[pulseaudio-tickets] [Bug 97799] Discontinuity in the interpolated delay after corking, flushing and uncorking.

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=97799

--- Comment #4 from Niklas Haas <bugs.freedesktop at haasn.xyz> ---
In case it helps at all to debug this issue, I've added some more attachments:

“no interpolation”: the same test as “regular playback” but with interpolation
disabled. This basically shows pulseaudio's internal timing. You can clearly
see that pulseaudio's auto-timing code keeps halving the timing update
frequency, up to a maximum of 1.5s

“no interpolation seeking”: same as “no interpolation” but with two seeks
inserted. Every seek event resets the timing update frequency.

“alternate timing hack”: In this test, INTERPOLATE_TIMING and
AUTO_TIMING_UPDATE were disabled and the call to pa_stream_get_latency was
replaced by the following:

1. pa_stream_update_timing_info + pa_stream_get_timing_info
2. latency = pa_bytes_to_usec(write_index - read_index)
3. return latency - transport_usec + sink_usec

Since this code doesn't do any smoothing, the result is much more rough
overall, but it's nonetheless fairly continuous.

“alternate timing hack seeking”: same as “alternate timing hack” but with seek
events included. There are still some small downwards blips in the reported
delay, but these are in sync with the fact that less data has been written (the
blue line is stable).

It's possible that I could work around this issue by using the “alternate
timing hack” code path and doing smoothing of the “ao-dev” (blue) variable
myself, but I would really rather avoid it.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/pulseaudio-bugs/attachments/20160913/cf400cd3/attachment.html>