Re: Few confusing things about first normal form

From: Roy Hann <specially_at_processed.almost.meat>
Date: Thu, 23 Oct 2008 04:56:51 -0500
Message-ID: <>

paul c wrote:

> wrote:
>> greetings
>> 1) For DB to be in 1NF there must be no multi-valued attributes, and
>> no repeating groups. When so, data is said to be atomic. ...
> When Codd first used the word "atomic", he may have intended it very
> casually, as some of his intended audience were decision-makers, but
> (just as they often are today) many of those were non-technical people.

I think we can be certain he did intend it casually because he never went on to make any argument based on what it (might) mean. So in effect his failure to define it is no more significant than Euclid failing to define or comment on the concept of colour.

Furthermore we know that relational algebra doesn't provide tools to discern the internal structure of any value. If we were going to explain that to someone we'd probably also say values are atomic. It's an excellent word for the idea.

The usual misunderstanding seems to be to think that RT and SQL tell us we have to do something to our designs to *make* our values atomic. But that is backwards. We can use any kind of value we want, including the classic list of pizza toppings, but RT (if not SQL) will only ever be able to treat it as an atom. (Having said that, I am still not at all happy with RVAs! :-)

Received on Thu Oct 23 2008 - 11:56:51 CEST

Original text of this message