Re: foundations of relational theory?
Date: Wed, 22 Oct 2003 13:09:29 +0100
Message-ID: <bn5s7p$1grg$1_at_gazette.almaden.ibm.com>
"Anthony W. Youngman" <thewolery_at_nospam.demon.co.uk> wrote in message
news:BhNnjiArEal$EwOo_at_thewolery.demon.co.uk...
> In article <bn33eu$12j0$1_at_gazette.almaden.ibm.com>, Paul Vernon
> <paul.vernon_at_ukk.ibmm.comm> writes
> >"Dawn M. Wolthuis" <dwolt_at_iserv.net> wrote in message
> >news:6db906b2.0310202027.58324c36_at_posting.google.com...
> >>
> >> I'm likely just getting old, Paul, but I'm not tracking with you on
> >> this question. What about my statement gives you a hint that if
> >> everything else is equal then an MV system will be less useful and
> >> just what is the everything else? You mean if someone has the same
> >> number of dollars to sink into an MV syste as an RDBMS system, then
> >> the MV system will be less useful? Why?
> >
> >Because a MV system is more complex (contains more constructs), for no
extra
> >power than a relational system.
>
> Except that we need far FEWER tables than you do.
> That means that MV is
> more SIMPLE than a relational system. How many tables do you need to
> model an invoice?
Wrong question. The question is, how many *concepts* do you need to model an invoice?
> In a well designed MV system we will have just *one*
> FILE, that maps EXACTLY to your invoice view.
But a FILE is more than just a set of tuples (aka a table). It is a more complex concept.
> All your integrity enforcement (cascading delete, can't save order
> detail without matching order, etc etc etc) just comes as "part of the
> package" with MV.
> By the way, there is *scientific* *proof* that you're wrong in assuming
> that a dollar spent on MV buys less than a dollar spent on relational.
It that buys less in 'the future' or 'on average weighted over all possible futures'
> Admittedly there is precious little proof, but what little there is came
> to the conclusion that a dollar spent on MV bought the same as two
> dollars spent on relational!
So marketing surveys are *scientific* *proof*, hey. Wow.
[snip]
> And nor do we. The MV model (properly used) treats all entities as
> equal. What it does NOT do is treat
> attributes as equal to the entity they describe (which relational does).
?
> Relational also treats relations as equal to their entities ...
??
[snip]
> Well, from reading Fabian (and I'd include others in this too), I came
> to the conclusion that he is incredibly blinkered and bigoted, he has a
> closed mind, and if he can't understand what you are saying he won't try
> to work out where you're wrong - he'll simply slag you off as deluded.
>
> That would be okay if he's right,
> but what's the point of trying to
> understand him if he can't be bothered to understand you?
Concentrating on 'educating' the more prolific (I'm not sure I should say 'more respected'), such as in this exchange http://www.dbdebunk.com/page/page/766119.htm, is a better bet in my book. However, I would agree that his style of persuasion is not to everyone's tastes...
> In those
> circumstances the only option open is to assume that he must be wrong.
> Since he has no interest in identifying the flaws in your argument (as
> opposed to you), then the only tenable conclusion is there are no flaws.
> Ergo, it's him that's wrong, not you.
If your arguments we novel and well constructed, then he might have more interest .
Regards
Paul Vernon
Business Intelligence, IBM Global Services
Received on Wed Oct 22 2003 - 14:09:29 CEST