I believe this is because signal works on a thread level. Once you
call signal from a thread, in this case the main thread and no
other resource has a lock on your semaphore, then the signal call
simply returns and puts a lock that belongs to your thread on the
semaphore. The second time you call signal from the same thread,
your thread already has the lock so signal does not block and you
are free to continue. I think you need to test this with at least
one other thread to see it in action. I haven't tested this myself
in a while, but I believe what I stated is correct. I'm sure
someone will chime in if I'm mistaken.
Kevin
You are describing a CriticalSection. The documentation for
Semaphore is explicit that the counter will decrement even in the
same thread. By inference, it's a CriticalSection that does not
decrement a second time in the same thread. To wit, the
documentation for Semaphore states:
The Semaphore class is different from the CriticalSection
and Mutex classes in this way: calling Signal in the same
thread will cause the counter to decrement.
The main thread (controlling the GUI) is a thread, too. So, I
still believe my code should block.
Jeff
You are right. I read this in the Language Reference PDF and
interpreted in the wrong way. It refers to "the" lock instead of "a"
lock and is somewhat confusing in light of the rest of the
documentation.
Call Signal to obtain a lock on a resource. When you obtain a
lock, you can use the resource without fear that another
thread will try to use it simultaneously. If the call succeeds,
this function returns immediately and you code continues to
execute. If the lock is not available, then the thread that
called this method will wait until the lock becomes available.
When the call returns, you will have the lock, but you may
have to wait until the lock becomes available. When you are
finished using the resource, call the Release method to give it
back to the Semaphore.
The built in Language Ref does seem to be more clear and what you see
would appear to be a bug.
_______________________________________________
Unsubscribe or switch delivery mode:
<http://www.realsoftware.com/support/listmanager/>
Search the archives of this list here:
<http://support.realsoftware.com/listarchives/lists.html>
