Re: Principle of Orthogonal Design
Date: Thu, 7 Feb 2008 10:02:03 -0800 (PST)
Message-ID: <89350c30-de5f-4e85-89b2-8162bc4d2225_at_y5g2000hsf.googlegroups.com>
On 7 feb, 18:26, Tegiri Nenashi <TegiriNena..._at_gmail.com> wrote:
> On Feb 7, 3:21 am, "Brian Selzer" <br..._at_selzer-software.com> wrote:
>
> > A join is not a constraint, is it
>
> There is little controversy what proper constraint definition is. In
> math we write
>
> x^2 + y^2 = 1
>
> in RA we write an equation with relational operators. Now, equality
> constraints are not general enough, so you need inequality as well,
> and there is indeed such a thing.
What makes you think adding inequalities increases expressive power? Can you prove this? And what if I have Table Dee and Table Dum in my algebra?
> Now, the thesis:
>
> Every constraint can be expressed as a system of relational
> inequalities.
That depends on your definition of constraint. If you restrict them to those expressible in first order logic, then this seems correct, but otherwise I don't think so.
- Jan Hidders