On Fri, 2009-07-03 at 11:29 -0700, David Fetter wrote:
I'm missing what you're doing here that foreign keys don't cover.
Could you send along your DDL?
No DDL yet... I'm just in the thinking stages. FKs technically would do it, but would become unwieldy. The intention was to have subclasses of each of the variant, association, and phenotype tables. That leads to the polymorphic key problem.
Just generally, I've only found table inheritance useful for
partitioning. "Polymorphic" foreign key constraints can be handled
other ways such as the one sketched out below.
That answers the question -- I do want polymorphic foreign keys. Dang.
Thanks,
Reece