I have a simple question:
When working in a transaction, should an HQL Named Query flush the session before returning a list of objects?


In pseudo-code (given an entity of type 'A'):

1. Begin transaction
2. Create a new instance of A
3. session.save(the new instance of A);
4. HQL to try to load A
5. Commit transaction


Should the HQL be able to find the new instance of A? I haven't been able to get it to work...

This is related to this other post which I found:
http://forum.hibernate.org/viewtopic.ph ... +flushmode


Thanks!

Hi,
>>When working in a transaction, should an HQL Named Query flush the session before returning a list of objects?
Yes. NO.
We have a say in it.
It depends on session's FLUSH mode.

>> Should the HQL be able to find the new instance of A?
Yes.

>> I haven't been able to get it to work.
Just check you session's flush mode.
In AUTO mode, it works fine. I have set up a small example for the same.

-----------------------------------------
Rate the reply if you find it helpful

Thanks for the reply.

I am still having problems, but am tracking them down.

Right now, the status is that the code works with hibernate 3.1.1 but does not work with 3.2.4 sp1.

In turning up the logging I have found that for 3.2.4, the following hibernate code is not executed:
-----------------------------------------------
2007-06-19 09:02:17,354 [main] DEBUG org.hibernate.event.def.AbstractFlushingEventListener - processing flush-time cascades
2007-06-19 09:02:17,354 [main] DEBUG org.hibernate.event.def.AbstractFlushingEventListener - dirty checking collections
---------------------------------------------------

I'm tracing into the source now to figure out why its being called in the old code, but not the new code.

I am using FlushMode.AUTO everywhere (and have verified via the debugger at runtime).

Thanks for any help

I tracked down the problem. In the newer version of hibernate, the session is only flushed if a transaction is active. I looked through my stack trace and the transactions were being created appropriately by Spring and wrapping my hibernate invocations.

The problem ended up being that I had by accident declared the wrong type of transaction manager. I had been using:
org.springframework.jdbc.datasource.DataSourceTransactionManager
when I (obviously) should have been using:
org.springframework.orm.hibernate3.HibernateTransactionManager

Just a mistake on my part in grabbing the wrong class... hopefully this helps someone else track down a similar problem.

Thanks