[FFmpeg-devel] FATE & Regressions (and PPC is broken)

Måns Rullgård mans
Wed Mar 12 23:43:31 CET 2008


Mike Melanson <mike at multimedia.cx> writes:

> Diego Biurrun wrote:
>> On Wed, Mar 12, 2008 at 02:40:04PM -0700, Mike Melanson wrote:
>>> The following regression test is broken on PowerPC (gcc 4.1.2):
>>>
>>> diff -u -w
>>> "/home/melanson/ffmpeg/ffmpeg-main"/tests/ffmpeg.regression.ref
>>> tests/data/vsynth.regression
>>> --- /home/melanson/ffmpeg/ffmpeg-main/tests/ffmpeg.regression.ref
>>> 2008-03-08 19:10:07.000000000 -0800
>>> +++ tests/data/vsynth.regression        2008-03-12 14:27:53.000000000 -0700
>>> @@ -197,7 +197,7 @@
>>>  353368 ./tests/data/a-flac.flac
>>>  c4228df189aad9567a037727d0e763e4 *./tests/data/flac.vsynth.out.wav
>>>  stddev: 33.31 PSNR:65.87 bytes:1040384
>>> -a1e56bb034c2773438ba1c830c4cea07 *./tests/data/a-wmav1.asf
>>> +59575b4756d674cd8c8d53d86c08e8cb *./tests/data/a-wmav1.asf
>>>  106004 ./tests/data/a-wmav1.asf
>>>  stddev:12251.50 PSNR:14.56 bytes:1056768
>>>  stddev:2106.00 PSNR:29.85 bytes:1048576
>>> make: *** [codectest] Error 1
>>>
>>> If anyone can suggest a revision where it might have last worked, I can
>>> test it.
>> 
>> I think it has never worked, i.e. it was broken ever since Michael added
>> WMA encoding.
>
> Thanks. I'm glad I didn't try to bisect the problem.
>
> This does make the spec a little less useful, though, in terms of FATE.
> Proposal: Remove this particular spec from the formal regression suite
> and move it to its own FATE test spec. It will still fail on PPC but at
> least the baseline regression suite will pass on PPC and we will be
> alerted to legitimate breakages.

Better yet, find out why it breaks on ppc.

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




More information about the ffmpeg-devel mailing list