separate files. Split primary buffer and secondary buffer interfaces
into separate implementations. Made the primary buffer state part of
the dsound object, so primary buffer objects can be created and
refcounted independently. Made the interfaces follow COM aggregation
rules more closely. Converted the dsound critical section to a R/W
lock to improve potential remixing concurrency. Fixed various bugs and
refcounting issues.
system supports the Wine implementation of DirectSound. Also added
some entries to the sample config file and changed the values of the
"EmulDriver" key to "Y"/"N".
debug channels definitions.
Made win32 the default type for spec files.
Ignore C compiler in winebuild so we can simply pass it $(DEFS).
Removed type win32 and debug_channels from spec files.
Fixed winebuild to always generate correct C identifiers (reported by
Vincent Béron).
Added -exe mode to generate a .spec.c for an exe without having to
provide a .spec.
Removed some no longer needed spec files.
Removed .dll extension in makefiles import specifications.
to prebuffering, with some basic prebuffer canceling support, to get
rid of mixing delays in Half-Life. Used a very small waveout buffer
queue in HEL mode, using a callback to queue additional buffers, to
get rid of playback delays in Half-Life. Fixed a couple of bugs.
last-written position of streaming buffers, it attempts to avoid
prebuffering past this position unless necessary; this may avoid a
number of sound problems in various situations. Merged HEL and HAL
mixing code, which should reduce the HEL mode's CPU usage slightly.
to be found on #WineHQ
- use strerror instead of errno at important places
- got rid of HAVE_STRERROR macro
- removed some #include:s (hopefully I didn't break anything)
10ms latency (not noticeable), and about 300ms of prebuffering. Fixed
a few bugs. Added a Wine-only WINMM flag for waveOutOpen for
DirectSound to use (so drivers like wineoss have some way of knowing
what to prepare itself for).
instead. Implemented the DirectSound HEL, with the ability to
prebuffer 300ms of sound, while maintaining play latency of 40ms, and
the exact playposition (Starcraft cinematics are lip-synched). Some
initial HAL support.
Alexandre Julliard