Re: Primary Key Theory Question

From: dataman <>
Date: Mon, 29 Mar 2004 13:57:21 -0000
Message-id: <40682910.12d.41_at_news2>

> So, if sequential, server generated surrogate keys are not
> inherently evil, why do I see so much negative press
> given to the Sybase/MS style Identity column? Outside of
> it being non-portable/non-standard, I can get the same
> functionality in MS and Oracle (my primary RDBMSs). Sure,
> I could replicate the functionality of a Oracle sequence
> for both using tables with one row and one field for each
> ID I need, and I could code up some handy SPs to make the
> coding easy, but, if the server will do it for me, why
> bother (I take laziness as the sign of a good coder in
> general).
> <snip>
> > other databases have similar mechanisms (except for
> MS-Access, which
> > isn't really a database...)
They're not evil. The point is, you should always consider a natural key before deciding on a surrogate key. It just so happens that people can be difficult to uniquely identify. You can change your name, SS#, your address, virtually anything that can be associated to you as an identifying attribute. Received on Mon Mar 29 2004 - 15:57:21 CEST

Original text of this message