Presumably the Java objects he is talking about stay behind the service
layer of the Session Beans. The clients interact with the service layer
which return "by value" representations of the domain mode.

This EB usage is called Entity Beans as Data Access Layer on the wiki web
(http://www.c2.com/).

I haven't thought about it enough to have a solid opinion, but it feels icky
at first blush. If we are going to do thing this way why not have the Java
objects mapped by a good OR tool anyway? I guess we are looking for CMP to
be our OR tool so that we gain portability.

-Chris.

> -----Original Message-----
> From: Tim Endres [SMTP:[EMAIL PROTECTED]]
> Sent: Friday, April 07, 2000 11:49 AM
> To:   [EMAIL PROTECTED]
> Subject:      Re: 'local' entity beans vs dependent objects
>
> > The point here is that EJB is an implementation technology and has so
> > many weird design quirks/flaws that its requirements should not make it
> > into your domain model.  Also, the common practice of somehow mystically
> > grouping domain objects together under, say, an entity bean facade is
> > not (a) easy or (b) practical in most cases (at least IMHO).  So all I'm
> > saying is: what if you did it the other way?  What if you stick an
> > entity bean under a regular java object "facade" (I use quotes because
> > it's not really a true Facade in the Design Patterns sense)?  And as
> > long as you're doing that, why not shove session beans under regular
> > java object facades as well?
>
> The one issue I see is this: the advantage of a SessionBean Intf is that
> I get one interface over what may be a slow socket - on the server side,
> where connections are fast, I can leverage all of the EJB overhead, then
> send the thin result back down the original slow connection.
>
> With your approach, if I put in my client the domain objects that wrap
> the EJB objects, then I have lots of EJB connections being made to the
> server over that slow link to service each of the EBeans.
>
> This is especially horrible if I have many objects involved in some
> transaction whose result may only be a number.
>
> As I see it, in that conext, you lose one of the primary advantages of
> the session bean interface. Of course, there are many conexts in which
> the difference will be small/negligible.
>
> tim.
> Tim Endres              [EMAIL PROTECTED]
> ICE Engineering, Inc.   http://www.trustice.com/
> "USENET - a slow moving self parody." - Peter Honeyman
>
> ==========================================================================
> =
> 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".

===========================================================================
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".

Reply via email to