yeah i mean that's it, in fact a stack of Id from the object will do, but
pffffffff tons of work for nothing (imho)
if weblogic wants to do that and 2 phi commit on distributed DBs and corba,
be my guest, yeah that is worth $20k/CPU
me? I keep it simple, I keep it free,
marc
|-----Original Message-----
|From: [EMAIL PROTECTED]
|[mailto:[EMAIL PROTECTED]]On Behalf Of marc fleury
|Sent: Friday, October 20, 2000 12:34 PM
|To: jBoss Developer
|Subject: RE: [jBoss-Dev] Release lock bug in JDK? Fixed ?
|
|
|seriously now the only test I see is passing the context in the
|invocations... which means the proxies somehow... then you build a stack of
|ctx in a thread local thing and you pass that in the invocations by proxy
|retrieval, hugh....
|
|BTW highly non portable since there is NO way to pass that information
|implicetely in a portable manner (period).
|
|marc
|
|
||-----Original Message-----
||From: [EMAIL PROTECTED]
||[mailto:[EMAIL PROTECTED]]On Behalf Of marc fleury
||Sent: Friday, October 20, 2000 12:21 PM
||To: jBoss Developer
||Subject: RE: [jBoss-Dev] Release lock bug in JDK? Fixed ?
||
||
||
||
|||-----Original Message-----
|||From: [EMAIL PROTECTED]
|||[mailto:[EMAIL PROTECTED]]On Behalf Of Rickard Oberg
|||Sent: Friday, October 20, 2000 12:02 PM
|||To: jBoss Developer
|||Subject: Re: [jBoss-Dev] Release lock bug in JDK? Fixed ?
|||
|||
|||> |EJBObject calls, true. Reentrant calls in one tx on an Entity
|marked as
|||> |non-reentrant should fail however. I could not see in code
||that this was
|||> |dealt with.
|||>
|||> That is correct but again the test that was there and you are
|||referring to
|||> considered every thread in a used bean as reentrant... this was wrong.
|||This
|||> is not what "reentrant" means and not only on EJBObject calls.
|||
|||Hm.... if bean A calls bean B which calls bean A, then that is a
|reentrant
|||call. If bean A calls bean A through the EJBObject, then that is a
|||reentrant
|||call. If the bean is marked as non-reentrant, an exception should
|||be thrown.
|||Da law of da spec.
||
||Right and it is not what the test did
||
||This is what the test used to do:
||
||if Client cA calls bean bA and another client cB calls the same
||bean bA with
||the same transaction then throw reentrant exception,
||
||in effect preventing "concurent" calls on a bean
||
||the correct reentrance definition is the one you provide above (of course)
||meaning instance to same instance, and that, I would be curious
|to hear how
||you plan on implementing it :) (ok let's stop this, I know it is
|going into
||a wall )
||
||marc
||
||"don't do it, be sincere!"
||-- MJ Cole, "Sincere" --
||
||
|||
|||/Rickard
|||
|||
|||
|||
||
||
||
|
|
|
