Dears,

I submitted a (probably) too articulate question yesterday.

I'm trying to reduce it to the following:

how much of the database structure must be described in the hibernate mappings and how much of it can be left described by my db-creating sql script? Ie: constraints. I have a table which must be unique on a cluster of two columns. I don't see how to describe it in the hibernate mapping, so I'm asking if I can just create the table with this contraint and let the db 'inform' (by a constraint violation error) the hibernate layer when something goes wrong. Also, I have table columns like 'name' whose uniqueness must be case-insensitive. Can I apply the same strategy?

Thank you,

Giampaolo

set unique="samename" on your both properties

Well, but this way I would encforge uniqueness on both columns. I instead want something like PostgreSQL 'unique(colA, colB)', which means that uniqueness is to be enforced for the clustered columns (ie: colA+colB), not for each one.

Thanks anyway,

Giampaolo

will give you exactly what you describe.

Actually it is unique-key="samename" :)

ah, damnit :)