[vlc-devel] Re: More clocking observations ...

[John Michael Zorko]

Gildas,

> You will find the real problem in modules/demux/mpeg/audio.c. Looking 
> at the
> way the mpeg audio demuxer emulates PCRs, you don't need to reset the 
> clock
> after a seek().
> Just try to put: p_pgrm->i_synchro_state = SYNCHRO_OK; just before the 
> call
> to input_ClockManageRef(). That should take care of the problem.

This helps quite a lot -- the wobble is far less pronounced than it was 
before.  It still appears, though, from time to time.  I've been 
wondering -- perhaps the extra time it takes to seek backwards in the 
file is something that should be compensated for when figuring how long 
to mwait() before reading again?  In other words, perhaps 
input_ClockManageRef() should be aware that we've seeked _and_ read, 
rather than just read, so it could make adjustments?

I say this because the issues I still hear are predominantly times 
where ff / rewind is active, and the audio gets a bit slower for a bit, 
then recovers, then gets slower again, etc.  I still think that seeking 
and input_ClockManageRef() should be connected somehow.  If I could 
measure the time it took to seek and read as a mtime_t, and somehow 
make input_ClockManageRef() aware of it, it could decrease the mwait() 
time by that amount.

Thoughts?

Regards,

John, who really wants to see VLC become the kick-ass cross-platform 
media player that it has the potential to be, and who wants to help 
bring it there

Falling You - exploring the beauty of voice and sound
New EP, "hope thrown down," available now at
http://www.mp3.com/fallingyou











-- 
This is the vlc-devel mailing-list, see http://www.videolan.org/vlc/
To unsubscribe, please read http://developers.videolan.org/lists.html
If you are in trouble, please contact <postmaster at videolan.org>