Re: Surrogate primary key plus unique constraint vs. natural primary key: data integrity?

From: <robur.6_at_gmail.com>
Date: Tue, 12 Mar 2013 09:47:31 -0700 (PDT)
Message-ID: <dab4e24f-f488-4c54-b248-42dab016843c_at_googlegroups.com>


On Monday, March 11, 2013 8:53:56 PM UTC+2, Wolfgang Keller wrote:
>
> Since I couldn't find an example like that in the (online as well as
> printed) literature, is there a textbook (or online documentation) about
> these issues? First, to quote it when I have to argue with people, and
> second, to improve my own education.
>

Unfortunately I cannot recommend any paper on this particular subject (dbdebunk is still a good reference), my observations are mainly based on practice, I use composite natural keys a lot.

The main point is that foreign keys do not represent references to rows, but subsets of values. And an overlapping foreign key actually denotes a subset of an intersection (or join) of some relations. This has no equivalence in OO world, that’s why the analogy between foreign keys and references is so dangerous (in spite it apparently works for single attribute keys). Received on Tue Mar 12 2013 - 17:47:31 CET

Original text of this message