I have already described this problem in a reply to another topic [NullPointerException in SessionImpl.getOrphans()], but nobody seems to look at it. So here it is again ;-)

In the other topic, Gavin asked me to put a line of code into Hibernate 2.1.2 in order to fix another issue:



The new line of code fixed my problem, but I believe it might be be responsible for another problem I now have in the constellation decribed below. I believe this because the new line of code appears in the stack trace.

Class A has a one-to-many collection of class B with cascade="all-delete-orphan", inverse="true". Class B has a many-to-one reference to class C, which in turn has a one-to-many collection of class B with cascade="all-delete-orphan", inverse="true".

The entire object graph consists of detached objects. I want to remove an instance of B by removing if from the collections in A and C. When I call saveOrUpdate(), the instance of C vetoes the update for application-specific reasons. This causes the operation to fail with
HibernateException: reassociated object has dirty collection (details below)

I do understand the general reason why this exception would be thrown, but I am not sure that it is appropriate in this case for several reasons:
1. The table containing C doesn't need to be updated, so no problem.
2. The collection has inverse="true", so it is not C's job to maintain it.
3. This was not a problem in 2.1 final (not sure that counts).

net.sf.hibernate.HibernateException: reassociated object has dirty collection
at net.sf.hibernate.impl.OnLockVisitor.processCollection(OnLockVisitor.java:47)
at net.sf.hibernate.impl.AbstractVisitor.processValue(AbstractVisitor.java:69)
at net.sf.hibernate.impl.AbstractVisitor.processValues(AbstractVisitor.java:36)
at net.sf.hibernate.impl.AbstractVisitor.process(AbstractVisitor.java:93)
at net.sf.hibernate.impl.SessionImpl.reassociate(SessionImpl.java:1654)
at net.sf.hibernate.impl.SessionImpl.doUpdateMutable(SessionImpl.java:1420)
at net.sf.hibernate.impl.SessionImpl.doUpdate(SessionImpl.java:1441)
at net.sf.hibernate.impl.SessionImpl.saveOrUpdate(SessionImpl.java:1364)
at net.sf.hibernate.engine.Cascades$4.cascade(Cascades.java:114)
at net.sf.hibernate.engine.Cascades.cascade(Cascades.java:436)
at net.sf.hibernate.engine.Cascades.cascadeCollection(Cascades.java:526)
at net.sf.hibernate.engine.Cascades.cascade(Cascades.java:452)
at net.sf.hibernate.engine.Cascades.cascade(Cascades.java:503)
at net.sf.hibernate.engine.Cascades.cascade(Cascades.java:482)
at net.sf.hibernate.impl.SessionImpl.doUpdate(SessionImpl.java:1446)
at net.sf.hibernate.impl.SessionImpl.saveOrUpdate(SessionImpl.java:1364)

I've also experienced this similar problem. Does anyone have a patch that will fix this issue or will I have to wait for Hibernate 2.2?

Thanks so much,
Dino

bumping the thread.

Hoping someone will look at it.... pretty please?

Looked at this and I am happy with the behavior.

Just to clarify: you are saying that if a collection is dirty on an object, the lifecycle method onUpdate() cannot veto the update, even if the relationship has inverse="true"?