[mythtv] mythtranscode should give up at some point ...

[Robert Tsai]

Last night's broadcast of "The Simpsons" (My Big Fat Geek Wedding) is
giving me problems manifested in my mythbackend.log:

[mpeg2video @ 0x2a96395030]ac-tex damaged at 12 34
[mpeg2video @ 0x2a96395030]ac-tex damaged at 9 37
[mpeg2video @ 0x2a96395030]ac-tex damaged at 9 38
[mpeg2video @ 0x2a96395030]ac-tex damaged at 11 40
[mpeg2video @ 0x2a96395030]ac-tex damaged at 10 41
[mpeg2video @ 0x2a96395030]invalid mb type in B Frame at 31 42
2005-05-03 07:59:14.153 Deleting /media/myth/recordedtv/1025_20050502233000_20050503000000.nuv.tmp
2005-05-03 07:59:14.222 "Low Quality" transcode of "The Simpsons" recorded from channel 1025 at Mon May 2 23:30:00 2005: Running.
2005-05-03 07:59:14.288 transcode: Transcode Running
2005-05-03 07:59:14.294 "Low Quality" transcode of "The Simpsons" recorded from channel 1025 at Mon May 2 23:30:00 2005: Starting.
2005-05-03 07:59:14.297 transcode: Transcode Starting
...

over and over again for the past 6 hours or so :P.

I haven't started digging yet, but it seems that mythtranscode ought
to recognize when it's killed itself (and return some special exit
status) so that the job queue knows not to do anything more with that
recording?

The recording in question is on my favorite Boston-area flaky digital
broadcast station, FOX-25.

I was going to start working on this kind of logic to prevent jobs
from repeating themselves when they fail (I'm guessing mythcommflag
can have this symptom, and userjobs as well, although I don't know
what can be done about userjobs). Is anyone already working on
something like this?

Is this just a symptom of perhaps some deeper problem in the
decoding/transcoding code?

--Rob