Re: Database design, Keys and some other things

From: JOG <jog_at_cs.nott.ac.uk>
Date: 26 Sep 2005 18:39:41 -0700
Message-ID: <1127785181.723733.131210_at_g14g2000cwa.googlegroups.com>


It boils down to this. artificial keys are a workaround. No data model should rely on the user having to fudge identifiers in to get the job done (Though I understand that Codd was a pragmatist, and his recommendation of surrogate keys is to get the system working)

The real world is stating a predicate such as:

"There is a car that is a buick, blue and built in 2000."

NOT
"There is a car that is a buick, blue and built in 2000 with an
invisible universal id of 24."

These key's are fudges, hacks, and indicate something lacking in a model, which should natively be able to handle what you are trying to do.

jog
(I believe I may have used the word fudge too many times) Received on Tue Sep 27 2005 - 03:39:41 CEST

Original text of this message