Re: Normalization and DBMS

"Leandro Guimarães Faria Corsetti Dutra" <leandro_at_dutra.fastmail.fm> wrote in message news:pan.2004.05.12.15.33.32.950178_at_dutra.fastmail.fm...
> Em Wed, 12 May 2004 09:08:37 -0500, Dawn M. Wolthuis escreveu:
>
> >> See how you want to have it both ways. You want to 'withstand [...]
> >> requirement changes', but also a less powerful, more complex, less
> >> flexible structure.
> >
> > Could you point me to any experiments that would yield such a conclusion
> > of graphs being "less powerful, more complex, less flexible structure"?
>
> What about 50 years of data processing practice?

Do we retrieve the loudest of these voices, the most prolific, the one that has grabbed the most dollars, anecdotes?

> The relational model was created precisely because former
> systems were -- and unfortunately still are -- a failure in power,
> simplicity and flexibility.
>
> Also it wins by Okham's razor: it is the simplest one to use
> (elements, operators) yet imposes no arbitrary restrictions.

Not buyin' it -- I have had software development teams using IMS, a SQL-DBMS, and a PICK database and there is simply no contest -- from what I have seen first-hand I would not put my money in the SQL-DBMS money-sucking machines without some really good reason that I don't have to date.

> >> This is trivial. Even SQL does it right.
> >
> > Not if is isn't specified.
>
> Obvious points' expression should generally be avoided.

Ah, but you see, in the one case the specification is a by-product so you necessarily get the feature of parent-child integrity where in the other you must rely on both the standards of a company on what to specify and how to specify it as well as individuals in that company to carry out the standards. --dawn Received on Wed May 12 2004 - 19:00:32 CEST