Re: Dawn doesn't like 1NF

From: Dawn M. Wolthuis <dwolt_at_tincat-group.comREMOVE>
Date: Sun, 17 Oct 2004 16:45:14 -0500
Message-ID: <ckup5m$5if$1_at_news.netins.net>


"Laconic2" <laconic2_at_comcast.net> wrote in message news:Q9-dnQ4jc7h_y_rcRVn-rg_at_comcast.com...
>
> "Marshall Spight" <mspight_at_dnai.com> wrote in message
> news:0qI9d.331735$mD.311492_at_attbi_s02...
> > "Laconic2" <laconic2_at_comcast.net> wrote in message
> news:CbydnaMpAdDJq_vcRVn-gQ_at_comcast.com...
> > >
> > > If you will recall, Dawn started a discussion in here a few months ago
> with
> > > the (somewhat wry) title of "Date's first great blunder." In that
> article,
> > > the difference between Date's formulation of 1NF and Codd's
formulation
> of
> > > 1NF was outlined pretty clearly. Equally clear, at least to me, was
> that
> > > Dawn's objection to 1NF was based on the requirement that column
values
> be
> > > atomic, and not based on the difference between a bag and a set. The
> > > details of that discussion covered an awful lot of the ground you are
> > > sending me off to Date to "learn".
> >
> > Found it.
>
> Thanks for finding it. After rereading the first 10 posts of that
> discussion, I'm starting to think that I made a mistake, and that it was
> some other thread where the subject of 1NF and atomicity was dealt with.
> Ah, memory.

I'm a bit late reading this thread, but it was a different one in which we discussed atomicity -- but I don't recall which one either. Date doesn't argue from the standpoint of atomicity like others do. The only way that atomicity is a term that is understandable is if we are talking about atomicity with repect to the database functions (a type or value that has no database functions that break it down further being atomic) and if those functions are not extensible. However, when you look at how folks are taught 1NF even today, it gets back to either the term "atomic" or "scalar values" or some such.

While the definitions of "relation" is very clear (at least from the original definitions from mathematics), I would like to see a clear definition of 1NF that has some basis in anything -- either mathematics or experience -- related to databases. I didn't even find one in Date's most recent edition IIRC (but I'll look again). Considering that 2NF & 3NF (which make a lot of sense to me) require that data first be in 1NF, I'd like to have some better rationale on 1NF. --dawn Received on Sun Oct 17 2004 - 23:45:14 CEST

Original text of this message