I just noticed something else that I should have caught before: In ThreadWaitSleeper.java, Thread.interrupted() is wrong; it should be Thread.currentThread().interrupt(). The former clears the interrupt flag and returns whether it was set; the latter sets the flag.
Same thing is true in BlockingMessageListener.receive() in the JMS code. You might also want to consider replacing the use of Object.wait/notify in that code with something else. Likewise in DefaultServerLoadMonitorHarness in the tests.
Apart from that, ThreadWaitSleeper looks a lot cleaner. I hope it isn't meant to be reusable, because CountDownLatches are one-shots: once the latch is open, it can't be closed. A second call to goToSleep() will return immediately. If you need to be able to use it repeatedly, I can suggest an implementation.
--tim
Joe Walker-3 wrote:
...indexes are good for keywords, and not for concepts that take a couple of lines to explain.
What would you have looked under? Maybe the next edition can add it.
--tim
