Hello,
we are trying to use Hibernate with optimistic concurrency control. Our entities have the "@version" annotation so that hibernate can check if there has been an update from user B during the timespan from when user A has loaded the entity until user A commits his changes.
So let's assume that user A loads the entity (with the version 1, then user B commits his changes (setting the version in the db to 2), the user A tries to commit. In that case, Hibernate throws the infamous StaleObjectStateException, because there is a conflict.

Now let's now assume that user A wants to accept the changes of user B. So he does something like:

try
{
Session session = transactionManager.getSession();
org.hibernate.Transaction tx = session.beginTransaction();
session.persist(entity);
tx.commit();

}catch(StaleObjectStateException ex){

String entityName = ex.getEntityName();
Serializable identifier = ex.getIdentifier();
tx.rollback();

BusinessObject modifiedObject = entity;
BusinessObject dataBaseState = (BusinessObject) session.get(entityName, identifier);
//System.out.println(modifiedObject.getVersion()); //displays 1
//System.out.println(dataBaseState.getVersion()); //displays 2

session.refresh(modifiedObject);
//System.err.println(modifiedObject.getVersion()); //displays 2

//Try to commit again
Session session = transactionManager.getSession();
org.hibernate.Transaction tx = session.beginTransaction();
session.persist();
tx.commit(modifiedObject); //Throws an StaleObjectStateException again, in spite of having identical version numbers!!!!!
}

Without any other user interfering, this again throws a StaleObjectStateException, although after session.refresh(entity), the entity is up-to-date to the database (in particular, the version is set to 2).

Why is it that Hibernate sees a conflict although the entity is up-to-date to the database?

Many thanks!