[MPlayer-dev-eng] Updated nailfps filter patches

[xiphmont at xiph.org]

On Nov 11, 2007 2:07 PM, Carl Eugen Hoyos <cehoyos at ag.or.at> wrote:
>  <xiphmont <at> xiph.org> writes:
> > ...or if the demuxer is one of the many that can't report A/V
> > presentation timestamps properly.  If the demuxer fudfes the values
> > out of laziness ("nothing uses them anyway") the code won't work.
> > I'll ad some more debug output so we can see if that's the case.  Is
> > there any output to the console?
>
> Yes, it tells that it shows the first frame repeatedly.
> (I already removed the hard-to-apply-patch, but I'm sure you can test the
> behaviour with any ts sample from samples.mplayerhq.hu)

I had a look; the audio pts reported by the demuxer is garbage.
Nailsync has no hope of making heads or tails of the ts test files I
found.

> > as in 'is unkillable'?
>
> No, as in "echo off".
>
> >  That's the mplayer sighandler biting you (I
> > completely disable it in my builds, it causes me nothing but trouble).
> >  The exit() is only called if nailsync is used without nailfps.
>
> I don't think so: Did you try above command line?

Yes.  On the ts test files I tried, mplayer always locks up and has to
be kill -9'ed (with the nailsync filter).  It deals rather badly with
having the returned PTS being a few thousand years apart.

Monty