Subj : win32 threads broken (was Win32 condition variables redux)
To   : comp.programming.threads
From : Joe Seigh
Date : Wed Aug 03 2005 10:14 am

Joe Seigh wrote:
> 
> Well if PulseEvent worked (it doesn't) then it would look like
> a broadcast only solution using SignalAndWait.  It would be easy
> enough to fix in the windows kernel but I guess Microsoft doesn't
> have the talent they think they have.
> 
Actually, if I understand it right, *all* of win32 Wait functions are
broken.  PulseEvent is probably considered broken since it's a really
fast SetEvent/ResetEvent relative to a kernel APC.  Or saying it the
other way around, SetEvent/ResetEvent is slow compared to kernel APC
so race conditions are somewhat rare. This is not likely to be as true
on a multiprocessor.  And I suspect this is true for other synchronization
objects as well that don't always remain in a signaled state.

-- 
Joe Seigh

When you get lemons, you make lemonade.
When you get hardware, you make software.

.