http://bugs.winehq.org/show_bug.cgi?id=28010
--- Comment #7 from Andrew Eikum aeikum@codeweavers.com 2011-08-11 08:44:53 CDT --- Thanks. That is... really bizarre. It seems to be loading the builtin winmm, as it should, but it never actually calls any waveOut* functions. That means MMDevAPI never gets loaded, and you have no sound. So the question is why Raptor decides not to call waveOut*.
This section of your +all log looks suspicious:
0021:Call KERNEL32.GetEnvironmentVariableA(68143000 "SDL_AUDIODRIVER",0012c2c8,0000000c) ret=68109245 0021:Ret KERNEL32.GetEnvironmentVariableA() retval=00000005 ret=68109245 0021:Call ntdll._stricmp(68148590 "dsound",0012c2c8 "pulse") ret=6810224a 0021:Ret ntdll._stricmp() retval=fffffff4 ret=6810224a 0021:Call ntdll._stricmp(68148500 "waveout",0012c2c8 "pulse") ret=6810224a 0021:Ret ntdll._stricmp() retval=00000007 ret=6810224a 0021:Call ntdll._stricmp(681471f0 "disk",0012c2c8 "pulse") ret=6810224a 0021:Ret ntdll._stricmp() retval=fffffff4 ret=6810224a 0021:Call ntdll._stricmp(681472b0 "dummy",0012c2c8 "pulse") ret=6810224a 0021:Ret ntdll._stricmp() retval=fffffff4 ret=6810224a 0021:Call ntdll.strlen(681432eb "No available audio device") ret=6810a0ca 0021:Ret ntdll.strlen() retval=00000019 ret=6810a0ca
That looks to me like SDL is reading your SDL_AUDIODRIVER environment variable, seeing that it contains the string "pulse," and is trying to load that driver with the Windows-based SDL, which obviously doesn't recognize a "pulse" driver (instead it recognizes dsound, winmm, disk, or dummy). Not finding a "pulse" driver, it concludes that there is "No available audio device."
What's baffling is that this worked before 1.3.25. Raptor never gets to the point where it tries to load any audio code, so my changes shouldn't have an effect. Did your setting the SDL_AUDIODRIVER environment variable coincide with upgrading to 1.3.25?
If you know that it worked before 1.3.25 with this environment variable set, then a +all log of 1.3.24 would be useful.