Platform: Hibernate 2.1 / JBOSS 3.2.2

1) Why must an EJB with CMT be marked as "Required" in order to do a session.load? Supports has always worked for JDBC datasources - for "read" use cases.

2) When I do this in a CMT session bean with required...

Session session = null;
Session session2 = null;
try
{
ctx = new InitialContext();
SessionFactory sessionFactory =
(SessionFactory)ctx.lookup("java:/hibernate/HibernateFactory");
session = sessionFactory.openSession();
session2 = sessionFactory.openSession();
MyObjectWithEqualsImpl obj = session.load .....
MyObjectWithEqualsImpl obj2 = session2.load ...
if (! obj.equals( obj2))
throw new SessionsAreNotEQualIDoNotUnderStandException();

why is this a failure?? Certainly these means loading deep object graphs will not work very well - the whole point of going with hibernate (for me) in the first place.

3) The javadocs show a package for net.sf.hibernate.jca. What is the significance of this package in terms of using Hibernate as a service in JBOSS?

4) If I want to also use JMS in EJB methods - will Hibernate factory play nicely in a XA transaction?

Thanks,
Joel Stewart

correction, on question 2, the .equals() DOES work (of course) but the == operator does not work.

Sorry about the confusion.

1) Hibernate uses callbacks from the transaction manager for dealing wth the cache management issues, eg, hibernate.transaction.manager_lookup_class which is needed when JVM-level caching is enabled in a JTA environment.

2) Session effectively operated in its own world and contains its own cache of objects (session level cache) thus the instances are not shared across sessions.

3) Hibernate can be created several ways in JBoss.
o The servlet level.
o The EJB Level via either MBeans (as the examples in the docs show)
or by using a standard JCA container. JCA advantage is that you can
use it for other application servers.

4) Yes.