[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Gcl-devel] Re: translate.c compiler bug in GCL 2.7.0

From: Camm Maguire
Subject: [Gcl-devel] Re: translate.c compiler bug in GCL 2.7.0
Date: 15 Jul 2006 13:28:23 -0400
User-agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.2


OK, just committed a crude fix -- do not yet understand how any build
option could affect this, simply added an extra newline at the
appropriate point.  May need to do so later at one other point, but
translate11 is now fine.

Also included support for mvprog1 mvcall, and mvlist to use the same
faster mv when the exact number of values is known.  This works here
with the usual quick tests, but if you want to wait on this, just keep
the (wt-nl) in c2multiple-value-bind in gcl_cmpmulti.lsp.  You can
accelerate the turnaround by simply

(cvs diff -u to see the patch if desired)
cvs update
(edit gcl_cmpmulti.lsp if desired)
cd cmpnew 
cd ../unixport
make saved_ansi_gcl

On an unrelated note concerning the gmp functions -- whenever I get a
request for one or two more, I simply extend the adhoc current
method.  To do all of them, it seems as though we need a gmp package,
we keep the same function names, and we do not attempt to specially
handle fixnums, though I am unsure still on the latter point.  The
idea is that the compiler can avoid making a few bignums if fixnums
are known to be in the right place.  Perhaps all functions should be
provided in gmp coercing all arguments to bignums, and then special
workarounds added later for the fixnums.  In any case, I don't yet
have a clear idea on a simple comprehensive user interface.  If anyone
else does, this would facilitate things.

Take care,

Robert Boyer <address@hidden> writes:

> > this one needs quashing.
> Great.  UTCS directory/script details below.
> > I take it it does not arise in vanilla acl2 3.0?
> Not when built with a vanilla gcl 2.7.0.  But this could be a 'static'
> problem, for example.
> I have created the directory /u/boyer/my-partition/forcamm on the UTCS
> file system to which you have, I hope, total read/write/execute access,
> including from elgin or any of the other machines on the UTCS 'production'
> (as opposed to 'research') network, as it is called.  Please feel free to
> make any changes you wish to any files in or under
> /u/boyer/my-partition/forcamm.  I did the following to give you access:
>   chgrp -R gcl /u/boyer/my-partition/forcamm
>   chmod -R g+rwX /u/boyer/my-partition/forcamm
> In the subdirectory /u/boyer/my-partition/forcamm/gcl-2.7.0 is a very
> recent build of GCL 2.7.0, built with the script
> /u/boyer/my-partition/forcamm/build2, which shows the configuration
> command.  The file /u/boyer/my-partition/forcamm/log shows a transcript of
> the GCL build output and the following GCL compile error with acl2-hons.
> In the subdirectory /u/boyer/my-partition/forcamm/v3-0-hons-Jul10 is a
> copy of the acl2/hons files that Warren and I are currently hacking on.
> The commands
>    cd /u/boyer/my-partition/forcamm/v3-0-hons-Jul10
>    /u/boyer/my-partition/forcamm/gcl-2.7.0/unixport/saved_ansi_gcl
>    (load "init.lisp") 
>    (in-package "ACL2")
>    (compile-acl2)
> will give rise fairly quickly to the translate.lisp compilation error.
> The gcl was built with the enable-debug flag.
> Thanks so much!
> Bob

Camm Maguire                                            address@hidden
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah

reply via email to

[Prev in Thread] Current Thread [Next in Thread]