Re: Normalize until neat, Automate until Complete

From: Kenneth Downs <firstinit.lastname_at_lastnameplusfam.net>
Date: Wed, 24 Nov 2004 17:29:17 -0500
Message-ID: <tief72-t74.ln1_at_pluto.downsfam.net>


Kenneth Downs wrote:

> Jan Hidders wrote:
> 

>> Kenneth Downs wrote:
>>> 
>>> I have a simple question for those who oppose automation on the grounds
>>> that
>>> it denormalizes.  If normalization is intended to ensure correctness,
>>> and your system disallows writes to automated columns, have you not
>>> preserved
>>> correctness while also improving the lot of your users?  If so, isn't
>>> that what it's all about?

>>
>> Yes, it is. In fact, that is exactly what normalization theory tells us.
>> If the redundant columns cannot be updated then there are no update
>> anomalies. Therefore theory only requires that the updatable part should
>> be normalized if you want to avoid update anomalies.
>>
>> Don't blame the theorists if the practicioners have only a shallow
>> understanding of the theory. :-)
>>
> 
> hmmm, I must be one of those practitioners who has only a shallow
> understanding of the theory.  As far as I can tell, relational algebra and
> relational calculus are both about querying existing data (though I am
> willing to be corrected).  Under what terms might I find the theoretical
> foundation for the *generation* of information out of other information?
> 

I should have pointed out when asking this question that I am aware of the use of derivations in views and selects, what I have never found is any systematic treatment of their inclusion in tables, except for the donning of scary masks and a lot of dancing around howling, "away evil spirits, begone!"

-- 
Kenneth Downs
<?php $sig_block="Variable scope? What's that?";?>
Received on Wed Nov 24 2004 - 23:29:17 CET

Original text of this message