Hi,
> Unless Christoph can make the event sound server give a nice result under > UNIX, I would suggest leaving that until Microsoft come up with > high-resolution timing commands, or another operating system needs that > server. DOS would. > Instead, a very similar sound server could be introduced, perhaps known as > the 'lightweight' sound server. It would be virtually the same as the > current event sound server, but without a call to do_sound() every 16ms. > This would be got around by parsing the entire song when > SOUND_COMMAND_PLAY_HANDLE is given, where the server would only process > SCI-specific MIDI commands and pass the rest directly as-is to the midiout > driver. The midiout driver would then be MCI as it currently is or possibly > DirectMusic. The server would not keep track of where we are in the song so > save and restore of song position would not be possible. An estimation would be possible (restoring by creating the song with zeroed delta times and velocities). Note that songs may consist of two parts (start and loop), of which the latter part may be repeated n times, with n being set at run-time. > However, I have a sneaking suspicion that Christoph will tell me that this > can't be done due to something tricky within SCI. :) IMHO, It can be done (see the discussions around the 10th of September), but at the expense of some accuracy, particularly with loops. Suffice it to say, this isn't going to be a perfect solution either, but it might help. llap, Christoph
