Re: Natural keys vs Aritficial Keys

From: Roy Hann <specially_at_processed.almost.meat>
Date: Sat, 16 May 2009 13:45:38 -0500
Message-ID: <>

Walter Mitty wrote:

> My big bugaboo with the ID field is that if two rows have different IDs but
> otherwise identical data, there tends to be a disconect between developers
> and analysts as to whether this is harmful duplication or not.

There's no two ways about it. If the row would be duplicated if not for the spurious distinction of a value that was generated only to ensure distinction and *for no other reason*, it's logically still a duplicate row.

By itself a duplicate is absurd but pretty harmless. The problem is when one copy gets updated. Then you've got a contradiction and no logical basis to prefer one over the other. That just can't ever be right and it's not something that can ever be a matter of opinion or taste.

Received on Sat May 16 2009 - 20:45:38 CEST

Original text of this message