[MPlayer-dev-eng] Potential source of A-V sync issues

[Guillaume POIRIER]

Hi,

On 2/18/06, Matthias Hopf <mat at mshopf.de> wrote:
> On Feb 17, 06 13:42:37 +0100, Guillaume POIRIER wrote:

> > However, I'd be interested if you could upload a _short_ sample
> > somewhere (you could try ftp://213.144.138.186/MPlayer/incoming/) to
> > see if your problem is reproductible on my machine.
>
> This will be difficult, as
> a) I'm working with HDTV material and

If there is actually a A/V sync problem it should happen no matter the
resolution, so if could could come up with a downscaled sample that
would exhibit the problem, it would really help.

Well, I suppose any sample out there could do the trick, but as I
don't have any A/V sync problem, it's hard for me to help you! :(


> b) this is more some theoretical problem ATM.
>
> I haven't encountered a stream with an issue WRT this yet, but AFAICS it
> is theoretically possible, that skip_flag is non-zero at the end of the
> main loop, and its value should be preserved for the next iteration.
>
> I just analyzed the mencoder a/v sync system, and stumbled upon this,
> and thought that one of the main authors of the main loop could
> immedeately accept or decline this patch. It's completely possible that
> I missunderstood something in the code, but my conclusions seem
> plausible (to me, that is ;^)

Unfortunately, to the best of my knowledge there's no MEncoder
maintainer. The ppl who would have a chance to help you are the ones
who have added features to MEncoder: Oded, Nico Sabi and Corey. I hope
they will help you out.


> > I'd also suggest trying Corey Hickey's patches to improve MEncoder:
> > http://archives.free.net.ph/thread/20060217.023945.cb6b19b8.en.html
> > http://archives.free.net.ph/thread/20060217.090827.3fadf13f.en.html
>
> I'm following this discussion closely, and they fix other much more
> important long-standing issues.

Then don't be shy, please try the patches and report if the patches
work all right or if you have issues with them. It's very important to
have ppl test patches and discuss the way the code has been written.
We need and welcome testers.
Testing patches is considered as less sexy than coding by lotsa ppl,
which is a bummer because testing improves software quality greatly.

Guillaume

--
Just because code is syntactically "valid" GNU C doesn't mean gcc can
always compile it.
  Steven Bosscher - 2005-01-01
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11203#c14