNOMEMMOVE -DNODIFFTIME -c postscript.c cc -DUNIX -DNEWGC -DBYTECODE -DGNUPLOT -O -DNOMEMMOVE -DNODIFFTIME -c unixstuff.c "unixstuff.c", line 619: identifier redeclared: memmove current : function() returning void previous: function() returning pointer to void : "/usr/include/string.h", line 105 *** Error code 2 make: Fatal error: Command failed for target `unixstuff.o' $ uname -a SunOS gauss 5.5.1 Generic sun4u sparc SUNW,Ultra-1 No time to try out things, any suggestion will be really appreciated. Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Ramon Trias Fargas, 25-27 | WWWeb: http://libiya.upf.es/ 08005 Barcelona SPAIN | or http://www.econ.upf.es/~udina ____________________________________-__________________________________ >> Please >> DON'T include this e-mail address in any commercial mailing list << From majordom@umnstat.stat.umn.edu Tue Aug 5 08:59:27 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04432; Tue, 5 Aug 1997 08:59:27 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04428; Tue, 5 Aug 1997 08:59:21 -0500 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.7.5/10.1) id JAA23477; Tue, 5 Aug 1997 09:59:16 -0400 (GMT+4:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Date: Tue, 5 Aug 1997 10:00:49 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: Forrest Young Subject: ViSta update Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk ViSta - The Visual Statistics System Announcing a major new release of ViSta for MSWindows and MacOS, in English, and of ViSta for MacOS in French. Version 4.28, August 1st, 1997 Copyright (c) 1991-97 by Forrest W. Young ViSta is a freely available visual statisics system designed for a wide range of users, particularly for students and teachers of statistics. ViSta features visual tools for revealing structure in data; for showing the results of statistical analyses; for guiding the analyst through an analysis; for structuring the analyst's data analysis process; and for designing statistical guidance. This major new release completes ViSta's guidance and help systems, provides a French translation for MacOS, and fixes several bugs. This release is available in English for MSWindows (95, 3.1, NT, and OS/2), and in English and French for MacOS (PowerPC and 680xx). The translation into =46rench was done by Louis Le Guelte, Laboratoire de Socio=E9cologie et Conservation, Univ. Lyon I., Lyon, France. All of the data analysis and visualization methods available in previous releases are available. ViSta 4.28 and its documentation are available from our WWW and FTP servers = at: http://forrest.psych.unc.edu/research/ USA home site http://www.psych.unc.edu/fwy USA alternate site http://micros.hensa.ac.uk/mirrors/vista UK (ac.uk only, 8AM-8PM GMT) http://sunsite.univie.ac.at/ViSta Austria http://www.sci.usq.edu.au/general/vista Australia The new version will be available later in the week at the following servers= : http://cons-dev.univ-lyon1.fr/vista France http://ssda.anu.edu.au/mirrors Australia GENERAL DESCRIPTION ViSta features visual tools for understanding data structure, for seeing the results of statistical analyses; for guiding the analyst during an analysis session; and for structuring the analyst's data analysis process. ViSta is designed for a wide range of users. It is particularly useful for those learning statistics and for their teachers. ViSta is being used to teach introductory statistics; introductory and advanced multivariate statistics; statistical graphics; and computational statistics. ViSta is also being used for research and development in statistical graphics and computational statistics. ViSta is designed for an audience of users having a very wide range of data analysis sophistication, ranging from novice to expert. ViSta provides seamlessly integrated data analysis environments specifically tailored to the user's level of expertise. Guidance is available for students and novices, and tools are available for teachers and other experts to create guidance for these novices. A structured graphical user interface is available for competent users, and a command line interface is available for sophisticated users. The complete Lisp-Stat (Tierney, 1990) programming environment is available to researchers, programmers and graduate students who wish to extend ViSta's capabilities. ViSta performs univariate and multivariate statistical data analysis and visualization. ViSta's data analysis and visualization capabilities include: exploratory graphics and descriptive statistics; univariate tests and visualizations; analysis of variance; regression analysis; principal components analysis; multidimensional scaling and correspondence analysis. All analysis methods have extensive visualization features. =3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-= =3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D Release 4.28, AUGUST 1st, 1997 - MAJOR RELEASE - MacOS and MSWINDOWS =3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-= =3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D-=3D New features: 1) GuideMaps and Help are finished (thanks to Louis Le Guelte for motivation= ) 2) New SHOW MENU HELP menu item for providing help with using menus. Bugs Fixed: 1) Sort-permute transformation has "sort-by-variable" as default 2) Sort-permute transformation always sorts selected variable 3) Sort-permute transformation can sort ordinal or category variables 4) Rank transformation now handles ties correctly. 5) Rank transformation can rank ordinal variables 6) Error message improved when can't import data 7) Preference to show guidemap no longer causes loading errors 8) Load Model menu item enables 9) Works more easily as a Netscape helper for MS-Windows 10) Summary stats window now shows last line under Unix and Windows 11) Open Data, New Data, Simulate Data, and Import Data can be used to author guidemaps 12) Load-vista and prepare-to-quit functions no longer write protected files 13) Eval-selection and Edit-selection no longer give error if listener not active 14) Compiling facility improved 15) Dialog boxes presented with guidemaps 16) GuideMaps errors reduced. ----- =46orrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: mailto:forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Voice: 1-919-962-5038 ViSta Software and Documentation site: http://forrest.psych.unc.edu/research/ USA Home Site ViSta Mirrors: http://ssda.anu.edu.au/mirrors Australia http://www.sci.usq.edu.au/general/vista Australia http://sunsite.univie.ac.at/ViSta Austria http://cons-dev.univ-lyon1.fr/vista France http://micros.hensa.ac.uk/mirrors/vista UK (ac.uk only 8AM-8PM GMT) http://www.psych.unc.edu/fwy USA From majordom@umnstat.stat.umn.edu Thu Aug 14 05:05:56 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06365; Thu, 14 Aug 1997 05:05:56 -0500 Return-Path: Received: from salmon.maths.tcd.ie by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06361; Thu, 14 Aug 1997 05:05:49 -0500 Received: from orange.stats.tcd.ie by salmon.maths.tcd.ie with SMTP id aa01440; 14 Aug 97 11:05 +0100 Received: from [194.125.148.127] by stats.tcd.ie (4.1/SMI-4.1) id AA03017; Thu, 14 Aug 97 10:51:11 BST Date: Thu, 14 Aug 97 10:51:06 BST X-Sender: dbearly@orange.stats.tcd.ie Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: stat-lisp-news@umnstat.stat.umn.edu From: "Dom. Early" Subject: Dialog box question - injecting text Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, I am currently working with dialogs in XLISP-STAT on a Mac and am wondering if there is any way of 'injecting' text into a edit-text-item-proto object in the dialog. Say a user clicks a button, this would cause text to be inserted into the current string of the edit-text-item-proto at the current insertion point. I have tried using the :TEXT method of the edit-text-item-proto but that only returns the current string, so you still have no way of knowing where the insertion point (the cursor if you like) is within the string. Other ideas included using WRITE-STRING with the dialog input stream but I cannot work out what stream it is using. Another idea I had was to use the clipboard and to simulate a paste command. Curiously paste works only with modal dialogs. With modeless dialogs the cut/copy/paste commands are available from the edit menu but they do not do anything. If I issue the :MODAL-DIALOG message to a modeless dialog cut/copy/paste then work! Is there a good reason why they don't work on modeless dialogs? Thanks for any help or comments, Dom. +--------------------------------------------------------------------------+ | Dominic Early e-mail: Dom.Early@tcd.ie | | A.31, Department of Statistics, Phone: +353-1-608 1917 | | Trinity College, Fax: +353-1-661 5046 | | Dublin 2, Ireland. | | | | http://www2.tcd.ie/Statistics/staff/dbearly.html | +--------------------------------------------------------------------------+ _/_/_/_/ _/_/_/ _/ _/ _/_/_/_/_/ _/_/_/ _/_/_/_/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/ _/_/_/_/ _/_/_/ _/ _/ _/ _/ _/_/_/ _/_/_/_/ From majordom@umnstat.stat.umn.edu Thu Aug 14 08:37:32 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA07022; Thu, 14 Aug 1997 08:37:32 -0500 Return-Path: Received: from wara.bolnet.bo by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA07018; Thu, 14 Aug 1997 08:37:23 -0500 Received: from default ([166.114.9.32]) by wara.bolnet.bo (8.8.5/8.8.5) with SMTP id JAA13500 for ; Thu, 14 Aug 1997 09:41:08 +0400 Message-Id: <33F12DE1.69E7@wara.bolnet.bo> Date: Tue, 12 Aug 1997 23:45:38 -0400 From: kjetil halvorsen X-Mailer: Mozilla 3.0Gold (Win95; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: Question: def and setf References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk What is really the differences between defining with def and with setf? After (def x '(1 2 3)) x is a variable; (variables) returns a list containing x, which doesnt occur with setf. Probably this means there is no way to free the space allocated with setf?, cannot be freed with undef. > (def x (list 2 4 6)) X > (variables) (X) > (setf y (list 1 2 3)) (1 2 3) > (variables) (X) Is there some situations where one can be used, and not the other? Thanks, Kjetil. From majordom@umnstat.stat.umn.edu Fri Aug 15 02:17:08 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10403; Fri, 15 Aug 1997 02:17:08 -0500 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10399; Fri, 15 Aug 1997 02:15:49 -0500 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA10434 (5.67b/IDA-1.5 for ); Wed, 13 Aug 1997 10:28:40 -0400 Received: from kp-w008118 ([150.103.229.15]) by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA8833 for ; Wed, 13 Aug 1997 10:26:49 -0400 Message-Id: <33F1C413.6231@kodak.com> Date: Wed, 13 Aug 1997 10:26:27 -0400 From: Jeff Morris Reply-To: morris@Kodak.COM Organization: Eastman Kodak Company X-Mailer: Mozilla 3.01 (WinNT; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: Date/Time Function? Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Is there a date/time function in XLS that I can use to stamp output? Something like the today() function in EXCEL is what I'm looking for. Thanks! Jeff Morris Eastman Kodak Co. Rochester, NY morris@kodak.com From majordom@umnstat.stat.umn.edu Fri Aug 15 10:51:02 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11527; Fri, 15 Aug 1997 10:51:02 -0500 Return-Path: Received: from gateway.ets.org by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11523; Fri, 15 Aug 1997 10:50:56 -0500 Received: from research.ets.org by gateway.ets.org with SMTP id AA25188 (5.65c/IDA-1.4.4 for ); Fri, 15 Aug 1997 11:50:55 -0400 Received: by research.ets.org id AA14309 (5.67b/IDA-1.4.3 for stat-lisp-news@umnstat.stat.umn.edu); Fri, 15 Aug 1997 11:50:53 -0400 Date: Fri, 15 Aug 1997 11:50:53 -0400 Message-Id: <199708151550.AA14309@research.ets.org> From: Russell Almond To: kjetil@wara.bolnet.bo Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <33F12DE1.69E7@wara.bolnet.bo> (message from kjetil halvorsen on Tue, 12 Aug 1997 23:45:38 -0400) Subject: Re: Question: def and setf Reply-To: ralmond@ets.org Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk What is really the differences between defining with def and with setf? After (def x '(1 2 3)) x is a variable; (variables) returns a list containing x, which doesnt occur with setf. Probably this means there is no way to free the space allocated with setf?, cannot be freed with undef. Def is a macro defined by Luke which does a number of things in addition to setf: 1) It adds the variable to the (variables) list. 2) It declares the variable to be special as with defvar. (This has scoping implications, see a good Lisp text, not just Luke's book.) Essentially, it allows you to refer to the global value within functions and methods without explicitly passing it. 3) It sets the value ala setf or setq. (This is an immediate set, in contrast to defvar which does not set the value of the symbol if it already has a value. This behavior of Lisp is somewhat surprizing the first time you change a bad initial value set using defvar, reload the file and don't see the value change.) I'm not sure exactly what the undef macro does. As far a freeing the storage for the variable goes, there is two things you can worry about: 1) freeing the storage for the value of the variable (e.g., a large data table or complex object). This is simple. In both cases, changing the value of the variable will free the storage for the array (as soon as no variable or other data structure points to it, it will become garbage and cleaned up in the next garbage collection). Thus to free up this storage, just set it to some convenient value, e.g. NIL. (Warning: getting NIL from a variable which was "destroyed" in this way were you were expecting a number or some other real value is a common source of Lisp bugs). 2) Freeing the storage for the symbol object itself is much more complicated. You need to make it unbound (not just set its value to nil), make sure it doesn't have function or property definitions, and then unintern it from its package before it becomes garbage. Few Lisp programmers bother. I don't know if undef does this, or simply makes the variable unbound and removes it from the list. --Russell Almond Educational Testing Service Research Statistics Group, 15-T Princeton, NJ 08541 Phone: 609-734-1557 FAX: 609-734-5420 Email: --almond@acm.org, --ralmond@ets.org http://bayes.stat.washington.edu/almond/almond.html From majordom@umnstat.stat.umn.edu Fri Aug 15 14:07:55 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12859; Fri, 15 Aug 1997 14:07:55 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12855; Fri, 15 Aug 1997 14:07:48 -0500 Received: from avery.med.virginia.edu by mail.virginia.edu id aa16057; 15 Aug 97 15:07 EDT Received: from bootp-17-17.bootp.virginia.edu (bootp-17-17.bootp.Virginia.EDU [128.143.17.17]) by avery.med.Virginia.EDU (8.8.5/8.6.6) with SMTP id PAA08212; Fri, 15 Aug 1997 15:07:19 -0400 Date: 15 Aug 97 15:17:17 -0400 Subject: Re: Date/Time Function? From: "Steven D. Majewski" To: morris@Kodak.COM, stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Cyberdog/2.0 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Wed, Aug 13, 1997 10:26 AM, Jeff Morris wrote: >Is there a date/time function in XLS that I can use >to stamp output? Something like the today() function >in EXCEL is what I'm looking for. > >Thanks! > At my request, Luke very kindly added a #'(get-decoded-time) function, which returns the time as a multiple-values list -- #'(get-internal-real-time) returns an internal tick count. I needed to add ascii timestamps to my data files to make them EMSA/EMMFF legal format ( and besides, it's a useful bit of information for your data! ) This was a fairly recent addition: I think it's only in 3.50 . > ( get-decoded-time ) 51 58 14 15 7 1997 5 NIL 4 I've written a time-stamp object which handles formatting: ; loading :timestmp.lsp ; finished loading "timestmp.lsp" ( setf time ( send time-stamp :new )) Fri Aug 15 14:56:14 1997 > time Fri Aug 15 14:56:14 1997 > ( type-of time ) OBJECT 'time' above is a time-stamp object whose :print method writes out an ascii string similar to what posix ctime() returns. ( The last three definitions are for the EMSA/EMMFF, which wants separate "#Date:" and "#Time:" headers, rather than ctime() format. FYI: EMSA/EMMFF is a standard ascii format used in microscopy for EDS/WDS/EELS spectra. ) - Steve Majewski ;; timestamp.lsp ;; ( get-decoded-time ) returns multiple value: ;; ( sec min hr date month year day-of-week DST TZ ) ( defproto time-stamp '( times ) '( weekdays months )) ( send time-stamp :slot-value 'weekdays (list "Sun" "Mon" "Tue" "Wed" "Thu" "Fri" "Sat" )) ( send time-stamp :slot-value 'months (list "Jan" "Feb" "Mar" "Apr" "May" "Jun" "Jul" "Aug" "Sep" "Oct" "Nov" "Dec" )) ( defmeth time-stamp :now () ( slot-value 'times ( multiple-value-list (get-decoded-time)))) ( defmeth time-stamp :isnew () ( send self :now )) ( defmeth time-stamp :weekday () ( elt ( slot-value 'weekdays ) ( elt (slot-value 'times) 6 ))) ( defmeth time-stamp :month () ( elt ( slot-value 'months ) ( elt (slot-value 'times) 4 ))) ( defmeth time-stamp :yr () ( elt ( slot-value 'times ) 5 )) ( defmeth time-stamp :asc-time () ( apply #'format Nil "~2,0d:~2,0d:~2,0d" ( reverse ( subseq ( slot-value 'times ) 0 3 )))) ( defmeth time-stamp :day () ( elt ( slot-value 'times ) 3 )) ( defmeth time-stamp :ctime () ( format nil "~A ~A ~D ~A ~D" ( send self :weekday ) ( send self :month ) ( send self :day ) ( send self :asc-time ) ( send self :yr ))) ( defmeth time-stamp :print ( &optional (str nil) ) ( format str "~A" ( send self :ctime ))) ( defproto emsa-time-stamp () () time-stamp "Adds methods to write out time & date in EMSA/EMMFF header format." ) ( defmeth emsa-time-stamp :date () ( format nil "~2,0d-~A-~D" ( send self :day ) ( send self :month ) ( send self :yr ))) ( defmeth emsa-time-stamp :time () ( apply #'format nil "~2,0d:~2,0d" ( reverse ( subseq ( slot-value 'times ) 1 3 )))) From majordom@umnstat.stat.umn.edu Fri Aug 15 15:17:14 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA13193; Fri, 15 Aug 1997 15:17:14 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA13188; Fri, 15 Aug 1997 15:17:09 -0500 Received: from avery.med.virginia.edu by mail.virginia.edu id aa14992; 15 Aug 97 16:17 EDT Received: from bootp-17-17.bootp.virginia.edu (bootp-17-17.bootp.Virginia.EDU [128.143.17.17]) by avery.med.Virginia.EDU (8.8.5/8.6.6) with SMTP id QAA76250; Fri, 15 Aug 1997 16:16:57 -0400 Date: 15 Aug 97 16:26:55 -0400 Subject: Re: your enhanced boxplots | Graphics gallery image From: "Steven D. Majewski" To: Gary McClelland , Michael Friendly Cc: stat-lisp-news@umnstat.stat.umn.edu X-Mailer: Cyberdog/2.0 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk [ CC-ed to stat-lisp-news, because these notes reminded my to upload fixed version of realgraf.lsp ] On Fri, Aug 15, 1997 1:25 PM, Gary McClelland wrote: >Steve, > From Micheal Friendly's website, I found your interesting enhanced >boxplots. I've been working on some similar ideas, but it looks like >you may already have the solution. Unfortunately, the link to your >boxplots from Michael's site was broken. Could you send me info on >where I might find your boxplots either on the web or a paper ref? On Fri, Aug 15, 1997 12:32 PM, Michael Friendly wrote: >Hi, Steve. > >I have one of your boxplots listed on my Graphics Gallery page, with a >link to > >HREF="http://minsky.med.virginia.edu/sdm7g/Projects/XlispStat/Ca- >conc.box.snap >.gif" > Full size (465 x 300) [6k] > > >Someone informed me that this link gives a 'server does not respond >message. > >Is the image available at another location? > Sorry -- the "temporary" relocation of my site to while my server was down has dragged on, as I've had no time to be a sysadmin! contains the images: and a link to the XlispStat implementation: jitterpl.lsp. However, jitterpl.lsp requires readgraf.lsp, and several other modules, so if you want to try to use this code, you need to look upstream at either: or mostly at the annotated soure index at: There was a bug in the code that was posted there until recently. These are snapshots of my working code, and the penultimate set of changes made composit graph objects non-working -- which I didn't notice until I tried the jitterplots again recently. I've just uploaded the latest fixed -- :redraw and :resize were required to migrate up to a super class. After a long spell of working on some other projects, I'm back to cleaning up and documenting this package, so I hope to have it in better shape soon. ( Including a more integrated distribution. ) I have done some nice graphs recently using the graph objects in realgraf.lsp to plot continuous data with error bands, instead of error bars around descrete points: for example, one set of data plotted as a dark blue line over a wider, light blue error band, the other set as a red line over a pink band. I'll try to add both code and graphic examples to my pages soon. I'm cc-ing this to the xlisp-stat mailing list, both to notify of the bug fix, and to take a chance to plug Michael's enormous list of statistics resource links: http://math.yorku.ca/SCS/StatResource.html and his Gallery of Data Viz: http://www.math.yorku.ca/SCS/Gallery/ Steve Majewski From majordom@umnstat.stat.umn.edu Sat Aug 23 02:22:44 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA25671; Sat, 23 Aug 1997 02:22:44 -0500 Return-Path: Received: from motgate.mot.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25667; Sat, 23 Aug 1997 02:22:37 -0500 Received: from pobox.mot.com (pobox.mot.com [129.188.137.100]) by motgate.mot.com (8.8.5/8.6.10/MOT-3.8) with ESMTP id CAA22369 for ; Sat, 23 Aug 1997 02:22:35 -0500 (CDT) Comments: ( Received on motgate.mot.com from client pobox.mot.com, sender skay@miel.mot.com ) Received: from hpux4.miel.mot.com (hpux4.miel.mot.com [217.1.84.89]) by pobox.mot.com (8.8.5/8.6.10/MOT-3.8) with SMTP id CAA29443 for ; Sat, 23 Aug 1997 02:22:28 -0500 (CDT) Received: from henna.miel.mot.com by hpux4.miel.mot.com with SMTP id AA21926 (5.67b/IDA-1.6 for stat-lisp-news@umnstat.stat.umn.edu); Sat, 23 Aug 1997 12:44:11 +0500 Received: by henna.miel.mot.com (SMI-8.6/SMI-SVR4) id MAA05322; Sat, 23 Aug 1997 12:52:20 +0530 From: skay@miel.mot.com (Santha Kumar V A ) Message-Id: <9708231252.ZM5320@henna> Date: Sat, 23 Aug 1997 12:52:20 +0000 Mail: -: From Desk of Shantha. Reply-To: skay@miel.mot.com X-Mailer: Z-Mail (3.2.1 10oct95) To: stat-lisp-news@umnstat.stat.umn.edu Subject: Problems encountered in Installing Xlisp-stat. Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi This is SanthaKumar from Motorola India Elecronics Limited (MIEL), Bangalore, India. To install R-Code software I downloaded xlispstat-3-50 from the site ftp://ftp.stat.umn.edu/pub/xlispstat/3-50/xlispstat-3-50.tar.gz But then I could not compile the stuff as I encounter some problems in 'make' which I could not figure out. I'm including the log so that u can figure out on whats wrong and find if you can help me out. I'm trying to install this in my Sun Sparc machine. Regards, Shantha. >make > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xsdynload.c > "xsdynload.c", line 702: warning: empty translation unit > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xssystem.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c unixprim.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlarray.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlbfun.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlbignum.c > "xlbignum.c", line 373: warning: semantics of "%" change in ANSI C; use > explicit cast > "xlbignum.c", line 374: warning: semantics of "/" change in ANSI C; use > explicit cast > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlcont.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xldbug.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xldmem.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xleval.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlfio.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlftab.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlglob.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlimage.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlinit.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlio.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlisp.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xljump.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xllist.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlpp.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlprin.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlrand.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlread.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlseq.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlstr.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlsubr.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlsym.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlsys.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlstruct.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlmath2.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlmath3.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xltvec.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlbcode.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlbcutil.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xlmodule.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c dummy.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c basics.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c common.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c compound.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c dists.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c ddists.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c hrdwrobs.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c mats1.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c mats2.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c math.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c obinit.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c objects.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c optimize.c > "optimize.c", line 301: warning: empty translation unit > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c sortdata.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c statinit.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c stats.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c utils.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c utils2.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c gamln.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c splines.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c kernel.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c lowess.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c cfft.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c betab.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c gammab.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c nor.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c ppnd.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c studentb.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c bivnor.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c linalg.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c cholesky.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c ludecomp.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c qrdecomp.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c svdecomp.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c makerot.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c eigen.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c minimize.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c blas.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xsiview2.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xsivint.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xsnewplt.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xsgraph.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xsiview3.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xsivwin.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xsiview.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xsivwin2.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c menus.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c dialogs.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c iviewdat.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c iviewscl.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c iview.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c iviewint.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xshist.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xsnames.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xssctmat.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xsspin.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c xssctplt.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c windows.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c stmem.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c graphics.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c myplot.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c term.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c postscript.c > cc -DUNIX -DNEWGC -DBYTECODE -DX11WINDOWS -O -DNOMEMMOVE -DNODIFFTIME > -I/usr/openwin/include -c X11BSDstuff.c > "X11BSDstuff.c", line 621: identifier redeclared: memmove > current : function() returning void > previous: function() returning pointer to void : > "/usr/include/string.h", line 105 > *** Error code 2 > make: Fatal error: Command failed for target `X11BSDstuff.o' From majordom@umnstat.stat.umn.edu Mon Aug 25 03:16:48 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19321; Mon, 25 Aug 1997 03:16:48 -0500 Return-Path: Received: from pompeiu.imar.ro by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19317; Mon, 25 Aug 1997 03:16:38 -0500 Received: from roimar.imar.ro (roimar.imar.ro [193.226.4.2]) by pompeiu.imar.ro (8.8.5/jtpda-5.2) with SMTP id LAA20336 for ; Mon, 25 Aug 1997 11:28:50 -0700 (PDT) Received: by roimar (MX V3.1C) id 15542; Mon, 25 Aug 1997 10:31:12 0200 Date: Mon, 25 Aug 1997 10:29:26 0200 From: ctambrea@roimar.imar.ro Reply-To: ctambrea@roimar.imar.ro To: STAT-LISP-NEWS@umnstat.stat.umn.edu Cc: ctambrea@roimar.imar.ro Message-Id: <009B94CA.39F95D60.15542@roimar> Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, I am Narcisa Tambrea from Romanian Academy. Do you have some information about an work group with focus on Time series? Thank you! Narcisa From majordom@umnstat.stat.umn.edu Wed Aug 27 12:59:19 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19335; Wed, 27 Aug 1997 12:59:19 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19327; Wed, 27 Aug 1997 12:59:12 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa00442; 27 Aug 97 13:59 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA17823; Wed, 27 Aug 1997 13:59:09 -0400 Date: Wed, 27 Aug 1997 13:59:08 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Lisp Statistics Cc: Luke Tierney Subject: XlispStat Colors ( There really *are* questions at the bottom! ) Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk [ There is actually a question here -- skip to the bottom if you're not interested in the background discussion. ] You get to appreciate Lisp's automatic memory management/garbage collection when you try to do something with the color tables in XlispStat. Color table entries have to be added add freed manually, and if you free a color that's used in a plot, you get an error when that plot attemps to :redraw itself. It would be nice if colors could be automatically managed, however, since they do refer to a finite system resource, it's probably not possible to combine explicit control over that resource with automagic management. ( Sigh! ;-) Anyway: I've been trying to clean up some of my graphics classes and I've been working on fixing some of the semi-working methods in colors.lsp. I had functions for generating colors lighter or darker than a given color. This was intended so that selected objects could change their colors and that composite graphics objects could manage their own "color-families". ( for example: for the "error-band" plots, I want the data shown with a colored line, and the +/- error-band drawn beneath the line as a lighter colored, but same hue, solid poly region, and it would be nice if the default colors could be selected automagically. ) However, to avoid having the color table get filled up with automatically generated colors, I wanted to find out if there is a 'close-enough" existing color, either in the (color-symbols) table, or in the RGB.txt generated list available to #'parse-color . ( My colors.lsp code initially started out as a Mac equivalent to X11's parse-color.) I had some problems with regression towards grey in my close-to color functions -- both with the original versions and my newer and better code. This seems to be mostly due to using the X11 RGB.TXT file as my test colortable -- grays are heavily represented, and the darker hues are particularly under-represented. ( see: I also now have some RGB <-> HSL color converter functions -- I'm experimenting with a stronger weight for Hue in the color comparison. ) are gifs of spin-plots of the RGB.TXT color file, plotted in the rgb color space, with some of the points colored their appropriate color. I had assumed that there was a 256 color limit in Mac XlispStat 3.50, however, when generating those color spin plots I realized that there are actually 256 + 8 color entries. The first two gifs and are plotted with that size color table. When I realized the error of my assumption, I tried rebuilding XlispStat with a 512 + 8 color table allocation. A similar color spin plot is However, you will note that some of the colors are colored wrongly. I don't think this was a bug in my code -- it was pretty simple and obvious and only a few lines of code -- and it mostly looks correct. Finally, the Questions: [1] Is this effect dues to ( I'm guessing, based on remembering some previous mailing list threads ... ) the guess that XlispStat does not actually manage color pallettes via #'add-color, but is going thru an existing pallette, and thus another "close enough" translation by the system ? [2] Any other known limitations of color graphics in XlispStat -- especially cross platform issues ? I would like to keep my code platform portable. Are there any fixed color-table sizes in the X11 or MS-Win versions ? ( Or, are the color bugs I'm seeing due to some Mac hardwired limitation I'm not aware of ? ) [3] Anybody have any other ideas of how to do this sort of color management better ? [ BTW: The new color code is not posted to the web yet, but it will be updated soon. ] ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Wed Aug 27 14:08:32 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19594; Wed, 27 Aug 1997 14:08:32 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19590; Wed, 27 Aug 1997 14:08:28 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.7.5/10.1) id PAA29612; Wed, 27 Aug 1997 15:08:18 -0400 (GMT+4:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 27 Aug 1997 15:06:12 -0500 To: vista@unc.edu, stat-lisp-news@umnstat.stat.umn.edu, edstat-l@jse.stat.ncsu.edu From: Forrest Young Subject: ViSta Version 4.28 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi folks: 1) The new version of ViSta for Unix (version 4.28) is now available. 2) If you missed the announcement, ViSta for Windows (version 4.28, August 13th, 1997) has been slightly revised so that it: a) unzips properly under Windows 3.1 (16 and 32 bit versions) b) works if moved to a new directory, or if the directory is renamed. It continues to unzip properly for Windows 95, NT and OS/2. The new version now works with directory changes under these systems. You can modifiy the statinit.lsp file in the previous version (4.28, August 1st, 1997) to allow changing and renaming directories. This is done by changing the line: #+msdos (working-directory *default-path*) to: #+msdos (working-directory (strcat (get-working-directory) separator)) 3) Also, ViSta for Mac and PowerMac (version 4.28) in either English or French, was announced at the beginning of August. The new Unix version and revised MSWindows versions are available from the following servers: http://forrest.psych.unc.edu/research/ USA Home Site http://www.psych.unc.edu/fwy USA Mirror Master Site http://sunsite.univie.ac.at/ViSta Austria SunSite Vienna http://micros.hensa.ac.uk/mirrors/vista UK (ac.uk only 8AM-8PM GMT) http://www.sci.usq.edu.au/general/vista Australia: Univ. S. Queensland http://ssda.anu.edu.au/mirrors Australia: ANU Site The following server is not yet updated: http://cons-dev.univ-lyon1.fr/vista France: Univ. Lyon. More information about ViSta is available at all of the servers given above. --- Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: mailto:forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Voice: 1-919-962-5038 From majordom@umnstat.stat.umn.edu Wed Aug 27 14:33:48 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19796; Wed, 27 Aug 1997 14:33:48 -0500 Return-Path: Received: from amber.ccs.neu.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19792; Wed, 27 Aug 1997 14:33:41 -0500 Received: from RISA ([192.127.136.201] (may be forged)) by amber.ccs.neu.edu (8.8.6/8.7.3) with SMTP id PAA12342 for ; Wed, 27 Aug 1997 15:33:30 -0400 (EDT) Received: by RISA with Microsoft Mail id <01BCB2FB.8E9276B0@RISA>; Wed, 27 Aug 1997 15:11:55 -0400 Message-Id: <01BCB2FB.8E9276B0@RISA> From: psksvp To: "'Xlisp-Stat List'" Subject: DDE interface to Visual C Date: Wed, 27 Aug 1997 15:08:43 -0400 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi!! I'd like to use XList-Stat with a C++ program using Visual C++. Does anyone ever use DDE to send Lisp string like "(car '(a b c))" to XLisp-Stat and get the result string back. eg. char r[255]; r = XLisp("(car '(1 2 3))"); // r contains result string from xlisp psksvp@ccs.neu.edu From majordom@umnstat.stat.umn.edu Wed Aug 27 18:47:18 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA20715; Wed, 27 Aug 1997 18:47:18 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA20711; Wed, 27 Aug 1997 18:47:12 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa14359; 27 Aug 97 19:47 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA17114; Wed, 27 Aug 1997 19:46:56 -0400 Date: Wed, 27 Aug 1997 19:46:55 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Lisp Statistics Subject: How to draw 3-d plot of RGB space & ignore-errors [was: XlispStat Colors ] In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk FYI: Here's how to plot the distribution of colors in the rgb color space cube. ( free-extra-colors and *colortable* , and maybe some other things here are defined in ) This will not load, however, as the ignore-errors wrapped around the add-color call that is expected to eventually overflow, seems to work properly when I cut and paste this code interactively, however, it fails and caused a break into the lisp debugger, when I load it from a file. I tried wapping it again in an unwind-protect clause, but with the same result. Am I doing something wrong ? Is there another way to get the same effect -- i.e. how do I continue to load in spite of errors -- in this case, it's an expected error. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson ( free-extra-colors ) ; start fresh (optional) defined in colors.lsp ;; remove some colors from the set of all in colortable ;; ( i.e. the grays and the standard fixed colors ) ( setf colors ( remove nil ( mapcar #'(lambda (x) (if (find x *colors*) nil x)) ( mapcar #'intern ( remove "GRAY" ( remove "GREY" ( mapcar #'car *colortable* ) :test #'string-search) :test #'string-search ))))) ;; and shuffle them so we will have a random selection ( setf colors ( shuffle colors )) ( add-color 'gray ) ( add-color 'lightgray ) ;; fill the color table -- we'll get an error when the table is full ( unwind-protect ( ignore-errors ( dolist ( c colors ) ( add-color ( print c )))) ) ( setf colors (color-symbols)) ( setf colors-used ( mapcar #'(lambda (x) (find x colors)) ( mapcar #'intern ( mapcar #'car *colortable* )))) ( setf s ( spin-plot ( transpose ( mapcar #'cdr *colortable* )) :point-labels ( mapcar #'car *colortable* ))) ( send s :use-color T ) ;; color the points whose color is in the allocated (color-symbols) table ( send s :point-color (iseq (length colors-used)) colors-used) ;; selected points show up with larger dots ;; the uncolored points will show up smaller and dimmer. ( send s :point-selected ( iseq ( length colors-used )) colors-used ) From majordom@umnstat.stat.umn.edu Wed Aug 27 19:00:35 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA20759; Wed, 27 Aug 1997 19:00:35 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA20755; Wed, 27 Aug 1997 19:00:31 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa18208; 27 Aug 97 20:00 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA19511; Wed, 27 Aug 1997 20:00:28 -0400 Date: Wed, 27 Aug 1997 20:00:28 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Lisp Statistics Subject: Re: How to draw 3-d plot of RGB space & ignore-errors [was: XlispStat Colors ] In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk -- I forgot: I meant to add to the previous that I now notice that even with the smaller number of rgb colors set in the source code, I get what look like a few misplaced colors in that spin plot -- I'm still not sure what the cause is. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@nokomis.stat.umn.edu Thu Aug 28 11:39:44 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA25515; Thu, 28 Aug 1997 11:39:44 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA25510; Thu, 28 Aug 1997 11:39:36 -0500 From: Luke Tierney Message-Id: <9708281639.AA25510@nokomis.stat.umn.edu> Subject: Re: How to draw 3-d plot of RGB space & ignore-errors [was: XlispStat Colors ] To: sdm7g@Virginia.EDU (Steven D. Majewski) Date: Thu, 28 Aug 1997 11:39:36 -0500 (CDT) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Steven D. Majewski" at Aug 27, 97 07:46:55 pm X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 732 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > > This will not load, however, as the ignore-errors wrapped around > the add-color call that is expected to eventually overflow, seems > to work properly when I cut and paste this code interactively, > however, it fails and caused a break into the lisp debugger, > when I load it from a file. > This appears to be due to a bug in the byte code interpreter which I'll track down and fix in the next release. -- Luke Tierney University of Minnesota Phone: 612-625-7843 School of Statistics Fax: 612-624-8868 206 Church Street email: luke@stat.umn.edu Minneapolis, MN 55455 USA WWW: http://www.stat.umn.edu From majordom@umnstat.stat.umn.edu Sat Aug 30 20:08:29 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28628; Sat, 30 Aug 1997 20:08:29 -0500 Return-Path: Received: from wara.bolnet.bo by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28622; Sat, 30 Aug 1997 20:08:21 -0500 Received: from pc.bolnet.bo ([166.114.9.72]) by wara.bolnet.bo (8.8.5/8.8.5) with SMTP id VAA22494 for ; Sat, 30 Aug 1997 21:12:28 +0400 Message-Id: <34076AB7.53EE@wara.bolnet.bo> Date: Fri, 29 Aug 1997 20:35:04 -0400 From: kjetil halvorsen X-Mailer: Mozilla 3.0Gold (Win95; I) Mime-Version: 1.0 To: lisp-stat Subject: getting data from histogram object Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Another simple(?) question: Is there a way to get the the values of the variables stored inside an histogram object? I suppose (send hist :slot-value 'histogram-internals) (# . #) this is the data, but cannot see a way to access them, as is possible with a regression model object. Thank you, Kjetil Halvorsen. From majordom@umnstat.stat.umn.edu Sat Aug 30 20:08:28 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA28626; Sat, 30 Aug 1997 20:08:28 -0500 Return-Path: Received: from wara.bolnet.bo by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA28617; Sat, 30 Aug 1997 20:08:19 -0500 Received: from pc.bolnet.bo ([166.114.9.72]) by wara.bolnet.bo (8.8.5/8.8.5) with SMTP id VAA22489 for ; Sat, 30 Aug 1997 21:12:23 +0400 Message-Id: <34075C93.6753@wara.bolnet.bo> Date: Fri, 29 Aug 1997 19:34:44 -0400 From: kjetil halvorsen X-Mailer: Mozilla 3.0Gold (Win95; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: duplication of :help output? why References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk In the below output hist is an object of type histogram (made be (def hist (histogram ... Sending it the message :slot-names, in the answer many slots are listed multiple. The output is edited to show this clearer. Why is this, and is there a way to make more reasonable output in this and similar situations? Thank you, Kjetil. (send hist :slot-names) (HARDWARE-ADDRESS SUBORDINATES GO-AWAY SIZE LOCATION TITLE INTERNALS MENU-TITLE HAS-V-SCROLL HAS-H-SCROLL BLACK-ON-WHITE MENU SLICERS MOUSE-MODE FIXED-ASPECT SHOWING-LABELS VARIABLE-LABELS NUMBER-OF-VARIABLES MENU-TEMPLATE OPTIONS-TEMPLATE SCALE-TYPE OVERLAYS HISTOGRAM-INTERNALS DOCUMENTATION HISTOGRAM-INTERNALS HARDWARE-ADDRESS SUBORDINATES GO-AWAY SIZE LOCATION TITLE INTERNALS MENU-TITLE HAS-V-SCROLL HAS-H-SCROLL BLACK-ON-WHITE MENU SLICERS MOUSE-MODE FIXED-ASPECT SHOWING-LABELS VARIABLE-LABELS NUMBER-OF-VARIABLES MENU-TEMPLATE OPTIONS-TEMPLATE SCALE-TYPE OVERLAYS PROTO-NAME INSTANCE-SLOTS DOCUMENTATION MODE-LIST OVERLAYS SCALE-TYPE OPTIONS-TEMPLATE MENU-TEMPLATE NUMBER-OF-VARIABLES VARIABLE-LABELS SHOWING-LABELS FIXED-ASPECT MOUSE-MODE SLICERS HARDWARE-ADDRESS SUBORDINATES GO-AWAY SIZE LOCATION TITLE INTERNALS MENU-TITLE HAS-V-SCROLL HAS-H-SCROLL BLACK-ON-WHITE MENU PROTO-NAME INSTANCE-SLOTS DOCUMENTATION MENU BLACK-ON-WHITE HAS-H-SCROLL HAS-V-SCROLL MENU-TITLE INTERNALS HARDWARE-ADDRESS SUBORDINATES GO-AWAY SIZE LOCATION TITLE PROTO-NAME INSTANCE-SLOTS DOCUMENTATION TITLE LOCATION SIZE GO-AWAY HARDWARE-ADDRESS SUBORDINATES PROTO-NAME INSTANCE-SLOTS SUBORDINATES HARDWARE-ADDRESS PROTO-NAME INSTANCE-SLOTS DOCUMENTATION PROTO-NAME INSTANCE-SLOTS) From majordom@umnstat.stat.umn.edu Tue Sep 2 12:21:39 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03508; Tue, 2 Sep 1997 12:21:39 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03504; Tue, 2 Sep 1997 12:21:34 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa05144; 2 Sep 97 13:21 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA13484; Tue, 2 Sep 1997 13:21:29 -0400 Date: Tue, 2 Sep 1997 13:21:28 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: kjetil halvorsen Cc: lisp-stat Subject: Re: getting data from histogram object In-Reply-To: <34076AB7.53EE@wara.bolnet.bo> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Fri, 29 Aug 1997, kjetil halvorsen wrote: > Another simple(?) question: Is there a way to get the the values > of the variables stored inside an histogram object? I suppose > > (send hist :slot-value 'histogram-internals) > (# . #) > > this is the data, but cannot see a way to access them, as is > possible with a regression model object. > All of the plot objects have a :POINT-COORDINATE method: > ( send h :help :point-coordinate ) POINT-COORDINATE Method args: (var point &optional value) Sets or retrieves coordinate for variable VAR and point POINT in the original coordinate system. Vectorized. VAR is the variable, i.e. which dimension of the plot. Histograms are unidimensional, so VAR = 1. "Vectorized" means that point can be a single point number or a list of point numbers. To get all of the data points: ( send hist :point-coordinate 1 ( iseq ( send hist :num-points ))) The y value of the histogram plots is not really an attribute of the data but of the view of the data. It is available via the :BIN-COUNTS method. ( I've complained in the past that Xlisp-Stat's plot objects don't make the distinction between data and views clear enough sometimes, but it looks like Luke was trying to maintain some distinction here. It may have been a sacrifice to performance that required the data slots to be internal to graphics objects. ) Re: your other question about :SLOT-NAMES : I'm not sure what your question is, however... If the problem is that the list is too long, sometimes you may want to try :OWN-SLOTS on the prototype object. :SLOT-NAMES returns the names of all slots inherited or owned by that object. :OWN-SLOTS returns only the slots owned by that object. ( note that you will most likely get a different list returned by sending that message to the prototype versus and instance of that prototype. ) > ( send histogram-proto :own-slots ) (DOCUMENTATION HISTOGRAM-INTERNALS HARDWARE-ADDRESS SUBORDINATES GO-AWAY SIZE LOCATION TITLE INTERNALS MENU-TITLE HAS-V-SCROLL HAS-H-SCROLL BLACK-ON-WHITE MENU SLICERS MOUSE-MODE FIXED-ASPECT SHOWING-LABELS VARIABLE-LABELS NUMBER-OF-VARIABLES MENU-TEMPLATE OPTIONS-TEMPLATE SCALE-TYPE OVERLAYS PROTO-NAME INSTANCE-SLOTS) > ( send hist :own-slots ) (HARDWARE-ADDRESS SUBORDINATES GO-AWAY SIZE LOCATION TITLE INTERNALS MENU-TITLE HAS-V-SCROLL HAS-H-SCROLL BLACK-ON-WHITE MENU SLICERS MOUSE-MODE FIXED-ASPECT SHOWING-LABELS VARIABLE-LABELS NUMBER-OF-VARIABLES MENU-TEMPLATE OPTIONS-TEMPLATE SCALE-TYPE OVERLAYS HISTOGRAM-INTERNALS) If what you want is different formatting of the output, you can try pretty-printing the returned list: > ( pprint ( send hist :own-slots )) Or define another function that formats and prints the list the way you like it. If the problem is finding the appropriate method or slot when you're not sure what it's called, you can try something like this -- an :apropos method for objects that's a sort of parallel to the apropos function: ( defmeth *object* :apropos ( str &key help ) ( dolist ( name ( send self :doc-topics )) ( when (string-search str (symbol-name name)) ( if help ( progn ( send self :help name ) (terpri)) (format t "~s~%" name )))) ( format t "~a : ~a SLOTS:~%" ( slot-value 'proto-name ) self ) ( dolist ( name ( send self :slot-names )) ( when (string-search str (symbol-name name)) ( format t "~s : ~a~%" name ( slot-value name ))))) > ( send histogram-proto :apropos "point" ) :POINTS-IN-RECT :DRAW-DATA-POINTS :POINTS-SELECTED :POINTS-SHOWING :POINTS-HILITED :UNSHOW-ALL-POINTS :POINT-SHOWING :POINT-HILITED :POINT-SELECTED :POINT-SYMBOL :POINT-LABEL :POINT-SCREEN-STATE :POINT-STATE :POINT-COLOR :POINT-MASKED :POINT-TRANSFORMED-COORDINATE :POINT-SCREEN-COORDINATE :POINT-COORDINATE :NUM-POINTS :CLEAR-POINTS :ADD-POINTS :ANY-POINTS-SELECTED-P :ALL-POINTS-UNMASKED-P :ALL-POINTS-SHOWING-P :SHOW-ALL-POINTS :UNMASK-ALL-POINTS :UNSELECT-ALL-POINTS :ADJUST-SCREEN-POINT :ADJUST-POINTS-IN-RECT :DRAW-POINT HISTOGRAM-PROTO : # SLOTS: NIL ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Tue Sep 2 13:32:29 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03877; Tue, 2 Sep 1997 13:32:29 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03873; Tue, 2 Sep 1997 13:32:23 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa17807; 2 Sep 97 14:32 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA14436; Tue, 2 Sep 1997 14:32:20 -0400 Date: Tue, 2 Sep 1997 14:32:19 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: lisp-stat Subject: point-coordinates [was: getting data from histogram object] In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Tue, 2 Sep 1997, I wrote (among other things): > > The y value of the histogram plots is not really an attribute of > the data but of the view of the data. It is available via the > :BIN-COUNTS method. ( I've complained in the past that Xlisp-Stat's > plot objects don't make the distinction between data and views > clear enough sometimes, but it looks like Luke was trying to maintain > some distinction here. It may have been a sacrifice to performance > that required the data slots to be internal to graphics objects. ) > To give the contrary example, one place where the xlisp-stat plot-object design doesn't work as well is with boxplots. :NUM-POINTS of a boxplot is zero. Which is clearly correct of the graphic view, but not of the data points of the data model. ( If you're going to accuse me of confusing the two, my point is that the design of Lisp-Stat graphics objects encourages this confusion somewhat. If you use scatterplots that allow some direct manipulation, then you are sort of forced to using plot-objects as the repository of their data. ) In my jitterplot code, , I was trying not just to change the presentation, but to make "smarter" boxplots: there are some methods added to handle grouped data. ( defmeth jitterplot-proto :group-points ( gn &optional msg (value nil set)) "Args: ( group-number &optional msg value ) - with group-number as sole argument, returns the point-numbers for data-group. - with msg, sends msg using point-numbers as an arg. ( msg may be a single keyword or a list of arguments ). Provide value arg for setter form of message. example: ( send boxplot :group-points 2 '( :point-coordinates 1 )) -- returns the y values of group #2 ( send boxplot :group-points 1 :point-color 'blue ) -- sets the points in group #1 to blue." ... However, these additions also fall short of a real and consistent Model-View type model: the boxplot overlays, for example, do not automatically respond to changes in the point data. ( Some of the inconsistency is due to trying to follow & integrate with standard lisp-stat graphics objects, and some is due to my own lack of consistency in my first stab at this problem. ) ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Mon Sep 8 16:57:13 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA21770; Mon, 8 Sep 1997 16:57:13 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA21766; Mon, 8 Sep 1997 16:57:07 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa26320; 8 Sep 97 17:57 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA27324; Mon, 8 Sep 1997 17:57:02 -0400 Date: Mon, 8 Sep 1997 17:57:01 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU Reply-To: "Steven D. Majewski" To: Lisp Statistics Subject: Testreport about mathematical/statistical software for data analysis ! (fwd) Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Xlisp-Stat is missing from this comparison. It consists of a series of tables of support for various types of math function, graphics, foreign function, etc. Maybe if someone can contribute the table entries for XlispStat, we could convince the author to include it. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson ---------- Forwarded message ---------- Date: Mon, 08 Sep 1997 12:22:01 +0200 From: Stefan Steinhaus To: s-news@utstat.toronto.edu Subject: Testreport about mathematical/statistical software for data analysis ! Dear users of S-Plus, I would like to announce my testreport "Comparison of mathematical programs for data analysis" which has just been finished. The testreport which is written in English is available at the following website : http://www.uni-frankfurt.de/~stst/ncrunch.html This testreport contains information about the actual versions of the programs - GAUSS (Version 3.2) - Macsyma (Version 2.2) - Maple (Version 4.0) - Mathematica (Version 3.0) - Matlab (Version 5.1) - MuPAD (Version 1.3) - O-Matrix (Version 3.2) - Ox (Version 1.11) - S-Plus (Version 4.0) grouped by the following sections - Comparison of the mathematical functionality - Comparison of the graphical functionality - Functionality of the programming environment - Data import/export options - Available operating systems - Speed comparison - Summary The weighting for the above listed sections are all explained in the testreport so that an individuel weighting could also be done. Sincerely, Stefan Steinhaus P.S. Please feel free to email me your impressions and critism about it. -- ##################################################################### ## Stefan Steinhaus ## ## ## ## University of Frankfurt / Germany ## ## Fachbereich Informatik (FB20) ## ##-----------------------------------------------------------------## ## EMAIL : stst@informatik.uni-frankfurt.de ## ## ## ## Homepage______ : http://www.uni-frankfurt.de/~stst/ ## ## Scientific Web : http://www.uni-frankfurt.de/~stst/science.html ## ##################################################################### From majordom@umnstat.stat.umn.edu Mon Sep 8 19:41:02 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22338; Mon, 8 Sep 1997 19:41:02 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22334; Mon, 8 Sep 1997 19:40:55 -0500 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.7.5/10.1) id UAA02848; Mon, 8 Sep 1997 20:40:49 -0400 (GMT+4:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 8 Sep 1997 20:40:57 -0500 To: Lisp Statistics From: Forrest Young Subject: Re: Testreport about mathematical/statistical software for data analysis ! (fwd) Cc: "Steven D. Majewski" Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Steven D. Majewski writes: Stefan Steinhaus writes: >>I would like to announce my testreport "Comparison of mathematical >>programs for data analysis" which has just been finished. The testreport >>which is written in English is available at the following website : >> >> http://www.uni-frankfurt.de/~stst/ncrunch.html >> >Xlisp-Stat is missing from this comparison. > >It consists of a series of tables of support for various types of >math function, graphics, foreign function, etc. > >Maybe if someone can contribute the table entries for XlispStat, >we could convince the author to include it. I have typed up a table containing all of the stat system features covered by the testreport. The features are the rows of the table. I have included my (initial) rating of LispStat using the testreport's rating categories of "+", "-", or "m". "+" means the feature is supported "-" means not supported "m" means it is available in an additional module (i.e., in code not included in the original distribution of LispStat, but available from the net). I've added "?" in places I'm not sure about my rating. I've added a few random comments. I'd appreciate anyone else's additions/corrections, especially about whether "-"s should be "m"s. Then I'll send it to Stefan Steinhaus. Forrest Young =-=-=-=-=-=-= 2.1 Standard Mathematical Functions Trig functions + Gamma function + Poly Gamma - Log-Gamma functions + Beta function + 2.2 Linear Algebra Eigenvalues + Eigenvectors + Cholesky Decomposition + Crout Decomposition - LU Decomposition + Singular Value Decomposition + Upper Hessenberg form - Toeplitz matrix - Schur form of quadratic matrix - Optimization (Unconstr. / Constr.) +/- (I'm not sure about this) Linear equation solver + (I'm not sure about this) Non-linear equation solver - Ordinary Differential Equation solver - Partial Differential Equation solver - Sparse matrices handling - Moore-Penrose pseudo-inverse - 2.3 Analysis Numerical integration - Double integration - Triple integration - Numerical differentiation - Fourier transf. (1D / 2D /multidim.) - Inverse Fourier transformation (1D / 2D / multidim.) - 2.4 Numerical Mathematics - Classical Interpolation - k-Spline Interpolation - B-Spline Interpolation - Newton method for finding roots + Bisection - Runge Kutta method for solving ODE - 2.5 Stochastic Markov models - Mean + Median + Mode - Variance + Beta Distribution + + + Chi-Squared Distribution + + + Gamma Distribution + + + Log-Normal Normal Distribution + + + Poisson Distribution + + + Uniform Distribution + + + More Distribution 2.6 Statistics Linear regression + Polynomial regression - Nonlinear regression + Loess regression m LOGIT regression - PROBIT regression - PSN regression - Event count models - Duration models - Goodness of fit test + T-Test + F-Test + Q-Test - Z-Test m Maximum Likelihood m ARIMA - Time series analysis - GARCH models - Wavelets - Cluster analysis m Survival analysis - 2.7 Other Mathematics - Does anyone know of functions/methods that have been written for any of these? Cointegration models Black Scholes model Dynamic rational expectation models Linear rational expectation models Non-linear rational expectation models Social network models Kalman filter Neuronal networks Regressive-autoregressive models Portfolio analysis State-space models 3.0 Graphics - It seems to me this is a woefully inadequate classification 2D Graphics Bar charts Other charts + Error charts High-Low-Average plot Histograms + Log Plot m Log-Log Plot m Polar Plot XY Plot + 3D Graphics Charts m Contour Plot Error bars Height Colors Surface Plot + XYZ plot + Special Graphics Animations + Bollinger bands Box & Whisker Plots + Candlestick charts Cluster graphs Dendograms Periodograms QQ Plot + 4.0 Functionality of the Programming Environment Editing Features Built-in Editor + External Editor Configurable + ource code formatting + Syntax highlighting Debugging Breakpoints + Function Tracer + ? Line Tracing Profiler Stack inspection Variable inspection + Language Features API interface m DDE support - GUI programming m N-dimensional arrays + Object Oriented programming + OLE support ? P code compiling + (If this means Byte-Code compiling) Language Interfaces Assembler - C/C++ + FORTRAN + GAUSS - LISP + (note that I added this category) Macsyma - Maple - Mathematica - Matlab - MuPAD - Q-Matrix - Ox - S-Plus - DLL-Calls - 5.0 Data Import-Export none except ASCII 6.0 Available Operating Systems Convex + (if this means unix) Cray + (if this means unix) DEC + + - HP 9000 + - IBM RISC + Intel Dos + Intel OS/2 + Intel Win 3.1x/95,NT + + Intel Linux + Morotola (Mac OS/Nextstep) + - SGI - SUN (solaris) + From majordom@umnstat.stat.umn.edu Mon Sep 8 23:23:35 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22920; Mon, 8 Sep 1997 23:23:35 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22916; Mon, 8 Sep 1997 23:23:28 -0500 Received: from [164.67.22.130] (ts25-16.wla.ts.ucla.edu) by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA06813; Mon, 8 Sep 97 21:23:17 PDT X-Sender: deleeuw@laplace.stat.ucla.edu Message-Id: In-Reply-To: References: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 8 Sep 1997 21:22:24 -0700 To: Forrest Young , stat-lisp-news@umnstat.stat.umn.edu From: Jan de Leeuw Subject: Re: Testreport about mathematical/statistical software for data analysis ! (fwd) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I have added modified suggested ratings between parentheses. --- Jan >=-=-=-=-=-=-= > >2.1 Standard Mathematical Functions >Trig functions + >Gamma function + >Poly Gamma - >Log-Gamma functions + >Beta function + > >2.2 Linear Algebra >Eigenvalues + >Eigenvectors + >Cholesky Decomposition + >Crout Decomposition - >LU Decomposition + >Singular Value Decomposition + >Upper Hessenberg form - >Toeplitz matrix - >Schur form of quadratic matrix - >Optimization (Unconstr. / Constr.) +/- (I'm not sure about this) unconstr + constr m >Linear equation solver + (I'm not sure about this) (o yes) >Non-linear equation solver - + >Ordinary Differential Equation solver - >Partial Differential Equation solver - >Sparse matrices handling - >Moore-Penrose pseudo-inverse - > >2.3 Analysis >Numerical integration - >Double integration - >Triple integration - >Numerical differentiation - >Fourier transf. (1D / 2D /multidim.) - >Inverse Fourier transformation (1D / 2D / multidim.) - > >2.4 Numerical Mathematics - >Classical Interpolation - >k-Spline Interpolation - >B-Spline Interpolation - (m) >Newton method for finding roots + >Bisection - (m -- and many others) >Runge Kutta method for solving ODE - (m) > >2.5 Stochastic >Markov models - (m) >Mean + >Median + >Mode - >Variance + >Beta Distribution + + + >Chi-Squared Distribution + + + >Gamma Distribution + + + >Log-Normal >Normal Distribution + + + >Poisson Distribution + + + >Uniform Distribution + + + >More Distribution (m - many more) > >2.6 Statistics >Linear regression + >Polynomial regression - (+ - this is a single statement) >Nonlinear regression + >Loess regression m (+ -- there is lowess) >LOGIT regression - (m) >PROBIT regression - (m) >PSN regression - (?) >Event count models - (m) >Duration models - (m) >Goodness of fit test + >T-Test + >F-Test + >Q-Test - >Z-Test m >Maximum Likelihood m (+) >ARIMA - >Time series analysis - (m) >GARCH models - >Wavelets - >Cluster analysis m >Survival analysis - (m) > >2.7 Other Mathematics - Does anyone know of functions/methods that have >been written for any of these? (This should really be "other econometrics") >Cointegration models >Black Scholes model >Dynamic rational expectation models >Linear rational expectation models >Non-linear rational expectation models >Social network models >Kalman filter (m) >Neuronal networks >Regressive-autoregressive models >Portfolio analysis >State-space models > >3.0 Graphics - It seems to me this is a woefully inadequate classification > >2D Graphics >Bar charts >Other charts + >Error charts >High-Low-Average plot >Histograms + >Log Plot m >Log-Log Plot m >Polar Plot >XY Plot + > >3D Graphics >Charts m >Contour Plot (+) >Error bars >Height Colors >Surface Plot + >XYZ plot + > >Special Graphics >Animations + >Bollinger bands >Box & Whisker Plots + >Candlestick charts >Cluster graphs (m) >Dendograms >Periodograms >QQ Plot + > > >4.0 Functionality of the Programming Environment > >Editing Features (+) >Built-in Editor + >External Editor Configurable + >ource code formatting + >Syntax highlighting (if run in emacs) > >Debugging (+) >Breakpoints + >Function Tracer + ? >Line Tracing >Profiler >Stack inspection (+) >Variable inspection + > >Language Features >API interface m >DDE support - >GUI programming m >N-dimensional arrays + >Object Oriented programming + >OLE support ? (only in the spanish version) >P code compiling + (If this means Byte-Code compiling) > >Language Interfaces >Assembler (m -- using shared libs) >C/C++ + >FORTRAN + >GAUSS - >LISP + (note that I added this category) >Macsyma - >Maple - >Mathematica - >Matlab - >MuPAD - >Q-Matrix - >Ox - >S-Plus - >DLL-Calls - > >5.0 Data Import-Export > >none except ASCII > > >6.0 Available Operating Systems > >Convex + (if this means unix) >Cray + (if this means unix) >DEC + + - >HP 9000 + - >IBM RISC + >Intel Dos + >Intel OS/2 + >Intel Win 3.1x/95,NT + + >Intel Linux + >Morotola (Mac OS/Nextstep) + - (+) >SGI - (+) >SUN (solaris) + --- Jan de Leeuw; Professor and Chair, Interdivisional Program in Statistics; US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw ============================================================================ No matter where you go, there you are. --- Buckaroo Banzai ============================================================================ From majordom@umnstat.stat.umn.edu Mon Sep 8 23:56:33 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22989; Mon, 8 Sep 1997 23:56:33 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22985; Mon, 8 Sep 1997 23:56:28 -0500 Received: from [164.67.22.130] (ts16-12.wla.ts.ucla.edu) by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA07472; Mon, 8 Sep 97 21:56:22 PDT X-Sender: deleeuw@laplace.stat.ucla.edu Message-Id: In-Reply-To: References: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 8 Sep 1997 21:56:20 -0700 To: Jan de Leeuw , Forrest Young , stat-lisp-news@umnstat.stat.umn.edu From: Jan de Leeuw Subject: Re: Testreport about mathematical/statistical software for data analysis ! (fwd) Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I missed a piece >2.3 Analysis >Numerical integration - (m) >Double integration - (m) >Triple integration - >Numerical differentiation - (+) >Fourier transf. (1D / 2D /multidim.) - (m) >Inverse Fourier transformation (1D / 2D / multidim.) - -- Jan --- Jan de Leeuw; Professor and Chair, Interdivisional Program in Statistics; US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw ============================================================================ No matter where you go, there you are. --- Buckaroo Banzai ============================================================================ From majordom@umnstat.stat.umn.edu Tue Sep 9 18:14:51 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06756; Tue, 9 Sep 1997 18:14:51 -0500 Return-Path: Received: from franz.stat.wisc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06752; Tue, 9 Sep 1997 18:14:43 -0500 Received: by franz.stat.wisc.edu id m0x8ZUO-000hreC (Debian Smail-3.2 1996-Jul-4 #2); Tue, 9 Sep 1997 18:14:32 -0500 (CDT) To: stat-lisp-news@umnstat.stat.umn.edu Cc: Kurt Hornik , Subject: Debian package for xlispstat From: Douglas Bates Date: 09 Sep 1997 18:14:32 -0500 In-Reply-To: Forrest Young's message of Wed, 27 Aug 1997 15:06:12 -0500 Message-Id: <6rsove6pwn.fsf@franz.stat.wisc.edu> Lines: 31 X-Mailer: Gnus v5.4.37/XEmacs 19.15 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk This message will probably only be of interest to Debian GNU/Linux users who also use xlispstat. I hope to prepare several packages of statistical software, including xlispstat, ViSta, R and ESS (Emacs Speaks Statistics), for the Debian distribution of Linux. The xlispstat package is the simplest of these. I have created a preliminary version of the package and made it available as ftp://franz.stat.wisc.edu/pub/Debian/xlispstat_3.50-1_i386.deb There is also a .rpm file for RedHat systems in that directory. The .rpm file was created from the .deb file using the "alien" program on Debian so it may look a little strange to RedHat users. I don't know. I don't use RedHat myself. If you are a Debian user, I would appreciate it if you could try to install the package and report back to me on any problems that you encounter. This is the first time that I have created a Debian package and I trust that there are several gliches for me to repair. If you have read this far and do not know what "Debian" is, check out http://www.debian.org/ Debian is a Linux distribution, i.e. a way of installing Linux on your computer. Generally people who try several different distributions end up preferring the Debian or the RedHat distribution. Both Debian and RedHat have powerful management systems for installing, removing and upgrading applications software and even system software. Those in the Debian camp believe that the Debian tools are superior. -- Douglas Bates bates@stat.wisc.edu Statistics Department 608/262-2598 University of Wisconsin - Madison http://www.stat.wisc.edu/~bates/ From majordom@umnstat.stat.umn.edu Wed Sep 10 17:07:57 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA20462; Wed, 10 Sep 1997 17:07:57 -0500 Return-Path: Received: from wara.bolnet.bo by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA20441; Wed, 10 Sep 1997 17:07:46 -0500 Received: from pc.bolnet.bo (cl35.entelnet.bo [166.114.10.67]) by wara.bolnet.bo (8.8.5/8.8.5) with SMTP id SAA07595; Wed, 10 Sep 1997 18:11:18 +0400 Message-Id: <3414427D.5B6E@wara.bolnet.bo> Date: Mon, 08 Sep 1997 14:22:53 -0400 From: kjetil halvorsen X-Mailer: Mozilla 3.0Gold (Win95; I) Mime-Version: 1.0 To: Jan de Leeuw Cc: Forrest Young , stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Testreport about mathematical/statistical software for data analysis ! (fwd) References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Jan de Leeuw wrote: > >OLE support ? (only in the spanish version) What is tis? where? its a joke? Kjetil. From majordom@umnstat.stat.umn.edu Wed Sep 10 17:37:17 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA20759; Wed, 10 Sep 1997 17:37:17 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA20755; Wed, 10 Sep 1997 17:37:09 -0500 Received: from [164.67.20.96] (ts17-3.wla.ts.ucla.edu) by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA13367; Wed, 10 Sep 97 15:36:42 PDT X-Sender: deleeuw@laplace.stat.ucla.edu Message-Id: In-Reply-To: <3414427D.5B6E@wara.bolnet.bo> References: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 10 Sep 1997 15:36:31 -0700 To: kjetil halvorsen , Jan de Leeuw From: Jan de Leeuw Subject: Re: Testreport about mathematical/statistical software for data analysis ! (fwd) Cc: Forrest Young , stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk At 2:22 PM -0400 9/8/97, kjetil halvorsen wrote: >Jan de Leeuw wrote: > >> >OLE support ? (only in the spanish version) > >What is tis? where? its a joke? > >Kjetil. Yes. Joke. Sorry about that --- Jan --- Jan de Leeuw; Professor and Chair, Interdivisional Program in Statistics; US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw ============================================================================ No matter where you go, there you are. --- Buckaroo Banzai ============================================================================ From majordom@umnstat.stat.umn.edu Thu Sep 11 19:22:49 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05671; Thu, 11 Sep 1997 19:22:49 -0500 Return-Path: Received: from maria.physics.lsa.umich.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05667; Thu, 11 Sep 1997 19:22:42 -0500 Received: by maria.physics.lsa.umich.edu (1.38.193.4/16.2) id AA12854; Thu, 11 Sep 1997 20:23:56 -0400 Date: Thu, 11 Sep 1997 20:23:56 -0400 (EDT) From: Rick Riolo Reply-To: rlriolo@umich.edu To: xlispstat Subject: simple scatterplot labeling questions Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I am new to xlispstat (at least in terms of trying to become a regular user), and I must be missing something basic about creating simple line plots with labels. What I would like is to create plots with 1, 2 or a few variables plotted vs time on the x axes, with a "legend" of sorts that indicates what each line corresponds to. From a quick read of the book, I thought that sending a :variable-label message is how to do it, but I tried: > (setf tList '( 1 2 3 4 5 6 7 8 9 10 )) (1 2 3 4 5 6 7 8 9 10) > (setf pList '( 2 4 6 8 10 12 14 16 18 20 )) (2 4 6 8 10 12 14 16 18 20) > (setf s (plot-points tList pList)) # > (send s :showing-labels t) T > (send s :variable-label 0 "fitness") "fitness" > and although the line plot appears (with one line in this case), no variable label ("fitness") appears. Can anyone please tell me what I am not doing that I should be, or how I should be doing this differently? I also would like to put a title in the graph (not on the window), and text labels on the axes (in addition to the numeric labels for the tick marks that one gets for free with the plot-points function). One way I got a title in the plot was with an overlay (similar to the description on pp286+). That worked fine for the title (e.g., it redraws and resizes correctly), and I can see how I could do the same for axes labels, but I wondered if a) there is an easier way to do it (eg some methods I am not aware of), or b) there there are libraries of user supplied codes out there that provide this kind of functionality? Thanks for any help and pointers you can provide. - r Rick Riolo rlriolo@umich.edu Program for Study of Complex Systems (PSCS) 4068 Randall Lab University of Michigan Ann Arbor MI 48109-1120 http://pscs.physics.lsa.umich.edu/PEOPLE/rlr-home.html From majordom@umnstat.stat.umn.edu Fri Sep 12 09:55:33 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA16871; Fri, 12 Sep 1997 09:55:33 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA16867; Fri, 12 Sep 1997 09:55:27 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa22901; 12 Sep 97 10:55 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA16771; Fri, 12 Sep 1997 10:55:23 -0400 Date: Fri, 12 Sep 1997 10:55:22 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: rlriolo@umich.edu Cc: xlispstat Subject: Re: simple scatterplot labeling questions In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Thu, 11 Sep 1997, Rick Riolo wrote: > > I am new to xlispstat (at least in terms of trying > to become a regular user), and I must be > missing something basic about creating simple line plots > with labels. ... > > (send s :variable-label 0 "fitness") > "fitness" > > > > and although the line plot appears (with one line in this case), > no variable label ("fitness") appears. > > Can anyone please tell me what I am not doing that I should be, > or how I should be doing this differently? See: > ( send p :help :y-axis ) Y-AXIS Method args: (&optional showing labeled ticks) Sets or retrieves current acis label state. SHOWING and LABELED should be true or NIL; TICKS should be a number. All three should be supplied for setting a new state. A list of the three properties is returned. -- the same for :x-axis. You need to set the second arg: ( let (( xa ( send s :x-axis )) ( ya ( send s :y-axis ))) ( send s :x-axis ( first xa ) T ( third xa )) ( send s :y-axis ( first ya ) T ( third ya ))) The variable labels will then appear next to the axis. > I also would like to put a title in the graph (not on the window), > and text labels on the axes (in addition to the numeric labels > for the tick marks that one gets for free with the plot-points > function). One way I got a title in the plot was with an overlay > (similar to the description on pp286+). > That worked fine for the title (e.g., it redraws and resizes correctly), > and I can see how I could do the same for axes labels, but I wondered if > a) there is an easier way to do it > (eg some methods I am not aware of), or > b) there there are libraries of user supplied codes > out there that provide this kind of functionality? > I have a bunch of graphics classes in: and partially documented and annotated in: which add some extensions to scatterplot objects and overlay objects. I don't know if any of them will do exactly what you want, but it might be a source to learn some of the not-as-well documented graphics tricks. I do have a text overlay class that uses real coordinates for it's placement -- this was meant so that I could automatically label spectral peaks in a line graph -- using a point label wouldn't work without a point, although I suppose I could have put in a "phantom" point to attach the label to. ( This is in realgraf.lsp ) You might also look at backgrid.lsp, which handles background grid lines. Not what you're asking for, but the code uses some axis attributes to figure out where to draw the lines, and it adds some methods and attributes to scatterplots to do this if GRID-COLOR is set. i.e. it might be a good example to look at if you want to do custom axes. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Fri Sep 12 11:06:11 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17292; Fri, 12 Sep 1997 11:06:11 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17288; Fri, 12 Sep 1997 11:06:05 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id ab05390; 12 Sep 97 12:06 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA20721; Fri, 12 Sep 1997 12:06:01 -0400 Date: Fri, 12 Sep 1997 12:05:59 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: rlriolo@umich.edu Cc: xlispstat Subject: Re: simple scatterplot labeling questions In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk BTW: On Fri, 12 Sep 1997, I wrote: > > I do have a text overlay class that uses real coordinates for it's > placement -- this was meant so that I could automatically label > spectral peaks in a line graph -- using a point label wouldn't > work without a point, although I suppose I could have put in a > "phantom" point to attach the label to. ( This is in realgraf.lsp ) > although the text labels use real graph coordinates for placement, there is a magic interpretation of #.positive-infinity and #.negative-infinity as up-against-the-(top,right,bottom,left) edge of the window. So if you want to label features that are at some set of real x coordinates in the graph, but want the labels out of the way in the upper edge of the window, use : (send plot :add-overlay (label text real-x #.positive-infinity :color 'red :just :below)) [ I'm not sure that the use of just keywords makes obvious sense -- I had to try it myself just now to see what works -- a sign that it's not obvious what it's in reference to. I'll have to change this and document it better! ] ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Fri Sep 12 13:01:46 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA17835; Fri, 12 Sep 1997 13:01:46 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA17831; Fri, 12 Sep 1997 13:01:40 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa10461; 12 Sep 97 14:01 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA20768; Fri, 12 Sep 1997 14:01:37 -0400 Date: Fri, 12 Sep 1997 14:01:36 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: rlriolo@umich.edu Cc: xlispstat Subject: Oops! [Re: simple scatterplot labeling questions] In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk While giving it a test, I found one bug in those label classes which may be related to MacOS 8 -- I don't have a 7.x machine nearby to test if this bug was in the old version and I just missed it, or if it's new behaviour: Running XlispStat under MacOS 8, on graphics windows, even without scrollbars, there is a region of the window where the scroll bars would be, on the right hand side, with no content. ( The same thing doesn't happen on the bottom for some reason.) labels placed with an X coordinate of #.positive-infinity, will be justified against the right edge of the window, and will be obscured by this content free region. The placement code should be modified to subtract some small value for this scroll area. ( I'll fix that in the online code Real Soon! ) ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Mon Sep 15 08:53:37 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA22371; Mon, 15 Sep 1997 08:53:37 -0500 Return-Path: Received: from maria.physics.lsa.umich.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA22367; Mon, 15 Sep 1997 08:53:31 -0500 Received: by maria.physics.lsa.umich.edu (1.38.193.4/16.2) id AA15366; Mon, 15 Sep 1997 09:54:54 -0400 Date: Mon, 15 Sep 1997 09:54:54 -0400 (EDT) From: Rick Riolo Reply-To: rlriolo@umich.edu To: xlispstat Subject: unix process number into xlispstat var Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Is there a way to get the unix process number of the running xlispstat into an list variable in that process? (For example, perl has a special variable with that value.) Or more generally is there a way to capture the stdout stream from a job started with (system ) For example, capture each line as the value of a entry in a list of strings, one string for each output line. That would open up a lot of possibilites. thanks. - r Rick Riolo rlriolo@umich.edu Program for Study of Complex Systems (PSCS) 4068 Randall Lab University of Michigan Ann Arbor MI 48109-1120 http://pscs.physics.lsa.umich.edu/PEOPLE/rlr-home.html From majordom@nokomis.stat.umn.edu Mon Sep 15 10:23:19 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA13281; Mon, 15 Sep 1997 10:23:19 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA13275; Mon, 15 Sep 1997 10:23:10 -0500 From: Luke Tierney Message-Id: <9709151523.AA13275@nokomis.stat.umn.edu> Subject: Re: unix process number into xlispstat var To: rlriolo@umich.edu Date: Mon, 15 Sep 1997 10:23:10 -0500 (CDT) Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Rick Riolo" at Sep 15, 97 09:54:54 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1399 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk Rick Riolo wrote: > > > Is there a way to get the unix process number of > the running xlispstat into an list variable in that process? > (For example, perl has a special variable with that value.) > > Or more generally is there a way to capture the stdout > stream from a job started with > (system ) > For example, capture each line as the value of a > entry in a list of strings, one string for each output line. > That would open up a lot of possibilites. > thanks. > > - r > > Rick Riolo rlriolo@umich.edu > Program for Study of Complex Systems (PSCS) > 4068 Randall Lab University of Michigan > Ann Arbor MI 48109-1120 > http://pscs.physics.lsa.umich.edu/PEOPLE/rlr-home.html > > An undocumented feature of the current system command is that it takes a stream to use as output for its second argument. So or example, > (with-output-to-string (s) (system "date" s)) "Mon Sep 15 10:15:54 CDT 1997 " The feature is undocumented because it may change once I figure out a better way to do it, so rely on this with caution. luke -- Luke Tierney University of Minnesota Phone: 612-625-7843 School of Statistics Fax: 612-624-8868 206 Church Street email: luke@stat.umn.edu Minneapolis, MN 55455 USA WWW: http://www.stat.umn.edu From majordom@umnstat.stat.umn.edu Thu Sep 18 03:26:40 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03762; Thu, 18 Sep 1997 03:26:40 -0500 Return-Path: Received: from hermes.fundp.ac.be by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03758; Thu, 18 Sep 1997 03:26:31 -0500 Received: from thales.math.fundp.ac.be by hermes.fundp.ac.be; (5.65v3.2/1.1.8.2/14Dec95-0443PM) id AA11850; Thu, 18 Sep 1997 10:28:16 +0200 Received: from sturm.math.fundp.ac.be by thales.math.fundp.ac.be (5.65v3.2/1.1.10.5/14May97-0614PM) id AA18057; Thu, 18 Sep 1997 10:26:39 +0200 Message-Id: <3420E575.501A1A6A@fundp.ac.be> Date: Thu, 18 Sep 1997 10:25:25 +0200 From: Isabelle REGINSTER Organization: FUNDP - MATH - GRT X-Mailer: Mozilla 3.01Gold (X11; I; Linux 2.0.18 i686) Mime-Version: 1.0 To: xlispstat Subject: to write at the end of a file Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, I open a file with this command : (set f1 (open "myfile" :direction :output)) I would like to write at the end of the file and no to overwritten. How can I do ? (I'm a beginner in xlispstat) Thanks in advance and have a nice day. Isabelle -- -------------------------------------- Isabelle REGINSTER Facultes Universitaires ND de la Paix Departement de Mathematique Groupe de Recherche sur les Transports (Pr. Ph.Toint) Rempart de la Vierge, 8 B.5000 Namur (Belgium) tel : 32-81-72.49.23 fax : 32-81-72.49.14 isabelle.reginster@fundp.ac.be -------------------------------------- From majordom@umnstat.stat.umn.edu Thu Sep 18 04:50:14 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA03908; Thu, 18 Sep 1997 04:50:14 -0500 Return-Path: Received: from zam149.zam.kfa-juelich.de by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA03904; Thu, 18 Sep 1997 04:50:06 -0500 Received: from hlrz9-f by fz-juelich.de (5.65v3.2/DEC-AXP) id AA25068; Thu, 18 Sep 1997 11:49:58 +0200 Received: from hlrz4.hlrz.kfa-juelich.de by hlrz9.hlrz.kfa-juelich.de (SMI-8.6/SMI-4.1) id LAA03411; Thu, 18 Sep 1997 11:49:55 +0200 Reply-To: jmu@hlrz9.hlrz.kfa-juelich.de Received: by hlrz4.hlrz.kfa-juelich.de (SMI-8.6/SMI-4.1) id LAA22068; Thu, 18 Sep 1997 11:49:57 +0200 Date: Thu, 18 Sep 1997 11:49:49 +0200 Message-Id: <7458-Thu18Sep1997114949+0200-jmu@hlrz.kfa-juelich.de> X-Mailer: emacs 19.34.1 (via feedmail 7 I) From: jmu@hlrz9.hlrz.kfa-juelich.de (Johannes Mueller-Gerking) To: majordom@umnstat.stat.umn.edu Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <3420E575.501A1A6A@fundp.ac.be> (message from Isabelle REGINSTER on Thu, 18 Sep 1997 10:25:25 +0200) Subject: Re: to write at the end of a file Content-Type: text Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk > I open a file with this command : > > (set f1 (open "myfile" :direction :output)) > > I would like to write at the end of the file and no to overwritten. > How can I do ? (I'm a beginner in xlispstat) For example: (with-open-file (of "myfile" :direction :output :if-does-not-exist :create :if-exists :append) OPEN has the same keyword arguments than with-open-file. Best Johannes ----------------------------------------------------------------------- Johannes Mueller-Gerking, PhD email : jmu@hlrz.kfa-juelich.de FZ Juelich, HLRZ 52425 Juelich Precode : [+49] (2461) office : 61-2318 Germany home : 346572 FAX : 61-2430 From majordom@umnstat.stat.umn.edu Thu Sep 18 14:24:18 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06394; Thu, 18 Sep 1997 14:24:18 -0500 Return-Path: Received: from amber.ccs.neu.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06390; Thu, 18 Sep 1997 14:24:11 -0500 Received: from RISA ([192.127.136.202] (may be forged)) by amber.ccs.neu.edu (8.8.6/8.7.3) with SMTP id PAA27121 for ; Thu, 18 Sep 1997 15:24:06 -0400 (EDT) Received: by RISA with Microsoft Mail id <01BCC443.BFD98D90@RISA>; Thu, 18 Sep 1997 15:01:32 -0400 Message-Id: <01BCC443.BFD98D90@RISA> From: psksvp To: "'Xlisp-Stat List'" Subject: calling naive c function Date: Thu, 18 Sep 1997 15:01:27 -0400 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk hi Does anyone know, how to implement a call to a native c function from xlisp-stat? psksvp@ccs.neu.edu From majordom@umnstat.stat.umn.edu Thu Sep 18 16:42:16 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA07052; Thu, 18 Sep 1997 16:42:16 -0500 Return-Path: Received: from neopath.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA07048; Thu, 18 Sep 1997 16:42:10 -0500 Received: from denali.neopath.com by neopath.com (SMI-8.6/SMI-SVR4) id OAA17882; Thu, 18 Sep 1997 14:34:54 -0700 Received: from pilchuck.neopath.com by denali.neopath.com (SMI-8.6/SMI-SVR4) id OAA05709; Thu, 18 Sep 1997 14:38:21 -0700 Received: by pilchuck.neopath.com (5.x/SMI-SVR4) id AA22643; Thu, 18 Sep 1997 14:35:51 -0700 Date: Thu, 18 Sep 1997 14:35:51 -0700 From: keith@neopath.com (Keith Frost) Message-Id: <9709182135.AA22643@pilchuck.neopath.com> To: psksvp@ccs.neu.edu Cc: stat-lisp-news@umnstat.stat.umn.edu In-Reply-To: <01BCC443.BFD98D90@RISA> (message from psksvp on Thu, 18 Sep 1997 15:01:27 -0400) Subject: Re: calling naive c function Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk hi Does anyone know, how to implement a call to a native c function from xlisp-stat? psksvp@ccs.neu.edu The functions you need are "dyn-load" and "call-cfun". There is some on-line documentation of these. Also you can look in the xlispstat book. The way to compile c code for dynamic loading varies between platforms -- consult the machine-dependent documentation in the xlispstat source distribution for details. From majordom@umnstat.stat.umn.edu Fri Sep 19 04:46:48 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA18024; Fri, 19 Sep 1997 04:46:48 -0500 Return-Path: Received: from anorien.mcs.surrey.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA18020; Fri, 19 Sep 1997 04:46:41 -0500 Received: from godel.ee.surrey.ac.uk by anorien.mcs.surrey.ac.uk with smtp (Smail3.1.29.1-ident) id m0xBze4-000S8cC; Fri, 19 Sep 97 10:46 BST Received: by godel.ee.surrey.ac.uk (Smail3.1.29.1 #8) id m0xBze2-0002nbC; Fri, 19 Sep 97 10:46 BST Message-Id: From: map1rs@ee.surrey.ac.uk (Richard Stevens) Subject: Re: calling naive c function To: stat-lisp-news@umnstat.stat.umn.edu Date: Fri, 19 Sep 1997 10:46:38 +0100 (BST) In-Reply-To: <9709182135.AA22643@pilchuck.neopath.com> from "Keith Frost" at Sep 18, 97 02:35:51 pm X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 2240 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk > > Does anyone know, how to implement a call to a native c function from > xlisp-stat? > > psksvp@ccs.neu.edu > > The functions you need are "dyn-load" and "call-cfun". There is some > on-line documentation of these. Also you can look in the xlispstat > book. The way to compile c code for dynamic loading varies between > platforms -- consult the machine-dependent documentation in the > xlispstat source distribution for details. Hi. In addition to the book and the on-line help, here are some extra notes that clarify a couple of points that aren't mentioned in the book. These are just notes I wrote for our system here - I don't know how universally they apply. I assume a UNIX setup with GNU's gcc compiler. Hope they help a little. Any corrections to this material welcome! Richard Stevens University of Surrey Notes on Binding C Functions into Xlisp-stat -------------------------------------------- Bear in mind that Xlisp-stat only offers dynamic loading of C functions, not static loading. Then: 1. If you recompile your object files, you must quit Xlisp-stat and re-enter it before loading them again. (dyn-load "myfunction.o") will _not_ overload the routine with the new version. 2. Although Luke Tierney's book points out that you can load libraries such as -lm for the maths library etc, I haven't been able to work that with libraries I've written myself. If that's right and not just my error, then if your c code needs external functions, you cannot make a library and then use the :libflags option. You must make a .so file instead, as follows: ld -dy -G myfunction.o mylibrary.lib -lm -o myfunction.so (You probably could leave the -lm for the :libflags option to cope with, but you can't do that with "mylibrary.lib".) The -dy option specifies dynamic instead of static code. -G specifies a shared object. You can then use (dyn-load "myfunction.so") and (call-cfun "myfunction" arg1 arg2 etc) as described in the manual. 3. You _do_ have to take note of argument types - unusually for Xlisp-stat. For example, you could write code where all the arguments are doubles, and forcing Xlisp-stat to represent your data as floats e.g. by writing 1.0 instead of 1. 10th Feb 1997 From majordom@umnstat.stat.umn.edu Fri Sep 19 10:58:58 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA19536; Fri, 19 Sep 1997 10:58:58 -0500 Return-Path: Received: from sulu.biostat.washington.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA19532; Fri, 19 Sep 1997 10:58:54 -0500 Received: from troi by sulu.biostat.washington.edu (8.7.1/UW-NDC Revision: 2.23 † id IAA18018; Fri, 19 Sep 1997 08:58:52 -0700 (PDT) Date: Fri, 19 Sep 1997 08:58:50 -0700 (PDT) From: Thomas Lumley X-Sender: thomas@troi To: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: calling naive c function In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Fri, 19 Sep 1997, Richard Stevens wrote: > You must make a .so file instead, as follows: > > ld -dy -G myfunction.o mylibrary.lib -lm -o myfunction.so > > (You probably could leave the -lm for the :libflags option to cope > with, but you can't do that with "mylibrary.lib".) > > The -dy option specifies dynamic instead of static code. -G specifies a > shared object. Also note that the proper incantation for creating shared objects can vary between different varieties of Unix and ld. Thomas Lumley ------------------------------------------------------+------ Biostatistics : "Never attribute to malice what : Uni of Washington : can be adequately explained by : Box 357232 : incompetence" - Hanlon's Razor : Seattle WA 98195-7232 : : ------------------------------------------------------------ From majordom@umnstat.stat.umn.edu Sat Sep 20 09:41:02 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02804; Sat, 20 Sep 1997 09:41:02 -0500 Return-Path: Received: from amber.ccs.neu.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02800; Sat, 20 Sep 1997 09:40:57 -0500 Received: from RISA (email.egat.or.th [202.44.210.41]) by amber.ccs.neu.edu (8.8.6/8.7.3) with SMTP id KAA01995 for ; Sat, 20 Sep 1997 10:40:38 -0400 (EDT) Received: by RISA with Microsoft Mail id <01BCC5AE.85930F00@RISA>; Sat, 20 Sep 1997 10:18:21 -0400 Message-Id: <01BCC5AE.85930F00@RISA> From: psksvp To: "'Xlisp-Stat List'" Subject: RE: calling naive c function Date: Sat, 20 Sep 1997 10:18:15 -0400 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi Thanks for all your answers, but I am using xlisp-stat on a pc running NT4. I don't seem to find any infos about this topic in the source distribution. Have any of you, aven done this on pc version of xlisp-stat? thanks in advance psksvp@ccs.neu.edu ---------- From: Thomas Lumley Sent: Friday, September 19, 1997 11:58 AM To: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: calling naive c function On Fri, 19 Sep 1997, Richard Stevens wrote: > You must make a .so file instead, as follows: > > ld -dy -G myfunction.o mylibrary.lib -lm -o myfunction.so > > (You probably could leave the -lm for the :libflags option to cope > with, but you can't do that with "mylibrary.lib".) > > The -dy option specifies dynamic instead of static code. -G specifies a > shared object. Also note that the proper incantation for creating shared objects can vary between different varieties of Unix and ld. Thomas Lumley ------------------------------------------------------+------ Biostatistics : "Never attribute to malice what : Uni of Washington : can be adequately explained by : Box 357232 : incompetence" - Hanlon's Razor : Seattle WA 98195-7232 : : ------------------------------------------------------------ From majordom@umnstat.stat.umn.edu Wed Sep 24 09:14:49 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA21502; Wed, 24 Sep 1997 09:14:49 -0500 Return-Path: Received: from anorien.mcs.surrey.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA21498; Wed, 24 Sep 1997 09:14:42 -0500 Received: from aragorn.ee.surrey.ac.uk by anorien.mcs.surrey.ac.uk with smtp (Smail3.1.29.1-ident) id m0xDsD7-000SARC; Wed, 24 Sep 97 15:14 BST Received: by aragorn.ee.surrey.ac.uk (Smail3.1.29.1 #8) id m0xDsD6-000406C; Wed, 24 Sep 97 15:14 BST Message-Id: From: map1rs@ee.surrey.ac.uk (Richard Stevens) Subject: Axis labels To: stat-lisp-news@umnstat.stat.umn.edu Date: Wed, 24 Sep 1997 15:14:36 +0100 (BST) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 497 Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, all. May I ask: does Lisp-stat let us control the tick marks on a scatterplot? My x data runs from 1-9. For visual reasons I don't want the x-axis going through the point x=1, for example, so I'm using :range 0 0 10 to give some extra room at each side: but that causes tick marks to appear for 0 and 10, which are entirely meaningless co-ordinates for this data. Is there some way I can stretch the range while leaving the outer tick marks at 1 and 9? Many thanks, Richard Stevens From majordom@nokomis.stat.umn.edu Wed Sep 24 10:18:10 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA20329; Wed, 24 Sep 1997 10:18:10 -0500 Return-Path: Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA21681; Wed, 24 Sep 1997 09:37:21 -0500 Date: Wed, 24 Sep 1997 09:37:21 -0500 Message-Id: <9709241437.AA21681@umnstat.stat.umn.edu> To: owner-stat-lisp-news@nokomis.stat.umn.edu From: owner-stat-lisp-news@nokomis.stat.umn.edu Subject: BOUNCE stat-lisp-news@stat.umn.edu: Admin request of type /\buns\w*b/i at line 4 Status: RO Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk >From bates@stat.wisc.edu Wed Sep 24 09:37:12 1997 Received: from franz.stat.wisc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA21673; Wed, 24 Sep 1997 09:37:12 -0500 Return-Path: Received: by franz.stat.wisc.edu id m0xDsYq-000hrZC (Debian Smail-3.2 1996-Jul-4 #2); Wed, 24 Sep 1997 09:37:04 -0500 (CDT) Sender: bates@stat.wisc.edu (Douglas Bates) Sender: bates@franz.stat.wisc.edu To: stat-lisp-news@umnstat.stat.umn.edu Subject: xlispstat-3.50 now an official part of Debian GNU/Linux From: Douglas Bates Date: 24 Sep 1997 09:37:04 -0500 In-Reply-To: map1rs@ee.surrey.ac.uk's message of Wed, 24 Sep 1997 15:14:36 +0100 (BST) Message-Id: <6rraaen5gv.fsf@franz.stat.wisc.edu> Lines: 10 X-Mailer: Gnus v5.4.37/XEmacs 19.15 The Debian distribution of the Linux operating system (see http://www.debian.org/) now includes xlispstat-3.50 as one of the packages available in the "math" section of the distribution. At present this package is only available for the "unstable" distribution which is the development version of Debian 2.0. I will also try to provide the package for the "stable" (Debian 1.3.1) version. -- Douglas Bates bates@stat.wisc.edu Statistics Department 608/262-2598 University of Wisconsin - Madison http://www.stat.wisc.edu/~bates/ From majordom@umnstat.stat.umn.edu Wed Sep 24 13:36:16 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23486; Wed, 24 Sep 1997 13:36:16 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23482; Wed, 24 Sep 1997 13:36:09 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa21553; 24 Sep 97 14:35 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA17354; Wed, 24 Sep 1997 14:35:10 -0400 Date: Wed, 24 Sep 1997 14:35:09 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Richard Stevens Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Axis labels In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Wed, 24 Sep 1997, Richard Stevens wrote: > > Hello, all. May I ask: does Lisp-stat let us control the tick marks on a > scatterplot? My x data runs from 1-9. For visual reasons I don't want the > x-axis going through the point x=1, for example, so I'm using > > :range 0 0 10 > > to give some extra room at each side: but that causes tick marks to appear > for 0 and 10, which are entirely meaningless co-ordinates for this data. > Is there some way I can stretch the range while leaving the outer tick marks > at 1 and 9? It's not as easy as it *ought* to be, but it's not TOO difficult. The routine that draw the axis and tickmarks is part of the Iview graphics package -- i.e. the part written in C underneath XlispStat. The :x-axis & :y-axis scatterplot methods just set parameters that are used by this code. There is no lisp visible :redraw-axes method you could override to change the behavious, however, you can back up a bit and redefine :redraw-background. The default method is also implemented in C, but if you redefine that method, it will be called instead. I redefine :redraw-background in so that I can draw colored grid lines -- palegreen :grid-color and mintcream :back-color makes a nice graph-paper effect -- but I still want the default axes draw, so it does a (call-next-method) so that the default :redraw-background is also called. The default method in C: VOID IViewStdRedrawBackground P1C(IVIEW_WINDOW, w) { int left, top, height, width; StGWWinInfo *gwinfo = IViewWindowWinInfo(w); StGWStartBuffering(gwinfo); if (IViewMouseMode(w) == brushing) IViewEraseBrush(w); StGWGetViewRect(gwinfo, &left, &top, &width, &height); StGWSetClipRect(gwinfo, TRUE, left, top, width, height); StGWEraseRect(gwinfo, left, top, width, height); IViewDrawAxes(w); StGWBufferToScreen(gwinfo, left, top, width, height); if (IViewMouseMode(w) == brushing) IViewDrawBrush(w); } If you skip the MouseMode part for now, basically clears the window and sets it to the background color ( which you can do from lisp by calling the windows :erase-window method ) and draws the axes. If you look at my backgrid.lsp code, it replicates some of the logic of the axes drawing routine. It gets the same axes parameters to draw the extended grid lines at the same points as the tic marks on the graph. If you change the graph axes parameters ( or if something else changes it automatically ) the grid lines will follow the tics. You could start from that code and modify it to: [1] allow setting of a different initial offset or setting of a list of tic coordinates; [2] draw tiny ticks instead of long grid lines. [3] take the (call-next-method) out of :redraw-background and insert :erase-window and a call to your axes method. ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Wed Sep 24 14:43:44 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA23981; Wed, 24 Sep 1997 14:43:44 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA23977; Wed, 24 Sep 1997 14:43:39 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa08123; 24 Sep 97 15:43 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA12654; Wed, 24 Sep 1997 15:43:25 -0400 Date: Wed, 24 Sep 1997 15:43:24 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU Reply-To: "Steven D. Majewski" To: Xlisp List Subject: XlispStat, dll's, foreign functions, native-objects & SWIG... Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I've recently had to interface what's been basically the same shared library (growing and changing over time) on the Mac, first to XlispStat, then to NIH Image, and now to Python. I'm doing the Python mostly because the XlispStat dll has a pretty "thin" interface -- it's not easy to code to a complex foreign function interface. A couple of observations and questions: It ought to be possible to export the wrapper functions to convert C data types to/from Lisp objects. i.e. build much of XlispStat as a core shared library, which the XlispStat application and other dynamic/shared library code can all link to. ( This is what Python does ). Shared library code could then use that XLisp API for wrapping and unwrapping args. Is this collection of wrapper functions documented anywhere ? Is there any info on how to add an new hardware-object written in C to the object system ? I have been learning how to use SWIG -- Simple Wrapper and Interface Generator to generate the "glue code" to make my functions Python callable. This is a very nice tool -- it takes an interface description file, or sometimes it can simply take an .h file, and it generates all of the necessary glue code to make the functions callable from Python, Perl, tcl, or guile. It is possible to add new language interfaces and extensions. (But I haven't gotten so far as to know how difficult this is.) Has anyone familiar with the XlispStat internals looked at this tool ? ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Wed Sep 24 15:27:18 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24222; Wed, 24 Sep 1997 15:27:18 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24218; Wed, 24 Sep 1997 15:27:10 -0500 Received: from [164.67.25.68] (ts32-7.wla.ts.ucla.edu) by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA10990; Wed, 24 Sep 97 13:26:12 PDT X-Sender: deleeuw@laplace.stat.ucla.edu Message-Id: In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 24 Sep 1997 13:23:26 -0700 To: "Steven D. Majewski" , Xlisp List From: Jan de Leeuw Subject: Re: XlispStat, dll's, foreign functions, native-objects & SWIG... Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk At 3:43 PM -0400 9/24/97, Steven D. Majewski wrote: >I've recently had to interface what's been basically the same shared >library (growing and changing over time) on the Mac, first to XlispStat, >then to NIH Image, and now to Python. I'm doing the Python mostly >because the XlispStat dll has a pretty "thin" interface -- it's not >easy to code to a complex foreign function interface. > >A couple of observations and questions: > > It ought to be possible to export the wrapper functions to convert >C data types to/from Lisp objects. i.e. build much of XlispStat as >a core shared library, which the XlispStat application and other >dynamic/shared library code can all link to. ( This is what Python >does ). Shared library code could then use that XLisp API for wrapping >and unwrapping args. > It would be great to have XlispStat in library form (similar to libtcl or libphp or libsiod or libguile). I think the slisp project at the University of Washington (http://betz.biostr.washington.edu/slisp/) did do something similar for Xlisp. This would make it possible, for instance, to have an XlispStat Apache module, and maybe an NSAPI XlispStat plugin, with not too much work. Also calling XlispStat from Java etc would be easier. And we can have embedded Xlisp interpreters in our C programs. With the new department and all, I have absolutely zero time to work on this, but I'll try to get some grad students interested. Also, what is the status of replacing code resources on the Mac with shared libraries. Is there a version for 3.50 ? --- Jan de Leeuw; Professor and Chair, Interdivisional Program in Statistics; US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw ============================================================================ No matter where you go, there you are. --- Buckaroo Banzai ============================================================================ From majordom@nokomis.stat.umn.edu Wed Sep 24 15:54:47 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA20676; Wed, 24 Sep 1997 15:54:47 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA20671; Wed, 24 Sep 1997 15:54:44 -0500 Received: from franz.stat.wisc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24361; Wed, 24 Sep 1997 15:38:34 -0500 Received: by franz.stat.wisc.edu id m0xDyCg-000hrZC (Debian Smail-3.2 1996-Jul-4 #2); Wed, 24 Sep 1997 15:38:34 -0500 (CDT) To: stat-lisp-news@nokomis.stat.umn.edu Subject: More on Linux packages of xlispstat From: Douglas Bates Date: 24 Sep 1997 15:38:34 -0500 Message-Id: <6rg1qumoqd.fsf@franz.stat.wisc.edu> Lines: 28 X-Mailer: Gnus v5.4.37/XEmacs 19.15 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk I'm not sure if my previous message about a Debian GNU/Linux package for xlispstat-3.50 made it to the entire list. I got an unusual message about it bouncing. In case it didn't make it, here is a synopsis. xlispstat is now an official part of the Debian distribution of Linux. See http://www.debian.org/ for more information on Debian Linux. The xlispstat-3.50 package is in the math subsection of the "unstable" (a.k.a. "hamm") distribution. This is the development version for Debian 2.0. If you are running Debian 1.3 (a.k.a. "stable" or "bo") you can pick up a copy from ftp://franz.stat.wisc.edu/pub/Debian/bo A .rpm package for RedHat Linux (possibly also suitable for Caldera or SuSE, I don't know) is in the directory ftp://franz.stat.wisc.edu/pub/Debian I would appreciate feedback on successes (or otherwise) in the installation process. P.S. If you are wondering about the names "hamm" and "bo", the code names for various releases of Debian are names of characters from the movie "Toy Story". Bruce Perens, the director of the Debian project, works for Pixar. -- Douglas Bates bates@stat.wisc.edu Statistics Department 608/262-2598 University of Wisconsin - Madison http://www.stat.wisc.edu/~bates/ From majordom@umnstat.stat.umn.edu Wed Sep 24 16:04:51 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24630; Wed, 24 Sep 1997 16:04:51 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24626; Wed, 24 Sep 1997 16:04:46 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa01829; 24 Sep 97 17:04 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA26270; Wed, 24 Sep 1997 17:04:27 -0400 Date: Wed, 24 Sep 1997 17:04:26 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Jan de Leeuw Cc: Xlisp List Subject: Re: XlispStat, dll's, foreign functions, native-objects & SWIG... In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Wed, 24 Sep 1997, Jan de Leeuw wrote: > Also, what is the status of replacing code resources on the Mac > with shared libraries. Is there a version for 3.50 ? I'm routinely using a version of 3.50 built with Bob Stine's patch for supporting shared libraries. However, since that code completely changes the function of #'open-resource-file, it is not compatible with code that might actually need to use something from a resource file, such as a cursor bitmap. It really needs to be renamed to keep both functions separate before it's a candidate for incorporating into the standard distribution. ( Something I was considering doing, however all of these other problems and concerns led me to stop and think about whether there is a better way. Besides SWIG, I've been experimenting a tiny bit with COM on the Mac -- which now gets installed on your Mac with MS Internet Explorer's Active-X support. ) It obviously ought to be a portable interface -- even if the implementation is different. What function handles loading shared libraries on the other platforms ? ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Wed Sep 24 16:11:33 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24665; Wed, 24 Sep 1997 16:11:33 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24661; Wed, 24 Sep 1997 16:11:28 -0500 Received: from [164.67.22.44] (ts34-15.wla.ts.ucla.edu) by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA12592; Wed, 24 Sep 97 14:11:21 PDT X-Sender: deleeuw@laplace.stat.ucla.edu Message-Id: In-Reply-To: References: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 24 Sep 1997 14:11:18 -0700 To: "Steven D. Majewski" , Jan de Leeuw From: Jan de Leeuw Subject: Re: XlispStat, dll's, foreign functions, native-objects & SWIG... Cc: Xlisp List Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Another thing to consider might be the triple The Gimp/Net-Fu/Script-Fu. The GIMP is an image generating machine (a bit Photoshop like), Script-Fu is a scripting language to drive it (actually Scheme from the SIOD library, but there are versions in Python and Guile as well), and Net-Fu is a Java interface that replaces Script-Fu over the net. See http://www.XCF.Berkeley.EDU/~gimp/. At 5:04 PM -0400 9/24/97, Steven D. Majewski wrote: >On Wed, 24 Sep 1997, Jan de Leeuw wrote: > >> Also, what is the status of replacing code resources on the Mac >> with shared libraries. Is there a version for 3.50 ? > > >I'm routinely using a version of 3.50 built with Bob Stine's >patch for supporting shared libraries. However, since that >code completely changes the function of #'open-resource-file, >it is not compatible with code that might actually need to use >something from a resource file, such as a cursor bitmap. > >It really needs to be renamed to keep both functions separate >before it's a candidate for incorporating into the standard >distribution. ( Something I was considering doing, however >all of these other problems and concerns led me to stop and >think about whether there is a better way. Besides SWIG, I've >been experimenting a tiny bit with COM on the Mac -- which now >gets installed on your Mac with MS Internet Explorer's Active-X >support. ) > >It obviously ought to be a portable interface -- even if the >implementation is different. > >What function handles loading shared libraries on the other platforms ? > > >---| Steven D. Majewski (804-982-0831) |--- >---| Department of Molecular Physiology and Biological Physics |--- >---| University of Virginia Health Sciences Center |--- >---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- >All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson --- Jan de Leeuw; Professor and Chair, Interdivisional Program in Statistics; US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw ============================================================================ No matter where you go, there you are. --- Buckaroo Banzai ============================================================================ From majordom@umnstat.stat.umn.edu Sat Sep 27 04:58:59 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05551; Sat, 27 Sep 1997 04:58:59 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05547; Sat, 27 Sep 1997 04:58:53 -0500 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.7.5/10.1) id FAA12524; Sat, 27 Sep 1997 05:58:51 -0400 (GMT+4:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sat, 27 Sep 1997 05:59:53 -0500 To: stat-lisp-news@umnstat.stat.umn.edu From: Forrest Young Subject: Stem & Leaf Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi LispStat folks: Has anyone out there written a stem-and-leave plot? ----- Forrest Young Professor, Psychometrics UNC Psychometrics Lab e-mail: mailto:forrest@unc.edu CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu Chapel Hill NC 27599-3270 USA Voice: 1-919-962-5038 Fax: 1-919-304-2329 ViSta Software and Documentation site: http://forrest.psych.unc.edu/research/ USA Home Site ViSta Mirrors: http://ssda.anu.edu.au/mirrors Australia http://www.sci.usq.edu.au/general/vista Australia http://sunsite.univie.ac.at/ViSta Austria http://cons-dev.univ-lyon1.fr/vista France http://micros.hensa.ac.uk/mirrors/vista UK (ac.uk only 8AM-8PM GMT) http://www.psych.unc.edu/fwy USA From majordom@umnstat.stat.umn.edu Sat Sep 27 09:13:59 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA05793; Sat, 27 Sep 1997 09:13:59 -0500 Return-Path: Received: from julia.math.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA05789; Sat, 27 Sep 1997 09:13:54 -0500 Received: from [164.67.22.47] (ts35-2.wla.ts.ucla.edu) by julia.math.ucla.edu (Sendmail 4.1/1.13) id AA28652; Sat, 27 Sep 97 07:13:42 PDT X-Sender: deleeuw@laplace.stat.ucla.edu Message-Id: In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sat, 27 Sep 1997 07:13:41 -0700 To: Forrest Young , stat-lisp-news@umnstat.stat.umn.edu From: Jan de Leeuw Subject: Re: Stem & Leaf Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Archive http://www.stat.ucla.edu/develop/lisp/xlisp/xlisp-stat/code/ in statistics/plotting/stem-and-leaf At 5:59 AM -0500 9/27/97, Forrest Young wrote: >Hi LispStat folks: > >Has anyone out there written a stem-and-leave plot? > > >----- > >Forrest Young >Professor, Psychometrics > >UNC Psychometrics Lab e-mail: mailto:forrest@unc.edu >CB-3270 Davie Hall WWW: http://forrest.psych.unc.edu >Chapel Hill NC 27599-3270 USA Voice: 1-919-962-5038 > Fax: 1-919-304-2329 > >ViSta Software and Documentation site: >http://forrest.psych.unc.edu/research/ USA Home Site > >ViSta Mirrors: >http://ssda.anu.edu.au/mirrors Australia >http://www.sci.usq.edu.au/general/vista Australia >http://sunsite.univie.ac.at/ViSta Austria >http://cons-dev.univ-lyon1.fr/vista France >http://micros.hensa.ac.uk/mirrors/vista UK (ac.uk only 8AM-8PM GMT) >http://www.psych.unc.edu/fwy USA --- Jan de Leeuw; Professor and Chair, Interdivisional Program in Statistics; US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw ============================================================================ No matter where you go, there you are. --- Buckaroo Banzai ============================================================================ From majordom@umnstat.stat.umn.edu Sat Sep 27 11:38:19 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA06336; Sat, 27 Sep 1997 11:38:19 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA06332; Sat, 27 Sep 1997 11:38:13 -0500 Received: from [152.2.85.134] by gibbs.oit.unc.edu (8.7.5/10.1) id MAA09086; Sat, 27 Sep 1997 12:38:03 -0400 (GMT+4:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: In-Reply-To: References: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Sat, 27 Sep 1997 12:37:14 -0500 To: Erich Neuwirth From: Forrest Young Subject: Re: Stem & Leaf Cc: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Erich Neuwirth writes: >hello forrest, > >who else could ask for stem and leaf ;-) > >BUT > >let me start a discussion. >don't you think that stem and leaf is >something which has been designed for paper and pencil technology? >it essentially is a histogram with >restricted choices for the borderlines of the bins. > >i think, with coumputers perhaps we should other representation >methods. > >i use sliding histograms with variable window widths. >and changing the 'rectangle' counter to a function >immediately gives kernel deinsity estimates. > >erich True... they are from the paper and pencil days, and are probably easier to do that way then in Lisp! But... they are popular in intro psych (and other social sciences) stat courses, and I feel as though I need one to have a complete system. I'm glad Jan has one and I don't have to write it. Perhaps there is a way a computerized stem and leaf can be "featurized" to take advantage of the computer and provide more than just paper and pencil. Forrest From majordom@umnstat.stat.umn.edu Mon Sep 29 09:14:15 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00243; Mon, 29 Sep 1997 09:14:15 -0500 Return-Path: Received: from hermes.fundp.ac.be by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00239; Mon, 29 Sep 1997 09:14:07 -0500 Received: from thales.math.fundp.ac.be by hermes.fundp.ac.be; (5.65v3.2/1.1.8.2/14Dec95-0443PM) id AA22356; Mon, 29 Sep 1997 16:13:51 +0100 Received: from sturm.math.fundp.ac.be by thales.math.fundp.ac.be (5.65v3.2/1.1.10.5/14May97-0614PM) id AA12766; Mon, 29 Sep 1997 16:14:12 +0200 Message-Id: <342FB76B.25E0910D@fundp.ac.be> Date: Mon, 29 Sep 1997 16:12:59 +0200 From: Isabelle REGINSTER Organization: FUNDP - MATH - GRT X-Mailer: Mozilla 3.01Gold (X11; I; Linux 2.0.18 i686) Mime-Version: 1.0 To: xlispstat Subject: sort number and string in the same time Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hello, I want sort a list with numbers and strings. But the sort-data function doesn't work with numbers and strings in the single list. For example : > (def toto ( list 5 "g" 1 7 "a")) > (sort-data toto) Error: can't compare these Happened in: # How can I do this ? Thanks in advance. Isabelle PS : I am a beginner in xlispstat. -- -------------------------------------- Isabelle REGINSTER Facultes Universitaires ND de la Paix Departement de Mathematique Groupe de Recherche sur les Transports (Pr. Ph.Toint) Rempart de la Vierge, 8 B.5000 Namur (Belgium) tel : 32-81-72.49.23 fax : 32-81-72.49.14 isabelle.reginster@fundp.ac.be -------------------------------------- From majordom@umnstat.stat.umn.edu Mon Sep 29 10:29:25 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA00887; Mon, 29 Sep 1997 10:29:25 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA00880; Mon, 29 Sep 1997 10:29:13 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa05850; 29 Sep 97 11:28 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA16385; Mon, 29 Sep 1997 11:28:46 -0400 Date: Mon, 29 Sep 1997 11:28:45 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Isabelle REGINSTER Cc: xlispstat Subject: Re: sort number and string in the same time In-Reply-To: <342FB76B.25E0910D@fundp.ac.be> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Mon, 29 Sep 1997, Isabelle REGINSTER wrote: > Hello, > > I want sort a list with numbers and strings. But the sort-data function > doesn't work with numbers and strings in the single list. > > For example : > > > (def toto ( list 5 "g" 1 7 "a")) > > (sort-data toto) > Error: can't compare these > Happened in: # > > How can I do this ? The first question is what do you expect the results of sorting ( list 5 "g" 1 7 "a" ) to be ? all the strings sorted followed by all the ints ? or the other way around ? or strings falling in between ints according to the ordinal value of their ascii codes ? Which is greater, 5 or "g" ? The procedure #'sort takes a list and a predicate. #'sort-data is a simpler packaging of: ( sort number-seq #'< ) #for numbers and ( sort string-seq #'string< ) #for strings If you know what you want the sorted order of ( list 5 "g" 1 7 "a" ) to be, you can define a predicate that will allow it to be sorted, but there is no defined ordering of that list. ( Python, for another example, does allow sorting of mixed type lists by default, but it's defined as an arbitrary sorting of types, with each item of type sorted among itself. No really very meaningful, but it can still be useful to have a defined order. ) ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Mon Sep 29 19:23:04 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04215; Mon, 29 Sep 1997 19:23:04 -0500 Return-Path: Received: from angel.heaven.net by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04211; Mon, 29 Sep 1997 19:22:58 -0500 Received: from [198.69.29.97] (josh.netheaven.com [198.69.29.97]) by angel.heaven.net (8.8.5/8.8.5) with ESMTP id UAA15380 for ; Mon, 29 Sep 1997 20:23:08 -0400 (EDT) X-Sender: josh@netheaven.com Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Mon, 29 Sep 1997 20:27:37 -0400 To: From: Josh Kuperman Subject: Has anyone compiled Xlisp-stat using CodeWarrior Pro Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I decided to give it go and managed to get most things to compile but I have some linking problems. Also, I remember Steve Majewski mentioning using some CFM libraries in an earlier version and wonder if it makes more sense to use the CFM libraries than not. I'm on a 68k machine so if anyone knows anything special in that regard it might help. Josh Kuperman josh@racing.saratoga.ny.us http://www.racing.saratoga.ny.us/~josh From majordom@umnstat.stat.umn.edu Mon Sep 29 22:07:00 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA04777; Mon, 29 Sep 1997 22:07:00 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA04773; Mon, 29 Sep 1997 22:06:54 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa15096; 29 Sep 97 23:06 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA13445; Mon, 29 Sep 1997 23:06:49 -0400 Date: Mon, 29 Sep 1997 23:06:47 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Josh Kuperman Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: Has anyone compiled Xlisp-stat using CodeWarrior Pro In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Mon, 29 Sep 1997, Josh Kuperman wrote: > I decided to give it go and managed to get most things to compile but I have > some linking problems. Also, I remember Steve Majewski mentioning using > some CFM libraries in an earlier version and wonder if it makes more sense > to use the CFM libraries than not. I'm on a 68k machine so if anyone knows > anything special in that regard it might help. Going by memory ( There may be some messages I wrote at the time in the archive. ): I've compiled XlispStat 3.50 for PPC under CWPRO1. [1] You may need to change the libraries and access paths -- I don't remember if it shipped with the MSL libraries set in the project. If you do, there is a file in the release notes folder that tells you what new libraries replace what old libraries. [2] The MSL libraries removed EACCESS, so I put that back in CWHeader.h: #define EACCES 38 /* removed from MSL * added here for back compatibility */ [3] I had to change the default settings for precompiled headers in MacHeaders.c ( in the MW Code Warrior:MacOS Support:MacHeaders folder.) These are what I have them set to currently. I think that is correct, but it's possible I changed them again for some other project. #define OLDROUTINENAMES 1 #define OLDROUTINELOCATIONS 0 I did not try compiling for 68K, but I would expect the same changed to work for that project. If it still doesn't work, let me know what error you're getting. ( My mention of CFM libraries didn't have anything to do with getting it to link succseefully -- it was dynamic loading of CFM libraries of external routines. Try to get the vanilla distribution to build before you try adding any patches. ) ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Wed Oct 1 04:39:02 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29604; Wed, 1 Oct 1997 04:39:02 -0500 Return-Path: Received: from newton.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29600; Wed, 1 Oct 1997 04:38:51 -0500 Received: from libiya.upf.es (libiya.upf.es [193.145.60.105]) by newton.upf.es (8.8.5/8.8.5) with SMTP id LAA13899 for ; Wed, 1 Oct 1997 11:35:57 +0100 (WET DST) Received: by libiya.upf.es (1.38.193.4/15.6) id AA09199; Wed, 1 Oct 1997 11:35:48 +0100 Date: Wed, 1 Oct 1997 11:35:48 +0100 (WETDST) From: Frederic Udina To: lisp stat list Subject: compiling in 3.50 Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi, I have a method, in a quite big project, that work fine in XLS version XLISP-STAT 2.1 Release 3.44 (Beta). and it works fine too in release 3.50 installed in a Sun Ultra-sparc-1. But when I compile it in this last environment, it fails, it gives me an error I can't trace back because it is compiled code. Any suggestion? =========================== (defmeth wkde-proto :do-mode-testing-click (x y m1 m2) "when click, identify the mode, draw coordinates and lines and mark the 'show-modes with the index of the mode then call the :test-a-mode kde-proto method" (kde-require "modetest" "modetest") (let* ((rect (send self :content-rect)) (xyscaled (send self :canvas-to-scaled x y)) finalxy (xclic (first xyscaled)) modl index finalindex curr (diff 9e999)) (when (and (< (first rect) x (+ (first rect) (third rect))) (< (second rect) y (+ (second rect) (fourth rect)))) (setf modl (second (send self :to-core :mode-list))) (dotimes (it (length modl)) (setf curr (abs (- (select modl it) xclic))) (when (< curr diff) (setf index it) (setf diff curr))) (setf finalxy (apply #'send self :do-show-coordinates (append (send self :scaled-to-canvas (select modl index) (second xyscaled)) (list m1 m2)))) ;;check that mouse is still in the mode (setf xclic (first (apply #'send self :canvas-to-scaled finalxy))) (setf diff 9e99) (dotimes (it (length modl)) (setf curr (abs (- (select modl it) xclic))) (when (< curr diff) (setf finalindex it) (setf diff curr))) (when (= index finalindex) (send self :to-core :slot-value 'show-modes index) (send self :to-core :test-a-mode index :print t) index) ))) Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Ramon Trias Fargas, 25-27 | WWWeb: http://libiya.upf.es/ 08005 Barcelona SPAIN | or http://www.econ.upf.es/~udina ____________________________________-__________________________________ >> Please >> DON'T include this e-mail address in any commercial mailing list << From majordom@umnstat.stat.umn.edu Wed Oct 1 06:18:37 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA29789; Wed, 1 Oct 1997 06:18:37 -0500 Return-Path: Received: from gibbs.oit.unc.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA29785; Wed, 1 Oct 1997 06:18:31 -0500 Received: from [205.139.153.206] by gibbs.oit.unc.edu (8.7.5/10.1) id HAA12700; Wed, 1 Oct 1997 07:15:36 -0400 (GMT+4:00) X-Sender: uluru@gibbs.oit.unc.edu Message-Id: In-Reply-To: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Wed, 1 Oct 1997 07:16:47 -0500 To: Frederic Udina From: Forrest Young Subject: Re: compiling in 3.50 Cc: stat-lisp-news@umnstat.stat.umn.edu Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Hi folks: I have exactly the same problem with ViSta, which is why I continue to use 3.44 as well. Forrest >Hi, > >I have a method, in a quite big project, that work fine in XLS version >XLISP-STAT 2.1 Release 3.44 (Beta). > >and it works fine too in >release 3.50 installed in a Sun Ultra-sparc-1. >But when I compile it in this last environment, it fails, >it gives me an error I can't trace back because it is compiled code. > >Any suggestion? > >=========================== > > >(defmeth wkde-proto :do-mode-testing-click (x y m1 m2) >"when click, identify the mode, draw coordinates and lines > and mark the 'show-modes with the index of the mode > then call the :test-a-mode kde-proto method" > (kde-require "modetest" "modetest") > (let* ((rect (send self :content-rect)) > (xyscaled (send self :canvas-to-scaled x y)) > finalxy > (xclic (first xyscaled)) > modl index finalindex curr (diff 9e999)) > (when (and (< (first rect) x (+ (first rect) (third rect))) > (< (second rect) y (+ (second rect) (fourth rect)))) > (setf modl (second (send self :to-core :mode-list))) > (dotimes (it (length modl)) > (setf curr (abs (- (select modl it) xclic))) > (when (< curr diff) > (setf index it) > (setf diff curr))) > (setf finalxy > (apply #'send self :do-show-coordinates > (append (send self :scaled-to-canvas (select modl index) > (second xyscaled)) > (list m1 m2)))) > ;;check that mouse is still in the mode > (setf xclic > (first (apply #'send self :canvas-to-scaled finalxy))) > (setf diff 9e99) > (dotimes (it (length modl)) > (setf curr (abs (- (select modl it) xclic))) > (when (< curr diff) > (setf finalindex it) > (setf diff curr))) > (when (= index finalindex) > (send self :to-core :slot-value 'show-modes index) > (send self :to-core :test-a-mode index :print t) > index) >))) > > > > >Frederic Udina >_______________________________________________________________________ > | voice: 34 - 3 - 542 17 56 >Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 >Universitat Pompeu Fabra | e-mail: udina@upf.es > Ramon Trias Fargas, 25-27 | WWWeb: http://libiya.upf.es/ > 08005 Barcelona SPAIN | or http://www.econ.upf.es/~udina >____________________________________-__________________________________ >>> Please >>> DON'T include this e-mail address in any commercial mailing list << From majordom@nokomis.stat.umn.edu Wed Oct 1 08:46:49 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA10934; Wed, 1 Oct 1997 08:46:49 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA10927; Wed, 1 Oct 1997 08:46:27 -0500 From: Luke Tierney Message-Id: <9710011346.AA10927@nokomis.stat.umn.edu> Subject: Re: compiling in 3.50 To: forrest@unc.edu (Forrest Young) Date: Wed, 1 Oct 1997 08:46:26 -0500 (CDT) Cc: udina@upf.es, stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Forrest Young" at Oct 1, 97 07:16:47 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 603 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk Forrest Young wrote: > > Hi folks: > > I have exactly the same problem with ViSta, which is why I continue to use > 3.44 as well. > > Forrest > Give me an example that is as simple as you can make it but complete enough so I can reproduce the problem and I'll look into it. luke -- Luke Tierney University of Minnesota Phone: 612-625-7843 School of Statistics Fax: 612-624-8868 206 Church Street email: luke@stat.umn.edu Minneapolis, MN 55455 USA WWW: http://www.stat.umn.edu From majordom@umnstat.stat.umn.edu Wed Oct 1 12:53:52 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02086; Wed, 1 Oct 1997 12:53:52 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02082; Wed, 1 Oct 1997 12:53:46 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa08081; 1 Oct 97 13:53 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA16285; Wed, 1 Oct 1997 13:53:40 -0400 Date: Wed, 1 Oct 1997 13:53:39 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Josh Kuperman Cc: Xlisp List Subject: Re: Has anyone compiled Xlisp-stat using CodeWarrior Pro In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Tue, 30 Sep 1997, Josh Kuperman wrote: > Thanks, for the reply. I'm still stuck on something. > > I believe something is definitely wrong either with my libraries or glue > code. I suspect it has to do more with something I need to do to rebuild > the libraries. I think I might want to simply contact MetroWorks as the > Link Errors I've listed below seem to be problems with their code working > with their code. Let me know if you have any ideas. > Don't worry about the warnings. They are the usual ones. > Link Error : printf.c: '__rt_cmpu64' referenced from 'longlong2str' is > undefined. > > Link Error : printf.c: '__rt_cmps64' referenced from 'longlong2str' is > undefined. > > Link Error : printf.c: '__rt_neg64' referenced from 'longlong2str' is > undefined. You're missing the MSL Runtime Lib. ( Choose the appropriate one from MacOS Support:Libraries:Runtime: ) > Link Error : console_io.mac.c: 'InstallConsole' referenced from > 'check_console' is undefined. > > Link Error : console_io.mac.c: 'RemoveConsole' referenced from > 'check_console' is undefined. And you may need to add console.stubs.c ( from MSL C:MSL Mac:Sources: ) ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Wed Oct 1 14:55:27 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA02880; Wed, 1 Oct 1997 14:55:27 -0500 Return-Path: Received: from galton.stat.ucla.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA02876; Wed, 1 Oct 1997 14:55:20 -0500 Received: (from deleeuw@localhost) by galton.stat.ucla.edu (8.8.3/8.8.3) id MAA15896 for stat-lisp-news@umnstat.stat.umn.edu; Wed, 1 Oct 1997 12:55:19 -0700 (PDT) From: Jan Deleeuw Message-Id: <199710011955.MAA15896@galton.stat.ucla.edu> Subject: mirrors To: stat-lisp-news@umnstat.stat.umn.edu (Xlisp List) Date: Wed, 1 Oct 1997 12:55:18 -0700 (PDT) X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk there is a new mirror running on ftp.stat.ucla.edu. we mirror xlispstat in pub/applications/xlispstat vista in pub/applications/vista clisp in pub/applications/clisp --- jan -- Jan de Leeuw; Professor and Chair, Interdivisional Program in Statistics; US mail: 8142 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554 phone (310)-825-9550; fax (310)-206-5658; email: deleeuw@stat.ucla.edu www: http://www.stat.ucla.edu/~deleeuw ============================================================================ Remember, wherever you go, there you are. --- Buckaroo Banzai ============================================================================ From majordom@umnstat.stat.umn.edu Thu Oct 2 09:21:58 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15328; Thu, 2 Oct 1997 09:21:58 -0500 Return-Path: Received: from angel.heaven.net by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15324; Thu, 2 Oct 1997 09:21:50 -0500 Received: from [198.69.29.97] (josh.netheaven.com [198.69.29.97]) by angel.heaven.net (8.8.5/8.8.5) with ESMTP id KAA11302; Thu, 2 Oct 1997 10:22:00 -0400 (EDT) X-Sender: josh@netheaven.com (Unverified) Message-Id: In-Reply-To: References: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 2 Oct 1997 10:25:54 -0400 To: "Steven D. Majewski" From: Josh Kuperman Subject: Re: Has anyone compiled Xlisp-stat using CodeWarrior Pro Cc: Xlisp List Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk >On Tue, 30 Sep 1997, Josh Kuperman wrote: > >> Thanks, for the reply. I'm still stuck on something. Ditto, though this time I believe I'm stuck on a lu-lu. >You're missing the MSL Runtime Lib. > ( Choose the appropriate one from MacOS Support:Libraries:Runtime: ) Yes, I needed that. >And you may need to add console.stubs.c > ( from MSL C:MSL Mac:Sources: ) And I needed that. So, it all compiles. Now I think, I am all set to go back to trying to build expert horse-player systems with belief networks. I check out my XLS68K881 ;========================================================================= ; bet you double or nothing this is the right log function for gambling ; (defun lg (x) (/ (log x) (log 2))) ;========================================================================= This blows me out of the water. Then I remember seeing a note from Luke last spring about some library problems with the self-same version I'm trying to compile. >From: Luke Tierney >Subject: 68K Mac 881 version of 3-50 is bad >To: stat-lisp-news@nokomis.stat.umn.edu >Date: Fri, 18 Apr 1997 14:45:55 -0500 (CDT) >Mime-Version: 1.0 >Sender: owner-stat-lisp-news@stat.umn.edu >Precedence: bulk > >The version of XLISP-STAT 3.50 for 68K Macintoshes with a coporcessor >crashes on calls to math functions because it was accidentally linked >with the wrong library. I have removed the executable file from the >distrtibution. I hope to replace it with a working one in a week or >two. Until then, please use the 020 version. Sorry about that. > >luke After noticing that there is still no 68K 881 version, my curiosity piques wondering if there is no 881 version becuase of the wrong library, the non-existence of the right library, or something else. Please note I will happily pursue this problem, as I believe in getting 10 years of use out of every computer and my Quadra 660av has another 5 to go. Josh Kuperman josh@racing.saratoga.ny.us http://www.racing.saratoga.ny.us/~josh From majordom@umnstat.stat.umn.edu Thu Oct 2 10:47:15 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA15926; Thu, 2 Oct 1997 10:47:15 -0500 Return-Path: Received: from newton.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA15921; Thu, 2 Oct 1997 10:46:58 -0500 Received: from libiya.upf.es (libiya.upf.es [193.145.60.105]) by newton.upf.es (8.8.5/8.8.5) with SMTP id RAA13496; Thu, 2 Oct 1997 17:41:57 +0100 (WET DST) Received: by libiya.upf.es (1.38.193.4/15.6) id AA09721; Thu, 2 Oct 1997 17:41:41 +0100 Date: Thu, 2 Oct 1997 17:41:41 +0100 (WETDST") From: Frederic Udina To: Luke Tierney Cc: Forrest Young , stat-lisp-news@umnstat.stat.umn.edu Subject: Re: compiling in 3.50 In-Reply-To: <9710011346.AA10927@nokomis.stat.umn.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Wed, 1 Oct 1997, Luke Tierney wrote: > > I have exactly the same problem with ViSta, which is why I continue to use > > 3.44 as well. > > > > Give me an example that is as simple as you can make it but complete > enough so I can reproduce the problem and I'll look into it. > Well Luke, thinks are not so easy. Trying to locate the problem I found first that using 3.50 under win95 gives me no problem. Compiled or not, my method works fine. Then I moved again to Solaris 5.2 and the problem was still there. I got the error: > Error: bad argument type - INFINITY Break level 1. when running the compiled code. Not with the interpreted one. Then, this INFINITY made me suspicious... I had a declaration like (setf diff 9e999) to make diff act as a flag. I wanted diff start with a bigggg value and I typed 9e999. It looks like the interpreter works fine with Infinity but compiled code fails. The function (defun big () (setf diff 9e999) (when (< 33 diff) (print 33)) diff) works fine but when compiled fails. OK, I don't need anything as big as 9e999, so by now some of my problems are solved. Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Ramon Trias Fargas, 25-27 | WWWeb: http://libiya.upf.es/ 08005 Barcelona SPAIN | or http://www.econ.upf.es/~udina ____________________________________-__________________________________ >> Please >> DON'T include this e-mail address in any commercial mailing list << From majordom@nokomis.stat.umn.edu Thu Oct 2 13:55:19 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA29209; Thu, 2 Oct 1997 13:55:19 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA29201; Thu, 2 Oct 1997 13:54:55 -0500 From: Luke Tierney Message-Id: <9710021854.AA29201@nokomis.stat.umn.edu> Subject: Re: Has anyone compiled Xlisp-stat using CodeWarrior Pro To: josh@racing.saratoga.ny.us (Josh Kuperman) Date: Thu, 2 Oct 1997 13:54:55 -0500 (CDT) Cc: sdm7g@Virginia.EDU, stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: from "Josh Kuperman" at Oct 2, 97 10:25:54 am X-Mailer: ELM [version 2.4 PL25] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1309 Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk Josh Kuperman wrote: > > After noticing that there is still no 68K 881 version, my curiosity piques > wondering if there is no 881 version becuase of the wrong library, the > non-existence of the right library, or something else. > The way Metrowerks does their 68K libraries there are either 8 or 16 -- I forget -- different libraries you need to choose from depending on flags in your project. They don't give you all combinations but do give you an apple script to build what it missing. Unfortunately getting the flags right in that build is tricky, at least for me, and if you don't the code blows up. I think I have this working in CW Pro, but I am tired of going through it every time they come out with a new release, so I will probably try to streamline the 68K distribution. Since the 881 version seems to be used very little these days, and there aren't a whole lot of non 020 or better machines left, I'll probably switch to just the 020 distribution for 68K machines. luke -- Luke Tierney University of Minnesota Phone: 612-625-7843 School of Statistics Fax: 612-624-8868 206 Church Street email: luke@stat.umn.edu Minneapolis, MN 55455 USA WWW: http://www.stat.umn.edu From majordom@umnstat.stat.umn.edu Mon Oct 6 09:24:07 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA08656; Mon, 6 Oct 1997 09:24:07 -0500 Return-Path: Received: from cheviot.ncl.ac.uk by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA08652; Mon, 6 Oct 1997 09:23:38 -0500 Received: from burnmoor.ncl.ac.uk by cheviot.ncl.ac.uk id (8.7.6/ for ncl.ac.uk) with ESMTP; Mon, 6 Oct 1997 15:23:35 +0100 (BST) From: D.J.Wilkinson@newcastle.ac.uk Received: from aidan.ncl.ac.uk (ndjw1@aidan.ncl.ac.uk [128.240.233.1]) by burnmoor.ncl.ac.uk (8.7.6/8.6.10-cf revision 2 for Solaris 2.x) with SMTP id PAA22936; Mon, 6 Oct 1997 15:23:34 +0100 (BST) Received: (ndjw1@localhost) by aidan.ncl.ac.uk (8.6.10/8.6.10-cf revision 1 for Solaris 2.x) id PAA14050; Mon, 6 Oct 1997 15:23:31 +0100 Message-Id: <199710061423.PAA14050@aidan.ncl.ac.uk> Subject: Nice PostScript from graph-window-proto? To: stat-lisp-news@umnstat.stat.umn.edu Date: Mon, 6 Oct 1997 15:23:30 +0100 (BST) X-Mailer: ELM [version 2.4 PL23] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I've written some software which draws lines, circles, circular arcs etc, on a prototype derived from graph-window-proto (NOT graph-proto). Does anyone have a "magic" method for production of nice PostScript for such drawings? Really, it is just the sort of thing PostScript is good at, so it ought to be possible. I'm most interested in a Unix solution, but any solution which gives me better than a PostScript bitmap is of interest. Just in case anyone is interested, the software is called BAYES-LIN, and is really for local computation in Bayes linear belief networks, via message-passing between objects representing clique-tree nodes. You can download it, and its (sparse) documentation from: http://www.ncl.ac.uk/~ndjw1/bayeslin/ I would like to be able to get nice PostScript output of the diagnostic plots it produces. Cheers, -- Darren Wilkinson d.j.wilkinson@newcastle.ac.uk http://www.ncl.ac.uk/~ndjw1/ From majordom@umnstat.stat.umn.edu Mon Oct 6 11:03:09 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA09773; Mon, 6 Oct 1997 11:03:09 -0500 Return-Path: Received: from maria.physics.lsa.umich.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA09769; Mon, 6 Oct 1997 11:03:03 -0500 Received: by maria.physics.lsa.umich.edu (1.38.193.4/16.2) id AA11815; Mon, 6 Oct 1997 12:05:08 -0400 Date: Mon, 6 Oct 1997 12:05:08 -0400 (EDT) From: Rick Riolo Reply-To: rlriolo@umich.edu To: stat-lisp-news@umnstat.stat.umn.edu Subject: gray-scale density plots? Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Has anyone written a more-or-less general tool for creating gray-scaled (or other colorscale) densitiy plots for 2D data? That is, what I'd like to send it is a list of lists representing a 2D matrix of density values and then have it put up a graph window with with little rectangles, one for each entry in the matrix, with its darkness (color) representing the value of the corresonding cell in the matrix. (other features, eg normalization control, etc, would be nice but I'd settle for any starting code.) thanks! - r Rick Riolo rlriolo@umich.edu Program for Study of Complex Systems (PSCS) 4068 Randall Lab University of Michigan Ann Arbor MI 48109-1120 Phone: 313 763 3323 Fax: 313 763 9267 http://pscs.physics.lsa.umich.edu/PEOPLE/rlr-home.html http://pscs.physics.lsa.umich.edu//pscs.html From majordom@umnstat.stat.umn.edu Mon Oct 6 16:01:35 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA11461; Mon, 6 Oct 1997 16:01:35 -0500 Return-Path: Received: from kodakr.kodak.com by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA11457; Mon, 6 Oct 1997 16:01:25 -0500 Received: from corpmail.kodak.com by kodakr.kodak.com with SMTP id AA15618 (5.67b/IDA-1.5 for ); Mon, 6 Oct 1997 16:59:20 -0400 Received: from kp-w008118 ([150.103.229.15]) by corpmail.kodak.com (post.office MTA v1.9.3b ID# 269-16266) with SMTP id AAA22953 for ; Mon, 6 Oct 1997 16:56:56 -0400 Message-Id: <34395179.7CEF@kodak.com> Date: Mon, 06 Oct 1997 17:00:41 -0400 From: Jeff Morris Reply-To: morris@Kodak.COM Organization: Eastman Kodak Company X-Mailer: Mozilla 3.01 (WinNT; I) Mime-Version: 1.0 To: stat-lisp-news@umnstat.stat.umn.edu Subject: Sorting x by y Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'm embarrassed to admit that I think I asked this question before, but I could find nothing in my mail logs, so here I go again... Given a factor x, and a response y, how does one sort x by y. That is, I sort x, but want to keep the original y value with the corresponding x value. Thanks as always! Jeff -- Jeffrey D. Morris, Ph.D. Eastman Kodak Co. Rochester, NY 14652 morris@kodak.com From majordom@umnstat.stat.umn.edu Mon Oct 6 19:35:38 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12492; Mon, 6 Oct 1997 19:35:38 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12488; Mon, 6 Oct 1997 19:35:30 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa23723; 6 Oct 97 20:35 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA24109; Mon, 6 Oct 1997 20:35:26 -0400 Date: Mon, 6 Oct 1997 20:35:26 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Jeff Morris Cc: Xlisp List Subject: Sorting x by y In-Reply-To: <9701232125.AA17668@nokomis.stat.umn.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Mon, 6 Oct 1997, Jeff Morris wrote: > I'm embarrassed to admit that I think I asked this question before, > but I could find nothing in my mail logs, so here I go again... > > Given a factor x, and a response y, how does one sort x by y. That is, > I sort x, but want to keep the original y value with the corresponding > x value. > > Thanks as always! > > Jeff > Lisp #'sort ( as opposed to xlisp-stat sort-data ) takes a predicate that is the function used to order the list. Thus, if you have a list of lists of ( x y ) pairs: [1]: ( sort pair-list #'( lambda (a b) (< (second a) (second b))) ) Will sort a list of pairs by the second (cadr) element. Common Lisp #'sort also takes a :key argument, and although ( help 'sort ) does not indicate a :key argument, the following does in fact work correctly: ( shorter and equivalent to #[1] ) [2]: ( sort pair-list #'< :key #'cadr ) But you're right -- you did ask this question ( in a slightly different form) once before, and ... On Thu, 23 Jan 1997, Luke Tierney wrote: > > sorting y according to x can be done as (select y (order x)) > > luke > ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Tue Oct 7 10:46:28 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24321; Tue, 7 Oct 1997 10:46:28 -0500 Return-Path: Received: from newton.upf.es by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24313; Tue, 7 Oct 1997 10:46:15 -0500 Received: from libiya.upf.es (libiya.upf.es [193.145.60.105]) by newton.upf.es (8.8.5/8.8.5) with SMTP id RAA12682 for ; Tue, 7 Oct 1997 17:23:36 +0100 (WET DST) Received: by libiya.upf.es (1.38.193.4/15.6) id AA12810; Tue, 7 Oct 1997 17:23:15 +0100 Date: Tue, 7 Oct 1997 17:23:14 +0100 (WETDST") From: Frederic Udina To: lisp stat list Subject: indices Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Is there a built-in function that return a list of the indices of elements in a list that satisfy a test? Frederic Udina _______________________________________________________________________ | voice: 34 - 3 - 542 17 56 Facultat de Ciencies Economiques | fax: 34 - 3 - 542 17 46 Universitat Pompeu Fabra | e-mail: udina@upf.es Ramon Trias Fargas, 25-27 | WWWeb: http://libiya.upf.es/ 08005 Barcelona SPAIN | or http://www.econ.upf.es/~udina ____________________________________-__________________________________ >> Please >> DON'T include this e-mail address in any commercial mailing list << From majordom@umnstat.stat.umn.edu Tue Oct 7 11:47:58 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA24899; Tue, 7 Oct 1997 11:47:58 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA24895; Tue, 7 Oct 1997 11:47:50 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa04292; 7 Oct 97 12:47 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA27669; Tue, 7 Oct 1997 12:47:19 -0400 Date: Tue, 7 Oct 1997 12:47:19 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: Frederic Udina Cc: lisp stat list Subject: Re: indices In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Tue, 7 Oct 1997, Frederic Udina wrote: > > Is there a built-in function that return a list of the indices of > elements in a list that satisfy a test? > Most of the tests in xlisp-stat are vectorized -- when applied to a list you get a list of (T|NIL)'s. #'WHICH will return the indices for which the element is non-nil: > ( iseq 10 ) (0 1 2 3 4 5 6 7 8 9) > ( oddp * ) (NIL T NIL T NIL T NIL T NIL T) > ( which * ) (1 3 5 7 9) > ( defun indices ( list pred ) ( which ( map 'cons pred list ))) whould give you a single function that does what you want (I think). ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@nokomis.stat.umn.edu Tue Oct 7 13:31:38 1997 Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA07340; Tue, 7 Oct 1997 13:31:38 -0500 Return-Path: Received: by nokomis.stat.umn.edu (1.38.193.4/16.2) id AA07335; Tue, 7 Oct 1997 13:31:35 -0500 Received: from zam149.zam.kfa-juelich.de by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA25019; Tue, 7 Oct 1997 12:05:29 -0500 Received: from hlrz9-f by fz-juelich.de (5.65v3.2/DEC-AXP) id AA16424; Tue, 7 Oct 1997 19:05:20 +0200 Received: from hlrz4.hlrz.kfa-juelich.de by hlrz9.hlrz.kfa-juelich.de (SMI-8.6/SMI-4.1) id TAA14375; Tue, 7 Oct 1997 19:05:18 +0200 Reply-To: jmu@hlrz9.hlrz.kfa-juelich.de Received: by hlrz4.hlrz.kfa-juelich.de (SMI-8.6/SMI-4.1) id TAA09964; Tue, 7 Oct 1997 19:05:17 +0200 Date: Tue, 7 Oct 1997 19:04:59 +0200 Message-Id: <3405-Tue07Oct1997190459+0200-jmu@hlrz.kfa-juelich.de> X-Mailer: emacs 19.34.1 (via feedmail 7 I) From: jmu@hlrz9.hlrz.kfa-juelich.de (Johannes Mueller-Gerking) To: majordom@nokomis.stat.umn.edu Cc: stat-lisp-news@nokomis.stat.umn.edu In-Reply-To: (message from Frederic Udina on Tue, 7 Oct 1997 17:23:14 +0100 (WETDST")) Subject: Re: indices Content-Type: text Sender: owner-stat-lisp-news@nokomis.stat.umn.edu Precedence: bulk > Is there a built-in function that return a list of the indices of > elements in a list that satisfy a test? > which example: > (def a '(1 2 3 4 5 4 3 2 1)) > (which (> 3 a)) (0 1 7 8) > (which (> a 3)) (3 4 5) > (help 'which) WHICH [function-doc] Args: (x) X is an array or a list. Returns a list of the indices where X is not NIL. NIL Best Johannes ----------------------------------------------------------------------- Johannes Mueller-Gerking, PhD email : jmu@hlrz.kfa-juelich.de FZ Juelich, HLRZ 52425 Juelich Precode : [+49] (2461) office : 61-2318 Germany home : 346572 FAX : 61-2430 From majordom@umnstat.stat.umn.edu Tue Oct 7 19:06:39 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27070; Tue, 7 Oct 1997 19:06:39 -0500 Return-Path: Received: from maria.physics.lsa.umich.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27066; Tue, 7 Oct 1997 19:06:33 -0500 Received: by maria.physics.lsa.umich.edu (1.38.193.4/16.2) id AA16376; Tue, 7 Oct 1997 20:08:41 -0400 Date: Tue, 7 Oct 1997 20:08:41 -0400 (EDT) From: Rick Riolo Reply-To: rlriolo@umich.edu To: stat-lisp-news@umnstat.stat.umn.edu Subject: hide windows until positioned, sized, etc... Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I would like to know if its possible to hide an entire window, eg, a window created by a plot-points or a histogram function call, until I get done resizing it, positioning it, etc. For example, I might do: (setf aPlot (plot-points xlocs ylocs)) (send aPlot :range 0 -1 17) (send aPlot :range 1 -1 17) (send aPlot :title title) (send aPlot :location 0 20) and so on. I also go on to set symbols for subsets of the points, etc. I have kludged things (via start-buffering, etc) so that the content doesn't show until all the points are added and symbols assigned. But I can't figure out how to just hide the entire window until everything is done and ready to see in its final state and place and size, so the user doesn't have to see the window jump around, resize, etc. Thanks for any help. - r Rick Riolo rlriolo@umich.edu Program for Study of Complex Systems (PSCS) 4068 Randall Lab University of Michigan Ann Arbor MI 48109-1120 Phone: 313 763 3323 Fax: 313 763 9267 http://pscs.physics.lsa.umich.edu/PEOPLE/rlr-home.html http://pscs.physics.lsa.umich.edu//pscs.html From majordom@umnstat.stat.umn.edu Tue Oct 7 20:52:06 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA27751; Tue, 7 Oct 1997 20:52:06 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA27747; Tue, 7 Oct 1997 20:51:58 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa16973; 7 Oct 97 20:52 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA29087; Tue, 7 Oct 1997 20:52:30 -0400 Date: Tue, 7 Oct 1997 20:52:29 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: rlriolo@umich.edu Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: hide windows until positioned, sized, etc... In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Tue, 7 Oct 1997, Rick Riolo wrote: > I would like to know if its possible to > hide an entire window, eg, a window > created by a plot-points or a histogram function call, > until I get done resizing it, positioning it, etc. > > For example, I might do: > > (setf aPlot (plot-points xlocs ylocs)) > (send aPlot :range 0 -1 17) > (send aPlot :range 1 -1 17) > (send aPlot :title title) > (send aPlot :location 0 20) > > and so on. > > I also go on to set symbols for subsets of the points, etc. > I have kludged things (via start-buffering, etc) > so that the content doesn't show until all the points > are added and symbols assigned. > > But I can't figure out how to just hide the entire > window until everything is done and ready to see in its > final state and place and size, so the user > doesn't have to see the window jump around, resize, etc. You can add ":SHOW NIL" to either the :ISNEW call for a graph-window ( including scatterplot-proto ) or to the #'PLOT-POINTS function. (setf aPlot (plot-points xlocs ylocs :show nil)) ... (send aPlot :show-window) ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Wed Oct 8 10:53:45 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA09902; Wed, 8 Oct 1997 10:53:45 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA09898; Wed, 8 Oct 1997 10:53:37 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa16600; 8 Oct 97 11:53 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA24213; Wed, 8 Oct 1997 11:53:32 -0400 Date: Wed, 8 Oct 1997 11:53:32 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: rlriolo@umich.edu Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: gray-scale density plots? In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Mon, 6 Oct 1997, Rick Riolo wrote: > > Has anyone written a more-or-less general tool for > creating gray-scaled (or other colorscale) densitiy plots > for 2D data? That is, what I'd like to send it is > a list of lists representing a 2D matrix of density values > and then have it put up a graph window with with > little rectangles, one for each entry in the matrix, > with its darkness (color) representing the value of the > corresonding cell in the matrix. > (other features, eg normalization control, etc, would be > nice but I'd settle for any starting code.) > You might look at a previous thread: "How to draw 3-d plot of RGB space". There's also a piece of code at: but the comments and discussion were all in that thread. This code was a 3-d plot of the colors in RGB.TXT in rgb-color space. For your use, you need to decide on a linear color mapping. ( IMHO: lots of gray levels are hard to distinguish. has rgb<>hsv functions that might help in generating a better pseudo color scheme. ) Then you have to add the colors in the scheme to the color table. Before you generate the colors, you need to know how many colors you can have in the color table. I did this ( the example was in that previous thread ) by adding colors until add-color failed when the table was full. This is a little awkward, since you need to know how many colors you can have to evenly space the color mapping. ( Maybe iview should export this value to xlispstat ? ) If you know what that value is, you can make it a constant for a temporary, non-portable hack. Sort the table of color symbols by your mapping scheme. Scale the values and map them into (iseq num-colors). Use those values as indexes into your sorted color-table, and send graph :point-color. You might want to select all of the points ( which is what I do in the rgbspace code ) and maybe set the point-symbol. ( Unselected points are "hollow" and don't show the colors as well as selected solid point symbols. ) ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Wed Oct 8 12:05:16 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10584; Wed, 8 Oct 1997 12:05:16 -0500 Return-Path: Received: from maria.physics.lsa.umich.edu by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10576; Wed, 8 Oct 1997 12:05:09 -0500 Received: by maria.physics.lsa.umich.edu (1.38.193.4/16.2) id AA17976; Wed, 8 Oct 1997 13:07:18 -0400 Date: Wed, 8 Oct 1997 13:07:18 -0400 (EDT) From: Rick Riolo Reply-To: rlriolo@umich.edu To: stat-lisp-news@umnstat.stat.umn.edu Subject: previous items (was Re: gray-scale density plots?) In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk Steve, Thanks for the pointers. One question: Your comment seems to imply there is an online source for old stat-lisp-news items. I didn't know that. Where might that be? thanks! - r Rick Riolo rlriolo@umich.edu Program for Study of Complex Systems (PSCS) 4068 Randall Lab University of Michigan Ann Arbor MI 48109-1120 Phone: 313 763 3323 Fax: 313 763 9267 http://pscs.physics.lsa.umich.edu/PEOPLE/rlr-home.html http://pscs.physics.lsa.umich.edu//pscs.html From majordom@umnstat.stat.umn.edu Wed Oct 8 12:52:07 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA10890; Wed, 8 Oct 1997 12:52:07 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA10886; Wed, 8 Oct 1997 12:52:01 -0500 Received: from elvis.med.virginia.edu by mail.virginia.edu id aa21106; 8 Oct 97 13:51 EDT Received: from localhost by elvis.med.Virginia.EDU (5.65c/1.34) id AA30778; Wed, 8 Oct 1997 13:51:45 -0400 Date: Wed, 8 Oct 1997 13:51:44 -0400 (EDT) From: "Steven D. Majewski" X-Sender: sdm7g@elvis.med.Virginia.EDU To: rlriolo@umich.edu Cc: stat-lisp-news@umnstat.stat.umn.edu Subject: Re: previous items (was Re: gray-scale density plots?) In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk On Wed, 8 Oct 1997, Rick Riolo wrote: > Your comment seems to imply there is an online source > for old stat-lisp-news items. > I didn't know that. Where might that be? Well -- I thought there was an archive. There's an archive file at: However, the README file says that it's updated irregularly by hand, and my browser shows the files data as 11/25/96 ! ( It looks like the same files, but gziped, on the ucla mirror. ) Sorry. Is there any more up-to-date archive ? ( If not, I'll see if I can sort out my own personal archive files and put them somewhere on my web. ) ---| Steven D. Majewski (804-982-0831) |--- ---| Department of Molecular Physiology and Biological Physics |--- ---| University of Virginia Health Sciences Center |--- ---| P.O. Box 10011 Charlottesville, VA 22906-0011 |--- All power corrupts and obsolete power corrupts obsoletely." - Ted Nelson From majordom@umnstat.stat.umn.edu Wed Oct 8 16:32:14 1997 Received: by umnstat.stat.umn.edu (1.38.193.4/16.2) id AA12108; Wed, 8 Oct 1997 16:32:14 -0500 Return-Path: Received: from mail.Virginia.EDU by umnstat.stat.umn.edu with SMTP (1.38.193.4/16.2) id AA12104; Wed, 8 Oct 1997 16:32:08 -0500 Received: from avery.med.virginia.edu by mail.virginia.edu id aa05544; 8 Oct 97 17:31 EDT Received: from bootp-17-17.bootp.virginia.edu (bootp-17-17.bootp.Virginia.EDU [128.143.17.17]) by avery.med.Virginia.EDU (8.8.5/8.6.6) with SMTP id RAA14758; Wed, 8 Oct 1997 17:31:51 -0400 Date: 8 Oct 97 17:43:21 -0400 Subject: RFC: Changes to XlispStat Mac CW Projects. From: "Steven D. Majewski" To: stat-lisp-news@umnstat.stat.umn.edu, Luke Tierney , Jan Deleeuw Cc: "Robert A. Stine" X-Mailer: Cyberdog/2.0 Mime-Version: 1.0 Message-Id: Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-stat-lisp-news@umnstat.stat.umn.edu Precedence: bulk I'll announce this now, before I get too far along. (especially in case anyone else has started on this or was about to start. ) Comments ? [1] Right now I'm in the midst of reorganizing the Mac xlisp-stat CW projects: Updating to CW-PRO1 / IDE-2 & updating libraries and paths. changing to a single project file with multiple targets. adding library projects for PPC and 68K ( most sources, minus xlisp.c, console.stubs.c and libraries ) changing 68K and PPC application targets to use libraries. ( eventually shared libraries ) [ The above, except for shared libraries is done. Other targets should be added for CFM68K, vanilla 68K, etc. ] [2] Next phase: Change Bob Stine's dynamic loading / shared library code ( which I have been happily (and essentially) using for some time now. ): change the names of the functions so that they don't replace Open/CloseResourceFile, which also have other functions on the mac. ( Load/Free_DLL are the windows names, I think. ) Put original Open/Close resource code back in. Add #'CALL_LFUN code to mac projects to call functions that use XLISP API. [3] Putting one and two together -- we'll be able to build dynamically loadable shared lib code that uses Xlisp calling conventions and uses the Xlisp APIs to get and return args and to handle symbols, lisp sequences, call-backs, etc. I'm going to be doing minimal work on the 68K and CFM-68K part -- all of the essential lab machines here have been upgraded to PPC. I have a 68K lib and a 68K881 target that appear to work. But if anyone else is willing to build and test some of the other flavors, I'll be happy to integrate these into this new project. I'm going to keep OpenResourceFile because it's needed for some other functions, like getting cursors or other non-code mac resources. I'm not real concerned with keeping both CFM and non-CFM dyn-load in the same image, but I will try to keep which syle of dynload as a compile time option. -- Steve Majewski