Completely confused about SessionSynchronization

Question

I'm having a tough time understanding what SessionSynchronization does and when it should be used. All I do know it can only be used with stateful session beans.
What can it do that requires its use? I've read things about synchronizing instance variables after rollbacks and commits...but I have no idea why and how SessionSychronization does this. And why it's not done automatically anyways.
Can somebody please clarify my confusion in detailed but understandable terms?
Thanks.

Answer 1

and whether a transaction commite or was rolled back
determines what the new values of those instance
variables would be!! Is this correct?Yes, when it comes to transaction lifecycle notifications, it's important
to be able to find out whether the transaction successfully committed
or not.
>
That explains afterCompletion(boolean), but I still
see no use for afterBegin() and beforeCompletion()afterBegin() is called the first time the SFSB is involved in the new
transaction. It's a convenient place to initialize the bean state
before business methods are called. afterBegin() is invoked within
the transaction so resource managers such as databases can be
accessed.
beforeCompletion() is called before the transaction is completed.
Like afterBegin, it is also called within the transaction context itself,
so e.g. it's the place where any pending resource manager updates
could be made. Those can't be done in afterCompletion(), because
there the transaction has already either committed or rolled back.
>
By the way, is there a difference between using JDBC
and using JDBC as part of BMP?Yes, BMP applies to EJB 1.x/2.x style Entity beans , which have
much different lifecycle and behavior than session beans. However,
going forward there is much less need to use either of these approaches
given the existence of the Java Persistence API.
--ken