Re: Nearest Common Ancestor Report (XDb1's $1000 Challenge)

> > C. J. Date states "the purpose of such reduction is
> > to avoid redundancy" in his chapter titled "Further Normalization..."
>
> There is no "general form" of normalization.

If one analyzes the central theme/goal of normalization in RM (and that in other data models), it is reasonable to extrapolate that the general goal of normalization is eliminating redundancy, just as C.J. Date did.

> Items like 'john' and 'o' (as in 'john', 'god', 'neo') are values.
> Normalization deals with the logical view of data.
> Normalization is concerned with information or "facts", not storage of values.

The problem with the above is that values are data.

> Non-redundant storage of values is a physical issue.

So how does this allow the provided solution to accommodate two things with the same name or a thing without a name or a thing with multiple names (without incurring NULLs).

> > Redundancy is plainly obvious to me, so we must be working from
> > different definitions of normalization. Redundancy makes a solution
> > less generic and more prone to problems over a border scope.
>
> This makes no sense.
> You're just making this up as you go along.

Please see other posts in this thread for examples of how redundancy can make a solution less generic and more prone to problems when the solution is extended to a boarder scope. Received on Thu May 27 2004 - 19:18:55 CEST