[webbeans-dev] Re: Producer fields on EJBs

[Andrew Lee Rubinger]

Pete Muir wrote:
> I thought through this some, and bounced ideas off Andy.
> 
> Gavin, I'm not sure that producer fields on SLSBs makes much sense - 
> SLSBs instance state is undefined I think (some EJB expert correct if 
> I'm wrong ;-). I think the spec should say that the behaviour of 
> producer fields on SLSBs is undefined.
> 
> Assuming the above, I would still argue for a getFieldValue SPI that 
> throws IllegalStateException if invoked on an SLSB reference - and I 
> think this removes the issue Ken has with a many-to-one instances.

UnsupportedOperationException perhaps?

> I'm slipping this for the upcoming release as we are too far from 
> consensus.
> 
> On 6 Apr 2009, at 18:22, Pete Muir wrote:
> 
>>
>> On 6 Apr 2009, at 18:00, Kenneth Saks wrote:
>>
>>>
>>> On Apr 6, 2009, at 12:31 PM, Pete Muir wrote:
>>>
>>>> Carlo, Ken, Andrew,
>>>>
>>>> JSR-299 supports the notion of a producer field. By adding the 
>>>> annotation @Produces to a field, you make it into a bean, which can 
>>>> be used as the source of injections. The field can also be 
>>>> associated with a scope. When you first look up an instance of a 
>>>> bean, the field value is accessed, and stored in the context for 
>>>> it's lifetime.
>>>>
>>>> As we deal with EJB references, we can't directly access the field 
>>>> value (using reflection) so I think we should use the SPI here.
>>>>
>>>> I propose we add a method with this signature to the 
>>>> SessionObjectReference we previously created for the purpose of 
>>>> casting/removing EJBs 
>>>> https://svn.jboss.org/repos/webbeans/ri/trunk/spi/src/main/java/org/jboss/webbeans/ejb/api/SessionObjectReference.java 
>>>>
>>>
>>> There isn't necessarily a one-to-one relationship between an ejb 
>>> reference and a bean instance.   For stateless session beans, any one 
>>> of the bean instances can be used for any invocation.
>>>
>>> Can't this behavior be handled by the web beans interceptor?  By 
>>> definition, each interceptor instance has the same life cycle as its 
>>> associated bean instance, and the bean instance is available via the 
>>> Interceptor InvocationContext.
>>
>> I did consider this, but not sure how to retrieve the result 
>> invocationContext.getTarget() at the time it is needed for accessing 
>> the producer field. We can't just store it from the construction point 
>> onwards (serializability) and I can't see a way to access it at some 
>> other point in the lifecycle - the only way to run an @AroundInvoke is 
>> to call a method, and in this case which method would you run, and how 
>> would you guarantee it has no side effects?
>>
>>>
>>>
>>>>
>>>> /**
>>>>  * Retrieve the value of a (possibly private) field
>>>>  *
>>>>  * @param field
>>>>  *           the field to retrieve the value of
>>>>  * @return the retrieved value
>>>>  * @throws IllegalArgumentException
>>>>  *            if the field doesn't exist on this session object 
>>>> instance
>>>>  */
>>>> public Object getFieldValue(Field field);
>>>>
>>>> Any comments?
>>>>
>>>>
>>>> On 29 Mar 2009, at 22:57, Pete Muir wrote:
>>>>
>>>>> Ok, I would think that some language in the spec to hint at this 
>>>>> might be good, as instances of session beans elsewhere in the spec 
>>>>> implies a EJB reference.
>>>>>
>>>>> On 29 Mar 2009, at 18:04, Gavin King wrote:
>>>>>
>>>>>> Yes, this is a good question - the 299 implementation would need
>>>>>> direct integration with the EJB container for this. (Not the only
>>>>>> place this is needed.)
>>>>>>
>>>>>>
>>>>>> On Sat, Mar 28, 2009 at 7:18 PM, Pete Muir <pmuir at redhat.com> wrote:
>>>>>>> Hey
>>>>>>>
>>>>>>> Not quite sure how this is supposed to work as everywhere we hold a
>>>>>>> reference (proxy) to the EJB, and therefore don't have access to the
>>>>>>> instance and therefore can't get hold of the fields. Any pointers?
>>>>>>>
>>>>>>> Thanks!
>>>>>>>
>>>>>>>
>>>>>>> -- 
>>>>>>> Pete Muir
>>>>>>> http://www.seamframework.org
>>>>>>> http://in.relation.to/Bloggers/Pete
>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> -- 
>>>>>> Gavin King
>>>>>> gavin.king at gmail.com
>>>>>> http://in.relation.to/Bloggers/Gavin
>>>>>> http://hibernate.org
>>>>>> http://seamframework.org
>>>>>
>>>>> -- 
>>>>> Pete Muir
>>>>> http://www.seamframework.org
>>>>> http://in.relation.to/Bloggers/Pete
>>>>>
>>>>> _______________________________________________
>>>>> webbeans-dev mailing list
>>>>> webbeans-dev at lists.jboss.org
>>>>> https://lists.jboss.org/mailman/listinfo/webbeans-dev
>>>>
>>>> -- 
>>>> Pete Muir
>>>> http://www.seamframework.org
>>>> http://in.relation.to/Bloggers/Pete
>>>>
>>>
>>> _______________________________________________
>>> webbeans-dev mailing list
>>> webbeans-dev at lists.jboss.org
>>> https://lists.jboss.org/mailman/listinfo/webbeans-dev
>>
>> -- 
>> Pete Muir
>> http://www.seamframework.org
>> http://in.relation.to/Bloggers/Pete
>>
>> _______________________________________________
>> webbeans-dev mailing list
>> webbeans-dev at lists.jboss.org
>> https://lists.jboss.org/mailman/listinfo/webbeans-dev
> 
> -- 
> Pete Muir
> http://www.seamframework.org
> http://in.relation.to/Bloggers/Pete
> 
> _______________________________________________
> webbeans-dev mailing list
> webbeans-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/webbeans-dev

-- 
Andrew Lee Rubinger
Sr. Software Engineer
JBoss, a division of Red Hat, Inc.
http://exitcondition.alrubinger.com