Hi ,
I have written a custom MergeEventListener by sub-classing the DefaultMergeEventListener and overriding its onMerge(MergeEvent event) method .

My goal is to compare the pre-update db snapshot of an entity with the currently changed entity(the entity that is currently registered with the session ) .My goal is that , if there is a difference some processing will be done .

I am elaborating this part a little bit ,
say I have a record Course ( course_id:1, name:"First Course" ) . This is already saved in DB .Now I have loaded the course object using hibernate session and made a change and invoked merge on it .
Lets assume now its state is ( course_id:1, name:"First Course Changed " ) . This is where I want to track the change .For this what I am doing is mentioned below :



As you can see above , persister.getDatabaseSnapshot is for getting the preupdated state and persister.getPropertyValues is providing changed entity state (both of them are proving object arrays ) and then finally I am invoking persister.findDirty method , which is returning me indices of all dirty properties.

This is working fine for me . But when my mapping is changed a little bit with another many to one attribute , I am facing an issue .

For example I have one parent entity named Node and the structure becomes like the following
Node [node_id,name] --->Course[course_id,name]

[Node is basically an entity with table per subclass mapping ]
Then naturally course has a many to one mapping to the parent node .

While a many to one mapping is added , my code throws the exception
--------Exception start ---------
org.hibernate.HibernateException: Unable to resolve entity name from Class [java.lang.Long] expected instance/subclass of [com.ibm.lk.domain.portfolio.NodeBO]

--------Exception end---------

I have checked the exception with Hibernate 3.5.3 source code , it is showing that the exceptions starts from :
org.hibernate.persister.entity.AbstractEntityPersister.findDirty(AbstractEntityPersister.java:3392).

The reason as far as I understand , the oldstate[] passed to findDirty method by using the method

has an entry of a Long object with value 2 [That is the node_id value of Parent object ] whereas the newState array has an object entry of NodeBO object .



This difference of just the Long object (i.e. the object type of the primary key of the parent entity ) and the Node object is leading to the exception .

If I do not have the many to one mapping everything is OK.But otherwise I face this problem.
Any suggestions , so that I can use the findDirty method preperly with many to one cases with table per subclass mapping also ?

Regards,
Ayan

Hi ,
Any update from anybody ? I am willing to know where I am doing the mistake . One thing also came to my mind , I am explaining that below ...
As the main problem seems to be (please correct me if am wrong ) the difference in the representation of a particular element inside Object[] oldState and Object[] newState.
oldState array has many to one entity representation as a Long object (i.e. id of the node object) whereas newState array contains the object itself i.e.(com.domain.portfolio.NodeBO) .
This is where the mismatch starts and it ultimately leads to the problem inside findDirty method . What if I can provide the findDirty with correct type of object arrays ?

Let me show you the mapping files ..
Course.hbm.xml

Node.hbm.xml


What I thought if I change the many to one association from to , then may be the line will fill up the oldState [] with the many to one object instead of its id .But that was not case .

Any idea what can be done ? or do we have any alternative of persister.getDatabaseSnapshot to get oldState [] filled up with actual objects instead of ids ?

Regards,
Ayan

Any update regarding this topic from anybody ?