Re: [jokosher-devel] Seek bug

[Jan Schmidt <thaytan noraisin net>]

On Thu, 2006-07-13 at 17:54 +0100, Jono Bacon wrote:
> Hi folks,
> 
> Well, we seem to have one bug left that is impacting 0.1 running
> smoothly, and it is the seek bug. See
> http://jokosher.python-hosting.com/ticket/119 for more details.
> 
> Basically, when seeking with Jokosher by clicking on the timeline
> while playing, after so many clicks Jokosher hangs. This only ever
> happens with more than one instrument (for a single instrument the
> problem does not exist). Because of this it smells like an adder
> problem.
> 
> I also thought it may be a race conditions issue, but it seems not. I
> can slowly click the timeline to seek, but it will still crash. The
> number of clicks required for crash seems random too, although I did
> notice some similarity at times (five and nine clicks).

Sounds like coincidence to me. It's more likely about the precise timing
between seeks and what state the pipeline is in at the time.

> I did a bit of debugging today to try and figure it out, and the debug
> log does have a load of extra lines around the time of the crash, but
> I have not had much time this week to poke into it. I thought I would
> just open it up now we have a mailing list and see what we can learn
> about the problem.
> 

Have you attached the log file to a bugzilla report somewhere? Don't
forget that GStreamer core and Plugins Base are frozen for new releases
at the moment, so unless the bug is found and fixed in the 2-3 days, it
won't make it into this release. In that case, you're probably looking
at fixing the bug later and having to temporarily include adder within
the jokosher tarball until a fixed GStreamer Plugins Base is released
about a month later. That is, if it is in fact an adder bug.

Cheers,
J.