Possible regression in aop?

[adrian.brock]

If you look at the following test in deployers:
http://viewvc.jboss.org/cgi-bin/viewvc.cgi/jbossas/projects/jboss-deployers/trunk/deployers-impl/src/tests/org/jboss/test/deployers/scope/test/LoaderMetaDataRepositoryUnitTestCase.java?revision=70660&view=markup
it uses the AOPProxyFactory to test that deployments are getting constructed
with the correct scope/metadata, but it is currently failing with the latest aop upgrade.

This is the deployer that creates the proxy
http://viewvc.jboss.org/cgi-bin/viewvc.cgi/jbossas/projects/jboss-deployers/trunk/deployers-impl/src/tests/org/jboss/test/deployers/scope/support/TestComponentCreator.java?revision=70660&view=markup
I've confirmed that the metadata passed to the proxy factory has the annotation,
but its not triggering the advice getting called when it invokes upon the proxy.

That this test fails is not a big issue, since nobody is really using the scoped
metadata setup by the deployers, but it could be indicitive of some other problem in aop
(assuming its not a broken test that just happened to work before by some fluke? ;-)

[kabirkhan]

The problem was that we moved the aop advisors to check MetaData.isMetaDataPresent()
http://www.jboss.com/?module=bb&op=viewtopic&t=129894
and TestComponentMetaDataLoader.retrieveMetaData() always returing null. I have modified it to check the annotations [url]JBDEPLOY-21/url]

[adrian.brock]

:-)

Ok I see the problem.
The real fix is to invoke the super method instead of returning null.

I don't even know why I override the method for this test :-)

[adrian.brock]

Actually the method is abstract because it is just retrieving a name.

Your implementation should move to be the default implementation in the AbstractMetaDataLoader.