Björn, On 5/24/13 4:08 PM, Björn Raupach wrote: > out of pure curiosity I was browsing the source of jdbc-pool and > ended up looking through FairBlockingQueue. FairBlockingQueue > utilises a global ReentrantLock for mutual exclusion. Common practice > is to always follow a call to lock with a try-finally block. The > methods poll(long, TimeUnit), pollAsync() don't use that practice. I > was wondering why. Attached is modified version of FairBlockingQueue. > I know FairBlockingQueue is a sensitive part of jdbc-pool and I am > certainly not trying to smart ass. Its just interest and probably > only a minor issue.
Your attachment has been stripped from the list. It's best to attach patches to Bugzilla entries, but I'd wait and see if anyone objects to the idea before filing the bug. If you can write a test that demonstrates a bug, that would be even better... "bug fixes" that come from pure code-reviews tend to be viewed skeptically around here, though this one seems to theoretically lead to deadlock. If it can theoretically lead to deadlock, can you force a situation where it *actually* leads to deadlock (or some other unfortunate condition)? -chris
signature.asc
Description: OpenPGP digital signature
