Re: SQL (was: Why using "Group By")

From: Bob Badour <>
Date: Fri, 14 Mar 2003 14:16:02 -0500
Message-ID: <Rlqca.49$>

"Mikito Harakiri" <> wrote in message news:IGpca.8$
> Bob,
> I would like to steer discussion away from narrow topics of duplicates and
> logical identity. Some folks in db field throw in (somewhat irresponsibly)
> new term "post relational". There is, obviously, little substance to their
> claims, because they have little to none novel technique to offer. On the
> other hand the term might be valid as soon as somebody suggests some new
> methods, that will expand SQL programming power.
> Here is an example demonstrating that ordering might have some rich
> symantics beyond a mere ordering of the end result:
> select
> case when prior ename is NULL
> then 0
> else prior sumsal+sal
> end AS sumsal
> from emp
> order by ename
> What is this: perverted low level SQL, or just plain procedural code? Any
> comments?

Post-relational will not evolve from SQL. It would require sufficient theoretical advances in mathematics or logic to develop a new logical model based on new theory.

I find the use of NULL perverse--though, characteristically SQL. We certainly do not need recursion to calculate SUM. I have no objection to the use of order based on some explicit collating sequence of data values--quota queries, for instance, require this.

Do you have an example problem requiring recursion for which transitive closure provides no solution?

(As an aside: Did you mean "then sal" where you have "then 0"?) Received on Fri Mar 14 2003 - 20:16:02 CET

Original text of this message