Re: Order details table reference live data

From: Bob Badour <bbadour_at_pei.sympatico.ca>
Date: Fri, 01 Sep 2006 21:07:02 GMT
Message-ID: <Wh1Kg.7144$9u.93552_at_ursa-nb00s0.nbnet.nb.ca>


Johan Sjöström wrote:
> The majority of the users of the system knows about the design and
> doesn't change data that affects orders. That's why management hasn't
> ordered a redesign -- it's expensive, and the users don't complain that
> much...yet. But they're beginning to.
>
> As I see it, one can either
>
> 1) Accept that a complete overhaul must be made and rewrite the Order
> tables so that they contain raw (no reference) data copied from e.g.
> the Product table as it was at the time of the order, and then rewrite
> all order queries.
>
> or
>
> 2) Something else. (Brainstorming appreciated). I'm having a hard time
> figuring out a good alternate solution. I guess there is none. ;)
>
> Cheers
> Johan

I can imagine all kinds of alternatives.

As a general rule, the earlier the error is introduced and the later the error is fixed, the higher the cost to fix it. You are facing an error made in the initial requirements and allowed to fester in production for some unspecified length of time > 0. This means the error propagated from the requirements into the analysis into the design into the implementation and into your company's operating procedures.

The only open question is: Has the error propagated into other systems yet?

People chastise me for taking the self-aggrandizing ignorants to task, but ultimately, they are responsible for these sorts of messes as evidenced by the misconception you suggested earlier that correctly recording historical information somehow violates a normal form.

> Bob Badour wrote:
>

>>Define minimal-impact. Does management prefer to minimize the future
>>repercussions of a system that doesn't work? Or does management prefer
>>to do nothing to fix the problem?
Received on Fri Sep 01 2006 - 23:07:02 CEST

Original text of this message