Responses inline.
Bill Shannon wrote:
> Sivakumar Thyagarajan wrote:
>> Hi Bill/Frank/Mark
>>
..
>> bundled JMS provider). The mappedName is product dependent and we
>> could choose it to mean what we want. Would this work?
>
> We should make mappedName work the same here as it does elsewhere
> in GlassFish. We shouldn't invent a new meaning for mappedName here.
> Still, that doesn't solve the problem in a portable way.
Thanks, I understand.
>
..
>> However, it should be noted, that both approaches (mappedName and
>> deriving ra-mid based on installed RAs) are non-portable.
>
> We should consider adding requirements to the spec to provide a portable
> solution. The spec could require the "matching" several of us have
> suggested. The spec could also define a way to explicitly "link" an
> MDB with a particular RA. We've talked about standardizing the use of
> JNDI names in a way that would allow use of mappedName to be portable,
> but I'm sure there are other possible solutions as well.
Agreed. The solution could be providing a way to specify ra-mid in the
MessageDriven annotation. I shall discuss this with Ken offline.
As a short-term solution, I have raised an enhancement request to
handle this situation the way you and Markus have described. Thanks.
Thanks
--Siva.
>