
8 Dec
2006
8 Dec
'06
9:42 a.m.
Steven Watanabe wrote:
I see no reason to require that a mutex be locked for notify. If you want to have it locked then you can just do so. Wait is different; it needs to release the lock first and upon notification reacquire it and check the predicate.
All you say is technically true, but there are other design considerations which I explained in my other posts. That's not to say that you are undoubtedly wrong. Many smart people agree with you and disagree with me...