Re: A Normalization Question

> > One can represent a quantity at the logical level in words ('ten'),
> > hex ('A'), decimal('10.0'), bits('1010'), etc and how these are stored
> > at the hardware level is irrelevant. A human brain represents all of
>
> Then why should anyone care about the physical represantation of
> quantities like a (hair) color, a street name or the like?

At the logical level, one shouldn't/doesn't care about the physical represenation of quantities like a (hair) color, a street name or the like?

> Isn't it irrelevant, if those are stored as one symbol (say chinese
> script) or a string like "brown".

From the logical layer, which is what most db attempt to present, its irrelevant how a symbol or string in stored at the physical layer.

> On the logical level this is irrelevant and so we don't have redundant
> data at the logical level in your example.

Wrong, most dbs (especially RM imlemenations) provide a logical interface and one can represent something redundantly at the logical layer. Below is an example:

T_Person
Name Age
brown 30

T_SoccerMember
Name Age
brown 30

T_ChurchMember
Name Age
brown 30

In the above data, a person (named by the string 'brown') is logically redundant, as is the string 'brown' (RM's defs makes this hard to see without further breaking the string into appropriate attribute values of tuples). Both the person and the string are subject to update anomaly. Received on Sun Jul 18 2004 - 04:27:21 CEST