Theory IS Practical.

From: Laconic2 <laconic2_at_comcast.net>
Date: Sat, 9 Oct 2004 09:03:19 -0400
Message-ID: <66KdncizAPo1fPrcRVn-qw_at_comcast.com>



I'll admit that I haven't read much of Date. When I came up to speed on relational databases, in about 1983, I never heard of Date. And later, I was too busy with practical matters to bother much with Date.But I do like some of the shorter things he's written.

And my favorite is this: "Theory is practical."

I'm more of a practical type myself, rather than a theoretician. And I have little use for dogma. If my karma and your dogma get into a fight, my karma is going to run over your dogma.

But theory IS, by golly, practical. I was designing indexed files at the time I was introduced to 1NF, 2NF, and 3NF. Some people have said that these rules are just "formalized common sense". There's some truth in that. But it HELPS to have some formalization of it. Once I learned the rules, I went back and looked at the normalization of my own files, and those of the systems that were feeding me data. And I found that, sure enough, the files that weren't following the rules were exhibiting precisely the anomalies that normalization theory would have predicted.

So the theory IS practical. It allows you a straightforward mechanism to predict access anomalies, or update anomalies.

In places where designers, for better or for worse, have not wanted to follow 3NF, the programming manager usually says, "yeah, but my people are real smart. They can program around those anomalies." And sometimes the programming manager is RIGHT, by golly! But sometimes the programming manager is WRONG, by golly!

I think that someone who only learns how to make a data design conform to 3NF has only learned half of the theory. What you also need to learn is what can happen if you DON'T follow any given normalization rule. Then, at least you know what you are going to have to program around.

Whenever you do the ETL for a data warehouse, you are going to have to program around some problems. Knowing the places where the DW deviates from full normalization helps you know where the pitfalls are. Theory IS practical. Received on Sat Oct 09 2004 - 15:03:19 CEST

Original text of this message