summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* Load backend libs on initChris Robinson2011-08-184-52/+28
|
* Make DSoundLoad return a booleanChris Robinson2011-08-181-9/+8
|
* Avoid some LP typesChris Robinson2011-08-181-7/+7
|
* Only attempt to load MMDevApi and DSound once, on initChris Robinson2011-08-182-10/+4
|
* Set the proper union member to 0Chris Robinson2011-08-171-1/+1
|
* Remove backends that fail to initializeChris Robinson2011-08-171-2/+14
|
* Allow the backend init functions to return failureChris Robinson2011-08-1716-29/+44
|
* Coalesce some mixer sample convertersChris Robinson2011-08-171-18/+6
|
* Avoid reading some more source parameters multiple times when updatingChris Robinson2011-08-171-12/+19
|
* Get rid of the remaining min/max usageChris Robinson2011-08-163-12/+1
|
* Rename minF/maxF/clampF to minf/maxf/clampf for consistencyChris Robinson2011-08-165-21/+21
|
* Use mini/maxi/clampi and minu/maxu/clampu to replace min/max callsChris Robinson2011-08-167-34/+41
|
* Avoid a float->int->float conversionChris Robinson2011-08-161-1/+1
|
* Use inline minF/maxF/clampF functions instead of the __min/__max macrosChris Robinson2011-08-165-38/+32
|
* Fix wet path conehf calculationChris Robinson2011-08-131-1/+1
|
* Make sure there's a buffer to render loopback samples toChris Robinson2011-08-121-1/+1
|
* Avoid a temporary throw-away buffer for mixing null outputChris Robinson2011-08-122-35/+20
|
* Use an existing function to clear and free the contextChris Robinson2011-08-121-7/+5
|
* Round the HRTF sample step countChris Robinson2011-08-111-4/+5
|
* Use ARM Neon vector ops to apply HRTFs when availableChris Robinson2011-08-113-7/+52
|
* Don't declare i in a for() statementChris Robinson2011-08-081-1/+2
|
* Use a message thread to handle MMDevAPI backend methodsChris Robinson2011-08-031-147/+309
| | | | | | Because MMDevAPI uses COM, CoInitialize must be called on the thread(s) where the device calls are made. OpenAL has no such requirement, thus no guarantee that COM will be (or remain) initialized on the calling thread(s).
* Avoid closing devices and deinitializing backends on process exitChris Robinson2011-08-031-20/+37
| | | | | They are still closed and deinitialized when the DLL is simply being detached (via FreeLibrary), however.
* Delay reading the config file on Windows, tooChris Robinson2011-08-031-7/+15
|
* Use specialized click removal loops for mono and stereo outputChris Robinson2011-07-301-7/+37
|
* Wait until all headers are unprepared before freeing the bufferChris Robinson2011-07-281-4/+6
|
* Make the thread event auto-resettingChris Robinson2011-07-281-2/+2
|
* Remove the unnecessary wave header eventChris Robinson2011-07-281-29/+2
| | | | | | It's only ever set before sending the quit message, which will set the thread event. It's also only waited on just before the thread event is used to signal the end of the thread.
* Use unnamed events for winmmChris Robinson2011-07-281-4/+4
|
* Make the DSound and MMDevAPI events initially non-signaledChris Robinson2011-07-242-2/+2
|
* Move an include to where it makes more sense to beChris Robinson2011-07-232-3/+3
|
* Don't cast for the float versions of math functionsChris Robinson2011-07-231-5/+5
|
* Use real-time priority by default on WindowsChris Robinson2011-07-231-0/+4
| | | | | It's actually "time-critical", but it seems normal priority doesn't cope as well when the system is under moderate load.
* Fix cmake checkChris Robinson2011-07-231-1/+1
|
* Check for the finite() function along with _finite()Chris Robinson2011-07-231-4/+8
|
* Look for ieeefp.h and include it when appropriateChris Robinson2011-07-233-0/+7
|
* Use a better method for determining if shared functions existChris Robinson2011-07-234-116/+100
|
* Round UpdateSize to the nearest multiple of MMDevAPI's period sizeChris Robinson2011-07-211-1/+1
|
* Reset the MMDevAPI and DSound events before starting playbackChris Robinson2011-07-212-3/+2
|
* Fix DSound UpdateSize calculation when NumUpdates is clampedChris Robinson2011-07-211-2/+2
|
* Make sure some state values are finiteChris Robinson2011-07-211-3/+3
|
* Create and close the DSound event handle with the deviceChris Robinson2011-07-211-21/+21
|
* Use an event to signal the mmdevapi backend to mixChris Robinson2011-07-211-15/+43
| | | | | Also, use a multiple of the default period size for the update size, since the event apparently triggers on period boundaries.
* Use DSound notifications to wake up the mixer threadChris Robinson2011-07-212-7/+57
|
* Add an option to emulate EAX reverb using standard reverbChris Robinson2011-07-204-1/+12
|
* Add a config option to boost reverb outputChris Robinson2011-07-204-1/+26
|
* Set real-time priority for the WaveOut backendChris Robinson2011-07-201-0/+2
|
* Scale reverb output based on the number of channelsChris Robinson2011-07-201-2/+7
| | | | | A scale of sqrt(2/numchannels) is used so the perceived volume matches as if it was stereo output, which seems to match other implementations.
* Make sure some effect property values are finiteChris Robinson2011-07-201-3/+3
|
* Make sure some listener property values are finiteChris Robinson2011-07-201-18/+35
|