Re: Date, Darwen, Pascal and the alternative to Nulls in the RM

falcon wrote:
> I have also been making my way through Date's books. I'm afraid some
> basic concepts still elude me. Any way, I have been also trying to
> read some logic books to get a deeper feel for RM. Isn't following a
> solution to the null problem:
>
> Consider a relation (relvar?):
> [social security number, name, age, grade]
> (where soc number is the primary key).
> Isn't this exactly the same (in logical terms) as:
> [soc number, name]
> [soc number, age]
> [soc number, grade]
> The primary key can't be a null. name, age and grade may be safely
> 'null' (they simply don't exist) in the second representation. The
> user may still see the first representation (perhaps views?), and if
> he/she enters a null for one of the non-primary-key fields, the actual
> logical model simply doesn't 'insert' the null value into its relevant
> table (eh...relvar).
>
> I've avoided using the term normalization because I'm even less sure of
> its meaning then other stuff I've written.
>
> Does this make any sense?

To me it does.

But ... to check wether it makes the sense to me you want it to mean: Non-existence of [soc number, age], would effectively be the the same as having [social security number<some number>, name<some name>, NULL, grade <some grade>].

Let's expose a humble cosmology:
1. All comprehension is a tiny part of what is real plus some stuff which isn't.
2. All language expressiveness doesn't give you the power to express comprehension.

It doesn't make sense to store [soc number, age]. You know it, I know it. Why doesn't it? Received on Thu Mar 23 2006 - 00:45:21 CET