i just uncovered a subtle bug in the audioengine inner loop. it might not affect you, but then again, if it does, it will be nasty.
when i wrote audioengine, the idea was to guarantee that all clients would never be asked to process more frames than was specified in the last call to their set_block_size() method. unfortunately, the engine does not honor this guarantee: in the event of a scheduling delay which causes us to be notified of N periods worth of data/space at once, rather than a single period's worth, the clients are asked to process it all at once, instead of subdividing into N calls to process(). i have fixed the sources, and a fix will be in CVS tonight. it would be worth thinking about whether any code you might be using that was based on this stuff is affected. i know that i have to fix JACK as well, for identical reasons. i'll put out a new version of JACK tomorrow. --p
