Hi,

On Sun, Jan 1, 2012 at 8:20 PM, Justin Ruggles <[email protected]> wrote:
> On 01/01/2012 11:06 PM, Ronald S. Bultje wrote:
>
>> Hi,
>>
>> On Fri, Dec 30, 2011 at 2:32 PM, Justin Ruggles
>> <[email protected]> wrote:
>>> On 12/30/2011 05:28 PM, Måns Rullgård wrote:
>>>
>>>>
>>>>
>>>> Diego Biurrun <[email protected]> wrote:
>>>>
>>>>> On Sat, Dec 31, 2011 at 02:06:03AM +0530, Gaurav Narula wrote:
>>>>>>
>>>>>> --- a/tests/fate-run.sh
>>>>>> +++ b/tests/fate-run.sh
>>>>>> @@ -69,6 +69,14 @@ md5(){
>>>>>>
>>>>>> +pcm24(){
>>>>>> +    avconv "$@" -vn -f s24le -
>>>>>> +}
>>>>>> +
>>>>>> +pcm32(){
>>>>>> +    avconv "$@" -vn -f s32le -
>>>>>> +}
>>>>>> +
>>>>>>  pcm(){
>>>>>>      avconv "$@" -vn -f s16le -
>>>>>>  }
>>>>>
>>>>> This is not a test for a 302M decoder, it's infrastructure for testing
>>>>> 24 and 32 bit PCM in FATE.  It's very welcome, but you need to split
>>>>> this patch into two pieces.
>>>>
>>>> I'm afraid there's more to it than that. Please don't commit this until 
>>>> I've had a closer look.
>>>
>>>
>>> This should work fine if the test uses CMP = diff. The oneoff tests
>>> obviously won't work until we add 24-bit and 32-bit support in tiny_psnr.
>> [..]
>>             if(len==2){
>>                 a= (int16_t)(a | (buf[0][++j]<<8));
>>                 b= (int16_t)(b | (buf[1][  j]<<8));
>>             }
>>
>> Can't be that difficult to change that to a loop...
>
>
> Yes, that's fine and needed for other things. But in this case we don't
> need a fuzzy test for a lossless codec. So I think just the md5 test
> should be ok.

Agreed.

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

Reply via email to