From: Computing Helpdesk
Subject: Free Common LISP compilers for Digital Alphas
Date:
Message-ID: <866871624.625949@cabal>
Does anyone know of a free, non-broken Common LISP compiler for Digital
UNIX? It must be able to run John Koza's simple Genetic Programming code
(I can send you a copy if you'd like to test it) without giving a bizarre
error or crashing.
So far, we have tried GNU Common LISP, which my sys. admin was unable to
compile. We have tried CLISP, which baulks at Koza's code with a bizarre
error that I was unable to solve.
We have tried CMU CL, which works up to a point. It will compile the code
successfully, and can solve some problems such as the
non-hamstrung-squad-car problem. But when you try some problems such as
symbolic regression, or anything else which uses a bit of memory, CMU CL
dies with a "Dead in fake_foriegn_function-call, context = 1ffff678"
error. It also seems to be using a huge amount of memory, given what I am
doing with it. Top reports the size as 259 megabytes, although I was just
using the sleep function. This would indicate that the Alpha port of CMU
CL is broken.
Apart from that strange, intermittent error, I didn't mind CMU CL. GCL,
which I have been using under Debian LINUX at home, runs Koza's code just
fine. So a binary distribution of gcl for Alphas or a non-broken version
of CMU CL would be very much appreciated.
--
*************** Department of Computing User Support Assistant ***************
* Email: ·······@st.nepean.uws.edu.au ** Tel: 047 360 773 (2773 internally) **
**** Office: Room Y232, K'wood *** Open: 9am - 9pm Mon - Fri, 9am-5pm Sat ****
········@st.nepean.uws.edu.au (Computing Helpdesk) writes:
>Apart from that strange, intermittent error, I didn't mind CMU CL. GCL,
>which I have been using under Debian LINUX at home, runs Koza's code just
>fine. So a binary distribution of gcl for Alphas or a non-broken version
>of CMU CL would be very much appreciated.
Ask cmucl-imp @ cons.org. There is a good chance that your problems
are fixed either by the newer sources (18a) anyway or that you hit a
compile-time memory limit. Unfortunatly, building 17f or 18a on Alpha
is highly nontrivial and the best change is to try to trigger someone
else on the mailing list above to spend the time to produce a 18a
binary for Alpha.
Martin
--
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
···············@wavehh.hanse.de http://cracauer.cons.org Fax.: +4940 5228536
"As far as I'm concerned, if something is so complicated that you can't ex-
plain it in 10 seconds, then it's probably not worth knowing anyway"- Calvin