Local invocation has fundamentally different semantics to remote
invocation, so allowing a single interface to be remote and local
simultaneously is disallowed by the spec.
-----Original Message-----
From: Dibyendu Majumdar [mailto:dibyendu_at_mazumdar.demon.co.uk]
Sent: Thursday, March 02, 2006 2:59 PM
To: Gavin King
Cc: ejb_at_glassfish.dev.java.net; ejb3-pfd-feedback_at_sun.com
Subject: Re: Enhancement to @EJB annotation
Hello Gavin,
Would you elaborate on the reasons?
Thanks
Gavin King wrote:
> This was discussed a number of times, and we decided not to go there.
>
> -----Original Message-----
> From: Dibyendu Majumdar [mailto:dibyendu_at_mazumdar.demon.co.uk]
> Sent: Thursday, March 02, 2006 2:45 PM
> To: ejb_at_glassfish.dev.java.net; ejb3-pfd-feedback_at_sun.com
> Subject: Enhancement to @EJB annotation
>
> Hi,
>
> I would like to suggest an enhancement to @EJB annotation so that the
> client can specify whether the local or remote interface is desired.
> This way, we can avoid the "antipattern" described by Sahoo
> (apologies,
> Sahoo!) in the discussion at:
>
> http://forums.java.net/jive/thread.jspa?threadID=13366&tstart=0
>
> This method will a) allow EJBs to be coded as POJOs, and b) mimic the
> approach in earlier versions of EJB where the client had to explicitly
> decide on the type of interface.
>
> Thanks and Regards
>
> Dibyendu
>
>
>
>