[pulseaudio-tickets] [Bug 65703] High latency in HTTP streaming

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=65703

--- Comment #60 from Tanu Kaskinen <tanuk at iki.fi> ---
I think I have now a bit better idea of how source rewinding works. I still
don't think that setting a rewind callback for the source output implementation
is a good idea. It does reduce the latency, because data doesn't get buffered
in pa_source_output.thread_info.delay_memblock if the callback is set. However,
the delay_memblockq exists for a reason: clients can't handle the stream time
going backwards, which is what happens on rewinds, so there must be a buffer
somewhere that hides the rewind events from clients. If you disable the buffer
in pa_source_output, you should implement a similar buffer in protocol-http,
but your patch doesn't implement that. The result is that your patch most
likely causes non-continuities in the audio, when rewinds occur.

If the latency caused by the delay memblockq is too large, the solution is to
request lower latency from the source output. This should reduce the maximum
rewind size, and therefore also the size of the delay memblockq.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/pulseaudio-bugs/attachments/20130909/f938aa5c/attachment.html>