[FFmpeg-devel] [PATCH] EA TGQ video fixes

[Mike Melanson]

Reimar D?ffinger wrote:
> On Wed, May 20, 2009 at 05:48:45PM -0300, Ramiro Polla wrote:
>> On Wed, May 20, 2009 at 5:41 PM, Reimar D?ffinger
>> <Reimar.Doeffinger at gmx.de> wrote:
>>> On Wed, May 20, 2009 at 05:31:48PM -0300, Ramiro Polla wrote:
>>>> On Wed, May 20, 2009 at 5:07 PM, Vitor Sessak <vitor1001 at gmail.com> wrote:
>>>> [...]
>>>>> Anyway, again, does anyone get a different output for:
>>>>>
>>>>> ffmpeg -idct simple -i
>>>>> http://samples.mplayerhq.hu/game-formats/ea-tgq-uv/tgq/v27.tgq -an -t 10 -f
>>>>> framecrc - | md5sum
>>>>>
>>>>> ?
>>>>> I get 2b84588602acda70269bd3e3aa1ce632
>>>> Wine gives:
>>>> e4c063ac0f26e12d63c3e293066eef10
>>> Have you time to test with the not yet applied patch?
>> The one that moves the block to context? It doesn't change anything.
>>
>> I guess it should be reproducible with any wine, so your 1) condition
>> on the previous message is satisfied =)
> 
> No, using Ubuntu 9.04 it is not reproducible (used a hack to avoid usleep due
> to outdated runtime), using
> ./configure --enable-gpl --cross-prefix=i586-mingw32msvc- --enable-memalign-hack --target-os=mingw32 --disable-demuxer=avisynth --disable-demuxer=vfwcap
> 
> results in the (probably) correct 2b8... result.
> This is on an Atom CPU.

Right now, all of the latest builds think that the answer should be 
2d6a5b44b2c86630cf5b5dae0bf2bf5c, according to my RPC data gathering 
tool. But it might be computing MD5 in a different way.

-- 
     -Mike Melanson