11:43 <padenot> Ford_Prefect, I'm also experimenting with various
setttings for the buffer_attr here
https://searchfox.org/mozilla-central/rev/b9041f813de0a05bf6de95a145d4e25004499517/media/libcubeb/cubeb-pulse-rs/src/backend/stream.rs#946
11:44 <padenot> I see conflicting information: Chrome's code seem to
work well and does something, but the docs say
something else
11:55 <Ford_Prefect> padenot: what values?
11:55 <padenot> the 5 values to put in pa_buffer_attr for input and output
11:56 <padenot> see this comment https://cs.chromium.org/chromium/src/media/audio/pulse/pulse_util.cc?type=cs&q=pa_buffer_Attr&g=0&l=480
12:09 <Ford_Prefect> padenot: huh, that seems to have changed since the
last time I saw it. minreq = tlength / 4 is good,
particularly for smaller values
12:11 <Ford_Prefect> I'm not sure why Chrome is doing * 3, they should
be okay with * 2 (so minreq will be buffer_size / 4)
12:11 <Ford_Prefect> er, tlength
12:12 <Ford_Prefect> I've never been able to come up with a good
works-for-everything heuristic on the PA side,
hence this divide-by-4 thumb rule
irc.mozilla.org#media, 2019/07/05, CEST