Hi, comments inline.
Matej ----- Original Message ----- > From: "Benjamin Confino" <benja...@uk.ibm.com> > To: firstname.lastname@example.org > Cc: "Emily Jiang" <emiji...@uk.ibm.com> > Sent: Monday, March 12, 2018 6:40:27 PM > Subject: [weld-dev] Potential bug with Typed not affecting if a > ProducerMethod is proxiable. > > Hello > > I can see something that might be a bug in weld. I attempted to work > around https://issues.jboss.org/projects/WELD/issues/WELD-2466 by using > the@Typed annotation on a Producer Method to exclude the unproxiable Types > and thus avoid getting caught by the failing isBeanProxiable check. > However the annotation did not affect whether or not the ProducerMethod is > proxiable. This might be a second bug. > > I have uploaded a recreate application which I have verified against > Wildfly 11.0.0 here: https://github.com/benjamin-confino/TypedProducerBug Thanks for reproducer, checking it now. > > To recreate you can download the pre-compiled binary from github, deploy > it to a version of Wildfly without the fix for WELD-2466, and visit the > url http://localhost:8080/WeldRecreateProducesExtendedAbstract/inheritance > this will trigger a call to getInjectableReference() which will fail > because the ProducerMethod for recreate.BeanProducer.produceBean() is not > proxiable. > > If you debug into constructor for ProducerMethod at line 76 you will see > that the proxiable flag is set using method.getTypeClosures() ; which > returns all the associated types, rather than just those specified with > the Typed() paramater. > > Is this the correct behaviour? Should a ProducerMethod be unproxiable even > with Typed limiting the bean types to proxiable types? I think this is correct behaviour because: * Weld needs to proxy whole class hierarchy, we cannot just skip one underlying class when creating the proxy object * @Typed is then used to determine where can this bean be injected, e.g. which bean types are used to match it against required types of each injection point Therefore, @Typed is not really connected to proxyability but rather to assignability of that produced bean to a given injection point. I'll double check this but I really think it's correct behaviour. > > Regards > Benjamin > Unless stated otherwise above: > IBM United Kingdom Limited - Registered in England and Wales with number > 741598. > Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU > > _______________________________________________ > weld-dev mailing list > email@example.com > https://lists.jboss.org/mailman/listinfo/weld-dev _______________________________________________ weld-dev mailing list firstname.lastname@example.org https://lists.jboss.org/mailman/listinfo/weld-dev