Ah. My apologies. My question was related to writing the implementation of acceptex. I have recently solved the problem and the question is now moot. The problem was related to a bug in the wineserver async implementation that appeared when a file descriptor handled its own polling events (like sockets) and garbage collection. Fixing the bug fixes the problem i was having earlier. Should I include those fixes in a separate patch or can I incorporate them into the acceptex patch? (Correct AcceptEx implementation relies on it).
Those fixes should be a separate patch, perhaps one in a series of patches. --Juan