Berin Loritsch wrote:
Peter Royal wrote:On Monday, January 27, 2003, at 04:16 PM, Berin Loritsch wrote:Ok. I made Event independent of the Excalibur ThreadPool implementations. Apparently Doug Lea's concurrent package has well tested thread pooling/executor framework. Fortress has one less dependency now. (well, two if you count Thread-Context).
ack! you mean you just removed the ability of having the event package sharing a common thread pool. (a phoenix app with a cornerstone thread pool block feeding several other blocks that host fortress containers or other blocks that have their own async queue's back by event).
I have no problem using Doug's library, but I would like to retain compatibility with our thread pools.
-pete
There is a major bug with the Excalibur Thread package that only
affects some systems. I am not sure what the final reason is, but all
the mutex/threadpool complexity created an infinite loop/deadlock
when the thread pool was destroyed.
I can confirrm this.
Have seen infinite loop/deadlock situations - but never been able to get to the cause.
Steve.
--
Stephen J. McConnell
mailto:[EMAIL PROTECTED]
http://www.osm.net
--
To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
