>>>> "SJT" == Stephen J Turnbull
<stephen(a)xemacs.org> writes:
>>>> "BAW" == bwarsaw
<bwarsaw(a)python.org> writes:
BAW> So what did RH /do/ to break this? Is the problem in the
BAW> XEmacs code, the X libraries, some place else? Any ideas, or
BAW> hints of places to look?
SJT> I'm not sure it's RH after all. Nix seems to think it's GCC
SJT> 3. See thread at
SJT>
http://list-archive.xemacs.org/xemacs-beta/200205/msg00189.html.
SJT> Now, of course there's a good chance that whatever broke in
SJT> GCC 3 is in Red Hat's "improved" version of gcc 2.9x.
Yup, that's the exact same problem I'm seeing, and just like the folks
in that thread, I'm also using "RH-enhanced" gcc 2.96:
% gcc -v
Reading specs from /usr/lib/gcc-lib/i386-redhat-linux/2.96/specs
gcc version 2.96 20000731 (Red Hat Linux 7.3 2.96-110)
Thanks for the pointer.
I'll try sucking down various gcc 3.x's and see if they make any
difference.
Note also that RH has released some updated libc's although their free
support line is severely overloaded so I'm unable to update. Perhaps
it's a libc problem, although some broken optimization in gcc seems a
more likely candidate.
Thanks,
-Barry