Hi, Alex: Sorry, I forgot to answer your last question.
[...] > > - and also if there are ports that need the device level right now, or will > > very soon > > > > That is, if there is something out there that will need the device level > > now, we may as well go the whole way. Is there? I'll first have to know what you mean with 'needing the device level' there. What I believe I can state safely is the following: a) There are at least two potential implementations for a "passive" song-player system, namely fork()/execvp()d timidity and MCI. b) There are at least two potential implementations for a synchronized song player that has its timing handled by FreeSCI, namely ALSA/OSS' MIDI devices with timing done by select(), and PCM output on MacOS X with the timing done by the SDL PCM layer's callback. [I hope this answers your question.] We need both, the question now is how to maximize code re-use, and how to model everything without introducing what I'll call architectural deficiencies, i.e. parts of the model which enforce (for certain drivers, architectures, operating systems etc.) inadequate (incorrect, unreliable) behaviour. llap, Christoph
