Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!uakari.primate.wisc.edu!zaphod.mps.ohio-state.edu!unix.cis.pitt.edu!dsinc!netnews.upenn.edu!grad2.cis.upenn.edu!aaron
From: aaron@grad2.cis.upenn.edu (Aaron Watters)
Newsgroups: comp.databases
Subject: Re: Entity Models and BCNF
Message-ID: <34942@netnews.upenn.edu>
Date: 17 Dec 90 15:10:21 GMT
References: <9129@cognos.UUCP>
Sender: news@netnews.upenn.edu
Reply-To: aaron@grad1.cis.upenn.edu (Aaron Watters)
Organization: University of Pennsylvania
Lines: 25

In article <9129@cognos.UUCP> nigelc@cognos.UUCP (Nigel Campbell) writes:
on behalf of colim@cognos.UUCP or try 
             3132050@mcimail.com which may reach him faster 
>Using Normalization to check your ERM is of course redundant, 
>but only if your ERM is perfect!
>
>In practice people are rarely perfect and shifting to another paradigm to look
>at the same problem another way is a standard way of getting the Human
>mind to stop overlooking things.

Okay, please give me an example where the paradigm shift actually
helps.  Someone previously mentioned something about removing
a foriegn key from an entity -- which seemed to me to be a silly
mistake in the first place -- the modeller should have asked
`Is this thing really an attribute of this entity?' to get the
right answer without worrying about dependancies at all.

Let me make a strong rhetorical claim:
	Any normalization that is NOT justified by an entity
	relationship (or similar) analysis is suspect and probably
	invalid -- ie, some user in the future will probably
	want to violate it.
comments?  (Keep in mind here that I'm talking about logical
modelling -- not performance tuning or anything else.)
		-aaron.