[Gcl-devel] Re: GCL vs. Fedora

[Camm Maguire]

Greetings, and thanks for your note!  I'll try to call in tomorrow.
In any case, the problem appears to be in the final call to unexec.
We've always known this part would be a problem that would have to be
resolved externally to GCL, as the unexec comes straight from emacs,
which also broke on Fedora, and which the Fedora developers told me
they were going to fix.  Perhaps you could find out the status here,
as that was FC1.

In any case, we have a work around which is slightly more difficult to
use.  --disable-statsysbfd --enable-dlopen.  This should build just
fine, but then you'll  have to follow the alternate acl2 build
instructions which you can find in the debian/rules of the debian acl2
package.  As using dlopen to relocate and load compiled objects is
awkward in lisp, we have focused on reducing/eliminating the need for
its use rather than making a dlopen build more transparent.  Perhaps
we could turn our attention to the latter in 2.7.

Just one other item -- It has just been pointed out in the last week
and a half that the latest binutils no longer retains symbol values
for undefined symbols in the executables it produces.  We've designed
a fix, but it is still a bit fluid.  Should be firmed up in a week or
two.  What's in cvs right now should be fine for these alternate
builds of maxima and acl2.  axiom needs a bit more work, which I hope
to commit in the next day or so.

Take care,


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