Re: Joins with nulls

David Cressey wrote:

<snip>

> Nulls in databases result from outer joins.

< snip >

Here we come to the crux of the problem.

Consider the example of a relvar for proper names.

How do you create a relvar for proper names. (To clarify, my proper name is John Eggert). Do we create two base relvars?: First Name; Last Name, or a single relvar, say: Proper Name. If we go with option 1 ( two base relvars), all base relvars must consist of (at most) two columns. In his case: First Name Index and First Name, etc.. This is the option that eliminates nulls (and duplicates). (This can be proved by induction). This is obviously absurd, hence by reducto ad absurdum, the theorum (no nulls in a relation) is rejected. This isn't rocket science. Why is it such a matter of contention?

Another way (less vigourous) of saying this is: Nulls in a database are inherent in the underlying data and ARE NOT a result of joins. As such, any system that cannot tolerate nulls will eventually come crashing against the solid wall of reality.

JE Received on Sat Nov 23 2002 - 05:40:29 CET