[mythtv-users] commercial detection speedup option .24

[Gavin Hurlbut]

On Sun, Nov 14, 2010 at 5:37 AM, Brian J. Murrell <brian at interlinx.bc.ca> wrote:
> Has a bug been filed with Ubuntu on this?  If so, do you know the bug
> number and if not, can you describe the symptoms in more detail, or
> better yet, how to reproduce it so that I can file a bug with them?

No I haven't.  I gave up on trying to hunt it down, and just put in a
local work-around.  When I tried to reproduce it outside of mythtv, I
couldn't, so I stopped trying for now.

The symptoms are...  when we get a return back from waitpid (with
status parameter correctly set to receive the status), and do:
  WIFEXITED(status) - it works as expected for exit(), but not for signals
  WIFSIGNALED(status) - works as expected for exit(), but not for signals

It appears that when the child exited, the status int is correct, but
when the child was signalled, we get it with the bytes swapped.  How
this happens... I dunno.  I fear it may be something stupid in eglibc,
but I couldn't find anything in the source of eglibc that looked
suspect.

These symptoms from within mythtv have only (so far) ever been
reported by Ubuntu users.  It manifests itself most commonly as a
large number of commercials flagged when mythcommflag was killed.  To
determine the signal number in such a case, subtract 128 from the
number of commercials it reported, and that's the signal number it was
killed with.

> Indeed, but a bug should be filed upstream also to make sure they fix
> this behaviour.

Good luck with that.  With our new rewrite of myth_system internals
that we just started testing today, who knows, the issue may (or may
not) be gone.  I'll know tomorrow when I test that.