I can't believe I am the only one here to encounter this, the simplest inheritance problem you can have. I posted my problem in this thread:

http://forum.hibernate.org/viewtopic.ph ... highlight=

Nobody came forward with suggestions.

In summary, I inherited my POJO in a class that holds my business logic. That new class won't persist. Why? It did under 2.x. Is there something I am doing wrong? I am happy to open a very detailed bug report if appropriate, but in any event, I cannot go forward because of this problem.

Can someone please advise?

I'm not a Hibernate expert, but I'm working on a project that uses Hibernate Synchronizer, which sets up a similar class hierarchy to yours (abstract base POJOs and extending 'business' classes). I recently migrated it from H2 to H3, and didn't have your problem. Maybe you have additional problems, and your class hierarchy isn't the cause.

I would suggest trying lazy="false" on your class mappings, to rule out cglib first.

Good luck.

you need to specify ALL classes you want to persist in the mappings.

in the simplets form is just a <subclass name="blah"/>

Max, using the <subclass> element just deleted my class with the business logic, which is exactly what I don't want.

How does one persist a class that extends a POJO? This used to work.

Nathanmoon:
adding the lazy = false didn't have any effect. Thanks for the suggestion though.

I might point out to others that <subclass> is not the solution to this problem, because it causes the hibernate system to generate the specified class. If you extend a class, the extended class *is* an instance of the superclass, so it makes no sense to me that you would have to do *anything* special. I am still in the dark as to how to solve this problem.

huh ? so your problem is with hbm2java, right ?

Use <meta attribute="generated-class">BaseXXX</meta> to have hbm2java generate a class you can extend to avoid overwritting your concrete class.

Well, no, my problem isn't with hbm2java, it's with hibernate for not recognizing that when A extends APOJO, that A really is APOJO and should persist it.

Hbm2java is a problem when I try to use the subclass attribute to solve that problem, because hbm2java then generates this subclass.

Is using a meta attribute going to cause regeneration? Seems like it would since you're telling me that it causes hbm2java to generate a class in the first place.

The mapping files and stack trace describing this can be found here:
http://forum.hibernate.org/viewtopic.ph ... highlight=

add a feature requst to allow no generation.

nathanmoon pointed out that hibernate synchronizer uses the same inheritance scheme that I do, and that he moved such a project from H2 to H3 with no problems.

So what am I doing wrong? Others, nathanmoon in particular, are getting the desired results from hibernate 3.

Just a shot in the dark - Hibernate Synchronizer declares the base POJOs to be abstract. You could see if making your POJOs abstract helps at all. I think you can do that with hbm2java using:

Thanks for the suggestion nathanmoon. It made the POJO abstract, but I am still stuck with the same error.