Re: Real world issue:- OT recreational interval
Date: Mon, 18 Sep 2006 11:44:04 GMT
Message-ID: <8EvPg.582$x11.244_at_trndny02>
"Marshall" <marshall.spight_at_gmail.com> wrote in message
news:1158529573.221832.120100_at_e3g2000cwe.googlegroups.com...
> Bob Badour wrote:
> > Marshall wrote:
> >
> > Marshall, your reply to Pamela was persuasively worded yet has two fatal
> > flaws:
>
> Bob,
>
> Thanks for the feedback. As to your point 1, I freely admit the
> failing. It is an area I am currently working on, and still need
> work in. I'll try to digest what you said. As to your point 2,
> sometimes one is more concerned with the wider audience,
> even when one is addressing a specific individual. I am
> particularly keen lately to knock down any hint that there
> is any advantage to ignorance.
>
>
> Marshall
>
You already pointed out that the lowest table in Pam's example is not in 1NF. The same is true of the two lowest tables in the original example: TRANS_A and TRANS_B. Pam's current theory is that GROUP BY produces the wrong groups.
I haven't attempted to use GROUP BY with data that is not in 1NF, but I would expect certain anomalies to appear. And I expect that summing an extensive measure is among them. I also expect that counting the rows in each group would produce anomalies. This has nothing to do with limitations in SQL. It has to do with inability to tell duplicates apart, inherent in data that is not in 1NF. Received on Mon Sep 18 2006 - 13:44:04 CEST