Re: MV Keys
From: Marshall Spight <marshall.spight_at_gmail.com>
Date: 28 Feb 2006 20:22:05 -0800
Message-ID: <1141186925.349435.160320_at_e56g2000cwe.googlegroups.com>
> Right now I'm treating 1NF as a 0 or 1. Either data is in 1NF or it is
> not. The RM requires 1NF and all other NF's require 1NF (today). If
> you bend this rule and no longer require normalization (1NF), you break
> the RM as we know it. So, what is this "close" thing?
Date: 28 Feb 2006 20:22:05 -0800
Message-ID: <1141186925.349435.160320_at_e56g2000cwe.googlegroups.com>
dawn wrote:
>
> Right now I'm treating 1NF as a 0 or 1. Either data is in 1NF or it is
> not. The RM requires 1NF and all other NF's require 1NF (today). If
> you bend this rule and no longer require normalization (1NF), you break
> the RM as we know it. So, what is this "close" thing?
Consider some schema that has many opportunities to apply 1NF. Apply most of them, but do not apply all of them. You are now close to having fully applied 1NF.
That's all I'm trying to say. My impression is that once we are in a position not to *require* 1NF, we'll still use it most of the places we could.
This only applies to a system that has a join operator. If you don't have join, you're hosed.
Marshall Received on Wed Mar 01 2006 - 05:22:05 CET