2018-08-23 14:51 GMT+02:00, Paul B Mahol <one...@gmail.com>:
> On 8/23/18, Carl Eugen Hoyos <ceffm...@gmail.com> wrote:
>> 2018-08-23 14:29 GMT+02:00, Paul B Mahol <one...@gmail.com>:
>>> On 8/23/18, Carl Eugen Hoyos <ceffm...@gmail.com> wrote:
>>>> 2018-08-23 11:11 GMT+02:00, Paul B Mahol <one...@gmail.com>:
>>>>> On 8/23/18, Carl Eugen Hoyos <ceffm...@gmail.com> wrote:
>>>>>> 2018-08-22 21:24 GMT+02:00, Paul B Mahol <one...@gmail.com>:
>>>>>>> On 8/22/18, Carl Eugen Hoyos <ceffm...@gmail.com> wrote:
>>>>>>>> 2018-08-22 18:00 GMT+02:00, Paul B Mahol <one...@gmail.com>:
>>>>>>>>
>>>>>>>>> + switch (avctx->bits_per_coded_sample) {
>>>>>>>>> + case 12:
>>>>>>>>> + s->stride = 3LL * FFALIGN(avctx->width, 8) >> 1;
>>>>>>>>> + break;
>>>>>>>>> + default:
>>>>>>>>> + return AVERROR_INVALIDDATA;
>>>>>>>>> + }
>>>>>>>>
>>>>>>>> Why are the condition and the error needed?
>>>>>>>
>>>>>>> Because only that is supported.
>>>>>>
>>>>>> Do valid samples with other values exist?
>>>>>
>>>>> No.
>>>>
>>>> Then I suggest to make the whole block above just:
>>>> s->stride = 3LL * FFALIGN(avctx->width, 8) >> 1;
>>>
>>> Nope, bits per coded sample must be checked.
>>
>> Could you explain why?
>
> Reference decoder checks it, so do we.
Shouldn't we try to decode files if at all possible?
>> Am I correct that no other decoder does that?
>
> Nope.
Which decoder does it?
Carl Eugen
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
