I have a schema like this


A:
A_ID PK

B:
B_ID PK
DATE

A_B:
A_B_ID PK
A_ID FK
B_ID FK

A_B_OVERRIDE:
A_B_ID PK
DATE_OVERRIDE

A and B have a many-to-many relationship through association table A_B. The association table itself is exposed as an entity class, because there is a one-to-one hanging off of it that allows the owning entity of the relationship, A, to override an attribute of the owned entity B for the given relationship.

The override table itself as you can see is trivial, and is only used to avoid a nullable override column in A_B. Instead of mapping it as its own entity class, I use a <join> mapping to pull it into the class that maps A_B. This seems to work well and makes my code simple and easy to write.

What do you think?

Ah crap.