Hi,

I'm getting this error in my program and don't know why...

I have a process implementing the Runnable interface which looks this way:

public class PerfectMatcherProcess extends HibernateProcess {

protected final ThreadLocal session = new ThreadLocal();

protected void initSession() throws DataException {
Session s = PersistenceLayer.getInstance().getSession();
session.set(s);
}

protected Session getSession() {
return (Session) session.get();
}

public void run() {
...
initSession();
...
Transaction tx = getSession().beginTransaction();
...
getSession.update(a);
getSession.save(b);
getSession.evict(b);
...
getSession().delete("from X where X.test = 'T'"); // here comes the exception!
...
}
}

a, b and c are objects, X is a classname. If I run my program, I'm getting the following exception:

net.sf.hibernate.AssertionFailure: possible nonthreadsafe access to session
at net.sf.hibernate.impl.SessionImpl.postInsert(SessionImpl.java:2274)
at net.sf.hibernate.impl.ScheduledInsertion.execute(ScheduledInsertion.java:29)
at net.sf.hibernate.impl.SessionImpl.executeAll(SessionImpl.java:2306)
at net.sf.hibernate.impl.SessionImpl.execute(SessionImpl.java:2259)
at net.sf.hibernate.impl.SessionImpl.flush(SessionImpl.java:2185)
at HibernateProcess.closeSession(HibernateProcess.java:50)
at HibernateProcess.processProcessingStopped(HibernateProcess.java:64)
at matcher.PerfectMatcherProcess.run(PerfectMatcherProcess.java:122)
at java.lang.Thread.run(Thread.java:536)

How can I handle this?

Thanks,

Thorsten

Call flush() between save() and evict(). Perhaps Hibernate is being a little bit too fussy by throwing an exception here, but I recall that there is another way that this can occur, which we are tryng to rule out.

I have the same problem as the one mentioned in the previous message. I cannot call flush between save and evict.

Is there any chance that hibernate do not throw this exception in future versions ?

Thanx

Seb

gavin thanks heaps - that soled the problem for something I'm working on.

It seems to be a design decision that evict() does not work on new objects - i.e. objects that have been moved to the state "persistent" using save().

This contradicts the state diagrams shown e.g. in "Java Persistence with Hibernate" or "Hibernate in Action", but it seems to be accepted. In Frameworks which separate the creation of objects, including doing a save (e.g. when a factory decides whether to load the object from the DB; or create a new one - afterwards, we want the postcondition "object is in state persistent" to hold) from actions which might do an evict (because later the object might be shipped somewhere else) and which for some reason do not want to use the "workaround" to call flush(), this is quite a PITA ... but so it goes. However, I confess I have only looked into NHibernate 1.0.x and 1.2 and Hibernate 2.x - for newer versions, the software might actually follow the documentation.

I just bumped into this when I changed my id generators from native to increment. I removed the evict() calls and it works, so I just call session.clear() periodically.

Any idea why this happens?