[FFmpeg-devel] fate : clang x86

Alex Converse alex.converse
Thu Sep 2 04:19:31 CEST 2010


On Wed, Sep 1, 2010 at 10:06 PM, Michael Niedermayer <michaelni at gmx.at> wrote:
>
> On Mon, Aug 30, 2010 at 06:02:04PM -0400, Ronald S. Bultje wrote:
> > Hi,
> >
> > 2010/8/30 M?ns Rullg?rd <mans at mansr.com>:
> > > "Ronald S. Bultje" <rsbultje at gmail.com> writes:
> > >> 2010/8/30 M?ns Rullg?rd <mans at mansr.com>:
> > >>> Alex Converse <alex.converse at gmail.com> writes:
> > >>>>>> Perhaps you should look at the yasm STRUC macro
> > >>>>>
> > >>>>> I am not going to recode MpegEncContext into yasm.
> > >>>>
> > >>>> You don't have to recode it, just write a little script to generate
> > >>>> the yasm STRUC from the C struct
> > >>>
> > >>> It's not that easy. ?It recursively contains other structs and has
> > >>> members with forced alignment. ?I wouldn't trust anything but the
> > >>> compiler to get it right.
> > >>
> > >> Compile a C source file using MpegEncContext that printf()s the offset
> > >> of relevant bits, run that script during compilation to output an
> > >> includable .asm file that contains the correct offsets and include
> > >> that .asm file in the yasm file.
> > >
> > > Breaks cross-compilation.
> >
> > Time to fix the DSP functions to stop taking huge structs then?
>
> that too but its not a solution here, this here really is a
> buggy compiler and so far no reasonable workaround has been suggested.
> Thus de facto that compiler does not work with ffmpeg in that configuration
> And i do think we should drop support for that compiler if its maintainers
> dont care enough to fix such bugs.
>

Clang has many benefits as a development compiler including much
better error and warning messages than GCC and is IMHO worth trying to
support if at all possible.

--Alex Converse



More information about the ffmpeg-devel mailing list