Re: So let me get this right: (Was: NFNF vs 1NF ...)

David Cressey wrote:
> "Jan Hidders" <jan.hidders_at_REMOVETHIS.pandora.be> wrote in message
> news:RurPd.10057$xU6.448623_at_phobos.telenet-ops.be...

>>David Cressey wrote:
>>
>>>3.  The elimination process referenced above and outlined later in the paper
>>>proves that this elimination does not reduce the expressive power of the
>>>resulting normalized collection of relations.
>>
>>Strictly speaking that is not exactly true. There is theoretical work
>>that shows that sometimes you cannot flatten relations without losing
>>information unless you cheat by introducing new domain values for
>>encoding the removed sets.

No. My mistake. Your formulation was quite precise but I still managed to read more into it than it actually said.

> Did the 1970 paper assert something that later work proves to be untrue
> (mathematically)?

No. No problem there. Codd was a good mathematician and the mathematics in the paper is trivial anyway. Which is not meant as criticism, in some sense this simplicity is actually one of the points of the paper. But when reading it one might get the impression that the proposed method allows you to flatten any nested relation without any problems.

> Does the theoretical work you refer to require tranformations not
> illustrated in the 1970s paper?
> Is it somethnig else?

It's something else. Codd's transformations are certainly correct but they only show that in *some* cases you can normalize to 1NF without losing information. That's more or less how I interpreted your "reduce expressive power" because otherwise I'm not sure what that term exactly means. The question then becomes if that is possble in *all* cases. You already know the answer. :-)

In case you're interested:

  http://portal.acm.org/citation.cfm?id=832

Drop me a mail if you can't get your hands on it.

• Jan Hidders