Comment 34 for bug 355846

1. After applying my patch, I have experimented with different latency settings, and I find that 32 ms is about the lowest that works on my machine, without xrun recovery.

2. Although my second patch largely removes the source of overruns and underruns, the first patch is still necessary to cover the case when the user presses the left or right arrow while recording with play-through (or during playback, if the latency is very low); without the first patch the recording (and playback) stops, with the patch it will continue after recovery. However, I think that should actually be a separate bug.