kjetil wrote this for VST. it works. its very cool. its a bit of a kludge. i reimplemented it to provide a JACK VST host (not released yet). to do it properly requires that WINE abandon its approach of providing its own incomplete pthreads implementation, either by completing it, or using the standard one, which they say they can't do. transgaming has taken a different approach with their WineX stuff, but that's not available in source form yet (they say it will be). under NPTL, the situation appears to have eased a little, because NPTL uses one layer of function call indirection through a pointer to access pthreads, making it possible for libraries like WINE to layer their own stuff on top of pthreads only where they need it. since i just moved to RH9 yesterday, i should give it a try ... BUT ... AFAIK, the PlanetCCRMA kernel with the best performance doesn't have NPTL support. my question about "how" wasn't mean as "how could you possibly do this?", but more as in "did these guys copy kjetil's work, or independently invent it, or did they actually use it?"

the last message i could find about this was from june 2003, and suggested it would be a while until they released that part. however, it sounded as if the mainstream wine guys (read: alexandre julliard) had no intention of using their work because they believe their own approach to be better. hah: try firing up ardour under wine and immediately getting "pthread_cond_wait: please implement me" :) they did? how did i miss that? --p

ah. i recall that now. too bad i was wrong in my optimism about 2.5/2.6 :( at least anyone who wants the HRT stuff can patch it into both 2.4 and 2.5 (so far, anyway).