Re: Newbie question about db normalization theory: redundant keys OK?
Date: Tue, 18 Dec 2007 22:03:53 -0000
"Brian Selzer" <brian_at_selzer-software.com> wrote in message
> Isn't it true that with FOL, there must be separate interpretations for
> tuple A and tuple B, since they belong to different database values,
> whereas with modal or temporal logic, the same interpretation can apply to
I don't know enough about modal logic to comment on that part. I don't see how you conclude that for FOL there MUST be separate interpretations just because the values have different valid times. It seems like there are temporal database models and even logic programming that don't require anything like modal logic to produce useful results.
> That assumes that relational assignment is primitive. I would argue that
> insert, update and delete are the primitive operations, and that
> assignment is a shorthand for a combination of the primitives delete and
> insert. Information is lost when an update is translated into an
> assignment, but not so the reverse: an assignment can always be translated
> into a delete and an insert.
I think it is self-evident that assignment can preserve exactly as much information as the user requires it to. If the information in question can be represented as values within relations then, ipso facto, a relational assignment is sufficient. Any form of DELETE or INSERT operator therefore doesn't add any expressive power. At best it would be a syntax shortcut of some kind.
-- David PortasReceived on Tue Dec 18 2007 - 23:03:53 CET