[FFmpeg-devel] [PATCH] QCELP decoder

[Baptiste Coudurier]

Reynaldo H. Verdejo Pinochet wrote:
> Hi Michael, thanks for your quick answer.
> 
> Michael Niedermayer wrote:
>> One point in favor of the parser though is copying the packets into
>> another container, in that case the trash should be discarded
>> somewhere along the way. This though could be done with a bitstream
>> filter or maybe the trash could even be discarded by the mov
>> demuxer (if its only a problem of mov files)
> 
> I still don't think this is the QCELP decoder's job, to account for a
>  given format or transport oddity that is.

For example, AAC decoders supports both mp4 and adts variants, some
even LATM.

Besides many codecs are supposed to work splitted and chunked (MPEG
audio, MPEG video, H264), the need for a parser is only in the case of
stream copying and rewrapping to extract each standalone frame.

> But now that you made the parser's 'raison d'etre' clearer I'd favor
> this change to take place on the mov demuxer (and in the rtp code
> when needed) as long as it can be implemented easily and in a clean
> way, if it doesn't the parser becomes appealing again. I did
> suggested the chance of implementing this change in the demuxer
> before, I just don't think it belongs to the decoder's code.

I didn't see a patch dealing with this, did I miss it ?

Mov demuxer is actually following the correct mechanism to extract
packets/chunks like it is told to, there is no more cruft to add to mov
demuxer.

-- 
Baptiste COUDURIER                              GnuPG Key Id: 0x5C1ABAAA
Key fingerprint                 8D77134D20CC9220201FC5DB0AC9325C5C1ABAAA
checking for life_signs in -lkenny... no