DomainKey-Signature: a=rsa-sha1; c=nofws; d=juffo.org; s=google;
	h=mime-version:in-reply-to:references:from:date:message-id:subject:to
	:cc:content-type;
	b=T+AwyntNQtAeJcJBFOoZBw9t5BDcTN1ew14TivC8AWOep6CfL1RrmA6VCUESRXpKhv
	M50N0iWks70WTXC6jQcxXxFuEcOIDpvNS53wIvy71Kj3v1cu0haMI+XSpKpQIDQ8FyUU
	AmzXe6XaEsfd4APY9hPzLdy33BU50B3iHUxwk=
MIME-Version: 1.0
In-Reply-To: <4DFA4D2F.8020101@peak6.com>
References: <BANLkTikntdSrhneCPzPutY9T-sAQpFXTMw@mail.gmail.com>
	<BANLkTikuFEyOse=sUoH0BXYrpUW=7Pp_tw@mail.gmail.com>
	<4DFA4D2F.8020101@peak6.com>
From: Marti Raudsepp <marti@juffo.org>
Date: Wed, 22 Jun 2011 12:55:46 +0300
Message-ID: <BANLkTim5saGCxhLCAx4nEUQ4tZuOSy65fg@mail.gmail.com>
Subject: Re: seq scan in the case of max() on the primary key column
To: sthomas@peak6.com
Cc: Magnus Hagander <magnus@hagander.net>,
	Svetlin Manavski <svetlin.manavski@gmail.com>,
	pgsql-performance@postgresql.org
Content-Type: text/plain; charset=UTF-8

On Thu, Jun 16, 2011 at 21:36, Shaun Thomas <sthomas@peak6.com> wrote:
> You can call that instead of max, and it'll be much faster. You can create
> an analog for min if you need it. So for this, you'd call:

Cool, I've needed this function sometimes but never bothered enough to
write it myself. Now I created a wiki snippet page for this handy
feature here:
https://wiki.postgresql.org/wiki/Efficient_min/max_over_partitioned_table

With Jim Nasby's idea to use regclass instead of relation names, the
function is now half its length and probably more reliable. There's no
need to touch pg_class directly at all.

I also changed it to return bigint instead of integer, as that's more
versatile, and the performance loss is probably negligible.

Regards,
Marti