Vince Kraemer wrote:
> It would be nice to have some feedback from the folks who have direct
> experience with the EJB implementation on this topic.
High level comment is that it would preferable to
incorporate the Java EE verifier code into this process since that's where
we have already placed the majority of these kinds of checks. That would
be better than reimplementing them within NetBeans. That applies to
all the Java EE functionality, not just EJB 3.0 assertions. We already
have hundreds of verifier assertions in Glassfish. That code has been
refined since the initial version of J2EE. E.g., the verifier already
includes basically
all the assertions listed on this EJBVerification wiki, plus many more.
These kinds of assertions are hard to get right and the maintenance
burden when new versions of the related specs are released can be high
since all the existing assertions have to be revisited. That's not
something we
should be doing in two different places.
cc-ing Sahoo, the module owner for the verifier.
Specific feedback :
1. SBSuperClassNotSB
This is not correct. Implementation inheritance is perfectly legal.
The point is just that there is no higher-level notion of "bean
inheritance". If a bean class extends a class that happens to also
be a bean class, the super-class will be treated like any other
inherited class.
2. BMnotPartOfRBIandLBI
This is not correct. It's true that there are differences in the
allowed types for Local and Remote interfaces, but it's too
restrictive to say there is no case where the same method can
be exposed through both. The given example void foo(); is
perfectly legal.
When it comes to local vs. remote recommendations it's best
for the verification to just stick to whether a given interface
meets the requirements of that interface type.
>
> Thanks,
> vbk
>
> ------------------------------------------------------------------------
>
> Subject:
> EJB Verification feature planning - action requested
> From:
> Tomasz Slota <Tomasz.Slota_at_Sun.COM>
> Date:
> Tue, 05 Jun 2007 15:56:28 +0200
> To:
> Martin Adamek <Martin.Adamek_at_Sun.COM>, Andrei Badea
> <Andrei.Badea_at_Sun.COM>, Erno Mononen <Erno.Mononen_at_Sun.COM>, Petr
> Jiricka <Petr.Jiricka_at_Sun.COM>
>
> To:
> Martin Adamek <Martin.Adamek_at_Sun.COM>, Andrei Badea
> <Andrei.Badea_at_Sun.COM>, Erno Mononen <Erno.Mononen_at_Sun.COM>, Petr
> Jiricka <Petr.Jiricka_at_Sun.COM>
> CC:
> "cdp-j2ee-dev_at_sun.com Iteam" <cdp-j2ee-dev_at_sun.com>
>
>
>
> Here is a list of assertions implemented in the EJB 3.0 area in NB 5.5:
>
> http://wiki.netbeans.org/wiki/view/EJBVerification
>
> Unfortunately we might not be able to migrate all the checks to
> Retouche and the EJB model. Not being EJB expert myself I am asking
> for help prioritizing the list. The highest priority obviously should
> go to the checks that could potentially have the highest user impact
> measured either by likeliness of occurrence in practice or ability to
> prevent a particularly tricky (difficult to diagnose) problem.
>
> Please take a few minutes to go through the list of checks and
> propose your own rating (doesn't have to be complete).
>
> Priorities:
> 1 - must have
> 2 - highly desirable
> 3 - will go to 6.0 only if trivial to implement
>
>
> -TS
>
>
>
>
>------------------------------------------------------------------------
>
>---------------------------------------------------------------------
>To unsubscribe, e-mail: dev-unsubscribe_at_glassfish.dev.java.net
>For additional commands, e-mail: dev-help_at_glassfish.dev.java.net
>
>