scratch team mailing list archive

[Derek O'Connell <doc@xxxxxxxxxxxxxxxxx>]

Hi John,

>What was the critical change?

Essentially nothing different from what I have been doing into the
cleaned-up version of the source code, just setting up PA buffering
parameters and supplying them when opening connections to PA... and I
had done that already in what feels like a million different ways. I'm
usually religious about not making any fundamental changes when I clean
up code but of course I had to extend some structs to accommodate the
buffer parameters so either there are some misaligned pointers or there
is some other simple mistake I have been overlooking. Either way, as
soon as it worked on Saturday I decided to stop while I was ahead of the
game ;-) Later today I will take a closer look at the "cleaned-up"
version to try and see what went wrong.

btw, in this working version I am not even setting up PA buffering for
playback, only for recording. In (PA) theory it's needed for both and in
addition the application is suppose to read back the resulting latency
and adjust accordingly. I'm not doing any of that yet. I won't tell
anyone if you don't :-)

re VMWare, I would have been surprised if it had worked because it
introduces too much delay. Might work for streaming audio output but not
for low latency app's.

Well, going by other emails this morning it looks like this version
works and time is tight for the next release. So if I get the clean-up
version working I will send it off to Ian for inclusion in subsequent VM
builds.

-D