- the Linux sound modules (ALSA and OSS) are now write-based, instead of

- the Linux sound modules (ALSA and OSS) are now write-based, instead of
mmap-based. It should fix the problems with inboard sound cards
- added 3 cvars controlling the sound output format: snd_speed,
snd_channels, and snd_width (default values: 48000, 2, and 2
respectively). They're saved in the config file.
- the checks for command line options and environment variables
modifying the sound output format are now common to all modules. The
command line options supported are: -sndmono, -sndstereo, -sndquad,
-sndspeed, and -sndbits. The environment variables supported are:
QUAKE_SOUND_CHANNELS, QUAKE_SOUND_SPEED, and QUAKE_SOUND_SAMPLEBITS.
- added a (still dumb) function to figure out a better sound format when
the initialization of the sound card failed
- sound modules can now suggest a sound format if the initialization of
the sound card failed
- you can now do a "snd_restart" while having modified snd_width and/or
snd_channel. Doing a snd_restart with a modified snd_speed isn't yet
supported, it will fall back to the previous speed
- The WGL video module was blocking the sound output when the
application window lost the focus. Now, this code is shared and the AGL,
GLX and SDL video modules implement such a mecanism
- A lot of minor fixes, changes, factorizations and rewrites all over
the sound engine code and its modules
- added CeilPowerOf2() in the math library
- fixed BSDmakefile for NetBSD

git-svn-id: svn://svn.icculus.org/twilight/trunk/darkplaces@6415 d7cf8633-e32d-0410-b094-e92efae38249