Re: 4NF is Where It Is At! [WAS Re: 1:1 relationships]

On 14 Feb 2001 10:13:49 GMT, hidders_at_REMOVE.THIS.win.tue.nl (Jan Hidders) wrote:

>What often works is looking at what happens if you denormalize.
>Suppose, for instance, you have the following normalized tables:
>
> Is_member_of(student, class) key: {student}
> Follows(class, course) key: {class, course}
>
> . . . snip . . .
>
> So, I hope this already helps you understand
>MVDs a little better and also when and why they should be split out of
>the table.
>
>--
> Jan Hidders

Thank you for the example, Jan. I have now worked through this and Date's Chapter 11 and the light has dawned about MVDs.

I think my problem is summed up by a note which Date made:

      "In practice it is usual - by separating independent repeating
       groups - to eliminate such MVDs _before_ applying [the
       normalisation rules] . . ." 

I have always done this under the impression that it was part of _1NF_ and that MVDs were something much more mysterious and esoteric. So having got to 3NF or BCNF my relations were already in 4NF without my realising it. :-)

Thank you very much for your help.

Now for 5NF! <g>

Regards

Peter Franklin Received on Wed Feb 14 2001 - 10:27:41 CST