[FFmpeg-devel] [PATCH] dca.c: In function 'dca_exss_skip_mix_coeffs': 'for' loop initial declaration used outside C99 mode

[Måns Rullgård]

Gianluigi Tiesi <mplayer at netfarm.it> writes:

> On Tue, Feb 01, 2011 at 01:28:10PM +0000, M?ns Rullg?rd wrote:
>> Gianluigi Tiesi <mplayer at netfarm.it> writes:
>> 
>> > On Tue, Feb 01, 2011 at 12:38:16PM +0000, M?ns Rullg?rd wrote:
>> >> Gianluigi Tiesi <mplayer at netfarm.it> writes:
>> >> 
>> >> > I'm still using gcc 4.2 on mingw, and I have to disable c99,
>> >> 
>> >> Why?
>> >
>> > unstable executables up to gcc 4.5
>> 
>> That's a non-answer.
>> 
>
> why not? I've tried to move all my environment to 4.3 but it had
> broken alloca() and a lot of unusable optimizations enabled by -O3,
> mainly ssa tree related, I've reported some to gcc bugzilla
>
> 4.4 suffered by the same problems, at least early releases I've tested
>
> 4.5 from trunk after a while fixed alloca() and some of problems
> with tree opts but e.g. ffmpeg test suite was still failing.
>
> I've tested some 4.6 from trunk but not yet enough usable
>
> I've switched at least three times my env to new gcc but I had to
> recompile back all packages because of executables unstability.
>
> I really don't known why no other one noticed high instability with
> mingw compilers, but I suspect not so much people is trying to
> compile ffmpeg and mplayer with specific cpu
> optimizations. Compiling ffmpeg and mplayer is always a stress for a
> compiler.
>
> It's enough as answer? :D

None of what you describe relates to C99, which gcc has supported
since some 3.x version.

-- 
M?ns Rullg?rd
mans at mansr.com