[cdi-dev] broken cdi generic rules

[Mark Struberg]

Hi folks!

It's about whether a 

@Produces <T> SomeClass<T> makeIt(InjectionPoint ip) {.. return (T) t;}

would be suitable for injection into
private @Inject SomeClass<Other> x;


We've had this discussion a few times already [1] and to me it seems that we are now d'accord that we must switch back to the CDI-1.0 behaviour of 'assignable to' for section 5.2.4.
See CDI-389 for more info on this topic.

The recent trigger was a discussion over at DeltaSpike [2] and also on some other projects like OmniFaces [3] which failed with the CDI-1.1 behaviour.

I hope we are all on the same boat here and we need to switch back to the CDI-1.0 wording for now, right?


LieGrue,
strub


[1] http://lists.jboss.org/pipermail/cdi-dev/2013-July/004258.html

[2] https://issues.apache.org/jira/browse/DELTASPIKE-405

[3] https://issues.apache.org/jira/browse/OWB-893