Re: header part of the value?

From: Tegiri Nenashi <TegiriNenashi_at_gmail.com>
Date: Wed, 27 Feb 2008 19:32:20 -0800 (PST)

On Feb 27, 7:08 pm, Marshall <marshall.spi..._at_gmail.com> wrote:
> It appears to me as if you are viewing the
> equation "x + 3 = y" as having 3 as the free variable, and
> x and y as constants! Whereas I would consider it much
> more natural to consider x and y as free variables, or
> parameters, or attribute names (they are all the same)
> and 3 and 5 as constants.

Well, the case with two variables in the header is tricky. First, define union and join then we can talk about it:-)

One more odd thing about the example is that an interesting mathematical case is a system of equations which is conjunction, and not disjunction!

> If I may select and project from your example, consider:
>
>   x>=
>   ---
>   1
>   2
>   5

This is invalid relation. Certainly, you have meant

x>=

```---
1

> That to me seems to be simply a special case, that of the natural join
> of the following two relations (specified via comprehension)
>
>   {(x, y) | x >= y}
> and
>   {(y) | y in {1, 2, 5}}

Now that relation collapsed to a single tuple can you please rework
the example ? For example, you can add one more attribute

x>=  z<=
---  ---
1    3
2    4
5    8

so that your original relation is a projection of that one. Once again
I find it fascinating that the min aggregate of the column x in the
relation

x    z
---  ---
1    3
2    4
5    8

(which is equal to 1) is just a projection of generalized relation

x>=  z<=
---  ---
1    3
2    4
5    8

onto the header "x>=". (The exact signature of the column z doesn't
matter).
```
Received on Wed Feb 27 2008 - 21:32:20 CST

Original text of this message