Hi.
I've had some advice on this from various folks, one of whom is a Pure
Data developer, and they pointed out that, as a separate process, I
can set audio to a higher priority than the GUI.  I've looked at a few
references, and now I think this is the way to go to avoid GUI-induced
dropouts in my Python sequencer:

Import the csnd library.

Run a dummy .csd whenever the user wants to query for audio devices;
parse the output for available devices.

When the user hits play(), start another instance of *python*, using
the Csound API within it to perform a realtime score.

Drop the priority (or raise the nice level, on Unix) of the parent process.

Pipe rt line events to the audio process from the parent process.

On stop(), kill the child process, like the godless Democrat I am, and
reset the GUI's priority/nice level.

The reason for using an instance of Python running the API instead of
just starting a Csound instance for audio is twofold: I've already
included the module in my app to query audio devices, and it's simpler
to distribute Csound with my app as a Python module than as a separate
executable.

Does anyone have any thoughts on how well this will work?  It'll be
some work to set it up, but hopefully I can at least test the priority
thing this weekend.  I've known for a while that Pure Data works as
several processes, one of which is basically Wish, but never quite
understood why.  And Python doesn't let one set threads to different
priorities.

-Chuckk

-- 
http://www.badmuthahubbard.com