Lets see if i understand you correctly.
Lets say there is a 3 party bean b1 and bean you develop as b2.
(b1 and b2 are impl)
Now rb1 and rb2 be their respective remote interface.
now if you say:
rb2 extends rb1{//your remote interface code}
b2 extends b1{//your impl}
Now if your client wants to access only b1 then cast it to b1 or leave it as
it is, its super.
Definitely you have to mention your own code in remote interface, but i
don't see why client needs to be changed.
(Assumtion:you can find out 3 party impl class)
Now definitely there is a twist if you want to extend more than one 3 party
because of missing muliple inheritance(but its doable)
May be I am missing something.. or didn't get your question.
Punit
>
>I'm investigating some problems found during a research project. One of the
>major downsides, according to the project conclusions, is that one cannot
>have multiple interfaces in EJB. According to the spec one can let the
>remote interface extend multiple interfaces, but that does not permit one
>to let a bean implement a new, possibly third-party, interface without
>making changes to the remote interface and the clients using the bean.
>
>Is there any way one can let the bean implement a new interface and have
>the clients the way they are? If I for example have a bean with interface
>X, the clients look up XHome and receives stubs implementing X. Now I also
>want the bean to implement Y, so I extend X and Y into XY (as I understand
>is the proposition made in the spec). Can I now let old clients, that want
>objects of type X, stay the way they are and just implement clients that
>want objects of type Y, or do I need to change the old clients so that
>they also know of XY (and thereby also Y) possibly casting to X directly?
>These questions arise from the fact that the developers in the project come
>from a COM+ background and (from what they have told me) one can specify
>what interface one wants returned when looking up an object. Thereby one
>can just let old clients be while implementing the new interface on the COM
>object.
>
>If it is possible to have the old clients the way they are, I propose the
>following changes to the spec. Please comment on the proposal (or even the
>possibility to do what I propose):
>* The Deployment Descriptor is given the possiblity to specify multiple
>interfaces (which only adds to the spec, still making old descriptors
>valid).
>* The deployment tools are responsible for making a remote interface that
>extends all specified interfaces (X and Y are combined into XY which is the
>interface returned by the home).
>* Optionally one might want to add the possibility to specify one of the
>interfaces in the DD as the "main interface", making it the interface
>returned by the home. Since it is a superclass to the combined interface I
>guess this should work.
>
>Thanks for any comments, big or small.
>
>/Kalle
>
>===========================================================================
>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".
>
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.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".
