After some conversations with David Blevins at ApacheCon in which he
pointed out that the jpa impl is not supposed to look for
persistence.xml files itself when deployed in an ee environment I
wrote a PersistenceUnitBuilder that can interpret persistence.xml
files included in a geronimo plan (where the gbeans go). This works
much better than leaving it out :-)
It looks like openjpa support is working fairly well in the sandbox/
javaee5 stuff. The toy jpa itest app deploys and can access the db
without errors.
I think the next step might be to make a version of daytrader that
uses jpa. I'm not quite sure what the best way to do this is. Here
are some ideas:
1. include the jpa stuff in the ejb jar, acessed through a session
bean and through something like tradeDirect. This forces daytrader
to run only on jdk 1.5
2. fork daytrader and do (1) in a fork
3. fork daytrader and only support jpa in the fork
4. have 2 ejb jars, one using jpa, the other not, and have 2 ears ,
one with jpa and the other not.
From a brief look at daytrader it looks to me as if the way to
proceed is to annotate and enhance the data holder classes in the ejb
jar. I don't know if using disconnected enhanced classes would have
any effect on performance for cmp or direct jdbc use, so I think that
(3) or (4) might be advisable to avoid possible interference.
Thoughts?
thanks
david jencks
