Why does the atomic_count ++ operator return a "long" for the win32,solaris and sync compilers, but returns "void" for all gcc compilers? The -- operator returns a "long" in all cases, so this would appear to be an annoying, anomalous behavior. But hopefully a rational explanation exists. Also, if it's the value of the ++ that is wanted, two threads could return the same value of this atomic counter, since another call is required to obtain it for the gcc case. Thank you, Doug