[MPlayer-dev-eng] [RFC] rc2 at the beginning of October

[Diego Biurrun]

On Wed, Sep 26, 2007 at 09:59:00PM +0200, Roberto Togni wrote:
> On Tue, 25 Sep 2007 09:23:32 +0200
> Diego Biurrun <diego at biurrun.de> wrote:
> 
> [...]
> > > I just quickly checked game formats and noticed the following issues and
> > > regressions:
> > > 
> > > * http://samples.mplayerhq.hu/game-formats/idroq/
> > >   used to work fine in rc1, now fails during colorspace negotiation.
> > 
> > OK, I found and fixed the colorspace issue, but it's still not working.
> > Now I get tons of the following messages and no picture:
> > 
> > [VD_FFMPEG] DRI failure.
> > [roqvideo @ 0x888fbf0]pic->data[0]!=NULL in avcodec_default_get_buffer
> > [roqvideo @ 0x888fbf0]  RoQ: get_buffer() failed
> > Error while decoding frame!
> > 
> > This problem was introduced by FFmpeg r9168:
> > 
> > ------------------------------------------------------------------------
> > r9168 | michael | 2007-06-01 23:17:05 +0200 (Fri, 01 Jun 2007) | 3 lines
> > Changed paths:
> >    M /trunk/libavcodec/roqvideo.h
> >    M /trunk/libavcodec/roqvideodec.c
> > 
> > fix skipped blocks
> > fixes decoding of
> > (http://samples.mplayerhq.hu/game-formats/idroq/bf2introseg.roq)
> > ------------------------------------------------------------------------
> > 
> > Reverting this commit fixes the problem, as does applying the following
> > hunk to my local tree:
> > 
> > --- libavcodec/roqvideodec.c     (revision 10571)
> > +++ libavcodec/roqvideodec.c     (working copy)
> > @@ -220,5 +220,5 @@
> >      NULL,
> >      roq_decode_end,
> >      roq_decode_frame,
> > -    CODEC_CAP_DR1,
> > +    NULL,
> >  };
> > 
> > Any further ideas?
> 
> The attached patch fixes it without touching ffmpeg, disabling direct
> rendering as already done for other codecs.

Then I guess you should apply your patch and we will have successfully
worked around this regression.

Diego