Lisp-stat and R? [was: Re: Status?]
8 Aug 2002 18:43:08 -0000

John Fox (see below) raises important questions for
the Lisp-stat community (and perhaps the R community) to consider.
This message thread was not cross-posted to,
so I do so now.

I have never been an active or particularly adept Lisp-stat programmer. But I  
have worked on or used several projects for which Lisp-stat seemed the ideal  
environment-- for implementing relatively portable, state-of-the-art  
applications, using dynamically-linked graphic windows to provide
and demonstrate leading-edge examples of highly-interctive environments
for the whole range from testing new ideas to delivering course content,
as has been done with Arc and ViSta.

My work with Forrest Young and Pedro Valero on ViSta implementations
of mosaic displays and other methods for categorical data has been
particularly gratifying, in that we've (well, largely Pedro) been able
to implement the kinds of interactive graphical analysis for contingency
tables that I always dreamed of.  This hinged on the built-in facilities
for object-oriented interactive graphic methods, abilities to link
graphic windows dynamically, and so forth, and Forrest's spreadplot
architecture for ViSta, which combines these facilities with a general
graph/window layout mechanism, so that all graph and text windows can
be made to work coherently.

But, as John points out, a number of people, formerly active in Lisp-stat
development have switched to R, for the reasons he outlines in his first
paragraph. I might add that for static graphs, and saving graphic
images, R now rivals, and, in some cases exceeds, what I have been
able to accomplish in SAS  -- but this (complete control of all graphic
parameters, text & math labels, etc.) is something that has always
been difficult and limited in Lisp-stat

On the other hand, R has very limited built-in facilities for
interactive graphics, and none that can be called 'dynamic'; nor does
it have any native GUI-building facilities, such as are used, with
great effect, in Arc and ViSta.  There is, however, a built-in mechanism
for R to interact with other software.  The Omega Project  
( details some of these, and there are also
other similar efforts (e.g., R-ggobi)

So, from the core of folks in the union (or intersection) of 

Lisp-stat and R developers: is there anyone to respond to John's
challenge -- ``to incorporate the Lisp-stat graphics and GUI-building 

capabilities into R''  I'll add a challenge of my own -- to incorportate
R modelling, static graphics and data manipulations into Lisp-stat.


On Tue, 06 Aug 2002 14:37:15 -0400, John Fox said:

I've spoken over the past couple of months to several individuals who, like 

myself, have more or less left Lisp-stat for R. There are several obvious 

reasons for doing so: Lisp-stat seems moribund, while R is alive and well; 

there is a very wide and growing range of statistical applications 

available for R; R has standards for data sets, models, documentation, 

etc.; and (perhaps more controversially) the S programming language is 

easier to use than Lisp.

Nevertheless, everyone to whom I've talked regrets losing some of the 

functionality of Lisp-stat, including the ability to draw dynamic, linked 

graphs, and the built-in facilities for constructing graphical interfaces.

Although I don't have the technical expertise either to answer the 

following question or, certainly, to implement a solution, I wonder whether 

it would be possible to incorporate the Lisp-stat graphics and GUI-building 

capabilities into R, ideally in the form of an add-on package that would 

work in all three supported environments -- Windows, Mac, and Linux/Unix. I 

know that there would be some obstacles -- and probably the components from 

Lisp-stat would have to run as a separate process -- but don't know whether 

these obstacles can be overcome.

If anyone has any ideas along these lines, I'd be interested in hearing 

them, and possibly in helping to organize a project to implement them.


At 12:44 AM 8/6/2002 +0200, Frederic Udina wrote:

>(just to be sure that somebody answers your question)
>Xlisp-stat, still being the only choice for some statistical applications,
>is not as active as it was some time ago.
>Some of the gurus have moved to R, some other are quite inactive. Some of
>us are still working with it but we obviously miss some lifting on the
>software... But apparently the efforts to undertake it have not been
>I have not looked to the official site for a while, it would be good to
>have it at least pointing to the existing resources.
>Somebody singin'  "We shall overcome..." out there?
>On Mon, 5 Aug 2002, Bill Harris wrote:
>~~~I just used XLISP-STAT again today after a bit of an absence, and I
>~~~thought I'd check to see what was new on the Web.  While I don't need to
>~~~see ongoing developments to find a tool useful, I was curious in that
>~~~ and the UCLA site that Google returned both
>~~~seemed offline.  What is up with XLISP-STAT these days?
>~~~Bill Harris                                  3217 102nd Place SE
>~~~Facilitated Systems                          Everett, WA 98208 USA
>~~~               phone: +1 425 337-5541
>  Frederic Udina
>   ____________________________________________________________________
>   <> Dept. D'Economia i Empresa   Universitat Pompeu Fabra
>                   Ramon Trias Fargas, 25-27      08005 Barcelona SPAIN
>   tel. 34- 935421756/1763 fax: 34- 935421746
>   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>   There are many here among us who feel that life is but a joke  B.Dy.

John Fox
Department of Sociology
McMaster University
Hamilton, Ontario, Canada L8S 4M4
phone: 905-525-9140x23604

r-devel mailing list -- Read
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: