From: Tom Lane <tgl@sss.pgh.pa.us>
To: Andreas <maps.on@gmx.net>
cc: pgsql-sql@postgresql.org
Subject: Re: How to access multicolumn function results?
In-reply-to: <51008731.8020006@gmx.net>
References: <51008731.8020006@gmx.net>
Comments: In-reply-to Andreas <maps.on@gmx.net>
	message dated "Thu, 24 Jan 2013 01:58:25 +0100"
Date: Wed, 23 Jan 2013 20:42:47 -0500
Message-ID: <17722.1358991767@sss.pgh.pa.us>
Precedence: bulk
Sender: pgsql-sql-owner@postgresql.org

Andreas <maps.on@gmx.net> writes:
> SELECT some_fct( some_id ) FROM some_other_table;
> How can I split this up to look like a normal table or view with the 
> column names that are defined in the RETURNS TABLE ( ... ) expression of 
> the function.

The easy way is

	SELECT (some_fct(some_id)).* FROM some_other_table;

If you're not too concerned about efficiency, you're done.  However this
isn't very efficient, because the way the parser deals with expanding
the "*" is to make N copies of the function call, as you can see with
EXPLAIN VERBOSE --- you'll see something similar to

	Output: (some_fct(some_id)).fld1, (some_fct(some_id)).fld2, ...

If the function is expensive enough that that's a problem, the basic
way to fix it is

	SELECT (ss.x).* FROM
	 (SELECT some_fct(some_id) AS x FROM some_other_table) ss;

With a RETURNS TABLE function, this should be good enough.  With simpler
functions you might have to insert OFFSET 0 into the sub-select to keep
the planner from "flattening" it into the upper query and producing the
same multiple-evaluation situation.

			regards, tom lane


-- 
Sent via pgsql-sql mailing list (pgsql-sql@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-sql