This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: GCC 3.0 Request
- To: Alexandre Oliva <aoliva at redhat dot com>
- Subject: Re: GCC 3.0 Request
- From: David Edelsohn <dje at watson dot ibm dot com>
- Date: Thu, 15 Mar 2001 17:53:29 -0500
- cc: Mark Mitchell <mark at codesourcery dot com>, gcc at gcc dot gnu dot org
>>>>> Alexandre Oliva writes:
Alexandre> On Mar 15, 2001, Mark Mitchell <mark@codesourcery.com> wrote:
>> This looks to be a regression from GCC 2.95.2, so we should see
>> about fixing this.
Alexandre> It isn't a regression. I recall it used to be a hard error, fixed
Alexandre> with:
Alexandre> 2000-02-22 Alexandre Oliva <oliva@lsd.ic.unicamp.br>
Alexandre> * config/obj-coff.c (add_lineno): Accept non-positive lineno with
Alexandre> warning, and bump it to 1.
Previously we were not using gas at all. Some (long?) C++ mangled
names seem to cause the AIX assembler to produce corrupt symbols in object
files, so we now need to use gas. Either GCC is producing incorrect
debugging information or GAS is not accepting correct debugging
information. Changing from an error to a warning and substituting an
innocuous value simply papers over the problem and prevents debugging of
that code.
Again, if there was a problem in the GNU toolchain, I would have
thought that the GNUPro development work would have fixed it and
contributed the patch to the public FSF sources. According to Geoff's
follow-up comment, the GNUPro patch "is not suitable for integration into
the public sources." So where does that leave the public sources and
developers who want to use the public sources on AIX?
David