Hi all,
I know it doesn't seem to make much sense that there are circular
dependencies in a data model, but the requirements make it so.
I'm coming across a problem where it becomes a "chicken or the egg" kind
of issue...
I will described a scaled-back version of our current model with the
following example:
(Obj1: User, Obj2: Specialization)
1) There is a Many to Many relationship between Users and
Specializations, because many users can have many specializations.
2) There is a Many to One relationship between Specializations and
Users, because each specialization will have a "primary contact" (aka
user). This is not nullable either.
Now how does one load the data?
1) Load Users first, but there are ConstraintViolationExceptions since
the Specialization objects have not been loaded.
2) Load the Specializations first, but it will complain about the null
FK since the User object hasn't been loaded yet.
Right now I am writing a workaround in java where I will load Users
first. If a User object breaks the constraint, I will make a clone and
load it without the Specializations set (clearing it manually with the
setter). But I will keep a list of the failed User objects to load for
later. I will load the Specializations second. Then finally, will try
to load the "bad apple" Users at the end, with the Specializations set
populated. Hibernate should simply update the Users with the
Specializations (and thus add records into the M-to-M junction table).
Is there something cleaner I can do to get around these circular
dependencies? Something I can leverage in my Hibernate mapping?
Thanks,
-Cheryl