RE: normalization

Quoting Steve Adams <steve.adams_at_ixora.com.au>:

> You are right.
> If it is a logical data model, then fight for 3NF, if not BCNF.
> But in physical database designs, denormalization should be normative
> if it will improve application efficiency and
> if it will not hinder development and maintenance.
> Just my opinion, of course.
>
> -----Original Message-----
> From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On
> Behalf Of Chris Stephens
>
> There is a discussion going on at work concerning calculated fields.
>
> I am claiming that any calculated field in a table is a violation of
> at least 3NF if not 2NF. I can find all sorts of references on the
> web that justify my position but nothing that directly says this
> violates normalization rules.
>
> The person who i disagree with is claiming that 'technically',
> calculated fields do not violate 3NF. They are just not recommended.
> I am unable to find anything on the web coinciding with this argument.
>
> Anyone know of a site with a direct statement that calcualted fields
> violate 2NF/3NF?

As far as I can recall calculated fields violate 3rd normal form as they are transitive values. However, like Steve said: if it helps speed up access with physical design, then it is probably fine to use them. I'd look at using MVs for that sort of functionality, though.

-- 
Cheers
Nuno Souto
from sunny Sydney
--
http://www.freelists.org/webpage/oracle-l