Jeff Trawick wrote: > Branko Äibej wrote: > >>> The only issue I know of which is presently considered a showstopper >>> for 1.0 is fixing some of the atomic APIs to return a value. >> >> >> >> And fixing the broken condition variable implementation on Windows. >> Sorry, I'm working on that, but very slowly due to lack of time... > > > code bug or API bug? > Definitely code bug. The current implementation has more races than the Formula 1 season. I have an almost-fix in my tree, but I'm still testing it.
Hm. Come to think of it, there's also a possible API enhancement that could make condition variables optionally faster on Windows NT 4.0 and higher, but would require a change in how we create thread mutexes. It's just that I'm too sleepy to write it up now, and have no time until Monday... -- Brane Äibej <[EMAIL PROTECTED]> http://www.xbc.nu/brane/
