Re: [jBoss-Dev] Release lock bug in JDK? Fixed ?

[Rickard Oberg]

> the piece of code does take care of the case you describe above but it
also
> takes care of valid cases clientA and clientB call beanA in the same Tx.
> yes?

No. You can't have two clients calling beanA simultaneously. That would
imply two threads of execution for one tx, which is not possible.

Or do you mean clientA call beanA, return, clientB call beanA, return? In
that case it is valid, but then it is not locked when clientB call comes in
so no problem!

> this test does more than is needed.  In fact does wrong stuff...
> (invalidates the above scenario)

No, as above.

> marc (yes? see it? I am completely off-base?)

AFAIK you're off-base.

/Rickard