current glibc vs debian sid

[Franz Sirl]

At 16:34 18.10.2002, Jakub Jelinek wrote:
>On Fri, Oct 18, 2002 at 04:29:52PM +0200, Franz Sirl wrote:
> > At 15:31 18.10.2002, Jack Howarth wrote:
> > >Roland,
> > >    Well on a ppclinux box with RH 8.0 locally built rpms I get the
> > >failure in configure using...
> > >
> > >GCC=gcc CFLAGS="-g -O3" ../configure --prefix=/usr --enable-add-ons=yes
> > >--withou
> > >t-cvs --enable-kernel=2.2.5 --without-tls powerpc-redhat-linux
> > >
> > >which of course isn't using --host or --build. This form of the configure
> > >command only passes if I manually create a powerpc-redhat-linux-gcc 
> symlink.
> >
> > It's a long-standing bug in newer autoconf, even though the naked stating
> > of 'ppc-redhat-linux' is supposed to be backwards compatible, it isn't.
> >
> > The closest possible emulation is "--host=ppc-redhat-linux
> > --build=ppc-redhat-linux --program-prefix=''", but that still lacks the
> > setting of $target_alias without triggering a cross-compile.
>
>Shouldn't we then after running autoconf in the makefile fix the generated
>configure by some sed command?

Not really, I've sent patches to the autoconf people a while ago, but it 
seems they haven't been applied yet. A lot of autoconf stuff may be bitten 
by this bug, not just glibc.


>This sucks badly.

Yeah, it does, looking at your gcc.spec it's quite clear you already had to 
workaround this autoconf behaviour too, since you install the 
${target}-gcc/g++/etc binaries/links :-(.
And on PPC it's even easier to trigger due to the ppc/powerpc naming duality.

Franz.