10:50 a.m.
Hi,
recently we had an issue with regard to Session Bean business interface inheritance when
running CDI TCK 1.0.4.SP3 on GlassFish - see also WELD-1133 [1]. The following discussion
revealed possible specification conflict (EJB 3.1 vs CDI 1.0). Pete Muir asked me to write
some brief summary. So here it is :-)
The problem: it seems that session bean business interface inheritance is not allowed in
EJB 3.1
(4.9.2.1 Session Bean Superclasses)
Let's assume this code:
@Local
interface Bar {}
@Local
interface Foo extends Bar {
}
@Stateless
class FooBean implements Foo {
}
According to the spec (EJB 3.1) the set of exposed client views of FooBean should only be
Foo. Not Foo and Bar.
However the CDI spec defines bean types for a session bean in a following way: "The
unrestricted set of bean types for a session bean contains all local interfaces of the
bean and their superinterfaces" (3.2.2. Bean types of a session bean).
Note that some EJB implementations DO support this kind of inheritance out of the box
(e.g. JBoss AS 7 EJB module).
Martin
[1]
https://issues.jboss.org/browse/WELD-1133
--
Martin Kouba
JBoss Quality Assurance Engineer
E-mail: mkouba(a)redhat.com
Web: www.cz.redhat.com
Red Hat Czech s.r.o., Purkyňova 99/71, 612 45, Brno, Czech Republic
5:21 a.m.
Hi
There's a bunch of small discrepancies like that between EJB and CDI. e.g
GF-16186 <http://java.net/jira/browse/GLASSFISH-16186> I reported one year
ago which ended up on WELD-921 <https://issues.jboss.org/browse/WELD-921>;.
And we also had isues with covariant methods on @Local interfaces in the
past.
With CDI gradually percolating through all EE, one day or another - from
EE7 I hope - EJB and CDI will have to behave the same way.
In the context of this WELD-1133 it would be preferable to align EJB on
CDI, unless of course some important usecase gets harmed (but I don't see
which). Else the spec clash will remain and in a few years users will keep
wonrdering what we were thinking.
fm.
On Tue, Jun 5, 2012 at 5:50 PM, Martin Kouba <mkouba(a)redhat.com> wrote:
Hi,
recently we had an issue with regard to Session Bean business interface
inheritance when running CDI TCK 1.0.4.SP3 on GlassFish - see also
WELD-1133 [1]. The following discussion revealed possible specification
conflict (EJB 3.1 vs CDI 1.0). Pete Muir asked me to write some brief
summary. So here it is :-)
The problem: it seems that session bean business interface inheritance is
not allowed in EJB 3.1
(4.9.2.1 Session Bean Superclasses)
Let's assume this code:
@Local
interface Bar {}
@Local
interface Foo extends Bar {
}
@Stateless
class FooBean implements Foo {
}
According to the spec (EJB 3.1) the set of exposed client views of FooBean
should only be Foo. Not Foo and Bar.
However the CDI spec defines bean types for a session bean in a following
way: "The unrestricted set of bean types for a session bean contains all
local interfaces of the bean and their superinterfaces" (3.2.2. Bean types
of a session bean).
Note that some EJB implementations DO support this kind of inheritance out
of the box (e.g. JBoss AS 7 EJB module).
Martin
[1]
https://issues.jboss.org/browse/WELD-1133
--
Martin Kouba
JBoss Quality Assurance Engineer
E-mail: mkouba(a)redhat.com
Web: www.cz.redhat.com
Red Hat Czech s.r.o., Purkyňova 99/71, 612 45, Brno, Czech Republic
_______________________________________________
cdi-dev mailing list
cdi-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/cdi-dev
--
http://www.suntriprecords.com
7:04 a.m.
Hi,
comments inline...
Martin
Hi
There's a bunch of small discrepancies like that between EJB and CDI. e.g GF-16186
<http://java.net/jira/browse/GLASSFISH-16186> I reported one year ago which ended up
on WELD-921 <https://issues.jboss.org/browse/WELD-921>;. And we also had isues with
covariant methods on @Local interfaces in the past.
WELD-921 is definitely related and should be fixed asap.
With CDI gradually percolating through all EE, one day or another - from EE7 I hope - EJB
and CDI will have to behave the same way.
+100 :-)
In the context of this WELD-1133 it would be preferable to align EJB on CDI, unless of
course some important usecase gets harmed (but I don't see which). Else the spec clash
will remain and in a few years users will keep wonrdering what we were thinking.
Further investigation showed up that the conflict is rather debatable or at least not so
serious. The most important question is whether methods from inherited interface should be
exposed as business methods through local client view of session bean. Not whether
inherited interface is exposed as local view. I suppose they should. But sometimes
it's a question of interpretation... :-(
fm.
On Tue, Jun 5, 2012 at 5:50 PM, Martin Kouba <mkouba(a)redhat.com
<mailto:mkouba@redhat.com>> wrote:
Hi,
recently we had an issue with regard to Session Bean business interface inheritance
when running CDI TCK 1.0.4.SP3 on GlassFish - see also WELD-1133 [1]. The following
discussion revealed possible specification conflict (EJB 3.1 vs CDI 1.0). Pete Muir asked
me to write some brief summary. So here it is :-)
The problem: it seems that session bean business interface inheritance is not allowed
in EJB 3.1
(4.9.2.1 Session Bean Superclasses)
Let's assume this code:
@Local
interface Bar {}
@Local
interface Foo extends Bar {
}
@Stateless
class FooBean implements Foo {
}
According to the spec (EJB 3.1) the set of exposed client views of FooBean should
only be Foo. Not Foo and Bar.
However the CDI spec defines bean types for a session bean in a following way:
"The unrestricted set of bean types for a session bean contains all local interfaces
of the bean and their superinterfaces" (3.2.2. Bean types of a session bean).
Note that some EJB implementations DO support this kind of inheritance out of the box
(e.g. JBoss AS 7 EJB module).
Martin
[1]
https://issues.jboss.org/browse/WELD-1133
--
Martin Kouba
JBoss Quality Assurance Engineer
E-mail: mkouba(a)redhat.com <mailto:mkouba@redhat.com>
Web: www.cz.redhat.com <http://www.cz.redhat.com>
Red Hat Czech s.r.o., Purkyňova 99/71, 612 45, Brno, Czech Republic
_______________________________________________
cdi-dev mailing list
cdi-dev(a)lists.jboss.org <mailto:cdi-dev@lists.jboss.org>
https://lists.jboss.org/mailman/listinfo/cdi-dev
7:18 a.m.
> There's a bunch of small discrepancies like that between EJB
and CDI. e.g GF-16186 <http://java.net/jira/browse/GLASSFISH-16186> I reported one
year ago which ended up on WELD-921 <https://issues.jboss.org/browse/WELD-921>;. And
we also had isues with covariant methods on @Local interfaces in the past.
WELD-921 is definitely related and should be fixed asap.
This requires API changes, which drags in a lot updates through all layers.
We'll be updating the API for OSGi support anyway, so this can be changed then as
well.
I'll grab Stuart to have a look at this once he returns from his European PTO. :-)
-Ales
4352
days inactive
4353
days old
3 comments
3 participants
participants (3)
-
Ales Justin -
Fabien Marsaud -
Martin Kouba