DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma;
	h=message-id:date:from:to:subject:mime-version:content-type:content-transfer-encoding:content-disposition;
	b=iRFhClOuSxDqTE07DMtLfHRULWYk/XDGieAuIJon7La/YPqD01KrnHhtgGQnpKHz5/C0pBhRbjVPg3tWdaPqgPyh2hWAyvOMC20k38K7g54IzMeT3xFbZNAnwgzVhi9JB0Sknj6B2IgvGJhpfh1opOm1AHULFL6V1Sg2GIZB/Pg=
Message-ID: <2d734e2f0804281447l5de57868na1a5415b0a95ee28@mail.gmail.com>
Date: Mon, 28 Apr 2008 17:47:59 -0400
From: "Lodewijk Voege" <lvoege@gmail.com>
To: pgsql-sql@postgresql.org
Subject: information_schema.referential_constraints permissions
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

hello,

I have some code that gets foreign key information from
information_schema.referential_constraints. I was puzzled about why it wasn't
returning anything for a while, until I read the information_schema.sql file
and the documentation carefully: it has a pg_has_role(c.relowner, 'USAGE') in
it and the documentation states "The view referential_constraints contains all
referential integrity (foreign key) constraints in the current database that
belong to a table owned by a currently enabled role".

Fair enough. But then I read the public draft of SQL 2003 and 200n on this
view, and it speaks only of "tables in this catalog that are accessible to a
given user or role", rather than ownership. the user I was working with most
definately had access. it could also read the pg_* system catalog, such that
if I copy/paste the view definition without the pg_has_role() line in it (and
massage it a bit to remove the CASTs to sql_identifier and character_data)
that user got just what I was expecting.

so, is that restriction correct?

Lodewijk