[gst-devel] media tests

[Thomas Vander Stichele]

Hi,

> > > There is a bugzilla tracker for this, it's a threading bug. I don't know
> > > how to fix it (yet). Basically, we EOS as we detect streams, which is
> > > not appreciated elsewhere. I'll look for a fix.
> > 
> > So why did it work before ?
> 
> I moved around some code in playbasebin that might be related to this.

So does it work for you ? Could you check what code got moved around ?

> Otherwise, I don't think anything was changed. Maybe discont handling in
> riff, although it's hard to imagine that being related (it'd be bad if
> it was). It's only for very small .wav files, so not a blocker.

Well,
a) still a regression
b) short wav files are typically used for system sounds
c) someone is working on finishing the patch for libgnome to use
gstreamer for these sounds
d) the gnome freeze is approaching

For all of these reasons I think it's a blocker.  It wouldn't be as bad
if the previous releases didn't do it right either, but they did.

 
> > Good luck - I'm running 0.8.8 :) Any other ideas ?
> 
> Tripple check that version 1.104 of gstoptimalscheduler.c was used to
> build your currently running version - also make 100% sure that the
> installed 0.8.8 RPM of -core is what your plugins are linked against,
> and they did not accidently pick up an uninstalled -core or another core
> (e.g. because your registry is using uninstalled plugins).

Hey, it's me :) I really do know what versions I'm running at any given
test.

> If you're unsure, rebuild your
> RPMs,

that's superstition :) If you don't trust my rpms, check the build logs
which are publically available.  the builds are 100% reproducable given
the tools I use to build them, so you can check for yourself.

>  remove your CVS checkouts (or move to a different dir) and re-run
> gst-register as root and remove ~/.gst*. I'm absolutely confident that
> the 1.104 patch on gstoptimalscheduler.c fixed the exact bug you're
> describing. Obviously, I cannot reproduce it. I've checked both 0.8.8
> and current CVS (of -core).

The exact bug I was describing seems to be the exact same thing you
fixed with that audioscale patch, btw.  One thing I have learnt in
GStreamer (and which you probably know too, from having a medical
background) is that similar symptoms don't necessarily mean the same
disease ...

Thanks for the audioscale patch, best patch all year !

Thomas


Dave/Dina : future TV today ! - http://www.davedina.org/
<-*- thomas (dot) apestaart (dot) org -*->
We'll walk right up to the sun
hand in hand
<-*- thomas (at) apestaart (dot) org -*->
URGent, best radio on the net - 24/7 ! - http://urgent.fm/