Date: Tue, 4 May 2004 11:25:39 -0500
From: Bruno Wolff III <bruno@wolff.to>
To: Svenne Krap <svenne@krap.dk>
Cc: Alain Reymond <alain.reymond@ceia.com>, pgsql-sql@postgresql.org
Subject: Re: Database structure
Message-ID: <20040504162539.GA10698@wolff.to>
Mail-Followup-To: Svenne Krap <svenne@krap.dk>,
	Alain Reymond <alain.reymond@ceia.com>, pgsql-sql@postgresql.org
References: <4092877B.2470.1E894F5@localhost> <4097C065.7050004@krap.dk>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <4097C065.7050004@krap.dk>
User-Agent: Mutt/1.5.6i

On Tue, May 04, 2004 at 18:10:13 +0200,
  Svenne Krap <svenne@krap.dk> wrote:
> 
> In the long run, being correct is usually better than being fast (at the 
> point of the implementation), as new hardware easily solves bottlenecks 
> for problems not scaling exponentially.

And it isn't even clear that denormalizing the schema will result in an increase
in speed.

If at some point the tests in various assessments can overlap you may not
want an assessment for each table.

I also noticed that the schema isn't enforcing consistancy between the
tests done and the assessment type being done. This may not really be a
business rule as much as something that might be flagged by the application
for attention as I can see cases where in reality the wrong test is done
and recording its results might be better than throwing the data away.