Re: So what's null then if it's not nothing?

Alexandr Savinov wrote:
> > Not at all the same thing because nulls violate the Information
> > Principle, which is a foundation of RM. Nulls as formulated in SQL also
> > create logical problems that are very hard to solve - at least SQL
> > hasn't succeeded in solving them very satisfactorily.
>
> Maybe, so what do you propose?

In place of SQL? The Relational Model would be nice to have one day... :-)

> > RM does not prohibit such questions. If it is more natural to give some
> > answer then the user can demand such an answer. The user can decide
> > what he wants to see in the case of a dog with colour or a dog without
> > colour. I would challenge your implication that users find nulls"more
> > natural". I bet most DBMS users wouldn't do a very good job of
> > explaining what a null is. In fact in many database projects the
> > developers go to a lot of trouble to hide nulls from regular users.
>
> User and developers are not part of a data model.
>

Neither are the questions. Your point was that certain questions were forbidden in the absence of nulls. Mine was that users decide on the questions and that null isn't an answer that they want or need.

> and the same because effectively it will not be visible. Why do we need
> to mark things/facts as absent by means of nulls if we can simply delete
> them physically? Because things exist in multidimensional hierarchical
> space so they exhibit themselves in a more complex manner than
> exist/non-exist. And of course it is differs from what RM teaches.
>

Can you give an example of how nulls could exhibit information information that is somehow "more complex" than can be represented without them? This is somewhat more important than your earlier assertions that nulls are merely "convenient". If you think nulls are _necessary_ to represent certain types of information then you ought to be able to demonstrate that by example. It seems plain to me that no such example can exist (unless you contrive some data model that is very stunted in the types of information it can represent - certainly not RM which represents multidimensional information very well).

Like JOG I thought the debate was over. We tolerate nulls, we don't like them.

-- 
David Portas 
SQL Server MVP 
--