Perhaps it would be helpful to include something along the lines of:
"When signaling a condition variable which no threads are waiting on,
the signal is lost. This may be something to investigate in the
event of a deadlock, which could be the result of a thread waiting
for a signal that has already been sent."
Also a note about whether the condition variables implement
signal-and-continue or signal-and-wait semantics might be appropriate.
(I'm assuming it's signal-and-continue.)
Johannes Groedem <OpenPGP: 5055654C>