Done.

So this is interesting and very instructive.

My first observation was that the annotation-based mapping is completely different from the XML-based mapping:

• in XML, we have single-table inheritance where one of the subclasses has a secondary table, but

• in annotations, we have a JOINED inheritance hierarchy, with a @DiscriminatorColumn tacked on, and two classes mapping to the same table.

These two things just aren’t equivalent. You can easily do SINGLE_TABLE inheritance and a @SecondaryTable with annotations, and you will get the same effect as with your XML mapping. This is and always has been the recommended way to solve this problem in Hibernate, since at least H3 and possibly earlier.

However, just for fun, I actually went through the motions of trying to make your code work and…to my surprise, it was actually pretty straightforward. H6 already had all the difficult parts of it built-in, and it was mainly just a matter getting better support for @DiscriminatorColumnin JOINED inheritance, which was easy to do (in particular, I needed to be able to “force” the discriminator, which was previously not supported).

Anyway, I’m not sure if I would recommend this approach, and we might still run into bugs with it, but I have got it working and I don’t hate it. It’s also poses some rather nice constraints on the implementation of H6, which I like, for robustness.