Re: Sixth normal form

From: Paul Mansour <paul_at_carlislegroup.com>
Date: Thu, 02 Aug 2007 13:42:43 -0700
Message-ID: <1186087363.584316.106350@m37g2000prh.googlegroups.com>

On Aug 2, 6:18 am, Jan Hidders <hidd..._at_gmail.com> wrote:

>Why would the DBMS not allow history to be rewritten? Of course it
>should allow you to specify that in certain cases it cannot, but in
>general I don't think that is a good idea, if only because you want to
>be able to correct mistakes afterwards.

"Allowing history to be rewritten" and "correcting mistakes afterwards" are, I think, two very different concepts, that are commonly confused because most DBMSs use destructive updates.

You can have a DBMS that never lets you "rewrite history", that is change the set of events that lead up the present state of a DB, but that certainly lets you make corrections to past mistakes.

One good reason to not allow destructive updates is that they are a large component of the complexity of the implementation of a RDMS. Another if fraud. Received on Thu Aug 02 2007 - 15:42:43 CDT