What's the rule for mapping in a situation where a concrete base class (ClassA) is simultaneously on the one side of a bidirectional one-to-one relationship with ClassX and a class derived from ClassA (lets say ClassB) is on the many side of a many-to-one relationship with ClassX. Both ClassA and ClassB navigate the relationship using the same property.
How is that property mapped? Do you map that property as one-to-one in the base class and many-to-one in the derived class (with the same mapping name?). Or do you map the property using many-to-one in the base class even though it doesn't fit? The former approach assumes NHibernate is smart enough to *override* the base class mapping in favor of the derived class mapping. I've actually tried both ways and have had issues both ways - just not sure what I should be expecting from NHibernate. Anyone?