Thanks, for the reply. Again, I was expressing my understanding of the
specification; certainly it's
imperfect. BTW I meant adherence (not coherence)... your points are well
taken nonetheless.

Here's the issue we are trying to deal with. Our EJB-centric architecture
ties in XML data via
a publish-subscribe mechanism. We use an RMI-IIOP client of a stateless
SESSION EJB (which is
also a subscriber) to (a) accept the published XML messages and (b) convert
them over to requests
for invocation on the business methods of the foregoing EJB. We need to able
to instantiate the
said EJB home interface (obviously). It's seems (IF I understand the
position on the EJB specification
articulate in the response) then that standard establishes no specification
for referencing EJB via
a naming space from RMI-IIOP client, but does establish such a mechanism for
referencing EJB within
other EJB (<ejb-ref/>)? We understand that EJB 2.0 has the "message-driven"
bean concept, but we are currently standardized
on EJB 1.1, not EJB 2.0 (and besides is EJB 2.0 a final public draft). We
want the solution for acquiring the
EJB home interface to be as vendor-neutral as possible.

Thus the specification in prejudicial against RMI-IIOP clients of EJB? The
logical conclusion is that the
specification treats non-EJB clients of EJB (including RMI-IIOP/COM-DCOM,
and even CORBA) pejoratively (EJB
referring to EJB are treated more kindly than other clients referring to the
same EJB bean).

As an architect, I would prefer equality rather that a preferential
treatment; although I understand there
are valid reasons for it in some cases, especially considering vendor
idiosyncrasies and legacy issues.
Perhaps this is one. Perhaps not.




-----Original Message-----
From: Rickard �berg [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, November 21, 2000 10:54 PM
To: [EMAIL PROTECTED]
Subject: Re: XML DD Question: Vendor XML DD and EJB XML DD (EJB 1.1
specificat ion compliance question)


On Tue, 21 Nov 2000 16:14:37 -0800, Richard Landon <[EMAIL PROTECTED]>
wrote:
>The following is a section from the final revision public draft of the EJB
>1.1 specification document (Section 16.5):
<snip>

>This seems to imply the the Vendor (or runtime) EJB deployment descriptor
>MUST support identification of the JNDI name
>or the requirement in the foregoing specification can not be satisified
>(that IS if  the JNDI name, say <jndi-name/> XML element, is not supportted
>by the vendor then the vendor has  "architected a (sic) relationship
between
>the ejb-name and the JNDI name".

Note that there is no requirement in the EJB specification for a JNDI
namespace at all. It is common to have one, but not mandatory.

>A particuliar vendor specific XML DD mapping does not support the JNDI name
>element except in the <ejb-ref/> DD. Thus
>since each session or entity EJB does not support an absolute JNDI name for
>the bean, how does a RMI-IIOP client
>of the bean access the home interface of  the EJB via the look-up of the
>initial context. In other words, there is
>no way for the vendor in question to support a name change during the
>assembly process, since the vendor does not
>support the absolute JNDI name (outside of the <ejb-ref/> DD).

This is a perfectly ok vendor decision. Not very flexible, but
definitely ok from a spec compliance point of view.

>I currently think this is a lack of coherence to the EJB 1.1 XML DD
>specification,  and would like any comments
>you may have on this area.

How can this be a lack of coherence? Again, note that the very notion of
binding the beans in a global namespace is proprietary in the first
place. And again, just because most (all?) vendors support it does not
mean that it is mandatory. It would be perfectly legal to not do it, and
I can see scenarios where it would even be a requirement that a global
JNDI namespace is not available (for example in an embedded
application).

>For the reference implementation of the J2EE (from Javasoft) the vendor
>(j2ee RI) XML DD specification requires
>the specification of the absolute JNDI name using the <jndi-name/> element.
>This is then accessed by performing
>a lookup of the "java:/comp/env/ejb" subcontext in the JDNI.

This is more counter-spec I think. Are you saying that the RI binds
*all* beans into java:comp/env based on their JNDI name? That is wrong:
it should only bind those that have been explicitly referenced to in
ejb-ref's in the XML.

BTW, when you say "lookup of java:/comp/env", is this by an application
client or a bean?

/Rickard

--
Rickard �berg

Email: [EMAIL PROTECTED]
http://www.telkel.com
http://www.jboss.org
http://www.dreambean.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".

===========================================================================
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