Re: In an RDBMS, what does "Data" mean?

Mikito Harakiri wrote:

>>Well in our database world we take tuples such as
>><name='Hrundi V. Bakshi', age=35> as self-evident truths.

That which we start with, that we take for granted, that aren't derived from anything else - our axioms really. Of course we know as viewers out side the system that these statements are motivated by things in the real world, but the database itself doesn't know that.

> OK, another important feature is generality, for example:
>
> for any x and y: x*y=y*x
>
> Constraints are general, database tuples are not.

OK so should we include the database constraints in the axioms? Well at one level I'd say not, because I'm considering a fixed snapshot of a database where all the tuples automatically satisfy the constraints. Of course you can also view a database as changing with time, when the constraints are necessary. Maybe that's the reason for the disagreement, I'm just modelling a simple case. I'm considering every single database instance as a separate system with its own axioms.

Are you thinking of a single axiom system for all databases in general? Or for one particular database as it evolves over time?

Are you saying that axioms should be general? i.e. include a "for any"? In normal axiomatic systems this is the case, but databases are slightly different because they work by explicitly listing a finite number of facts.

Maybe the point I'm making is so trivial and obvious that you're thinking I'm claiming more than I am?

> But fantasy world certainly don't obey laws of logic!

They don't have to, but they could do. I'm just thinking of the ones that do obey the laws of logic.

Paul. Received on Sat Jun 05 2004 - 13:07:01 CEST