that's the reason why i suggested a (central) spi for such cases. per default we can pass the tck and then you can decide on your own what works for you. (e.g. skipping checks because they are broken or skipping checks e.g. in project-stage production because they just introduce a significant overhead).
regards, gerhard 2013/4/12 Mark Struberg <[email protected]> > Not nice != broken. > > In the case of the serialization rule it was really broken and got > 'clarified' in cdi-1.1 ;) > > LieGrue, > strub > > > > > ----- Original Message ----- > > From: Gerhard Petracek <[email protected]> > > To: [email protected] > > Cc: > > Sent: Thursday, April 11, 2013 9:05 PM > > Subject: Re: Yan: CDI 1.0 TCK Problem + validatePassivationDependencies > > > > hi @ all, > > > > we also have other parts which are required by the tck and aren't that > nice. > > e.g. the check in EventUtil#checkEventBindings costs quite a lot of > > performance. > > > > imo we should introduce something like a "(Tck)RuleService" which > > passes > > the tck by default -> we can provide an implementation which changes such > > (simple) rules easily. > > > > regards, > > gerhard > > > > > > > > 2013/4/10 Mark Struberg <[email protected]> > > > >> Oki time to explain this now. > >> > >> > >> Think about a method > >> > >> @Produces @SessionScoped > >> > >> public EntityManager createEM() { return emf.createEntityManager(); } > >> > >> > >> And this would fail, even if the EntityManagerImpl being returned > might be > >> perfectly Serializable. This can actually only be verified at runtime > and > >> not at scanning time. The only exception is if a return type is final > - but > >> then it cannot get proxied anyway... > >> > >> LieGrue, > >> strub > >> > >> PS: I would not make a session scoped EM, but thats another story... > >> > >> > >> ----- Original Message ----- > >> > From: Mark Struberg <[email protected]> > >> > To: "[email protected]" > > <[email protected]> > >> > Cc: > >> > Sent: Wednesday, April 10, 2013 1:10 PM > >> > Subject: Re: Yan: CDI 1.0 TCK Problem + > > validatePassivationDependencies > >> > > >> > Don't you remember how many tests we challenged/excluded until the > > TCK > >> was > >> > finally ok? > >> > Well, this is another of those issues - but it got catched only > pretty > >> late. > >> > > >> > LieGrue, > >> > strub > >> > > >> > > >> > > >> > > >> >> ________________________________ > >> >> From: Gurkan Erdogdu <[email protected]> > >> >> To: "[email protected]" > >> > <[email protected]>; Mark Struberg > > <[email protected]> > >> >> Sent: Wednesday, April 10, 2013 11:23 AM > >> >> Subject: Yan: CDI 1.0 TCK Problem + > > validatePassivationDependencies > >> >> > >> >> > >> >> Hi Mark > >> >> > >> >> 1.1.8 branch > >> >> > >> >> > >> >> > >> >> Broken means that it is not necessary to pass this in TCK for CDI > > 1.0, > >> why > >> > this test exist in TCK? > >> >> > >> >> > >> >> Thks. > >> >> > >> >> > >> >> > >> >> Gurkan > >> >> > >> >> > >> >> > >> >> ________________________________ > >> >> Kimden: Mark Struberg <[email protected]> > >> >> Kime: "[email protected]" > >> > <[email protected]> > >> >> Gönderildiği Tarih: 9 Nis 2013 21:47 Salı > >> >> Konu: Re: CDI 1.0 TCK Problem + validatePassivationDependencies > >> >> > >> >> because it's broken! > >> >> It's broken in the CDI-1.0 spec and we clarified the correct > > behaviour > >> > in CDI-1.1. > >> >> > >> >> Btw, which branch do you > >> > speak of? > >> >> > >> >> LieGrue, > >> >> strub > >> >> > >> >> > >> >> > >> >> > >> >> ----- Original Message ----- > >> >>> From: Gurkan Erdogdu <[email protected]> > >> >>> To: openwebbeans-dev <[email protected]> > >> >>> Cc: > >> >>> Sent: Tuesday, April 9, 2013 11:17 AM > >> >>> Subject: CDI 1.0 TCK Problem + > > validatePassivationDependencies > >> >>> > >> >>> Hi > >> >>> > >> >>> In AbstractProducerBean below method is commented out but TCK > > 1.0 > >> still > >> > checks > >> >>> ProducerMethod's Serializable return type and fields. > >> >>> > >> >>> public void validatePassivationDependencies() > >> >>> { > >> >>> // don't call > > super.validatePassivationDependencies()! > >> >>> > >> > // the injection points of producers are the parameters of > the > >> >>> producermethod. > >> >>> // since CDI-1.1 we must not check those for is > > serializable > >> > anymore. > >> >>> } > >> >>> > >> >>> > >> >>> In CDI 1.1 this is corrected but TCK 1.0 still check this. > > Why is this > >> > commented > >> >>> out? > >> >>> > >> >>> > >> >>> Gurkan > >> >>> > >> >> > >> >> > >> >> > >> >> > >> > > >> > > >
