Re: Surrogate Keys: an Implementation Issue

paul c wrote:

> I don't see that immutability matters to the various purposes of most
> db's if those identifiers aren't exposed.

I think the immutability matters if you are trying to have a rollback database or some built in audit support. How else can I track a row over time?

>Also don't see that the term "Surrogate keys" applies.

Granted. That was a bad choice of terms. I guess I was thinking that many of the arguments that people employ in support of surrogate keys are in fact good arguments, but only at the physical level, not at the logical level. Received on Thu Jul 20 2006 - 14:40:51 CEST