[FFmpeg-devel] WMA regression still broken / decouple regression rules

[Robert Swain]

2008/9/2 Michael Niedermayer <michaelni at gmx.at>:
> On Tue, Sep 02, 2008 at 04:51:52PM +0200, Aurelien Jacobs wrote:
>> On Tue, 2 Sep 2008 15:51:43 +0200
>> Michael Niedermayer <michaelni at gmx.at> wrote:
>>
>> > On Tue, Sep 02, 2008 at 02:10:03PM +0200, Benjamin Larsson wrote:
>> > > Robert Swain wrote:
>> > > > 2008/9/2 Robert Swain <robert.swain at gmail.com>:
>> > > >
>> > > >> 2008/9/2 Michael Niedermayer <michaelni at gmx.at>:
>> > > >>
>> > > >>> On Mon, Sep 01, 2008 at 10:11:14PM -0700, Mike Melanson wrote:
>> > > >>>
>> > > >>>> Hi,
>> > > >>>>
>> > > >>>> For most of the x86_32 FATE configurations, (all except icc and gcc
>> > > >>>> 2.95.3), the FFmpeg regressions are breaking on the WMA test. It was
>> > > >>>> last working on 14980:
>> >
>> > 14980 does not work on x86-64
>> >
>> > more preissely:
>> >
>> > @@ -205,7 +205,7 @@
>> >  353368 ./tests/data/a-flac.flac
>> >  95e54b261530a1bcf6de6fe3b21dc5f6 *./tests/data/flac.vsynth.out.wav
>> >  stddev:    0.00 PSNR:999.99 bytes:  1058444/  1058444
>> > -4435d87463cd6c5407bd88cca241ca56 *./tests/data/a-wmav1.asf
>> > +6f7f3116b801ea641ce14f827de05cce *./tests/data/a-wmav1.asf
>> >  106004 ./tests/data/a-wmav1.asf
>> >  stddev:12251.38 PSNR: 14.56 bytes:  1056812/  1058444
>> >  stddev: 2105.96 PSNR: 29.85 bytes:  1048620/  1058444
>> >
>> > when was it last working on x86-64 ?
>>
>> IIRC the last working version on x86-64 was r14756.
>> And I think this is the last version which worked with all arch/compiler.
>> r14757 which broke regtests is:
>>   Change wma.c to use the ff_sine_window_init() from mdct.c
>
> and does reverting it fix it again ? if so lets revert it!

There have been other commits to the same code since and the
ff_sine_window*[] in mdct.c become redundant if they aren't used in
wma.c. And those used by aac.c don't even need to be shared. So it's
basically reverting the entire suggestion to share the window
coefficients.

Are we concluding that this is a GCC bug in the more commonly used GCC
versions (i.e. newer than 2.9*) on x86_32?

Rob