Jan Harkes <jaharkes_at_cs.cmu.edu> writes:

>     http://sourceware.cygnus.com/pthreads-win32/
> 
> Now that looks interesting, LGPL licenced. Maybe we should help these
> guys debugging it a bit, it sure beats hacking up our own preemptive lwp
> implementation for Win32.

I'm not a kernel-hacker on NT, but I think there is a problem with (only)
preemptive lwp on NT. When you wait for message from the kernel-module you
have to do it blocking (WaitOnObject()).

For arla, guess the same for coda, we want to use cygwin (speed is a
no-issue) since there is so much unix-ism in arla. Now that mean that we
will have to parts of the user-land program, one that is preemptive and one
that is not. And the not preemtive have to someway have a way to wakeup the
preemptive.

There is code that almost is supposed to do this in arla. Guess I just have
to motivate myself (or someone else :) to make it actually work.

Love