Date: Sat, 25 Mar 2006 07:38:29 -0500
From: Daniel CAUNE <d.caune@free.fr>
Subject: Re: Index on nullable column
In-reply-to: <12727.1143244097@sss.pgh.pa.us>
To: 'Tom Lane' <tgl@sss.pgh.pa.us>, 'Daniel Caune' <daniel.caune@ubisoft.com>
Cc: pgsql-sql@postgresql.org
Message-id: <0IWO00E7BOG2MP90@VL-MH-MR001.ip.videotron.ca>
MIME-version: 1.0
Content-type: text/plain; charset=UTF-8
Content-transfer-encoding: 7BIT
Thread-index: AcZPnXsTCsdqkOq1TaiQr6c4/tgv2AAasNxw

> > Is an index on a nullable column useful for retrieving rows having that
> > column null?
> 
> Nope, because IS NULL isn't an indexable operator.
> 
> You can make an end-run around that with a partial index, eg
> 
> 	create index fooi on foo(f1) where f1 is null
> 
> This can be used to satisfy queries using "where f1 is null", but it's
> not any good for any other purpose.
> 
> If you often do "where f1 is null and something-about-f2",
> you might get better mileage with
> 
> 	create index fooi on foo(f2) where f1 is null
> 
> but it's still a very specialized index.
> 

Thanks Tom.  I will use such an index even if it's very specialized; OLAP world is somewhat full of specialized index anyway... :-)

--
Daniel