On 12/08/13 11:08, Hendrik Leppkes wrote:
> On Mon, Aug 12, 2013 at 10:48 AM, Christian Schmidt <[email protected]> wrote:
>> On 10/08/13 18:47, Diego Biurrun wrote:
>>> A merge makes one thing out of two or more. This does not seem to be 
>>> happening here.
>>
>> That is exactly what it does.
>>
>>> Am 09.08.2013 um 20:08 schrieb Christian Schmidt <[email protected]>:
>>>
>>>> This patch removes the header decoding for PCM audio from
>>>> libavformat/mpeg.c and the 20/24bit parts from libavcodec/pcm.c,
>>>> and merges them into a new decoder in libavcodec/pcm-dvd.c.
>>>
>>> This sentence tells you to leave out "this patch". Or wait, just leave out 
>>> "this patch". ;-)
>>
>> It tells you that the patch takes the header parsing and mode decision
>> from the demuxer (where it does not belong by any means), and the
>> 20/24bit packed sample decoding from the PCM decoder (where it is kind
>> of broken as samples can span multiple packets, which the decoder does
>> not handle, and instead errors out with "invalid PCM packet"). Both
>> parts together make up the pcm_dvd decoder, and are merged in a new file.
>>
> 
> One thing i've been wondering about the header parsing move, the
> content of the AVPackets returned from the demuxer will change, right?
> Previously it didn't include a header, and in the future i presume it
> will? That change may bite people in the ass that use lavf without
> lavc or the other way around. Such changes should always be considered
> carefully.

Yes, you are right about that.

In my opinion the old way was incorrect however - a demultiplexer should
demultiplex, and not do the first part of decoding, for a single codec.

Regards,
Christian


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

Reply via email to