DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma;
	h=references:in-reply-to:mime-version:content-transfer-encoding
	:content-type:message-id:cc:x-mailer:from:subject:date:to;
	b=A9dzINEdGJ0Gh0JqrGSZf6WEnhlua/BcpA6oT5bD/wi/3l3+u8a5fe4pCqaHjAMCxf
	I9/kAJJy/S0Cds1rkkjhmyOsYxmK1etNIGX6TUgadH2HDPrv41h4iJRgVhBsvTBmk7Ob
	4VLV2d6fq0y3nKnvymHWNI/TFqB4srws+DNLQ=
References: <AANLkTi=SwWco0mJcU549tzMXJ0kmBwCsNHabH8OUjg1M@mail.gmail.com>
	<BANLkTimuvQr3Lvyh6T6fbKDs1v34Bi5=aQ@mail.gmail.com>
	<BANLkTimPgtYKLyEE8K0NKoE=DAvgm7p8+A@mail.gmail.com>
In-Reply-To: <BANLkTimPgtYKLyEE8K0NKoE=DAvgm7p8+A@mail.gmail.com>
Mime-Version: 1.0 (iPhone Mail 8A293)
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain;
	charset=us-ascii
Message-Id: <4B594466-0554-4AF8-A8BC-8FDA9C393DF9@gmail.com>
Cc: "pgsql-performance@postgresql.org" <pgsql-performance@postgresql.org>
From: Robert Haas <robertmhaas@gmail.com>
Subject: Re: big distinct clause vs. group by
Date: Sat, 23 Apr 2011 15:34:11 -0400
To: Uwe Bartels <uwe.bartels@gmail.com>

On Apr 18, 2011, at 1:13 PM, Uwe Bartels <uwe.bartels@gmail.com> wrote:
> Hi Robert,
>=20
> thanks for your answer.
> the aggregate function I was talking about is the function I need to use f=
or the non-group by columns like min() in my example.
> There are of course several function to choose from, and I wanted to know w=
hich causes as less as possible resources.

Oh, I see. min() is probably as good as anything. You could also create a cu=
stom aggregate that just always returns its first input. I've occasionally w=
ished we had such a thing as a built-in.

Another option is to try to rewrite the query with a subselect so that you d=
o the aggregation first and then add the extra columns by joining against th=
e output of the aggregate. If this can be done without joining the same tabl=
e twice, it's often much faster, but it isn't always possible.  :-(

...Robert=