[Standards-JIG] JEP-0060: 1.8pre18

Ralph Meijer jabber.org at ralphm.ik.nu
Wed Jun 14 13:18:57 UTC 2006

On Fri, Jun 09, 2006 at 11:03:59AM -0400, Brian Raymond wrote:
> In working with hierarchal nodes the idea of permission inheritance seems to
> make sense as an addition to the spec.
> Currently when you have a collection node with children all of the node
> permissions need to be managed independently. If all the nodes belonging to
> a parent need to be made available to the same users it's much more
> convenient to manage the permissions of just the top level node and have all
> of the children inherit those permissions. To accomplish this a
> configuration option could be added to a node so it can be configured to
> inherit permissions from the specified node.
> <field var='pubsub#innherit_perms'
>          type='test-single'
>          label='Inherit permissions from specified node'/>
> A boolean would be easier in this case since it could simply be set to true
> but that leaves some ambiguity when a node has multiple parents. Maybe two
> options make sense, one a boolean when there is a single parent and this for
> nodes with multiple parents?
> This is flexible enough to work with any nodes but the primary use case is
> for managing permissions on a group of nodes that allow the same access.

The way collections are setup currently is that the reference to the
parent node is held in the configuration of the child node. This works
great for notification, but not for top-to-bottom stuff like permission
checking. In the end it is an implementation detail, of course, but I'm
not sure if such application specific configuration options need to go
into JEP-0060.

Note that section 3.4 of JEP-0068 (Field Standardization for Data Forms)
describes how custom options (== field names) should be named. The MUST
always begin with 'x-'. So in your example: 'x-inherit-perms'.

The use of 'pubsub#' as a prefix is redundant because we always have a
FORM_TYPE field that indicates JEP-0068 semantics. We should have gotten
rid of it long ago.

To Peter: can we still do that?



More information about the Standards mailing list