On Fri, Apr 9, 2010 at 5:09 AM, Pete Muir <pmuir@redhat.com> wrote:

Hi Jordan,

I don't quite understand what you are doing, but I do wonder why you are making the topic definition a qualifier at all. IMO it is simpler to just treat the topic annotation a normal annotation, meta annotated with @JmsDestination. You can then use the injection point metadata to query the annotation and find out the destination name.

WDYT?

That's a much cleaner way to handle it. The reason I went with qualifiers is to require them on all injection points that had to look up a topic or queue (injection of Topic, Queue, MessageProducer/MessageConsumer). That way it was clear which destination was being used and from where instead of inferring the name from the InjectionPoint.Annotated. I suppose observing ProcessInjectionTarget and adding definition errors for missing annotations would work.

The other issue is that I would like to be able to support programmatic lookup via Instance.select(...). That requires the annotations be qualifiers.

Just brainstorming here... If they were qualifiers I'd need some way of creating unique producers for each declared one. I could dynamically create producer beans during start up for each identified qualifier. That seems like it's more work than should be necessary. If I were able to define some sort of generic qualifier that could be used to fulfill the injection request I wouldn't have to worry about creating beans.

Jordan

Pete

On 7 Apr 2010, at 13:30, Jordan Ganoff wrote:

> All,
>
> I'd like some feedback on how I've implemented JMS destination injection. For reference see the design notes at http://sfwk.org/Seam3/JMSModule.
>
> Generic destination injection is realized by decorating the AnnotatedTypes which define any transitive annotations of @JmsDestination. The annotations are replaced with the defined @JmsDestination and a simple pair of producer methods handle their production (http://github.com/jganoff/seam3-jms/blob/master/src/main/java/org/jboss/seam/jms/impl/inject/DestinationProducer.java). Although this works it removes the annotations. I'd much prefer another solution which leaves all annotations intact and somehow provides producers for each (if not already defined).
>
> Any thoughts on the matter would be greatly appreciated. Am I correct in assuming the original annotations should be preserved?
>
> Code base is currently on github until the svn repo is set up: http://github.com/jganoff/seam3-jms
> Relevant test cases are here: http://github.com/jganoff/seam3-jms/tree/master/src/test/java/org/jboss/seam/jms/test/inject/
>
> Thanks!
>
> --
> Jordan Ganoff

> _______________________________________________
> seam-dev mailing list
> seam-dev@lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/seam-dev