Re: Modelling and the element of time

From: <MSherrill_at_compuserve.com>
Date: Wed, 26 Sep 2001 16:02:23 GMT
Message-ID: <3bb1f345.8614156_at_news.compuserve.com>


On Tue, 25 Sep 2001 07:35:45 +0200, "Drago Ganic" <drago.ganic_at_in2.hr> wrote:

[snip]
>Namely, almost always a 1:N relation turns to an M:N relation, when I =
>think about time and/or about the relationships that exist in 10 - 20% =
>of entity occurrences.
>
>For example, the classic relation EMPLOYEE <--> DEPARTMENT is a M:N =
>relation *over time*. Rarely the relation is M:N also *in time*. I =
>resolve the relation with an intersection entity with two attributes =
>(the time attributes):
>
> * From date
> o Thru date
>
>So the user can always see the history in the production system.
[snip]

If the application requires you to store history, you don't have many options besides storing history. Before you go too far down this road, you really need a copy of "Developing Time-Oriented Database Applications in SQL", by Richard Snodgrass.

-- 
Mike Sherrill
Information Management Systems
Received on Wed Sep 26 2001 - 18:02:23 CEST

Original text of this message