Re: the relational model of data objects *and* program objects
Date: Fri, 15 Apr 2005 00:59:15 GMT
Message-ID: <DlE7e.11704$5F3.4927_at_news-server.bigpond.net.au>
"Kenneth Downs" <knode.wants.this_at_see.sigblock> wrote:
in message news:99d2j2-rkq.ln1_at_pluto.downsfam.net...
> erk wrote:
...[trimmed]...
>>> On the other hand, if you materialize the values, they are protected
>> against
>>> shifting definitions.
>>
>> But you don't always want that protection.
>
> Believe it or not, I agree. Having made the case as best I can above, I
> woulld offer that you probably don't want stored values for things like
> summary tables (although I do have a nifty trick to do that in real-time
> w/o update conflicts).
>
> Here is a case where I have not made up my mind. Consider a derived
> value,
> customer total exposure. It is built up through composition from orders
> and invoices. There is another derived value, customer credit limit. By
> making use of simple formulas and composition, we can begin to establish
> constraints as simple comparisons of these values, so that "customer total
> exposure" must be < = "credit limit". Materializing up to the customer
> table runs the risk of update conflicts, but then again how often are
> multiple people writing to the same customer?
Pete Brown
Falls Creek
Oz
www.mountainman.com.au
Received on Fri Apr 15 2005 - 02:59:15 CEST