Re: A Normalization Question

From: Paul <paul_at_test.com>
Date: Tue, 13 Jul 2004 00:38:10 +0100
Message-ID: <40f320e1$0$39747$ed2e19e4_at_ptn-nntp-reader04.plus.net>


Neo wrote:

>>relational model: separates out the logical and the physical. 
>>It sounds like you're trying to roll two concepts of "redundancy" into 
>>one when it's more powerful and useful to have them separate.

>
> I am only taking about redundancy of things being represented in the
> db at the logical level not the physical level when I talk about the
> duplicate string 'brown' or duplicate symbols.

The string 'brown' in this context is just a much a single logical thing as the number 10 is. They are both just tokens to represent the colour we call "brown".

Imagine some ficticious language where the word brown is spelt "10". Would you then still replace occurrences of the string "10" with the number 10 that "points" to the string "10"?

Is the problem that you see a string as an array of chars so not a "fundamental" type?

Would you "normalise" the number 19 by looking at its decimal expansion as ("1","9") ? Or by its binary expansion as ("1","0","1","1") ? Ultimately you could reduce anything down to binary so your database would only need one type.

Maybe if a language like Chinese was your first language it would make more sense to you because then you'd be used to words being denoted by single symbols that aren't decomposable into an alphabet?

Paul. Received on Tue Jul 13 2004 - 01:38:10 CEST

Original text of this message