Re: Why relational algebra is algebra

In article <9iku1q$qj0$1_at_news.tue.nl>, Jan Hidders says...
>
>Mikito Harakiri wrote:
>> I'm looking into the 'Restrict' operation in Relational algebra
>> (called 'Selection' at
>> http://www.cis.ohio-state.edu/~gurari/course/cis670/cis670Ch4.html)
>> and don't understand why it is an algebra. 'Restrict' definition contains a
>> propositional formula inside, so that calculus crept into the back door!
>
>Well you *can* define the restrication that way but you don't have to.
>If you only allow the propositions 'column1 = column2' and 'column =
>constant' your algebra will have the same expressive power.

Somehow, the fact that this part of the 'Restrict' definition is ambiguous disturbed me. Now I see, better definition must allow only atomic predicates, that are part of the domain spec anyway; not formulas.

>> Therefore, relational agebra doesn't seem to be self-contained: it can't
>> possibly be axiomatized without bringing in propositional logic at least.
>
>What is the purpose of your axiomatization? How complete do you want
>your axiomatization to be?
>

I shouldn't bring up "axiomatization": I remembering that just several months ago you initiated the thread under that name. I have no idea what was I talking about:-( Received on Sun Jul 22 2001 - 01:35:00 CEST