[
https://issues.apache.org/jira/browse/OWB-706?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Joe Bergmark resolved OWB-706.
------------------------------
Resolution: Fixed
Delivered the long overdue unit tests. This does appear to be working
correctly in trunk.
> Interceptor bindings on methods that come from an interface with a generic
> type may not work
> --------------------------------------------------------------------------------------------
>
> Key: OWB-706
> URL: https://issues.apache.org/jira/browse/OWB-706
> Project: OpenWebBeans
> Issue Type: Bug
> Components: Interceptor and Decorators
> Affects Versions: 1.1.5
> Reporter: Joe Bergmark
> Assignee: Joe Bergmark
> Priority: Minor
>
> If you apply an interceptor binding to a method that implements an interface
> that has a generic type, the interceptor might not be executed.
> I'll write some unit tests, but I think the scenario is something like the
> following:
> public interface Foo<T>{
> public void doSomething<T>
> }
> public class Bar implements Foo<String>{
> @MyInterceptorBinding
> public void doSomething<String>
> }
> When a contextual reference of Bar is cast to Foo, or even to Foo<String>,
> calls to doSomething are no longer intercepted.
> What seems to be happening is that due to type erasure, Bar really has to
> implement Foo<Object>, so the JDK adds a bridge method doSomething(Object) to
> Bar that under the covers just calls doSomething(String). The handler
> doesn't recognize that doSomething(Object) will call doSomething(String) and
> therefore doesn't run the interceptor.
> While the problem is complex, I believe the solution is pretty easy (at least
> for the Javassist case). We can just add a check for bridge methods to the
> new MethodFilter (probably rename it to something less specific than
> FinalizeMethodFilter) so that doSomethign(String) will get called on the
> proxy and we will properly run the interceptor then rather than passing the
> call to doSomething(Object) down to the proxied instance.
> This problem goes back to version 1.0, and its a very edge case, so I don't
> think it has to go into 1.1.6.
--
This message was sent by Atlassian JIRA
(v6.1.4#6159)
