From shriram Mon May  1 12:11:04 1995
Date: Mon, 1 May 1995 12:10:58 -0500
From: shriram (Shriram Krishnamurthi)
To: matthias
Subject: Uses of lisp
Organization: Computer Science, Rice University


> Path: rice!newsfeed.rice.edu!nb.rockwell.com!lazrus.cca.rockwell.com!cacd.rockwell.com!news-feed-1.peachnet.edu!gatech!swrinde!cs.utexas.edu!uwm.edu!vixen.cso.uiuc.edu!news.ecn.bgu.edu!chicagokent.kentlaw.edu!iitmax.acc.iit.edu!iitmax.acc.iit.edu!not-for-mail
> From: sanders@iitmax.acc.iit.edu (Gregory A. Sanders)
> Newsgroups: comp.lang.ada,comp.edu
> Subject: Re: Academic CS Losers? First languages, etc
> Date: 28 Apr 1995 21:55:44 -0500
> Organization: Illinois Institute of Technology
> Lines: 69
> Message-ID: <3ns9rg$37r@iitmax.acc.iit.edu>
> References: <3nfjhd$7d2@kaiwan009.kaiwan.com> <dewar.798724813@gnat> <3nmink$jhh@kaiwan.kaiwan.com>
> NNTP-Posting-Host: iitmax.acc.iit.edu
> Xref: rice comp.lang.ada:25999 comp.edu:9532

In article <3nmink$jhh@kaiwan.kaiwan.com> jmartin@kaiwan.kaiwan.com (Jay M Martin) writes:
>In <dewar.798724813@gnat> dewar@cs.nyu.edu (Robert Dewar) writes:
>>Jay seems to think that functional and declarative programming languages
>>are strictly academic toys relevant to possible future technology only.
>
>>This just shows a narrow awareness, there are many examples of large
>>industrial projects written in functional languages. Nothing to challenge
>>the use of Fortran or COBOL in numbers, but don't make the mistake of
>>taking too extreme a position on this without bothering to find out who
>>really is using SML, Lisp and Prolog.
>
>Examples, what examples?  The only people using this stuff from what I
>knows is the AI industry which is a total failure.  These languages
>are admitted to not being commercial successes even by their
>proponents.

Five examples come to mind off the cuff (obviously there are a lot more).

    1. Bedford wrote their entire composition (as in text editing, layout,
       and typesetting) system in Lisp.  It's the central product of the
       company.  It works well.  Particularly notable is that this was the
       most complex composition system available, and it appears that this
       was made possible by the advantages conferred by writing it in Lisp
       rather than a more typical (imperative) programming language (i.e.,
       the system was too complex for development in an imperative language
       to be a commercially viable effort).

    2. The extension-language for one of the major CAD companies (I forget
       who) is Lisp.

    3. I recently got a help-wanted ad from Bellcore seeking someone for
       a system written in Lisp that does "workflow management and
       visualization."  As I understand this, the product already exists.

    4. MACSYMA is still in everyday use for advanced symbolic math.

    5. Arthur Anderson (big accounting and business-consulting firm)
       funds a lot of semi-research work done in Lisp at ILS.

It appears to me that it is fairly common to use Lisp in large scale
human-interface systems.  It is definitely used at times for writing
exploratory or demonstration systems, which benefit from the strengths
of Lisp.  Among these key strengths are flexibility, extensibility,
bottom-up programming style, interactive development environment,
ease of creating embedded languages, and ease of doing source-code
transformations, or text transformations in general.  In fact, one
can say that when embedded languages are needed, development in
traditional imperative languages will be very difficult (a great 
deal of work indeed) and that development in Lisp will be downright easy.
In the case of advanced math, Common Lisp has support exceeding most
other languages and the resulting compiled program often have better
performance than programs written in most languages like C or C++.

Regarding Jay Martin's earlier characterization of Scheme and Common Lisp
as "legacy languages,"  Scheme was invented/defined in the mid 1970s.
Common Lisp was first invented/defined in the early 1980s.  I'm wondering
what "legacy language" means to him.  Certainly the languages themselves,
let alone the way in which one writes programs in them, have no real
resemblance to the sorts of programming techniques used in the early Lisps.

Regarding his characterization of Common Lisp (if I understand him
correctly) as being used only in businesses that are failures, it is
interesting that ANSI is currently creating a standard for the language
(ANSI committee X3J13).  One wonders why Jay thinks ANSI would regard
this as worth their efforts and expense if the language is not of some
commercial interest?

  -- Greg