Re: How to find Brothers and Sisters?

From: Cimode <cimode_at_hotmail.com>
Date: 4 Dec 2006 01:11:25 -0800
Message-ID: <1165223484.906729.52620_at_j72g2000cwa.googlegroups.com>


Lennart a écrit :

> Cimode wrote:
> > Lennart a écrit :
> > >drop table persons;
> > >create table persons (
> > > name char(10) not null primary key,
> > > father char(10) references persons on delete restrict,
> > > gender char(1) not null check (gender in ('M','F'))
> > > );
> > While getting demanded results, the above schema is not 1NF.
>
> Really, in what way do you think the schema violate 1NF?
A RTable is a set of facts, true propositions. All predicates must be validated by TRUE and represent the SAME segment of reality. In the design proposed, it is highly likely that the highest level proposition (also known as root) have the root node either be NULL or point to itself as opposed to the non root node elements. Therefore it is likely to be unormalized design. In a word, there's a 1:0 cardinality between parent and sibling. A parent may or may NOT have sibling. A sibling may or may NOT have a parent (the root does not have a parent by definitions)...

Regards..;)
> /Lennart
Received on Mon Dec 04 2006 - 10:11:25 CET

Original text of this message