[Bug 754851] aggregator: Use the whole aggregator latency for the queue size, not just the latency property

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=754851

--- Comment #4 from Nicolas Dufresne (stormer) <nicolas.dufresne at collabora.co.uk> ---
Just for more context. Most issues happens when the input buffer size are
slightly bigger then the queue size. The aggregator will slowly consume the
data in the buffer, but will leave the buffer in the queue, which causes
upstream to stay blocked. When we reach the end of that buffer, we process the
last bits, remove the buffer and wait for the timeout. This gives very little
time for upstream to send the next buffer (which was most likely ready before
we decided to unblock).

I thought initially that the solution was to take the buffer out of the queue
immediately, but Olivier's patch has exactly the same effect and is simpler to
implement. With this patch, using only the output buffer duration as latency is
now sufficient for the case your sources have no jitter. Low-latency live
mixing is now possible also.

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