Re: What does this NULL mean?

From: dawn <dawnwolthuis_at_gmail.com>
Date: 10 Dec 2005 19:28:48 -0800
Message-ID: <1134271728.159733.37470_at_g43g2000cwa.googlegroups.com>


paul c wrote:
> dawn wrote:
> > Eric Junkermann wrote:
> >
> >>...
> >>
> >>The best way to look at a NULL is as a sort of denormalisation. If we
> >>have a table X {A, B, C} where A is the key, B is a column we are not
> >>currently interested in, and C is the column which is NULL in at least
> >>one row, then we are really talking about two tables X1 {A, B} and Y {A,
> >>C}, where at least one row in X1 does not have a corresponding row in Y.
> >
> >
> > Within relational theory, I think you are exactly right that it is a
> > normaliztion issue and there is no place for nulls in a fully
> > normalized model. That is the position I have taken within the context
> > of relational theory. Outside of the RM (where I prefer to live), I
> > see it differently.
> >
> >
>
>
> So nulls are okay in a not-"fully normalized' relation, eg. one in 2NF?

By most definitions it isn't even in 1NF, right? I incorrectly used "fully" to refer to the breadth of the normalization across a domain rather than the depth.

If the dbms tools permit implementations that are not normalized then someone will make a call at some point that this "feature" is worth exploiting. There should be an interpretation for every possible implementation, I would think. --dawn Received on Sun Dec 11 2005 - 04:28:48 CET

Original text of this message