Date: Thu, 12 Jul 2007 06:51:36 -0600
From: Michael Fuhr <mike@fuhr.org>
To: Heikki Linnakangas <heikki@enterprisedb.com>
Cc: Marcin =?utf-8?Q?St=C4=99pnicki?= <mstepnicki@gmail.com>,
	pgsql-performance@postgresql.org
Subject: Re: one column from huge view
Message-ID: <20070712125136.GA37198@winnie.fuhr.org>
References: <f74p19$1gb$1@sea.gmane.org> <4695EB62.7060307@enterprisedb.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <4695EB62.7060307@enterprisedb.com>

On Thu, Jul 12, 2007 at 09:50:42AM +0100, Heikki Linnakangas wrote:
> Marcin Stępnicki wrote:
> >Let's say I've got a view with 100 columns and 1mln rows; some of them are
> >calculated "on the fly". For some reason I want only one column from
> >this view:
> >
> >select col1 from huge_view;
> >
> >Now, does PostgreSQL skip all the calculations from other columns and
> >executes this query faster then select * from huge_view?
> 
> In simple cases, yes. But for example, if you have a LEFT OUTER JOIN in 
> the view, the join is performed even if your query doesn't return any 
> columns from the outer relation. Also, if the calculation contains 
> immutable functions, it's not skipped.

Don't you mean "if the calculation contains VOLATILE functions,
it's not skipped"?

-- 
Michael Fuhr