Hi again-
Having fooled with the Pi a few more hours I think I now have a better idea how the thing behaves.
The main difficulty I've had is that having a USB keyboard and/or mouse connected to the Pi seems to conflict with USB audio devices. The main symptom is that Pd's GUI freezes. This seems to happen whether the GUI is running on the Pi's own X server or over an ssh connection.
The best mileage I can get is to ssh over from another machine, so that the only USB devices on the pi itself are the audio device and the network adapter (which is implemented as a USB 2.0 device.) I'm suspicious that the addition of a keyboard forces USB down to 1.1, but things break even if I plug in a keyboard via a USB 2.0 hub. (It's an old hub though, maybe a newer one would work better?)
I was then able to get audio output going with only occasional glitches (3 or 4 per minute I think.) I was never able to get input and output both working at once without manifold glitches.
I'm using Pd 0.44 (test version) for this, which I've just fixed up to not require the tcl and tk h files to compile Pd (why I didn't clean that out earlier I have no idea :) It's still necessary to run:
apt-get install libasound2-dev
before Pd wil be able to use the ALSA audio. Then grab Pd from git:
git clone git://pure-data.git.sourceforge.net/gitroot/pure-data/pure-data
(Maybe you'll have to apt-get install git first), then cd to pd/src, ./configure and make. Takes about 17 minutes to compile.
And by the way, don't use OSS audio - it's not running by default but if youturn it on and try it out; it instantly crashes the Pi.
WHAT I DON"T KNOW YET: I don't have an HDMI monitor and so can't test the "HDMI audio". I imagine it might be more robust than using USB audio device (and, by the way, in 0.44 you can sort of get away with using separaate devices for input and output, although there's no correction for clock drift).
cheers Miller