Re: UNIQUE and NULL in SQL

David Cressey <david_at_dcressey.com> wrote in message news:XH0Y7.9$Nq6.589_at_petpeeve.ziplink.net...

> Where I would take issue with you is when you call "Not Applicable" a
> "value". It isn't a value, IMHO. It's a conclusion drawn from the fact
> that a value is not present at a given intersection of a row and a column.

Maybe in some cases. But other 'values' that could also be valid in place of a value not present at a given intersection of a row and a column (what is currently implemented by many designers) could also include 'Not Known', 'Not Given', 'Not Confirmed', 'Irrevalent', etc. Any one of these classifications, in my opinion, are values if they are considered part of the domain of any particular attribute as defined by the data administrator or user. They are values in my opinion because they represent true facts, and are therefore reflected in a manner consistent with something like: The person identified by NAME 'Jan' is of a gender type that is 'NOT KNOWN'. A case can be made for all the various variations that are often confused with the NULL.

Why are they facts? Because they meet the constraints of a domain defined for a relation attribute and they are true predicates as reflected in the body of a relation variable.

> And I would maintain that NULL is never a value, regardless of the
> conclusion drawn from it. And I would further maintain that this is not
a
> matter of opinion, but a matter of definition.

I'll agree.
>
> --
> Regards,
> David Cressey
> www.dcressey.com
> "Mark Preston" <mark_at_mpreston.demon.co.uk> wrote in message
> news:s4313usa6t9oof1m0tv36tqusv7sq2qiqv_at_4ax.com...
>
>
Received on Tue Jan 01 2002 - 00:33:12 CET