Re: Dreaming About Redesigning SQL

"Marshall Spight" <mspight_at_dnai.com> wrote in message news:nkEmb.29665$Tr4.57087_at_attbi_s03...
> Well, there's a weird thing going on: order encodes information.
> Now, us relationalists (or whatever you call 'em) explicitly say
> we don't care about that, and we gain some optimization benefits
> and some flexibility by doing so. A significant amount, in fact,
> so that it's easy enough to say "it's worth it."

It's fair to say that relationists don't quite understand what the order is and how to treat it mathematically. The confusion goes back into math definition of ordered pair. Kuratovski's definition of ordered pair

(a,b) = {{a},{a,b}}

is very unsatisfactory.

> Because of my exposure to the relational model, I'm used to
> thinking of data as being exclusively expressed as values or
> sets. But there's that nagging feeling: there's a whole different
> *kind* of information besides sets.

Certainly, you may even say that intricacies of the set theory are irrelevant to practical mathematition (sounds familiar, huh?). It is often claimed that it's the connection of algebra and geometry that is the most exciting in math.

> Put another way: a set contains less information than an
> ordered set with the same elements. That just strikes me
> as bizarre, and although I think I understand the situation,
> I certainly don't understand all the implications of it.
> This different kind of information is truly weird, in
> that it cannot exist without a set to piggyback on.
> An ordered empty set contains the same amount
> of information as the regular empty set. Hmmm,
> well, maybe it doesn't. See how weird?

I'm not sure if I share your feeling, but relational theory certainly has some "uncompleted" flavor.

From practical perspective, I find it odd that one can't describe GUI relationally. This is why we still use old procedural languages on client side, right? Received on Sun Oct 26 2003 - 19:55:09 CET