[mythtv-users] Using commflag list while it is still

[Chris Pinkham]

> > COMMFLAG_REQUEST event which the flagger receives.  The flagger should then
> > occasionally recompute the commercial skip list and send out a
> 
> How often would be "occasionally"? And do I have to continue watching 
> (stay in playing)?

This appears to have been broken in SVN and should be fixed now with my commits
earlier today.

When the flagger knows that someone is playing the recording being flagged, the
flagger will regenerate the commercial skip list every 500 frames and if the
skip list is different than it was the last time it was generated, the new list
will be sent out via the COMMFLAG_UPDATE event to the frontend.

> Also, I noticed that commflagging starts only 19 minutes after recording 
> started (was scheduled with -10 minutes for begin). Isn't this kinda long?

The show you were recording was scheduled to start at 10:00 and you started
recording at 9:50.  The flagger knows that you started recording earlier
than the scheduled start time, so it will look for the logo starting at a
point which is 10 minutes into the video.  So, this requires at least a
10 minute delay since you started 10 minutes early.  The logo detection
code jumps forward in the video and examines frames across a 6 minute period
so that is another 6 minutes that the flagger has to wait to start.  Tack on
an extra minute or two for a little headroom and you get your 18-19 minutes
that you saw.  The flagger will try to catch up to within 2 minutes of realtime
by sleeping less between every frame until it catches up.  Once it catches up, it
sleeps more or less so that it stays around 2 minutes behind.  Then, once the
recording finishes, the flagger will revert to the normal speed it would run at
if the recording was not live (ie, if you have the JobQueue CPU usage set to High,
the flagger will start running full-speed, if you have the CPU usage set to Low,
the flagger will sleep a little every frame so it still doesn't try to consume
100% of the CPU).
 
> > I'll try to reproduce on my own as well, but please email me the above
> > gzipped (or bzipped) logs.
> 
> Email will be underway in a moment.

Thanks, these helped diagnose and fix the problem.  It was especially nice
that your servers are all synced to the same time so the logs were easy to
follow.

-- 
Chris