Re: Relational vs. PICK/Object DBMS

From: Dawn M. Wolthuis <>
Date: Thu, 22 Apr 2004 10:31:31 -0500
Message-ID: <c68ogs$4h7$>

"Timothy J. Bruce" <> wrote in message news:mxLhc.1321$
> Ross:
> > For the sake of someone like me that lacks the rigour to readily read
> > your relational set notation, could you possibly restate the problem
> > at hand in English, which tends to be the language of choice of my
> > users (and me!)
> Relational Mathematics requires a degree of precision that is lacking in
> English, hence the `Mathematics' in `Relational Mathematics'. English
> will not suffice.
> > Can you map this to a real world requirement, or is this an obtuse
> > theoretical device that has little relevance to the everyday world of
> > the programmer that you are trying to measure the productivity of, as
> > your relationships would surely be part of the DBMS in the first
> > instance (though become relevant in some environments when trying to
> > optimise an arbitrary query against Rik(A,Aik)
> Are you aware you are currently treading in comp.databases.THEORY?

Let's take a problem statement with more than one way to address it.

Prove that f(x) = x2 is a continuous function on the real numbers

Now take two ways of approaching this -- with epsilon-delta proofs in "real analysis" and using infinitesimals in "non-standard analysis". It is completely legit to ask folks using either approach to "solve this problem" (of proving continuity).

What will not yield good results is if you start the epsilon-delta proof and then ask the non-standard analyst to complete it their way. For example real analyst says to non-standard anaylst "Now, for any epsilon greater than zero, how would you show that there is a delta ..." The non-standard analyst could indicate that the epsilon-delta "game" is one of real analysis, so he won't play and would like the original problem statement instead. [And if this were done in this forum, then perhaps the real analyst would say -- "well, clearly you are an idiot" to the non-standard analyst, thus showing his true intelligence ;-) ]

Additionally, if the problem is one of mathematics, then it is fine to state the problem in mathematical notation. If the mathematics is simply one team's approach to solving the actual problem, such as in the case of a problem in business data processing (yup, I'm old, but at least I didn't say "electronic data processing") and another team uses Latin as a means of specifying the solution, then you need to have a common language at least for the problem statement.

So, Ross was asking for the problem statement in a common language and without a start at a solution (terms such as "base table" are clear indicators of a solution in relational theory).

Make sense? (I'm not talking to Mr. Bruce here bz he filters me out, knock on wood). Cheers! --dawn Received on Thu Apr 22 2004 - 17:31:31 CEST

Original text of this message