I have been involved in an EJB debate involving transaction declarations and
entity beans. In the systems that I have been designing, the client
(application) will never have direct access to any entity beans. To simplify
design and deployment of my entity beans, I have decided to mark all methods
on entity beans as SUPPORTS. My session beans will control all transaction
semantics throughout the system. For the most part, bulk accessors and
readonly data will not be accessed within a transaction. All other methods
in my session beans are marked as REQUIRED.
So much for the background.
The problem is: The spec states (as far as entity beans are concerned) that
ejbLoad() will be invoked before a business method is called on an entity
bean, but only the first time it is used in a transaction. This sounds
reasonable enough. However, what happens when the bean is not called within
a transaction? The same vaguary exists for ejbStore().
Shouldn't some entity bean accessors be able to be called outside of a
transaction? Is this type of access "undefined" by the spec. Should EJB 2.0
shore this up before it is final?
I could ask, "How does server XXX handle this?", however this would not help
me deploy a portable bean. I'm more interested in, "How does the spec
mandate this be handled?".
thanks,
jim
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff EJB-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
