[FFmpeg-devel] [PATCH] avformat: Support merging side data before the muxer

[Nicolas George]

Le jour de la Récompense, an CCXXII, Michael Niedermayer a écrit :
> user applications and libs which interface to FFmpeg or libavformat
> through a URLProtocol or AVIOContext receive the AVPacket.data but
> not AVPacket.side_data but the side data is often essential

Yes, I know that. And this is fact slightly wrong: the applications receive
the muxed binary stream, which contains probably (but not always, seem
framecrc) the packet data and possibly all or part of the packet side data.

What I ask is what the actual use case: if an application wants to access
the packet data and side data, why would it use a muxer and capture its
output instead of examining the packets directly?

> With this patch, such applications can set the flag and would
> receive the complete data stream. The alternative would be for such
> libs to be redesigned to interface to FFmpeg or libavformat
> differently

You mean to redesign the applications, I suppose? In that case, I think this
is the only proper solution: when I read this patch, my immediate reaction
was: someone is misusing the API, does not manage to achieve the desired
result and thus wants to misuse the API even more.

Regards,

-- 
  Nicolas George
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <https://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20140921/f8a9af0b/attachment.asc>