almixer_isolated: log

log

less more | rev 51: (0) -30 -10 -7 +7 +10 tip

age	author	description
Wed, 18 Jan 2012 12:17:42 -0800	Eric Wing	The evil Apple iOS 5.0 OpenAL regression bug strikes again. A user managed to encounter a case where the OpenAL buffer never unqueues and the workaround gets stuck in an infinite loop which results in the app getting killed by watch dog (crashing). I've added a timeout that will abort the unqueue attempt after 200 milliseconds. But if the buffers could be unqueued but the timeout we picked is too short, then we will regress back to the original problem. All iOS 5 audio users need to test this to make sure their apps don't break again. Please remember to file your Apple bug report if you haven't (bug:10145018), test with iOS 5.1 beta to check for new problems, and report results as necessary.
Fri, 21 Oct 2011 17:14:19 -0700	Eric Wing	Workaround for terrible iOS 5.0 regression bug in Apple's OpenAL implementation with regards to resetting streaming sources. The fix uses a version check to determine if the workaround is employed. Once Apple fixes the problem, an upper bounds version check should be added.
Fri, 30 Sep 2011 17:48:23 -0700	Eric Wing	Fixed bug with seeking channels with predecoded samples. The OpenAL 1.1 spec specifies that a playing source should automatically jump when already playing. The code was unnecessarily recalling alPlaySource again which is not necessarily the correct thing to do. I think the previous code is another remnant of the Open 1.0 era. Also fixed bugs in the return codes for seek and rewind. For single channels, the return code was 0 channels processed in ALmixer which resulted in returning false to Lua. The implementation now should return 1/true for single channels, but 0/false for no channels. Errors will return -1/false.
Fri, 30 Sep 2011 16:44:08 -0700	Eric Wing	Workaround: There is a terrible OpenAL regression bug in iOS 5 dealing with streaming sources. alSourcei(source_id, AL_BUFFER, AL_NONE); fails to clear queued buffer queues on a streaming source. The workaround involves manually dequeuing the individual buffers before calling alSourcei(source_id, AL_BUFFER, AL_NONE);. But there is an additional race condition bug where the unqueue fails to take, so the included workaround keeps looping until the buffers finally report as cleared.
Thu, 29 Sep 2011 11:33:34 -0700	Eric Wing	Added addition query to get the current number of buffers processed in the new inner buffer decoupling loop because I've been seeing a lot of 59TestingError warnings. I think it is possible the value may become stale after the first iteration of the loop resulting in assumption failures which result in triggering this error.
Tue, 13 Sep 2011 18:11:34 -0700	Eric Wing	Added AL_SOURCE_STATE to luaal
Tue, 13 Sep 2011 18:11:12 -0700	Eric Wing	merged

less more | rev 51: (0) -30 -10 -7 +7 +10 tip