Re: A Normalization Question

Neo wrote:
>
> The central issue is not why we normalize, but whether the string
> "brown" (not its role as a unsystematic reference) is redundant.

The two are connected. Once you understand why we normalize you will understand why the notion of redundancy is defined in database theory as   it is and why multiple occurrences of the same string in a data structure are not necessarily redundant.

> As you are alluding, one way to know if something is redundant is to
> change one of its attributes and check for corruption. This is
> difficult to see with strings, because strings typically don't change
> and if they change we consider them to be a completely different
> string.

Actually, that is very simple to see with strings. You replace the occurrence of the string with a variable X. If the value of X can be derived from the integrity constraints then this occurrence of the string is redundant. If you cannot then it isn't.

> [...] Suppose, the world is taken oven by islamic terrorist. As part
> of their spoils, they want every word in every computer to be spelled
> backwards, thus 'brown' needs to be update to 'nworb'. [...] In the
> above tuple, updating one and not the others, corrupts the db.

Yes, but only when you start from an instance that is not already an instance from the new database schema is concerned. To show that either instances of the new or the old database schema have redundancy you have to demonstrate that there are update anomalies that stay within that particular database schema.

• Jan Hidders