Re: Codd and non-simple domains

devajyoti.sarkar_at_gmail.com wrote:
> ...Codd:
> "...> (1) The graph of interrelationships of the nonsimple
> domains is a collection of trees.
> ..."
> My question is is this true for modern RDBs? Essentially, if the graph
> is a tree,

/collection/ of trees,

> then there would always be a single path between two table
> in terms of key inclusion in the primary key of child tables.

So no, there may very well be multiple paths, and hence, in practise, a choice to determine the primary key out of the candidate keys.

> Is this safe to assume to be true for any
> set of tables in 3NF or even 1NF?

So no. Received on Wed Aug 31 2005 - 19:35:42 CEST