Message-Id: <s1543456.028@mail.tpchd.org>
Date: Fri, 24 Sep 2004 14:49:56 -0700
From: "Ian Harding" <iharding@tpchd.org>
To: <gsstark@mit.edu>, <tgl@sss.pgh.pa.us>
Cc: <Karsten.Hilbert@gmx.net>, <pgsql-general@postgresql.org>,
	<pgsql-sql@postgresql.org>
Subject: Re: [SQL] need ``row number``
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline

Right.  Sequence within a group.  I did this in a SRF resetting the sequenc=
e after each subgroup.  I would love to see a better way.

<<< Greg Stark <gsstark@mit.edu>  9/24  1:42p >>>
Tom Lane <tgl@sss.pgh.pa.us> writes:

> Karsten Hilbert <Karsten.Hilbert@gmx.net> writes:
> > I am not convinced I'll need a SRF. I am not trying to
> > calculate something that isn't there yet. I am just trying to
> > join two views appropriately. I might have to employ some
> > variant of Celko's integer helper table but I'm not sure how
> > to proceed.
>=20
> A fairly common hack for this is to use a sequence:
>=20
> 	create temp sequence tseq;
> 	select nextval('tseq'), * from (select .... order by ...) ss;

But I thought she wanted to get the row number within a group. Not the row
number for the entire result set. A sequence can't do that. Or at least, I
suppose it could but it would be mighty strange to see setval() in a SELECT
query. And I can't think of how to detect the level break in a select query
either.

--=20
greg


---------------------------(end of broadcast)---------------------------
TIP 8: explain analyze is your friend

=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20=20