Re: patches from the graphite branch for review

[Paolo Bonzini <bonzini at gnu dot org>]

From http://cri.ensmp.fr/people/pop/gcc/1142_build.diff:

> --- gcc/Makefile.in     (.../trunk)     (revision 138275)
> +++ gcc/Makefile.in     (.../branches/graphite) (revision 138569)
>
> ...
>
> +# How to find POLYLIB
> +POLYLIBLIBS = @POLYLIBLIBS@
> +POLYLIBINC = @POLYLIBINC@
> +

does not refer to PPL yet.

> BACKEND = main.o @TREEBROWSER@  @GRAPHITE@ libbackend.a $(CPPLIB) $(LIBDECNUMBER)

I would prefer to always link a stub graphite.o and avoid the GRAPHITE 
substitution.

> +
> +# Check for PPL
> +ppllibs=
> +pplinc=

It would be great to factor this code, Cloog's, GMP's and MPFR's into an 
autoconf macro.

I think you shouldn't revert the patch to support in-tree Cloog and PPL. 
 I looked at it, and even if it doesn't work I think it is better to 
have it than not.

BTW, --enable-stage1-languages=c,c++ should not be necessary, because 
anyway the previous-stage C++ compiler is *not* used for CXX in 
subsequent bootstrap stages.  Furthermore, if it was we'd have to build 
3 copies of libstdc++ too.  In the future it might be a good idea, but 
for now no thanks. :-)

Paolo