This is the mail archive of the
gcc-regression@gcc.gnu.org
mailing list for the GCC project.
Re: GCC build failed with your patch.
- To: Zack Weinberg <zack at wolery dot cumb dot org>
- Subject: Re: GCC build failed with your patch.
- From: Alexandre Oliva <aoliva at redhat dot com>
- Date: 31 Jul 2000 23:39:14 -0300
- Cc: GCC regression checker <geoffk at cygnus dot com>, jakub at redhat dot com, gcc-regression at gcc dot gnu dot org
- Organization: GCC Team, Red Hat
- References: <200008010039.RAA25395@sloth.cygnus.com><20000731182607.G4155@wolery.cumb.org>
On Jul 31, 2000, Zack Weinberg <zack@wolery.cumb.org> wrote:
> On Mon, Jul 31, 2000 at 05:39:52PM -0700, GCC regression checker wrote:
>> (This is an automated message.)
>>
>> With your recent patch, GCC does not compile. The last 20 or so
>> lines of the build output are attached. Also attached is a diff of
>> the ChangeLog entries since the last time the regression tests passed,
>> on 2000-07-30T18:20:00Z.
> This is almost certain to be caused by this change at top level:
> 2000-07-31 Alexandre Oliva <aoliva@redhat.com>
> * configure.in (qCXX_FOR_TARGET): Quote `&' characters in
> CXX_FOR_TARGET for sed.
Looks like a good suspect, indeed. But I don't see why it might
fail. Geoff, does your build system sets CXX_FOR_TARGET in the
environment? What do you get if you run:
expr "x-x -x" : 'x\(.*\)' | sed 's,&,\\\&,g'
?
> I know & is a metacharacter on the right hand side of a sed
> s/// expression, but I don't know why there would be an & in
> $CXX_FOR_TARGET to begin with
libstdcxx_flags may contain it when using libstdc++-v3.
> In an i386-linux native build, $(CXX) is set correctly, so the problem
> is likely to be specific to cross compilation.
I'm trying a target=powerpc-eabi build right now.
--
Alexandre Oliva Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist *Please* write to mailing lists, not to me