Re: Normalization, Natural Keys, Surrogate Keys

From: Kai Ponte <>
Date: 29 May 2002 09:24:20 -0700
Message-ID: <>

Bernard Peek <> wrote in message news:<>...
> In message <
> >, Costin Cozianu <> writes
> >For example, in the above, because you allow for that 999999999 for
> >situations that are very rare (let's say <1% or <5% of the cases), you
> >fail to protect the vast majority of cases, because you won't be able
> >to declarer the UNIQUE constraint on the SSN. There are better way to
> >deal with this, one example that I suggested is create an
> >SSN_Exceptions table where you put the data collected for the persons
> >who don't have a SSN or their SSN has problems.
> There's an alternative solution, add an extra integer field to the main
> table with a default value of 0. Make the SSN and the new field a
> composite key. When you have an unknown or invalid (or duplicate) SSN
> insert a sequential number into the integer field. Any record with a
> non-zero value should be considered an ongoing problem.
> If you want you can create a view onto that table that will perform the
> same functions as the SSN_Exceptions table.

LOL! I read this one after I posted my reply. I think this thread is getting too long.


KP Received on Wed May 29 2002 - 18:24:20 CEST

Original text of this message