[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Simple] Presence Authorization Discussion B: Composition/Aut h Sequencing





Goeman Stefan wrote:
Hello Paul,

I understand what you mean, and you are right.
But, I would like to argue that there are two topics in authorization.
1) First, determining if a watcher is allowed to subscribe, and determine
what to do (allow, block, polite block, ...).
2) Secondly, if he is allowed, you can do all this element-by-element
authorization things you mention below.

What I mean is that it is probably useful to have a feature like a black
list, a white list and a gray list (you don't need all three, a black list
and gray list will do or a white list and gray list).
Maybe this is just an implementation issue. I think we can do with the
current presence authorization rules. It should be possible for the PS
(Policy Server) to construct these lists from the authorization policy.

There clearly is a distinction between whether a subscription is to be accepted, and what is to be disclosed once it is accepted. These need to be processed separately because they are handled at different points in time. (Obviously the decision to accept the subscription or not happens first, and typically only once per subscription. The decision about what to disclose must be repeated every time there is new presence state available.)


Exactly how the policy is implemented is not something the IETF should decide. The data model provides an abstract model of how it *might* be implemented, but isn't normative. As an implementor, you are perfectly free to partition the policy processing in a way that you find more optimal, as long as you preserve the result.

Here we need to be clear about the semantics of the policy so that different implementations are consistent. And it would be good to define things in such a way that an efficient and compliant implementation is possible.

	Paul

P.S.: Maybe the idea of black/white/gray lists have been discussed before. I
don't know. I only recently subscribed to the mailing list.

Greetings,
Stefan.


-----Original Message-----
From: Paul Kyzivat [mailto:pkyzivat at cisco.com] Sent: woensdag 13 oktober 2004 18:14
To: Markus.Isomaki at nokia.com
Cc: Goeman Stefan; simple at ietf.org
Subject: Re: [Simple] Presence Authorization Discussion B: Composition/Auth Sequencing



Markus,

I don't follow you - I think the point you make is backward.

First I want to clarify that we are talking about authorization on an element-by-element basis within the document, as has been proposed.

I also want to assume that that the composer has broad lattitude it what it does, and in particular that it is free to construct elements that didn't exist in any of the inputs. (E.g. it might construct an <activity> of on-the-phone from other inputs that didn't include any <activity> element.)

With those assumptions, an authorization policy that precedes and is independent of composition policy can be ineffective. It may not authorize inclusion of an <activity> element, but may authorize the elements that the composer will use to infer it.

To get the intended result in this case, authorization must follow composition.

	Paul

Markus.Isomaki at nokia.com wrote:

Hi Stefan and all,

I agree that it is actually hard to separate composition

and authorization from each other if both of them are watcher-dependent. (And I think there is some value that even composition can be watcher-dependent, such as the ability to show different person information to different watchers.) But if you assume that the same rulemaker makes rules for both functions, the rulemaker can at least in theory keep things consistent even if composition happens first and authorization after that. But if the composition policy is not understood by the auth policy rulemaker AND composition happens first, he can't make even auth rules that have a deterministic outcome. At the moment, before the details of composition policy are defined, we are at this kind of situation.


Markus



-----Original Message-----
From: simple-bounces at ietf.org [mailto:simple-bounces at ietf.org]On Behalf
Of ext Goeman Stefan
Sent: 13 October, 2004 13:08
To: 'simple at ietf.org'
Subject: RE: [Simple] Presence Authorization Discussion B:
Composition/Auth Sequencing



Hello,

I would agree on the order of first composition and then authorization in
the case that composition would be watcher independent.


But in light of the discussion of watcher dependent composition I would
favor the other way around (first authorization and the

composition).

Because, does it make sense to do a watcher dependent composition and spent
time and processing power on it, to find out later that you will block this
watcher.


Greetings,
Stefan.



-----Original Message-----
From: simple-bounces at ietf.org [mailto:simple-bounces at ietf.org] On Behalf Of Jonathan Rosenberg
Sent: maandag 11 oktober 2004 9:43
To: Simple WG
Subject: [Simple] Presence Authorization Discussion B: Composition/Auth Sequencing



We had a lot of list discussion at some point about whether composition came first or whether authorization came first, and about where the line between one ended and the next began.


I think this is much clearer now in light of the data model, which very explicitly separates the roles of composition (correlation,

conflict



resolution, merging and splitting) and privacy filtering based on authorization rules. The former happens first. Indeed, I think the data model gives us a good grounding to start a discussion later

on about



what we might want composition policy documents to look like. But lets hold that off until we finish our basic work.

Thanks,
Jonathan R.
--
Jonathan D. Rosenberg, Ph.D. 600 Lanidex Plaza
Chief Technology Officer Parsippany, NJ

07054-2711


dynamicsoft
jdrosen at dynamicsoft.com                     FAX:   (973) 952-5050
http://www.jdrosen.net                      PHONE: (973) 952-5000
http://www.dynamicsoft.com


_______________________________________________ Simple mailing list Simple at ietf.org https://www1.ietf.org/mailman/listinfo/simple


_______________________________________________ Simple mailing list Simple at ietf.org https://www1.ietf.org/mailman/listinfo/simple



_______________________________________________
Simple mailing list
Simple at ietf.org
https://www1.ietf.org/mailman/listinfo/simple



_______________________________________________ Simple mailing list Simple at ietf.org https://www1.ietf.org/mailman/listinfo/simple



_______________________________________________
Simple mailing list
Simple at ietf.org
https://www1.ietf.org/mailman/listinfo/simple