> I have changed the jobqueue.cpp code, see results below. I then
> started mythcommflag with the same options from the command line, and
> noticed that mythcommflag segfaults when used with the -l option. So
> unless it's obvious that mythcommflag should segfault when used this
> way from the command line, I guess the next step is debugging with
> GDB? I'll try to look into that in a couple of days if you're
> interested.
The -l and -j options are undocumented command line options to
mythcommflag and only meant to be called when mythcommflag is run
from the JobQueue. "-l" tells mythcommflag that the recording being
flagged is live (ie, still recording) so mythcommflag can throttle
itself. "-j" tells mythcommflag that it was run from the JobQueue
so that mythcommflag will update the status and comment fields in
the jobqueue table. Neither should cause a segfault, so it would be
nice if you could trace this and see where it is faulting at. I
have test scripts that run mythcommflag with those options and I
don't get segmentation faults.
> 2005-04-26 21:50:18.171 Finished, 0 break(s) found.
I'm still perplexed by this. When mythcommflag is run from the
JobQueue, the exit value of mythcommflag is the number of commercial
breaks found. There are only 2 places that mythcommflag would
explicitly return a 0 and neither of those are executed when
mythcommflag is run from the JobQueue. So, it appears that
mythcommflag is really finding 0 breaks. Can you add the following
lines to the bottom of NuppelVideoPlayer::FlagCommercials() right
before the return(comms_found)?
VERBOSE(VB_COMMFLAG, QString("NVP::FlagCommercials found %1 breaks")
.arg(comms_found));
This will verify the flagger is actually returning 0 breaks found.
If that's the case, it seems like the flagger may not be sleeping
like it should in order to let the recorder get ahead. If you want,
you can insert some other "VERBOSE(VB_COMMFLAG" type lines inside
NuppelVideoPlayer::FlagCommercials() to see where the flagger is
at what times and that might possibly help us determine what is
going on.
--
Chris
_______________________________________________
mythtv-dev mailing list
mythtv-dev@mythtv.org
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev
