[FFmpeg-devel] [PATCH] Set pix_fmt in svq3_decode_init

Måns Rullgård mans
Mon Dec 14 01:35:41 CET 2009

Mike Melanson <mike at multimedia.cx> writes:

> M?ns Rullg?rd wrote:
>> Mike Melanson <mike at multimedia.cx> writes:
>>> M?ns Rullg?rd wrote:
>>>> Michael Niedermayer <michaelni at gmx.at> writes:
>>>>> On Sat, Dec 12, 2009 at 04:49:46PM +0530, Jai Menon wrote:
>>>>>> Hi,
>>>>>> pix_fmt was earlier set in h264.c:decode_init but was later moved to
>>>>>> h264.c:decode_slice_header. So currently, the svq3 decoder doesnt set
>>>>>> avctx->pix_fmt (issue 1591 on roundup). Attached is a trivial patch to
>>>>>> fix it but please correct me if i'm missing something.
>>>>>> -- 
>>>>>> Jai Menon
>>>>>>  svq3.c |    1 +
>>>>>>  1 file changed, 1 insertion(+)
>>>>>> 73df4de9dbca537d210deb564768f9c82e920413  svq3_decode_fix.patch
>>>>> Should be ok in principle, i have something similar in my tree, the
>>>>> reason why i didnt commit it was that fate should be failing and it
>>>>> seems it does not.
>>>>> Once the fate issue is understood this can be commited of course
>>>> The SVQ3 test is disabled because it was causing fate to fall over
>>>> due
>>>> to non-utf8 characters in the metadata printed from that file.  Now
>>>> that we know this to be cause of those fate errors, that should be
>>>> dealt with one way or another so the test can be turned on again.
>>>> Where's Mike?
>>> Always listening. Since SVQ3 is supposed to work again (or at least,
>>> not make FATE fail in unusual ways), I have enabled the corresponding
>>> test spec.
>> What did you do?
> I flipped the switch from "don't test SVQ3" -> "test SVQ3". I see that
> icc 11.0 on x86_64 is back to having the same problem where it only
> inserts 127 tests but all the rest of the configs are fine.

So what happened to the death by non-ascii problem?

M?ns Rullg?rd
mans at mansr.com

More information about the ffmpeg-devel mailing list