For the IBM JDK there is a bug in version 1.5.0 SR4 (and SR5, which we
don't "officially" support yet) with the concurrent utilities, here is
an excerpt from the "Known Issues" section of
http://www.terracotta.org/confluence/display/docdraft/Integrations+IBM+JDK:

*IBM JDK SR4*: the
java.lang.util.concurrent.locks.AbstractQueuedSynchronizer.fullGetQueuedThread()
method could potentially run into an infinite loop (refer to Sun bug
6241823^ <http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6241823>
for details). This class is being used by several classes such as
java.util.concurrent.locks.ReentrantReadWriteLock. This is still a
problem in the SR5 release of the IBM JDK as well (the latest at the
time of writing), hopefully in a later service release this bug will be
fixed.

Presumably people using the IBM JDK will already be aware of this and it
is doc'd as a known issue, but folks shouldn't think this problem will
go away if they use DSO.  I believe our supported version of the Sun JDK
is a high enough version that this bug has been fixed and isn't a
problem, so it could appear to be a subtle problem if people move [back]
to the IBM JDK.

Nat

Orion Letizi wrote:
> Folks,
>
> I'm working on a doc page called "Gotchas" which is an enumeration of  
> a number of subtle issues that can crop up when using DSO.  This is  
> different than the troubleshooting guide in that it's meant to  
> highlight potential issues with DSO before they crop up, not a  
> symptom analysis doc.
>
> I'd like feedback from everyone on what stuff might go in there.
>
> Right now I have:
>
> - Lock Then Share pattern
> - Mutable enums
> - Collections not autolocked by default
>
> I'm sure there are others.
>
> --Orion
> _______________________________________________
> tc-dev mailing list
> [email protected]
> http://lists.terracotta.org/mailman/listinfo/tc-dev
>   
_______________________________________________
tc-dev mailing list
[email protected]
http://lists.terracotta.org/mailman/listinfo/tc-dev

Reply via email to