[FFmpeg-devel] [RFC/PATCH] New metadata conversion API

Michael Niedermayer michaelni
Sat Oct 9 22:46:11 CEST 2010


On Sat, Oct 09, 2010 at 09:31:05PM +0200, Anton Khirnov wrote:
> Before this thread falls into oblivion again, could I please hear
> everybody's opinion on how should we fix metadata conversion? (that's
> mainly Michael, Aurelien and Baptiste; of course anybody else is welcome
> to comment as well)
> 
> So far the proposals are:
> 1) Modify the conversion API to fix its problems. This involves adding
> an AVMetadataConv field to AVMetadata to track current format, changing
> the conversion function to work on a single AVMetadata instead of the
> whole AVFormatContext (there's at least one bug in ffmpeg.c related to
> this) and also adding callbacks to conversion tables, since we have to
> do advanced stuff like splitting/merging tags during the conversion
> step.
> This is what my first patchset in this thread does.

If we didnt need the callbacks this would look nice but the callbacks IMHO
make it really messy and defeat the whole idea of having the convertion
vissible. As you again have black box callbacks


> 
> 2) Remove the public conversion API completely. Only generic metadata
> format will be exported by demuxers/expected by muxers. This is
> somewhat simpler than 1), since advanced conversion can be done directly
> in (de)muxers. However the users can't access metadata in the native
> format.
> This is what my second patchset in this thread does.
> 
> 2a) Same as 2), but export both metadata sets.
> 2b) Same as 2), but add an AVOption to choose native format.
> 
> 3) ????
> 
> I am in favor of 2), since IMO accessing native format isn't ever really
> needed. We can later switch to 2b if users start complaining.

Iam also in favor of 2)
also if baptiste thinks the AVMetadataConv tables are usefull we can still
export them for example in AVMetadata but they would not be correct
for splited/merged/convertet fields
That is the tables would just be "interresting information" that wouldnt be
needed or used during normal operation


[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Into a blind darkness they enter who follow after the Ignorance,
they as if into a greater darkness enter who devote themselves
to the Knowledge alone. -- Isha Upanishad
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20101009/e1c6da95/attachment.pgp>



More information about the ffmpeg-devel mailing list