Why is an AuditLog class used, instead of putting the persistence code directly in the postFlush method of the AuditLogInterceptor ?

I'm wondering if this was done some technical reasons, which I don't see, or for design reasons. If for design reasons, it looks like this is done for the situation where there may be multiple instances of the Interceptor, as opposed to a global (singleton) Interceptor (such as one wired into Spring's Hibernate LocalSessionFactoryBean).

Can you tell me why it was done this way ?

Thanks.

Because it looks better.

I've implemented an AuditLogInterceptor similar to the one in HIA.

Like the example, I have an explicit flush call, so that the audit data will be created in postFlush, and then will be persisted at the end of the transaction.

There is one minor annoyance I am encountering with this model. When the transaction ends, not only is the audit data begin persisted as desired, but the domain object is being updated again even though it has not been updated since the explicit call to flush. I believe that this is happening because the domain object has a <timestamp> for managed versioning, and so Hibernate is detecting that the entity is dirty simply because the timestamp was updated during the last update.

Is there any way I can get rid of this unnecessary update ?

Thanks.

No idea, use the regular user forum.

fair enough --- I will

let me just ask one question strictly related to the book --- given that it states that flush() is rarely used except for such things as triggers and mixing with JDBC, there is no reason given in the AuditLogInterceptor example for the use of an explicit flush() --- am I correct in assuming that this is done to create the audit data in postFlush, so that the audit data itself will be persisted when the transaction ends ? or is there some other reason why an explicit flush is used in the example ?

thanks

The method has all kinds of uses, esp. if you write application transactions with a FlushMode.NEVER and long Session. In the example it was used to show that flush has to happen for postFlush() and onFlushDirty() to run, of course...