[FFmpeg-cvslog] r10823 - in trunk/libavformat: rm.h rmdec.c

Kostya kostya.shishkov
Thu Nov 1 11:03:40 CET 2007


On Thu, Nov 01, 2007 at 03:02:16AM -0400, Rich Felker wrote:
> On Thu, Nov 01, 2007 at 08:57:37AM +0200, Kostya wrote:
> > On Wed, Oct 31, 2007 at 10:39:35PM +0100, Michael Niedermayer wrote:
> > > Hi
> > > 
> > > On Sun, Oct 21, 2007 at 07:17:28PM +0200, kostya wrote:
> > > > Author: kostya
> > > > Date: Sun Oct 21 19:17:28 2007
> > > > New Revision: 10823
> > > > 
> > > > Log:
> > > > Demux full frames instead of sliced for RealVideo.
> > > > Some changes by Roberto Togni and blessed by him on IRC.
> > > 
> > > i will revert this and the later patches depending on it in 24h if noone
> > > fixes the rm/rv regression tests failure this change caused
> > 
> > I'm trying to fix RM muxer ASAP as it does not produce correct data at all.
> > For now the easiest solution is temporary disable regression tests for .rm
> > I will try to produce solution today though.
> 
> I hope you got the comment from Michael that if the demuxer is
> crashing, there's a bug in the demuxer not just the muxer. Even if
> data is totally invalid or pulled from /dev/urandom or from Gabu's
> encrypted cp volumes the demuxer should not crash but rather fail as
> gracefully as possible. Fixing this crash is essential to making
> Michael (and everyone, I think) happy; without the fix he will still
> want to revert.
> 
> Rich

I've taken liberty to commit the fix for demuxer but I still have to find
how to handle such situations correctly.




More information about the ffmpeg-cvslog mailing list