Hi, Well, ALSA seems to be working on my box /without/ bringing it to it's knees, so the problem must be something in the OSS emulation layer. However, sound is played much too slowly. I guess that this is because of the explicit flush operation; IIRC, fixing this without removing the flush would require a kernel recompile with HZ set to at least the MIDI bps rate (or something along the lines of that). This is not an option for everyone, of course, so I removed the flush; it still worked quite well, except for some things that didn't quite sound right (some notes were being played too late, IIRC). Is this a problem specific to ALSA, or does it happen with /dev/midi00, too? Also, several pieces of music, including the sound effects for answering a question in LSL3, were looped even though they definitely shouldn't be. Invalid looping is a known problem, of course, but I'd like to point out that this, unlike the SQ3 pod door cue thing, affects me as well. Those sound effects explicitly set their loop point (according to the debug output), so I guess that this is not what differentiates songs that should loop from those that shouldn't. Also, Iceman, HQ, and QfG1 don't play any music, despite loading their patches. As pointed out earlier, this may be because they detect the sound hardware in some way we haven't implemented yet, and believe that the PC speaker or Tandy is being used (this would cause sound.1?? to be played in HQ, which don't generate any actual output if the MT-32 output mask is in use). llap, Christoph
