Re: Is relational theory irrelevant? (was Re: Dreaming About Redesigning SQL)

"Mikito Harakiri" <mikharakiri_at_iahu.com> wrote in message news:d_atb.21$6v2.34_at_news.oracle.com...
>
> "Bob Badour" <bbadour_at_golden.net> wrote in message
> news:6oidnZwKgZJQryii4p2dnA_at_golden.net...
> > "Mikito Harakiri" <mikharakiri_at_iahu.com> wrote in message
> > > Thes numbers would help to the query front end to arrange query
results
> in
> > > order. A user would focus his attention on the row ranked #1 first,
them
> > > perhaps perform some physical action depending on the row value, next
> > shift
> > > his focus to row #2, and so on. Doing the procedure in the out of
order
> > > might be less optimal.
> >
> > What advantage would these numbers have over the current ORDER BY
> > functionality?
>
> I thought you objected ORDER BY in the beginning of the exchange.

Earlier, I objected to the ordered structure nested within the query. You now postulate a rank function to allow a user to control the order of external processing. The external processing posited requires ORDER BY with or without the rank function, which would seem to make the rank function superfluous.

> > Considering that the ORDER BY would still be required to
> > guarantee the results come out in the order desired, I would consider
> ORDER
> > BY the required feature.
>
> Given that it would be a no-brainer for the query front end to add rank,
> maybe.
>
> Consider a rank within a group, however.

Again, what advantage does rank give that quota queries do not already give? I can think of one: It represents the information as a value in a relation.

> > What
> > advantage to the logical model does the rank offer that is not offered
by
> > other logical constructs that already exist?
>
> I can join 2 tables by rank? "How many top 10 contenders switched the
ranks
> between 2002 and 2003"? Rank in the competition event database is not
> necessarily stored, as it might be redundant if there is a physical
metrics.

Agreed. By representing the information as a value of some appropriate ordinal type, one can use the value anywhere one can use a value of the type. It seems the information principle favours a rank function. Received on Sat Nov 15 2003 - 04:06:39 CET