Mailing-List: contact users-help@openjpa.apache.org; run by ezmlm
Precedence: bulk
Reply-To: users@openjpa.apache.org
Received-SPF: pass (nike.apache.org: local policy)
From: Henno Vermeulen <henno@huizemolenaar.nl>
To: "'users@openjpa.apache.org'" <users@openjpa.apache.org>
Date: Wed, 13 Apr 2011 14:59:32 +0200
Subject: RE: two issues when not using a DetachedStateField
Thread-Topic: two issues when not using a DetachedStateField
Thread-Index: Acv50fQ8QfQ96xU6SjG8Yo2VhAafLAABcO3gAACV/BA=
Message-ID: <1C448C478A6B4743AF19DBC3C3DCE13201B7FB4DF055@HMS.hm.local>
References: <1C448C478A6B4743AF19DBC3C3DCE13201B7FB4DF04F@HMS.hm.local>
 <1C448C478A6B4743AF19DBC3C3DCE13201B7FB4DF052@HMS.hm.local>
In-Reply-To: <1C448C478A6B4743AF19DBC3C3DCE13201B7FB4DF052@HMS.hm.local>
Accept-Language: nl-NL
Content-Language: nl-NL
acceptlanguage: nl-NL
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0

Even though the merge succeeds, there is one big issue with removing the ca=
scade from the bidirectional relation namely I get that=20

     event !=3D event.activities.get(0).event

Could this mean that event is unnecessarily fetched and extra unnecessary S=
QL gets executed (I've seen issues on this)?

-----Oorspronkelijk bericht-----
Van: Henno Vermeulen [mailto:henno@huizemolenaar.nl]=20
Verzonden: woensdag 13 april 2011 14:40
Aan: 'users@openjpa.apache.org'
Onderwerp: RE: two issues when not using a DetachedStateField

Issue 2 also does not occury when Activity to Event is not set to cascade.
The issue also doesn't occur when StringI18N is removed and the object grap=
h is only three entities deep. So it is quite a complicated combination of =
factors: no detached state field, four entities and a cascade from Activity=
 to Event...

-----Oorspronkelijk bericht-----
Van: Henno Vermeulen [mailto:henno@huizemolenaar.nl]=20
Verzonden: woensdag 13 april 2011 14:30
Aan: 'users@openjpa.apache.org'
Onderwerp: two issues when not using a DetachedStateField

We use OpenJPA 2.1.0 with build time enhancement. We have a client/server s=
ituation so we always detach + serialize entities on commit using fetch gro=
ups:

<property name=3D"openjpa.DetachState" value=3D"fetch-groups(DetachedStateF=
ield=3Dfalse)" />
            <property name=3D"openjpa.AutoDetach" value=3D"commit" /

We currently don't use a DetachedStateField or DetachedStateManager because=
 everything worked fine without them until now.

I found two issues that don't occur when I use DetachedStateField=3Dtrue. (=
According to the user guide OpenJPA should function just as well without on=
e except that it may be less efficient. "OpenJPA can take advantage of a de=
tached state field to make the attach process more efficient. This field is=
 added by the enhancer and is not visible to your application.")


Issue 1:
When I add a new entity to a List of a detached (but not Serialized) entity=
 and then merge, OpenJPA throws:

org.apache.openjpa.persistence.InvalidStateException: The context has been =
closed.  The stack trace at which the context was closed is held in the emb=
edded exception.
FailedObject: java.lang.IllegalStateException
            at org.apache.openjpa.kernel.BrokerImpl.assertOpen(BrokerImpl.j=
ava:4615)
.......

We don't experience this as a problem because we always Serialize an entity=
 before adjusting it!!!


Issue 2:
We have an object graph of four entities with autogenerated Id columns:

Event (OneToMany bidirectional) Activity (OneToMany) ProductOrderLine (OneT=
oOne) StringI18N

All relations are FetchType.EAGER and Cascade.ALL.

When I add a new Activity (with one new ProductOrderLine with one new Strin=
gI18N) to an existing Event and then merge the Event, OpenJPA tries to inse=
rt a null value into the StringI18N Id column. This seems to be only detect=
ed on transaction commit.

This problem does not occur when:

-          I make the Event - Activity relation unidirectional, i.e. Activi=
ty does not know it's parent Event.

-          The Event is new as well.

-          (workaround) I first add the new Activity to the Event but with =
no ProductOrderLines, then merge the Event, then add the ProductOrderLine t=
o the Activity in the and merge the Event again.

Are these known issues? What is the best way for me to go? Should I do my w=
orkaround or should I try to use a detached state field?
Shall I report this to JIRA? I have a test case which shows these issues.


Regards,

Henno Vermeulen
Huize Molenaar