This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: objc failures on branch
On , 9 March 2001, 17:55:47 +0100, aj@suse.de wrote:
> Manfred Hollstein <manfredh@redhat.com> writes:
>
> > On , 5 March 2001, 13:06:57 +0100, ro@TechFak.Uni-Bielefeld.DE wrote:
> >
> > > Manfred Hollstein <manfredh@redhat.com> writes:
> > >
> > > > On , 3 March 2001, 16:15:44 -0300, aoliva@redhat.com wrote:
> > > >
> > > > > On Mar 3, 2001, Manfred Hollstein <manfredh@redhat.com> wrote:
> > > > >
> > > > > > I'd say because gcc/testsuite/lib/g++.exp sets up LD_LIBRARY_PATH properly,
> > > > >
> > > > > I don't see where it adds the gcc build directory to LD_LIBRARY_PATH.
> > > > > Only the libstdc++-v3 build directory. But libstdc++-v3 alone isn't
> > > > > enough.
> > > >
> > > > In g++.exp you'll find this:
> > > >
> > > > proc g++_link_flags { args } {
> > > > global rootme
> > > > global srcdir
> > > > global ld_library_path
> > > >
> > > > ...
> > > > append ld_library_path ":${rootme}"
> > > > ...
> > > >
> > > > gcc/testsuite/site.exp contains this:
> > > >
> > > > set rootme "/home/gnu/work/GNU/gcc-3.0-i586-redhat7-linux-gnu/gcc"
> > > >
> > > > I'd say, we need something similar in the objc*.exp files. If I find
> > > > some time in the next week, I'll send a patch unless someone else beats
> > > > me to it...
> > >
> > > Already done. I've discovered this independently and posted a patch:
> > >
> > > http://gcc.gnu.org/ml/gcc-patches/2001-03/msg00125.html
> >
> > This is cool. Has it been approved by anyone so far? If not, I'd suggest
> > you split your patch into the testsuite and the compiler related bits and
> > re-submit it. The testsuite one should then go in under the "fixes an
> > obvious bug" rule. I agree with Gaby, though, that we should try to find
> > a more generic setup which we can use for all languages. But, that would
> > be a mid-term task IMHO.
>
> That patch does not fix the bug I have. We need the correct path at
> link time - but with Rainer's we only have it at execution time :-(.
I don't see how libgcc_s.so's location is different between link and
execution time. It's in ${builddir}/gcc and this is what $rootme points
at.
> I've posted a patch to fix this to gcc-patches (Subject: "Fix objc
> testsuite (was: Re: objc failures on branch)" and that discussion
> showed where the real problem is.
>
> All of this should not imply that Rainer's patch is wrong - it just
> fixes a different problem.
As I said, I'm not an Objective-C expert at all, but the small patch
to add $rootme to ld_library_path fixes the problem with not being
able to find libgcc_s.so at execution time.
>
> Andreas
> --
> Andreas Jaeger
> SuSE Labs aj@suse.de
> private aj@arthur.inka.de
> http://www.suse.de/~aj
Cheers.
l8er
manfred