Re: model inherited object

From: Erwin <e.smout_at_myonline.be>
Date: 23 Jun 2006 03:43:15 -0700
Message-ID: <1151059395.654885.315610_at_c74g2000cwc.googlegroups.com>


> So I record tuple X as (ID,Sq Ft, Toilets, ValidFrom) as LOC01 - 200 -
> 2 - 2005.
> Because of the building work I will record tuple Y as LOC01 - 400 - 2 -
> 2005 but in fact those 2 toilets are inherited and should be reflected
> if I ever update the 2005 record. ALSO I want the 2006 database to show
> me the relevant details (which in this case is 200 sq ft and 2 toilets
> because it hasn't changed since 2005).

You seem to take it as axiomatically true that any change recorded for year x must necessarily also be recorded for all years x+n. I disagree.

If the user wants to change some recorded facts about the past, then it is *his* responsibility to specify *precisely* what the period of time is for which he wants to change those recorded facts. And the system must both allow and force the user to do exactly that. It's not the system's task to decide on the timespan for which some update is valid.

> Thanks for any advice

I'm afraid there's only one really useful advice to be given here : if you're looking at *data*, then put off those damn OO glasses they gave you when you were born. It will help you to see clearer. E.g. in a database context, there is little true meaning in statements such as "toilets are inherited".

And if "putting off those darn OO glasses" implies that you are left without a vocabulary to express yourself, well, that's a strong indication that you still have a lot of relevant vocabulary to learn. Date's "Introduction to ..." is not that bad a place to start and learn it. Received on Fri Jun 23 2006 - 12:43:15 CEST

Original text of this message