[MPlayer-users] (no subject)

[Uoti Urpala]

On Sat, 2010-11-27 at 19:19 +0000, Carl Eugen Hoyos wrote:
> Uoti Urpala <uoti.urpala <at> pp1.inet.fi> writes:
> 
> > > > > http://roundup.ffmpeg.org/issue2246
> > > > 
> > > > Where did you get that "This is a regression since r22301" in the bug
> > > 
> > > Maybe testing?
> > 
> > I wonder since I tested an earlier version and saw no difference.
> > 
> > > > report from? Can you play this file with ffplay from r22301?
> > > 
> > > Neither do I claim this (but the opposite) nor do I ever test ffplay (except
> > > when explicitly mentioned).
> > 
> > What do you mean by "but the opposite"? That what you said in the bug
> > report should be interpreted to mean it did _not_ work in r22301?
> 
> I don't see how it could be interpreted differently (but I am not a native
> speaker, so who knows).

You meant that it's been broken since r22301? That revision didn't cause
the problem though. I looked into the issue and seems it's caused by a
large value of the "dts_shift" field in mov.c; the "raw" dts values
differ a lot and are then corrected when calculating pts, but various
code uses the "raw" values and so fails.