Newsgroups: comp.lang.lisp
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!newsfeed.internetmci.com!EU.net!peer-news.britain.eu.net!newsfeed.ed.ac.uk!edcogsci!jeff
From: jeff@cogsci.ed.ac.uk (Jeff Dalton)
Subject: Re: How do the different Lisps compare in terms of X3J13 compatibility?
Message-ID: <DKo8Cs.Lyz.0.macbeth@cogsci.ed.ac.uk>
Organization: Centre for Cognitive Science, Edinburgh, UK
References: <82wx7s4jsb.fsf@manic.eecs.umich.edu>
Date: Thu, 4 Jan 1996 19:19:40 GMT
Lines: 37

In article <82wx7s4jsb.fsf@manic.eecs.umich.edu> hucka@engin.umich.edu writes:
>We're trying to decide on whether to continue our Franz Allegro CL site
>license, and have been looking into freely available Lisp implementations.
>One thing we don't have a good idea of is how well the different Lisps
>conform to CLtL2 and/or X3J13.  Can others among you comment about that for
>the following?
>
>* Allegro CL
>* CLISP
>* CMU CL
>* GNU CL
>* Medley
>* WCL
>
>Also, any comments about the quality of the implementations and the
>programming environment provided by these different Lisps would be
>appreciated.

GCL-1.1 is still largely 1st edition CLtL.  However, you can easily
add defpackage, loop, and st least some of the condition system.
(I know because I've done it.)  The CLOS, if you add it, is PCL,
and it can be a pain because PCL compiles some things at run-time,
and GCL compiles to C.  So the C compiler is run, which can be
annoyingly slow.

GCL is moving towards X3J13 CL, but I'm not sure how far along
that road later versions (e.g. GCl-2.2) are.  There are no 
lightweight processes / stack groups / etc.

On the whole, I find GCL (and the earlier AKCL and KCL, which I've
been using since 1985) pretty good as an implementation, though
it's less optimizing than Lucid CL (the other CL I use most) and
less good for debugging compiled code.  I haven't used Allegro
very much recently, but my recollection is that it was also better
for debugging compiled code.

-- jeff
