This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: PR 9479 (was Re: GCC-3.2.2 pre-release (third iteration))
On Fri, Feb 07, 2003 at 03:32:10AM -0500, Jakub Jelinek wrote:
> On Fri, Feb 07, 2003 at 01:39:12PM +1100, Greg Schafer wrote:
> > On Wed, Feb 05, 2003 at 02:07:04PM +1100, Greg Schafer wrote:
> > > On Tue, Feb 04, 2003 at 06:59:14PM -0800, Joe Buck wrote:
> > > > On Wed, Feb 05, 2003 at 12:01:49PM +1100, Greg Schafer wrote:
> > > > > HJ's realeases have a date string. The configury uses the date string in
> > > > > preference to the version string. So no worries there..
> > > >
> > > > Maybe they do now, but the one that was packaged with Red Hat 7.2 did
> > > > not. It says
> > > >
> > > > GNU ld 2.11.90.0.8
> > > >
> > > > with no date.
> > >
> > > Ooops. But luckily the version check is checking for 2.12.1 or later so in
> > > theory this shouldn't present an issue. It would be nice to sort out the 5
> > > digit case once and for all though. I'll try and get the patch reworked to
> > > cope with 5 digits then attach it to PR 9479 then maybe we could get it
> > > applied to all branches..
> >
> > Ok, I have now attached a new patch to PR 9479 which should finalise this
> > matter once and far all.
> >
> > Joe/Jakub, could you pls review the patch(es) in the PR? If ok, pls then
> > apply to 3.2, 3.3 & 3.4 and close the PR.
>
> The code should stay as is, it definitely should not
> handle 5 digit version numbers without date.
> binutils 2.12.90.0.xx without date string were not having proper
> hidden support, although 2.12.90 > 2.12.1.
Ahh, the situation is more complicated than I first thought. Now I
understand the comment in the code that says:-
"GNU LD versions before 2.12.1 have buggy support for STV_HIDDEN. This is
irritatingly difficult to feature test for. Look for the date string after
the version number."
Ok, the new patch is wrong and should not be applied.
But I think it would be a good thing if the patch you applied to the 3.2
branch could be applied to the 3.3 & 3.4 branches and thus fix the immediate
problem with FSF binutils-2.13.2.1
Greg