Re: Date's First Great Blunder
Date: 21 Apr 2004 14:37:23 -0700
> > If NULLs are not necessary and increase complexity too much, why doesn't
> > Date drop the chapter on NULL/Missing Information?
> From which book?
An esoteric series of books titled "An Introduction to Database Systems". It is Chapter 20 in his 6th Edition.
Even if one builds a RDM db without an SQL layer, the db will still generate NULLs in certain circumstances. A tuple's attributes demands values specified by the relation's header. But reality says I don't care what your relation header demands, I can't give you that value. Reality can say this for almost any attribute of a tuple. In such a case, RDM can 1) refuse to add the tuple 2) add the tuple but create a NULL for the "missing info" 3) change the schema and seperate the column to a new table to avoid the NULL (which can become impractical in many cases).
True, NULL is what RDM creates when reality can't/won't provide a value demanded a tuple's attribute. (Reality generally wins this argument).
> He does deal with missing information quite nicely, with the proposal to
> include special values into each domain needing them.
> Looks like you never understood the special values proposal.
Special values are frequently masks for NULLs and still lead to 3VL. If john's age is "UNAVAILABLE" and mary's age is "UNAVAILABLE". Is john.age equal to or not equal to mary.age? (Unfortunately, understanding special values won't ever clear up this problem)
> > Per Date, Intro to Db Sys, "3VL suffers from the very serious
> > ('showstopper') problem that it does not match reality - that is, results
> > that are correct according to 3VL are sometimes incorrect in the real
> > world. NULLs and 3VL undermine the entire foundation of the relational
> > model."
> do think he's mixing things here.
Date is correct on this topic. Received on Wed Apr 21 2004 - 23:37:23 CEST