Re: Multiple specification of constraints
Date: Mon, 1 Mar 2004 23:07:22 -0500
Message-ID: <lYCdnTPOSZDplNndRVn-hA_at_golden.net>
"ben brugman" <ben_at_niethier.nl> wrote in message
news:4042f8ce$0$268$4d4ebb8e_at_read.news.nl.uu.net...
>
> > > Then there are constraints which are more difficult or impossible
> > > to implement in the RDBMS, and although the constraints should
> > > be 'kept' centralised.
> >
> > No such constraint exists. All you have to do to disprove my contention
is
> > demonstrate a single constraint in a well designed schema that
contradicts
> > my assertion.
> >
> 1. (Data outside the one database).
i.e. One made an arbitrary decision not to manage the data. This does has nothing to do with difficulty or impossibility.
> 2. (Security is kept outside the database.)
i.e. One made an arbitrary decision not to manage the data's security.
> (The encryption and the codes are often kept out
> of the server system).
I suggest you switch to a secure server. Communicating codes from another source only adds an opportunity for interception.
> 3. (Situation is to complex (performance) to be resolved inside the
> database).
If it is too complex to resolve within the dbms, it is necessarily too complex to resolve anywhere else.
> 4.(Snapshot isolation does not prevent other transactions to alter read
> rows).
Poor choice of dbms does not relate to difficulty or impossibility.
> Back to the argument, I think that in a lot of environments the
> communication
> between the database and the calling code the error handling and exception
> mechanism could be improved.
Or removed entirely.
> If no checking would be done by the calling code and all checking would be
> done by the database.
A database is a set of facts. It does nothing on its own. A dbms does things. A file processor does things. An application does things. A database just is. Received on Tue Mar 02 2004 - 05:07:22 CET