Re: Domains, types, and application engineering ....

From: Bernard Peek <bap_at_shrdlu.com>
Date: Tue, 9 Jan 2007 19:21:14 +0000
Message-ID: <hPHJiMHqs+oFFwKo_at_delta.shrdlu.com>


In message <1168368757.898108.24190_at_k58g2000hse.googlegroups.com>, DBMS_Plumber <paul_geoffrey_brown_at_yahoo.com> writes
>Adding a note (and yes, I am still puzzling about the definition of
>Plus, so relax!) ...
>
>To drag this whole discussion out into the machine shop, pin it down
>with a set of vice clamps, and go at it hard with the drill press and
>arc-welding gear, the POINT of a relational DBMS is to accurately model
>some real world problem domain.
>
>So what "ought" the behavior of the domains in your relation schema be?
>Precisely the interesting behavior of the real world phenomenon we're
>modeling. Not something some standard dreamed up (though that's
>probably going to be useful in a lot of cases, and re-use will
>doubtless improve your the quality of your information system and the
>productivity of your engineers).
>
>If your application calls for you to organize instances of a concept
>like '10 mandarins', then you need first, to document and nail down how
>these things behave, and then second, implement software to encapsulate
>(*gasp*) that behavior.
>
>The function of the DBMS is to take this implementation and to present
>you with a body of relational machinery that can be used to organize
>and reason about instances of this domain, according to the behavior
>you've defined for it.

The relational model locates a datum on an n-dimensional grid. Each table has two dimensions - domain and key-value.

-- 
Bernard Peek
back in search of cognoscenti
Received on Tue Jan 09 2007 - 20:21:14 CET

Original text of this message