JBossDeveloper

@Steve

IDENTITY works perfectly with MySQL auto_increment and Hibernate.

Hibernate executes the INSERT on EntityManager.persist() so the ID is available.

As the lead developer of Hibernate, yes I understand it works   But I understand when it does and when it does not.  Please read ALL of my reply, thanks.

I'll create the jira for allowing hibernate.id.new_generator_mappings=false to be specified as a property in the persistence.xml.

Created AS7-2460 which was inspired by this thread.

Just to be clear...  Personally I do not see AS7-2460 as the "solution" to this discussion.  AS7-2460 is just about giving AS7 users the ability to override the "hibernate.id.new_generator_mappings" setting.  But that just reverts Hibernate to its legacy behavior, which was changed for a reason.  We needed to allow specifying "hibernate.id.new_generator_mappings" to allow for backwards compatibility for pre-existing applications.

For new applications, specifying true for hibernate.id.new_generator_mappings (the default in AS7) is definitely the recommended set up. 

What Ove would be more interested in is a RFE against Hibernate itself to allow users to request that Hibernate create a sequence per entity in the case when no explicit sequence name is given.  Basically a org.hibernate.id.enhanced.SequenceStyleGenerator corollary to the org.hibernate.id.enhanced.TableGenerator#CONFIG_PREFER_SEGMENT_PER_ENTITY setting.  That RFE has not been created yet afaik.

Ove, another thing you might want to ask yourself is why you care that there is a single sequence instead of one per entity.  I have worked on huge systems in the past where this was the case and never had issues with running out of values. 

Here is that Hibernate RFE...

Steve,

I'm kind off piggy backing on this thread, but after encountering an issue that was solved by setting hibernate.id.new_generator_mappings=false i was wondering if someone can direct me to the place that explains why in AS 7 the out off the box value is true, while Hibernate documentation says that the value is false for reasons of backsword computability.

I didn't explore the new behavior much, but I did need to migrate code from 4.2.3 and setting the value to false kept me with the behavior I expected.

Did is just a question for academic knowledge.

thanks,

Noa

Libraries always need to allow for backwards compatibility (at least for some period of time).  That is the only purpose of (a) having ' hibernate.id.new_generator_mappings' and (b) allowing ' hibernate.id.new_generator_mappings' to be false.  Otherwise, these improvements in how identifier generation happen would cause problems with older applications that used Hibernate before those improvements were in place.

But, they are called improvements for a reason...

I just added some notes here about the properties that AS automatically adds to the persistence unit. 

Thanks, I have some time constraints so I can't look into it right now - but I will explore the issue more once I have the chance