On 09/29/2012 02:28:40 AM, Karl O. Pinc wrote:
> On 09/29/2012 01:16:51 AM, Karl O. Pinc wrote:
> 
> > The attached documentation patch further describes the 
> > additive nature of the permission system.
> 
> Attached a second patch, applied after the
> first, to extend the discussion
> further regards roles.

Found a typo in the second patch.  Resending
the original first patch and a v2 of the second.




Karl <k...@meme.com>
Free Software:  "You don't pay back, you pay forward."
                 -- Robert A. Heinlein

diff --git a/doc/src/sgml/ref/grant.sgml b/doc/src/sgml/ref/grant.sgml
index fb81af4..73f88e0 100644
--- a/doc/src/sgml/ref/grant.sgml
+++ b/doc/src/sgml/ref/grant.sgml
@@ -429,11 +429,27 @@ GRANT <replaceable class="PARAMETER">role_name</replaceable> [, ...] TO <replace
    </para>
 
    <para>
-    A user may perform <command>SELECT</>, <command>INSERT</>, etc. on a
-    column if he holds that privilege for either the specific column or
-    its whole table.  Granting the privilege at the table level and then
-    revoking it for one column will not do what you might wish: the
-    table-level grant is unaffected by a column-level operation.
+    Permission granted at any level of the
+    <database>database</><literal>.</><database>schema</><literal>.</><database>table</><literal>.</><database>column</>
+    object hierarchy grants permission to all contained objects.
+    E.g. a user may perform <command>SELECT</>, <command>INSERT</>,
+    etc. on a column if he holds that privilege for either the
+    specific column or its whole table.  Granting the privilege at the
+    table level and then revoking it for one column will not do what
+    you might wish: the table-level grant is unaffected by a
+    column-level operation.  Granting a privilege to some columns of a
+    table and denying the privilege to the table's other columns is
+    done in the same way as regards all other hierarchically organized
+    database objects (e.g. granting a privilege to some tables in a
+    schema and denying the privilege to the schema's other tables):
+    deny privilege to the protected columns, to their table, their
+    schema, and their database; grant privilege to the permitted
+    columns.  As <link
+    linkend="sql-grant-description-objects">described above</link>,
+    permissions granted to the roles assigned a user are likewise
+    entirely additive.  Permissions throughout
+    <productname>PostgreSQL</productname> combine in this uniform
+    fashion.
    </para>
 
    <para>


diff --git a/doc/src/sgml/ref/grant.sgml b/doc/src/sgml/ref/grant.sgml
index 73f88e0..0e878ba 100644
--- a/doc/src/sgml/ref/grant.sgml
+++ b/doc/src/sgml/ref/grant.sgml
@@ -453,6 +453,22 @@ GRANT <replaceable class="PARAMETER">role_name</replaceable> [, ...] TO <replace
    </para>
 
    <para>
+    Further, roles having the <literal>INHERIT</literal> attribute
+    that are assigned to other roles in a hierarchical fashion are
+    additive of permission in the fashion of the
+    <database>database</><literal>.</><database>schema</><literal>.</><database>table</><literal>.</><database>column</>
+    hierarchy.  E.g. a user's login role can be assigned a role of
+    <literal>accountant</> which is in turn assigned a role of
+    <literal>employee</>.  The user would have the permissions of an
+    <literal>accountant</> and, by virtue of the role hierarchy, also
+    all permissions granted to <literal>employee</>s.  Unlike the
+    fixed
+    <database>database</><literal>.</><database>schema</><literal>.</><database>table</><literal>.</><database>column</>
+    hierarchy the <productname>PostgreSQL</> user is free to fashion
+    roles into arbitrary hierarchical structures.
+   </para>
+
+   <para>
     When a non-owner of an object attempts to <command>GRANT</> privileges
     on the object, the command will fail outright if the user has no
     privileges whatsoever on the object.  As long as some privilege is


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

Reply via email to