[concurrency-interest] Request for feedback -- Bill/Brian/Cliff JavaOne talk

Bill Pugh pugh at cs.umd.edu
Sun Apr 8 22:51:40 EDT 2007


On Apr 6, 2007, at 12:54 PM, Oliver Zeigermann wrote:

> Hey, that is really cool!
>
> I was concerned about testing concurrent code myself some time ago.
> That led to some helper classes which do something similar to MTC (at
> least if my understanding of MTC is correct):
>
> http://jakarta.apache.org/commons/transaction/apidocs/org/apache/ 
> commons/transaction/util/TurnBarrier.html
> http://jakarta.apache.org/commons/transaction/apidocs/org/apache/ 
> commons/transaction/util/RendezvousBarrier.html
>
> I have never been quite sure if they actually make sense and could  
> be useful.
>
> What do you think?


I can understand them, and see where they might be useful for  
somethings. The nice thing about the tick counter
in MultithreadedTestCase is that it works very well for the case  
where you expect a thread to block and you want
to verify that it does block.

However, MultithreadedTestCase requires some deeper magic: a  
monitoring thread calling getState on all of the
test threads. Figuring out which threads should be monitored and such  
is a little tricky, and thus I'm not sure
how general purpose it is. We haven't really tried using it in a case  
where you have executors and/or dynamically
spawned threads.

	Bill


More information about the Concurrency-interest mailing list