[Bug 751311] rtp: Dynamic dropout / reorder limits

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=751311

Sebastian Dröge (slomo) <slomo at coaxion.net> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #306447|none                        |reviewed
             status|                            |

--- Comment #39 from Sebastian Dröge (slomo) <slomo at coaxion.net> ---
Review of attachment 306447:
 --> (https://bugzilla.gnome.org/review?bug=751311&attachment=306447)

::: gst/rtpmanager/rtpstats.c
@@ +48,3 @@
+
+  new_ts = ctx->last_ts;
+  new_ts = gst_rtp_buffer_ext_timestamp (&new_ts, ts);

Return value is unnecessary here, new_ts is already updated by the function

@@ +62,3 @@
+  } else {
+    ctx->avg_packet_rate = (ctx->avg_packet_rate + new_packet_rate + 1) / 2;
+  }

Add a comment here why you do this special averaging. The goal being, that
higher packet rates "win". If there's a sudden burst, the average will go up
fast, but it will go down again slowly.


Because otherwise it would happen that a bursty stream (think of 0.1fps
stream... lots of packets every 10 seconds and then nothing) would have a very
low packet rate... but during the bursts we have a lot of packets close to each
other and should allow a higher reorder/dropout there. Otherwise it might lead
to false positives too easily.


This part here needs to be tested a lot under all possible workloads to not
introduce regression, I would defer this to after 1.6.

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