On Mon, Sep 29, 2014 at 07:27:02PM +0200, Clément Bœsch wrote:
> On Mon, Sep 29, 2014 at 05:42:09PM +0200, Nicolas George wrote:
> > L'octidi 28 fructidor, an CCXXII, Clément Bœsch a écrit :
> > > ---
> > >  libavcodec/srtdec.c       | 33 +++----------------------------
> > >  libavcodec/srtenc.c       | 49 
> > > ++---------------------------------------------
> > >  libavformat/matroska.c    |  1 -
> > >  libavformat/matroskaenc.c | 45 
> > > +------------------------------------------
> > >  libavformat/srtenc.c      | 11 ++++-------
> > >  5 files changed, 10 insertions(+), 129 deletions(-)
> > 
> > I am in favour of doing this, and a quick glance at the code seems ok.
> > 
> > I have a small doubt about the handling of "move": it seems that with the
> > timed srt case, \move(x,y) is translated into "X: Y:" in the timestamp line
> > of the packet payload, while in the clean srt case it is dropped: should it
> > not be translated into side data? That is not something changed by the
> > patch, but the patch removes the "move" callback altogether, making the
> > issue less obvious, if there is one.
> > 
> 
> Ah you are right. But it's indeed not a regression, and fixing that would
> mean to rework the subtitles encode API: it currently takes a buffer and
> not an AVPacket, so the encoder is not able to set the side data...
> 

Added a TODO in the empty callback and pushed.

-- 
Clément B.

Attachment: pgp0NMpoo621T.pgp
Description: PGP signature

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Reply via email to