Hi,

On Thu, Sep 15, 2011 at 2:35 PM, Alex Converse <[email protected]> wrote:
> 2011/9/15 Måns Rullgård <[email protected]>:
>> Alex Converse <[email protected]> writes:
>>
>>> ---
>>>  tests/fate.mak              |    2 +-
>>>  tests/ref/fate/ea-tqi-adpcm |    1 -
>>>  2 files changed, 1 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/tests/fate.mak b/tests/fate.mak
>>> index a349e5c..a049783 100644
>>> --- a/tests/fate.mak
>>> +++ b/tests/fate.mak
>>> @@ -93,7 +93,7 @@ fate-ea-tgv-ima-ea-eacs: CMD = framecrc  -i 
>>> $(SAMPLES)/ea-tgv/INTRO8K-partial.TG
>>>  FATE_TESTS += fate-ea-tgv-ima-ea-sead
>>>  fate-ea-tgv-ima-ea-sead: CMD = framecrc  -i $(SAMPLES)/ea-tgv/INTEL_S.TGV 
>>> -pix_fmt rgb24
>>>  FATE_TESTS += fate-ea-tqi-adpcm
>>> -fate-ea-tqi-adpcm: CMD = framecrc  -i 
>>> $(SAMPLES)/ea-wve/networkBackbone-partial.wve
>>> +fate-ea-tqi-adpcm: CMD = framecrc  -i 
>>> $(SAMPLES)/ea-wve/networkBackbone-partial.wve -vframes 26
>>>  FATE_TESTS += fate-ea-vp60
>>>  fate-ea-vp60: CMD = framecrc  -i $(SAMPLES)/ea-vp6/g36.vp6
>>>  FATE_TESTS += fate-ea-vp61
>>> diff --git a/tests/ref/fate/ea-tqi-adpcm b/tests/ref/fate/ea-tqi-adpcm
>>> index c79eed5..4fbc117 100644
>>> --- a/tests/ref/fate/ea-tqi-adpcm
>>> +++ b/tests/ref/fate/ea-tqi-adpcm
>>> @@ -49,4 +49,3 @@
>>>  0, 144000, 115200, 0x65fd5e60
>>>  1, 144000, 5936, 0x2174304d
>>>  0, 150000, 115200, 0x0c256424
>>> -0, 156000, 115200, 0xa9cdd8d2
>>> --
>>
>> What bug is this really hiding?
>>
>
> We fall back to a previous frame data if a frame is truncated which is
> an implementation detail due to the use of av_fast_malloc(). This
> looks kind of funny http://imgur.com/a/PySOc#GXrJ2 , isn't correct and
> therefore shouldn't be something we test for.

I agree we shouldn't decode it.

I also think we should remove the data from the fate sample, so that
the -vframes option is no longer necessary. Is that possible?

Ronald
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to