Re: A question for Mr. Celko

"Jan Hidders" <jan.hidders_at_REMOVETHIS.pandora.be> wrote in message news:pan.2004.07.19.00.05.20.833431_at_REMOVETHIS.pandora.be...
> On Sun, 18 Jul 2004 16:19:35 +0000, Marshall Spight wrote:
> > "Jan Hidders" <jan.hidders_at_REMOVETHIS.pandora.be> wrote in message
> > news:pan.2004.07.18.10.53.04.846560_at_REMOVETHIS.pandora.be...
> >>
> >> It's not so much the "allowing any type" pary that they (and I) have
> >> problems with but that you no longer consider them as atomic.
> >
> > Could you expand on that?
>
> Yes, but it would be more efficient if you had a specific question I could
> answer. That would also tell me more about what is and what is not clear
> to you.

Okay. Date talks about atomic values, and how relvar updating is always essentially relation-assignment. Values are always atomic, whether the value is simple (3) or compound ("brown"). If we have a simple variable of type string, and we update "brown" via s/br/cl, we are substituting the entire value "clown" for "brown" and we should not think of our variable as having 5 sub-variables of type char. Likewise with relations, if we update one value of one row, we have reassigned the entire relation value to the relation variable.

In light of this view, I don't see anything in Date's interpretation of 1NF that means that values aren't atomic anymore. On the other hand, I could imagine you describing this interpretation of "atomic" as trivializing the concept of atomicity as well, which it kind of is.

In any event, I didn't understand your comment "I] have problems with but that you no longer consider them as atomic." in its context. I'm not at all clear as to what it referred to, so I just asked for a generic expansion.

I realize I still haven't asked anything specific, but perhaps I have written enough for you to respond.

Marshall Received on Mon Jul 19 2004 - 04:41:36 CEST