[Bug 739868] rtpmanager: rtpjitterbuffer fixes and improvements

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=739868

--- Comment #37 from Sebastian Dröge (slomo) <slomo at coaxion.net> ---
(In reply to Miguel París Díaz from comment #31)
> (In reply to Sebastian Dröge (slomo) from comment #15)
> > Comment on attachment 290316 [details] [review] [review]
> > 0005-rtpstats-RTP_MAX_MISORDER-as-3000.patch
> > 
> > There seems to be a real problem here but instead of increasing the number
> > we should only reset when we're supposed to: i.e. not when a) we lost 100
> > packets in a row, or b) got a rtx packet with a seqnum gap >100.
> > Resetting the stats in these two cases seems wrong.
> > 
> > 
> > Question now is how we can distinguish between these cases, and the case
> > when we actually want to reset the stats (because we think a new session
> > started?)
> 
> Could we accept this patch for the moment and then think about a better
> solution?

No, but I think I have a reproducible testcase for this now... and also think I
know what the problem is. The fix will only work properly if you use the
rtxsend/rtxreceive elements, not rtxqueue as you currently do (which is also
not conforming to the RFC).

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.