[Bug 780446] gst_rtp_buffer_ext_timestamp() math is incorrect for big jumps

GStreamer (GNOME Bugzilla) bugzilla at gnome.org
Wed Jul 26 12:45:18 UTC 2017


https://bugzilla.gnome.org/show_bug.cgi?id=780446

--- Comment #15 from Nicolas Dufresne (stormer) <nicolas at ndufresne.ca> ---
I think you missed that the current implementation does not support backward
wraparound (which is really important if you want to use it for disordered
packet). Supporting both direction reduces the maximum distance you can
tolerate between two timestamp, though I haven't referred to the spec to see
what's should be tolerated (left as an exersize).


(In reply to Miguel París Díaz from comment #14)
> In conclusion, I would update the implementation of
> gst_rtp_buffer_ext_timestamp and the doc.
> What do you think?

I think it's per use case. From jitterbuffer perspective, it should in my
opinion reorder first, and then set timestamp. The monotonic behaviour of 
gst_rtp_buffer_ext_timestamp () need fixing, so if the timestamp are
effectively backward after reordering, we just force them forward.

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


More information about the gstreamer-bugs mailing list