Date: Tue, 29 Mar 2005 09:58:23 -0400
From: Alvaro Herrera <alvherre@dcc.uchile.cl>
To: T E Schmitz <mailreg@numerixtechnology.de>
Cc: Bruno Wolff III <bruno@wolff.to>,
	Scott Marlowe <smarlowe@g2switchworks.com>, pgsql-sql@postgresql.org
Subject: Re: cost of CREATE VIEW ...  AS SELECT DISTINCT
Message-ID: <20050329135823.GA23852@dcc.uchile.cl>
References: <42487A6D.6020506@numerixtechnology.de>
	<1112047463.22988.25.camel@state.g2switchworks.com>
	<424928D8.7060705@numerixtechnology.de>
	<20050329132941.GA27251@wolff.to>
	<4249564B.1000807@numerixtechnology.de>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <4249564B.1000807@numerixtechnology.de>
User-Agent: Mutt/1.5.8i

On Tue, Mar 29, 2005 at 02:21:15PM +0100, T E Schmitz wrote:
> Bruno Wolff III wrote:
> >On Tue, Mar 29, 2005 at 11:07:20 +0100,
> >  T E Schmitz <mailreg@numerixtechnology.de> wrote:
> >
> >>Would the "SELECT DISTINCT origin" always cause a sequential table
> >>scan regardless whether there is an index on the origin column or
> >>not?
> >
> >It's worse than that, SELECT DISTINCT cannot use a hash aggregate
> >plan and will need to do a sort to eliminate duplicates. Unless the
> >view is used in a way that restricts the candidate rows, this
> >probably isn't going to be very fast. You might be better off
> >changing the view to use GROUP BY instead of DISTINCT.
> 
> As far as I can see (via EXPLAIN), both DISTINCT and GROUP BY will
> lead to a sequentail scan. Is that correct?

That (GROUP BY using a seqscan) may be caused by the small size of the
table.  Try populating it some more.

-- 
Alvaro Herrera (<alvherre[@]dcc.uchile.cl>)
"¿Qué importan los años?  Lo que realmente importa es comprobar que
a fin de cuentas la mejor edad de la vida es estar vivo"  (Mafalda)