Re: The Fact of relational algebra (was Re: Clean Object Class Design -- What is it?)

From: Jan Hidders <hidders_at_REMOVE.THIS.win.tue.nl>
Date: 3 Oct 2001 17:53:11 GMT
Message-ID: <9pfja7$n5s$1_at_news.tue.nl>


Daniel Poon wrote:
>
> Ive often heard that rdbms are based on relational algebra, a branch
> of mathematics.

What do you mean by "based on"? The relational algebra plays a role in query optimization, but that is about it. It helped the introduction of the relational model because there are some nice links with logic and model theory, but I wouldn't call it fundamental and by itself it is certainly not a branch of mathematics.

> When I studdied mathematics, a long time ago I admit, I never
> encountered relational algebra. The nearest thing I can hazily
> remember are 'relations', which are what you guys would call binary
> tuples.

The relational model is based upon a generalization of that: the n-ary tuples.

> The algebra we used was quite different than the one put about by the
> rdbms guys.

The term algebra can refer to all kinds of mathematics that deal with expressions and rules to rewrite them to other equivalent expressions.

> So where can I learn about 'relational algegra'. Does it really exist?

As a theoretical concept it exists and is important, even for practical issues such as query optimization. You can find the relevant literature in the books by Date (Introduct. to Database Systems), or Ullman (Principles of Database and Knowledge-Based Systems) and if you really feel like it (Foundations of Databases) by Abiteboul, Hull and Vianu.

> When applying relational algebra to computing, is it a fair
> assumption to say that it is a value based system? Mathematics always
> seemed more like an identity based system to me.

No, definitely not. The relational model is value based, as opposed to object-oriented data models which are usually more identity based. Since mathematics by itself is not really a data model it doesn't really make sense to call it identity based or value based unless you give first a definition of the term that could be meaningful here.

-- 
  Jan Hidders
Received on Wed Oct 03 2001 - 19:53:11 CEST

Original text of this message