Re: Why using "Group By"
Date: Thu, 13 Mar 2003 16:49:54 -0500
"Mikito Harakiri" <mikharakiri_at_ywho.com> wrote in message
> "oferbu" <junkbu_at_hotmail.com> wrote in message
> > Hello All,
> > This is a theoretical question. Why do I need to add the "Group By" in
> > the following SQL:
> > SELECT
> > CUSTOMER_CITY, COUNT(*)
> > FROM
> > CUSTOMER_TABLE
> > Isn't it clear that I want to get the number of rows per city, so why
> > is it necessary to add:
> > GROUP BY
> > CUSTOMER_CITY
> > I mean to say that those added lines didn't give more information, and
> > any resonable person (or a good sql parser...) could have understand
> > what I really want at the first sql.
> One more reason: you can omit aggregate and emulate "distinct" like this
> select empno, name, sal, deptno from emp
> group by empno, name, sal, deptno
> Therefore, either "distinct" is redundant as we can express it via "group
> by" or, alternatively, "group by" is redundant and can be expressed via
> corellated scalar subquery in the select clause, but we need "distinct"
> then. I wonder what theoretical reason behind this fact is.
Assuming logical identity, both distinct and group by are superfluous. Received on Thu Mar 13 2003 - 22:49:54 CET