Hi,
from the 2.0 spec 4.1.2 Flexible component model:
"Enterprise beans that are remotely accessible components are intended to be
relatively coarse-grained business objects...Fine-grained objects (e.g.,
line item on a purchase order, employee's address) should not be modeled as
remotely accessible enterprise bean components, but rather as locally
accessible enterprise beans or as the dependent classes of enterprise
beans."
I have a few related questions:
* which is the preferred way? to use local entity beans or to use dependent
classes, i.e., will the overhead of local entity beans be small enough to
model most of your domain objects as local entity beans.
* since we "always" access entity beans through remotely accessible session
beans when would we ever make an entity bean non local (or am I not
understanding what local entity beans really are)
* for ejb 1.0 and 1.1 all app servers i know optimize calls in the same
jvm...so we can think of entity beans within the same jvm as "local." how
much less overhead do these optmizations provide? i.e., because of these
optimizations can we model fine grain objects in ejb 1.1. as entity beans
(contrary to the 1.1 specs recommendation) ? there's probably a
philosophical debate here, but that aside, if i wanted to model address and
line item in ejb 1.1 can it be reasonable done without performance kill.
I've tried searching the archives but can't seem to fine these answers
readily. Much thanks.
--
Tinou Bao
www.tinou.com
===========================================================================
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".
