gcl-devel
[Top][All Lists]
Advanced

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

RE: [Gcl-devel] Re: [Maxima] 5.9.0 release candidate 1


From: Billinghurst, David (CRTS)
Subject: RE: [Gcl-devel] Re: [Maxima] 5.9.0 release candidate 1
Date: Tue, 3 Sep 2002 11:10:38 +1000

I managed to build a working maxima with gcl/mingw/msys last night,
using current CVS of both.  Testsuite results were OK.

I found and resolved Problem 1, and stumbled over a couple of other
minor gnits.  I will work through it again and refine my notes.
Also tried using gcc-3.2, but the raw_gcl.exe choked.  I may 
investigate further.  

        David

-----Original Message-----
From: Mike Thomas [mailto:address@hidden
Sent: Monday, 2 September 2002 6:24 
To: Stephen Leake
Cc: address@hidden; Maxima List
Subject: [Gcl-devel] Re: [Maxima] 5.9.0 release candidate 1


Hi Stephen.

PROBLEM 1 - gcl path

> sh-2.04$ gcl
> /e/usr/local/bin/gcl:
e:/Gnu/e:/usr/local/lib/gcl-2.5.0/unixport/saved_gcl.exe: No such file or
directory
> /e/usr/local/bin/gcl: exec:
e:/Gnu/e:/usr/local/lib/gcl-2.5.0/unixport/saved_gcl.exe: cannot execute: No
such file or directory
>
> for some reason, sh is prepending the current directory in the call to
> exec.

To fix this, just edit the path to "saved_gcl.exe" (JUST the path, NOT the
arguments) in the gcl script so that it uses MSYS format rather than native
Windows eg:

======================================
#!/bin/sh
exec /g/gclm/lib/gcl-2.5.0/unixport/saved_gcl.exe \
   -dir g:/gclm/lib/gcl-2.5.0/unixport/ \
   -libdir g:/gclm/lib/gcl-2.5.0/ \
   -eval '(setq si::*allow-gzipped-file* t)' \
     "$@"
# other options: -load /tmp/foo.o -load jo.lsp -eval "(joe 3)"

======================================

Here I changed the "g:/" to "/g/" in the second line.

This is a known bug in MSYS and Earnie Boyd is looking at it.  I may put a
workaround into the GCL make system if a fix is not forthcoming shortly.


PROBLEM 2 - complaint about truename

> But, no maxima executable is built. Make trundles along, creating lots
> of .o files. Then we get:
>
> ;      - Loading source file "max_ext.lisp"
> Loading max_ext.lisp
>
> Error: Cannot get the truename of "e:/usr/local/bin/.exe/".
> Fast links are on: do (si::use-fast-links nil) for debugging
> Error signalled by SETQ.
> Broken at TRUENAME.  Type :H for Help.
> MAXIMA>>
>
> I can exit this by typing "(by)".
>
> I have no experience debugging Maxima at this level. If someone can
> offer suggestions, I'm willing to try.
>
> I'm doing this at work, and won't be back until Tuesday.

I had this problem with the CVS version of Maxima late last week and it has
now gone away.

My belief is that it was caused by a failure in the GCL install target due
to non-existent TK object modules (ie some parts of the fresh GCL build were
installed, some weren't) - there have recently been changes in pathname and
other code which presumably weren't being installed due to the problem.

I have a fix for this and hope to check it into GCL CVS this evening.



ADVICE - clean up between builds under under CYGWIN and MSYS

One other thing to note when you come to building Maxima with MSYS hosted
GCL, is that you must remove the junk associated with the previous Cygwin
hosted build eg:

   ./bootstrap
   ./configure --enable-gcl --prefix=g:/maxima >configure.log 2>&1
   make clean
   make >make.log 2>&1

A symptom of not cleaning up properly between the Cygwin and MSYS builds
will be complaints about "/bin/bash" not being found.  For example - if you
try to "make clean" before doing the "configure" you will get this problem.

Cheers

Mike Thomas.





_______________________________________________
Gcl-devel mailing list
address@hidden
http://mail.gnu.org/mailman/listinfo/gcl-devel




reply via email to

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