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

[GStreamer (GNOME Bugzilla)]

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

--- Comment #13 from Nicolas Dufresne (stormer) <nicolas at ndufresne.ca> ---
I'm not sure why a sequence like:

  G_MAXUINT32 + 1 -> G_MAXUINT32 + 1
  G_MAXUINT32     -> 3 * G_MAXUINT32 + 2.

Is more valid then:

  G_MAXUINT32 + 1 -> G_MAXUINT32 + 1
  G_MAXUINT32     -> 2 * G_MAXUINT32 + 2.

Both are going forward, so that's compliant with the doc. In practice, there
was no wraparound, so maybe returning G_MAXUINT32 + 1 would be valid. Now, the
theoretical aspect of this issue has no importance. It's a completely useless
to try and fix a behaviour for another just because it seems better. A study of
the function usage is the only useful rational to change this function.
rtpjitterbuffer is the main user, and I think studying that aspect, and
verifying if it may not be using it the way the function was designed is
important. From there, we will be able to construct a unit test that covers
real usages, and potentially fix a real bug. The best test, would be to trigger
a bug from the jitter buffer cause by the current behaviour.

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