Re: Inconsistent permission enforcement for schemas - Mailing list pgsql-general

From Tom Lane
Subject Re: Inconsistent permission enforcement for schemas
Date
Msg-id 1846618.1659372591@sss.pgh.pa.us
Whole thread Raw
In response to Inconsistent permission enforcement for schemas  (Marcin Barczyński <mba.ogolny@gmail.com>)
List pgsql-general
=?UTF-8?Q?Marcin_Barczy=C5=84ski?= <mba.ogolny@gmail.com> writes:
> Why can a user access a table from a forbidden schema if the table is
> explicitly specified in the definition of a view in an allowed schema? And
> not if the table is accessed by a function used in the view.

(1) Whether a view can access a table is determined by the ownership of
the view, not of the user using the view.

(2) It might be nice if the view's permissions were inherited by functions
called by the view, but we've not made that happen.  You might be able to
approximate it by marking such functions SECURITY DEFINER; although that
makes the function's permissions those of the function's owner not the
view's owner.

            regards, tom lane



pgsql-general by date:

Previous
From: Marcin Barczyński
Date:
Subject: Inconsistent permission enforcement for schemas
Next
From: "sivapostgres@yahoo.com"
Date:
Subject: Re: Creation of FK without enforcing constraint for existing data