Re: auto group by

From: "Andrew G(dot) Hammond" <drew(at)xyzzy(dot)dhs(dot)org>
To: Markus Bertheau <twanger(at)bluetwanger(dot)de>
Cc: postgres sql list <pgsql-sql(at)postgresql(dot)org>
Subject: Re: auto group by
Date: 2002-01-24 18:34:35
Message-ID: 1011897276.3198.2.camel@xyzzy
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-sql

On Thu, 2002-01-24 at 13:00, Markus Bertheau wrote:
Hello again,

here's another question:

suppose there's a table personen:

personen_id int primary key,
data text

and a table orders:

order_id int primary key,
personen_id int,
data text

Might as well do it the "Postgres Way".

CREATE TABLE personen (
personen_id SERIAL PRIMARY KEY,
name TEXT NOT NULL
);

CREATE TABLE order (
order_id SERIAL PRIMARY KEY,
personen_id INT REFERENCES personen
ON DELETE CASCADE ON UPDATE CASCADE NOT NULL,
data TEXT
);

. Now I want to know how many orders each person has, who has at least
one order. I would use the following SQL query:
select personen.personen_id, personen.text, count(order_id) from
personen join orders on personen.personen_id = orders.personen_id group
by personen.personen_id, personen.text

Suppose you had some more data in personen, like 10 extra fields, and
you want them to be included in the result. You would have to mention
each of them in the group by clause, not only say the personen_id. But
this information that is given to the db server seems a bit redundant to
me. If the personen_id is the same (which qualifies these rows for one
group), every single other field is the same also. Why isn't it
automagically included in the group by clause? You cannot do anything
useful with them but to group by them.

select [1] from tbl1 join tbl2 on tbl1.i=tbl2.i

Like: every row from tbl1 that is mentioned as is (that is not used in
an aggregate function) in the [1] marked location is automagically
included in the group by clause.

Why?

First off, I think this query will acheive what you're looking for

SELECT * FROM personen
NATURAL JOIN
SELECT personen_id, count(order_id) AS orders FROM order GROUP BY
personen_id

Assuming of course there's no "orders" column in personen.

Enlightenment appreciated.

I believe that there was a posting on this very topic a week or two
ago. From what I recall, it's allowed under the latest SQL (98?), but
not by any of the previous flavours. Try browsing back through the
archives. :)

--
Andrew G. Hammond mailto:drew(at)xyzzy(dot)dhs(dot)org
http://xyzzy.dhs.org/~drew/
56 2A 54 EF 19 C0 3B 43 72 69 5B E3 69 5B A1 1F
613-389-5481
5CD3 62B0 254B DEB1 86E0 8959 093E F70A B457 84B1
"To blow recursion you must first blow recur" -- me

In response to

Browse pgsql-sql by date

  From Date Subject
Next Message Tom Lane 2002-01-24 19:00:02 Re: Keeping information changes history
Previous Message Marius Andreiana 2002-01-24 18:32:57 Keeping information changes history