This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: SCO OpenServer / gcc-3.2 note
- From: Zack Weinberg <zack at codesourcery dot com>
- To: William Bader <william at nscs dot fast dot net>
- Cc: ac131313 at ges dot redhat dot com, kevinb at redhat dot com, gcc at gnu dot org
- Date: Fri, 16 Aug 2002 20:10:56 -0700
- Subject: Re: SCO OpenServer / gcc-3.2 note
- References: <200208162239.aa11405@mip486.nscs.fast.net>
On Fri, Aug 16, 2002 at 10:39:47PM -0400, William Bader wrote:
> Since you do SCO builds, the change below in gcc/ChangeLog makes
> gcc-3.1 builds fail with "machmode.def: undefined symbol: BITS_PER_UNIT".
>
> >2002-04-24 Robert Lipe <robertlipe@usa.net>
> > * config/i386/sco5.h(i386.h, att.h): Eliminate double inclusion.
>
> When I played with the bounds checking patches for gcc-3.1,
> I added the patch below. My SCO OpenServer system is a 66 MHz 486,
> so I never bother building gcc without the bounds checking patches,
> and since the config.gcc patch applied cleanly in gcc-3.2,
> I didn't try to build without it, but I do not see anything in
> the ChangeLog that would indicate that it was fixed.
>
> I posted the patch to the gcc list when gcc-3.1 came out,
> and someone said that they would fix the problem in gcc-3.1.1,
> but I do not think that they did. RJL said on June 4 that he
> would revert the patch to sco5.h.
>
> I'm not sure what they are planning, but if you try a SCO build,
> it could save you some time to be aware of this issue.
File a bug report in GNATS, describing this as a regression from
3.0.x, and attach this (correct) patch to that bug report. That will
prevent it from dropping off the radar for 3.2.1.
zw