Someone from Franz stated in a recent post that (I paraphrase)
LISP is still profitable, and adapting to real-world needs
just fine.
I'm in a research lab where AI-type programs
are the norm. The actual programming on these usually is
done by undergraduates and grad students, as is
true of most university labs. The language most often used
used to be LISP (we even have an Explorer), but simply because
of the preference of the programmers, the language of choice
is more often C now. I seldom hear of anyone using LISP anymore,
and having worked in both industry AI labs and university AI research,
am becoming convinced that LISP is indeed (to my dismay) fading away,
albeit slowly.
I think it would be quite interesting and appropriate at this
juncture for someone who is in a position to know (ie, Franz)
to tell us just who IS using LISP these days, and for what.
Thanks,
--Jim Kroger
well, for what it is worth, the University of Illinois has several large
groups in the "software engineering" area. (or systems, or whatever your
favorite term is). Common Lisp (usually with CLOS) is heavily used by my
research group, and is used by at least two others. This doesnt include
the AI groups here, but I'm pretty sure they use a mix of CL and Prolog.
We used to do all our programming in C, and have now switched almost
exclusively to Lisp, using C as a portable assembler when the need is dire.
We used to use AKCL+PCL, then switched to Allegro CL+PCL, now using that
and Allegro CL 4.0, which included CLOS.
Of course this is just one datum, but there it is.
-- Simon
--
Simon Kaplan ······@cs.uiuc.edu
Department of Computer Science ··············@uunet.uu.net
University of Illinois at Urbana-Champaign
1304 W. Springfield Ave phone: +1 217 244 0392
Urbana Illinois 61801, USA fax: +1 217 333 3501
In article <···@mara.cognet.ucla.edu> ······@scarecrow.cognet.ucla.edu
(James Kroger) writes:
> I'm in a research lab where AI-type programs
> are the norm. The actual programming on these usually is
> done by undergraduates and grad students, as is
> true of most university labs. The language most often used
> used to be LISP (we even have an Explorer), but simply because
> of the preference of the programmers, the language of choice
> is more often C now. I seldom hear of anyone using LISP anymore,
Hm. Why don't you force these students to learn lisp? It isn't good to
grow up thinking C is all that there is to programming. Being sort of at
the fringe of the AI Society (doing natural language semantics /
computational linguistics), I wonder what kinds of "AI programs" it is
that someone ever could prefer to write in C. I guess the least you need
is a pile of C code to simulate some things that come for free in lisp:
For instance _atoms_ (symbols), which is the most basic need for doing
symbolic programming. Does these guys reinvent atoms every time they need
them (or worse, are they doing it all with strings??), or are there some
nice libraries around which gives you some of the functionality of lisp
inside C??
-----------------------------------------
Espen J. Vestre
Department of Mathematics
University of Oslo
P.o. Box 1053 Blindern
N-0316 OSLO 3
NORWAY ·····@math.uio.no
-----------------------------------------
From: Laurence Kramer
Subject: Re: So who's really using LISP?
Date:
Message-ID: <18736@brahms.udel.edu>
Anybody in a research AI lab type environment that is programing in
C rather than Lisp needs their head examined. Lisp is clearly a better
prototyping/research environment due to its interactive nature. (Besides
other benefits).
It is my opinion that the grads and undergrads that you mentioned are prob-
ably programming in C because they don't know any better (i.e., have done
most of their academic coursework in C, and been only briefly exposed to
Lisp).
Enough of the diatribe. As to your question: We at Quantum Software (a
smallish software house specializing in AI applications) do 90% of our
coding in Lisp. The other 10% is in C, primarily for the systems' related
aspects of our programs. We also use C to "hand compile" some functions
within our Lisp deliverables which are particularly time critical.
Larry
From: Pete Grant
Subject: Re: So who's really using LISP?
Date:
Message-ID: <58415@hqda-ai.UUCP>
In article <·····@brahms.udel.edu>, ·······@brahms.udel.edu (Laurence Kramer) writes:
> Anybody in a research AI lab type environment that is programing in
> C rather than Lisp needs their head examined. Lisp is clearly a better
> prototyping/research environment due to its interactive nature. (Besides
> other benefits).
>
Right on. Lisp is still the King when it comes to rapid software
development, especially in those situations where the goals are not
clearly defined. At the Army AI Center, we use Lisp almost exclusively
for our work. There's no way we could achieve anywhere near the same
productivity using C or any other "conventional" systems.
> It is my opinion that the grads and undergrads that you mentioned are prob-
> ably programming in C because they don't know any better (i.e., have done
> most of their academic coursework in C, and been only briefly exposed to
> Lisp).
>
Yes, to be productive in Lisp, one must be proficient in it. Proficiency
in any language takes time and practice -- and Lisp is no exception.
I'm going into my 6th year of Lisp programming and am still learning
(and getting better), although the learning curve has flattened considerably.
From: Mark Pearson
Subject: Re: So who's really using LISP?
Date:
Message-ID: <895@mephisto.edu>
I've enjoyed following this discussion.
Just to assure you that LISP is not "dead", we
are currently implementing a LISP programming
environment on top of Clouds, a distributed operating
system in development here at Georgia Tech.
The implementation is written in (C)C++, a Clouds
derivative of C++. In this implementation, instances of
user programming environments and knowledge bases which
encapsulate distinct behaviors are stored as large-grained
persistent objects, enabling many users on many
machines to share these environments and knowledge
bases via inter-environment evaluations. Further, these
LISP objects plug and play with other large grained objects
implemented in CC++ and Eiffel (a heap-based, object-oriented
language).
The fact that a group of Operating System researchers
are interested in LISP, and the things it can do in a
distributed system, should put to rest all notions
that LISP, and its derivatives are on the way out.
-Mark Pearson
Mark Pearson, Distributed Systems Group/Clouds Project, College of Computing
Georgia Institute of Technology, Atlanta, Georgia, 30332-0280 404-853-9390/1
uucp : ...!{akgua,decvax,ihnp4,ulysses}!gatech!helios!mpearson
Internet: ········@cc.gatech.edu
Mark Pearson, Distributed Systems Group/Clouds Project, College of Computing
Georgia Institute of Technology, Atlanta, Georgia, 30332-0280 404-853-9390/1
uucp : ...!{akgua,decvax,ihnp4,ulysses}!gatech!helios!mpearson
Internet: ········@cc.gatech.edu