---------- Forwarded message ----------
From: Gavin King <gavin(a)hibernate.org>
Date: Thu, Oct 30, 2008 at 2:09 AM
Subject: Several proposals
To: Java Community Process JSR #299 Expert List <JSR-299-EG(a)jcp.org>
Hi, everyone, here's a list of things I have on my todo list that we
either did not have time to discuss for the public draft, or that I've
Currently the only place that Web Beans may be declared using XML is a
file named web-beans.xml. I can see that third-party extensions would
like to allow web bean declarations to live in other files. So I would
like to propose the addition of a public parse() method to the Manager
interface that would load and parse a file with a given path from the
classpath and install all web beans declared therein.
Initial field values
I think we should probably add BigDecimal/BigInteger to the list of
types that accept initial field values in 9.2.5. I *definitely* think
we should add the java.lang primitive wrapper types to the list.
Currently the spec is no explicit as to whether the deployment type of
an interceptor/decorator has to be enabled for the
interceptor/decorator to be enabled. I suggest that we say explicitly
that the interceptor/decorator must be (a) listed in web-beans.xml and
(b) have an enabled deployment type.
"EJB 3-style" language
Currently the spec only requires support for "EJB 3-style" session
beans, without really saying what that means. Really what it means is
that we don't require support for EJBs with home interfaces. We have
two options here:
(1) be explicit about what "EJB 3-style" means
(2) support EJBs with home interfaces if they have a create method
with no parameters or annotated @Initializer
I'm not sure which is best. For completeness, I don't see why we can't
support EJBs with home interfaces, but on the other hand it seems that
this feature is now basically deprecated, and I assume that support
for EJB home interfaces will not be required in EJB lite.
What do you guys think?
Interceptor stack creation
At the very last minute I made a change to section 5.8, after a
discussion with the EJB EG, and introduced a silly bug. The spec
currently says that interceptor/decorator stacks are created after
@PostConstruct is called, which of course is wrong, since the
interceptor might have a @PostConstruct. Let's fix that bug.