From: Tom Lane <tgl@sss.pgh.pa.us>
To: Achilleas Mantzios <achill@matrix.gatewaynet.com>
cc: pgsql-sql@postgresql.org
Subject: Re: bug in 9.2.2 ? subquery accepts wrong column name : upd
In-reply-to: <1641558.Xf5bk256pC@smadev.internal.net>
References: <1641558.Xf5bk256pC@smadev.internal.net>
Comments: In-reply-to Achilleas Mantzios <achill@matrix.gatewaynet.com>
	message dated "Thu, 14 Mar 2013 17:29:07 +0200"
Date: Thu, 14 Mar 2013 12:17:46 -0400
Message-ID: <17320.1363277866@sss.pgh.pa.us>
Precedence: bulk
Sender: pgsql-sql-owner@postgresql.org

Achilleas Mantzios <achill@matrix.gatewaynet.com> writes:
> dynacom=# SELECT DISTINCT ON (qry.setid) qry.setid, qry.arragg[1:2] FROM 
> 	(select distinct sst.setid,(array_agg(vsl.name) OVER (PARTITION BY sst.setid ORDER BY character_length(vsl.name))) as arragg 
> 	FROM sissets sst, vessels vsl WHERE vsl.id=sst.vslid ORDER BY sst.setid) as qry 
>      WHERE array_length(qry.arragg,1)>1  AND qry.setid IN (SELECT setid from sis_oper_cons) ORDER BY qry.setid,array_length(qry.arragg,1);
> [ works ]

> however, there is not column setid in sis_oper_cons,

If not, that's a perfectly legal outer reference to qry.setid.

Probably not one of SQL's better design features, since it confuses
people regularly; but it's required by spec to work like that.

			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