http://bugs.winehq.org/show_bug.cgi?id=28723
--- Comment #95 from Andrew Eikum aeikum@codeweavers.com 2011-12-16 10:42:15 CST --- (In reply to comment #90)
I put back set_period_near into the last patch. But I'm not satisfied with the results. My render test produces a spurious xrun notification with plug:dmix that disappears when running with full logs or when using set_period_min. Where does that xrun come from? OTOH, set_near allows PA to produce better position values in Intrepid. Oh well, it really looks like min&max is not over yet.
I'm working on Bug 14717 and testing the changes there on top of your latest patchset. In Darwinia, I get frequent (at least 1 every 5s) xruns in ALSA with _near(). If I change it to _min(), the xruns go away. Using _min() and _near() together also makes the problem go away. I don't know how to test for the improved PA behavior you suggest that _near() provides.