Re: Newbie question about db normalization theory: redundant keys OK?

From: Ross Presser <rpresser_at_gmail.com>
Date: Fri, 14 Dec 2007 09:52:36 -0800 (PST)
Message-ID: <68d6ef06-3978-4950-8ce8-f711afdb45a0_at_i12g2000prf.googlegroups.com>


On Dec 14, 10:49 am, "Tony Rogerson" <tonyroger..._at_torver.net> wrote:

> I'm interested - how would you solve this problem in the real world? Would
> you prevent the teacher from changing their name?

If teachers are permitted to change their names, then names are not appropriate keys for teachers. They are (excuse me if I get the language wrong) mutable attributes of these entities. "Ms. Fred" *isn't* exactly the same person as "Mrs. Sid" -- she's married now, she has a fancy new diamond ring, new taxable status, etc., etc. She shares history with Mrs. Sid -- but an omelet shares history with an egg, too.

If marital status is not significant to your model of teachers and classes, then you should choose a key that is not affected by a change of marital status. As another example in the USA, the SSN is not affected by a change of name, but the driver's license number generally is! Received on Fri Dec 14 2007 - 18:52:36 CET

Original text of this message