|-----Original Message-----
|From: [EMAIL PROTECTED]
|[mailto:[EMAIL PROTECTED]]On Behalf Of Rickard Oberg
|Sent: Friday, October 20, 2000 11:51 AM
|To: jBoss Developer
|Subject: Re: [jBoss-Dev] Release lock bug in JDK?
|
|
|
|> |Agree completely. When we do clustered stateful this will be even more
|> |important since activation will happen on all calls. Ah well, since we
|> |already have this going with EntityBeans it should be pretty
|> |straightforward.
|>
|> yes, I believe your entity layout could apply to the stateful stuff and I
|> had missed the clustered point, so it is a "two birds one stone" kinda
|thing
|> I think :)
|
|I have looked into the entity cache. Actually the above is only half-true.
|The ejbLoad calls are done outside of the cache, but ejbActivate is done
|inside the cache. Perhaps that too should be put into a separate
|interceptor, but before the EntitySynchInterceptor. Yes? That
|would minimize
|contention on cache, but the passivation would have to be more
|careful since
|it would mean that just because an id is in the cache does not mean that
|there is an active instance associated with it.
|
|Probably the right thing to do, but a little tricky.

No I know that (the fact that "activation" is done in cache) but it is a
"soft" activation there is no "state loading" in that activation so.....
(afaik)

marc

|
|/Rickard
|
|
|
|


Reply via email to