DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta;
	h=received:to:subject:date:user-agent:references:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:message-id:from;
	b=Pa/kVWV/Uk4t0dedUbOJMOHioCkANC3V2WzsyWe+l1iwK6rQQkmLBMxtyLI9CEi7b2aotqz+IUbmV1rUiI/owPEgKzOWdE1LJ3UO2AKsuQ1DPsQl5RnkVO/GAkVA1JhXwUg3q3sJhKBOwFkMC2QXYzDpQ4ulILn/5JnqRbLM4DI=
To: pgsql-performance@postgresql.org
Subject: Re: Weird performance drop
Date: Fri, 30 Mar 2007 08:36:16 +0200
User-Agent: KMail/1.9.5
References: <018c01c77257$cb0bd520$2e00a8c0@tridecap.com>
In-Reply-To: <018c01c77257$cb0bd520$2e00a8c0@tridecap.com>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200703300836.16601.Vincenzo.Romano@gmail.com>
From: Vincenzo Romano <vincenzo.romano@gmail.com>

On Friday 30 March 2007 01:12 Dave Dutcher wrote:
> > From: pgsql-performance-owner@postgresql.org
> > [mailto:pgsql-performance-owner@postgresql.org] On Behalf Of
> > Vincenzo Romano
> >
> > I thought that the query planner usually did a bad job on
> > function bodies
> > because they'd appear opaque to it.
> > In this case it seems to me that the body is opaque only if I
> > use the "like"
> > operator.
>
> If you run explain on a query that looks like "select * from a_table where
> a_column like 'foo%'" (and you have the appropriate index) you will see
> that postgres rewrites the where clause as "a_column >= 'foo' and a_column
> < 'fop'".  I think your problem is that the query is planned when the
> function is created, and at that time postgres doesn't know the value you
> are comparing against when you use the like operator, so postgres can't
> rewrite the query using >= and <.  The problem doesn't happen for plain
> equality because postgres doesn't need to know anything about what you are
> comparing against in order to use equality.
>
> Somebody else can correct me if I'm wrong.
>
> Dave

Is there any "workaround"?

In my opinion the later the query planner decisions are taken the more
effective they can be.
It could be an option for the function (body) to delay any query planner
decision.

-- 
Vincenzo Romano
----
Maybe Computers will never become as intelligent as Humans.
For sure they won't ever become so stupid. [VR-1987]