On Fri, Sep 10, 2010 at 2:16 PM, Manuel López-Ibáñez <lopeziba...@gmail.com> wrote: > On 10 September 2010 11:42, Richard Guenther <richard.guent...@gmail.com> > wrote: >> On Fri, Sep 10, 2010 at 11:38 AM, Steven Bosscher <stevenb....@gmail.com> >> wrote: >> >>> Not that I want to discourage anyone. Just practical considerations... >>> ;-) I can't believe I'm saing this but: It may be better to spend >>> some effort on making clang work as a GCC front end. >> >> Oh, indeed - I'd welcome patches making "frontend plugins" possible >> and plugging clang. > > I wonder what would actually be needed to implement this? Since clang > keeps around more information than GCC's FEs. New plugin hooks? I > don't think you need FE plugins but a LLVM->GIMPLE converter plug to > the gimplifier. > > Is adding code to GCC to handle LLVM intermediate language acceptable?
I think clang has its own intermediate language (aka parse tree). I'd write a clang -> GENERIC translator. The plugging place would be the various langhooks called by cgraphunit.c. Richard. > Cheers, > > Manuel. >
