RE: New draft of Metadata document - was RE: 2007-01-15 Teleconference canceled due to low attendance <EOM>

May I suggest that we re-word the examples slightly in order to indicate 
'best practice' and avoid confusion amongst implementers?

Each example could be of the form:
      In order to indicate <capability> then the following policy is 
recommended: <policy>

So example 3-4 might look something like:

In order to indicate WS-Addressing required with anonymous and 
non-anonymous responses supported by provider,  the following policy is 
recommended:
<wsp:Policy>
    <wsam:Addressing>
        <wsp:Policy>
            <wsp:ExactlyOne>
                <wsam:AnonymousResponses wsp:Optional="true"/>
                <wsam:NonAnonymousResponses wsp:Optional="true"/>
            </wsp:ExactlyOne>
        </wsp:Policy>
    </wsam:Addressing>
</wsp:Policy>

This would obviously need to be non-normative.

Katy





"Yalcinalp, Umit" <umit.yalcinalp@sap.com> 
Sent by: public-ws-addressing-request@w3.org
15/01/2007 19:22

To
David Illsley/UK/IBM@IBMGB
cc
"Bob Freund" <bob@freunds.com>, "[WS-A]" <public-ws-addressing@w3.org>, 
<public-ws-addressing-request@w3.org>, "Rogers, Tony" <Tony.Rogers@ca.com>
Subject
RE: New draft of Metadata document - was RE: 2007-01-15 Teleconference 
canceled due to low attendance <EOM>







 

> -----Original Message-----
> From: public-ws-addressing-request@w3.org 
> [mailto:public-ws-addressing-request@w3.org] On Behalf Of 
> David Illsley
> Sent: Monday, Jan 15, 2007 1:21 AM
> To: Yalcinalp, Umit
> Cc: Bob Freund; [WS-A]; public-ws-addressing-request@w3.org; 
> Rogers, Tony
> Subject: RE: New draft of Metadata document - was RE: 
> 2007-01-15 Teleconference canceled due to low attendance <EOM>
> 
> 
> Hi Umit,
> I'm in agreement with Tony, the intersection will fail. 
> Inclusion of the 
> empty nested policy is for use when either side wants to 
> allow another 
> mechanism to determine which Response EPRs are supported (one 
> suggestion 
> was trying one and switching if it faulted). If either party wants 
> behaviour guaranteed by policy (my preferred mode of operation), they 
> should not include an empty nested policy and the 
> intersection will fail. 
> My view on how this will work in practice is that servers 
> will probably be 
> permissive and include the empty policy element (with optional 
> AnonymousResponses and/or NonAnonymousResponses assertions) 
> while most 
> clients will specify exactly what they want.

Ok, that is my expected design pattern as well. I am glad we are in
agreement there. Are you also saying that this is the intended
interpretation of the paragraph I asked for clarification in this
thread? 

BTW, the example 3.2 with the title "no-statement on supported
responses" actually is "no-supported responses", because it is going to
block a specific type of response unless both sides do not care and
include an empty policy. Perhaps you guys should think about clarifying
this specific point, because just reading the document as it is, the
points we just clarified in this email exchange are not obvious. 

I just do not want service providers or clients to use the nested
assertions in a way to get unintended results, thus the guidance should
be clear. Perhaps the WS-Addressing nested assertions may be a good
example to use in the WS-Policy Guidelines document to clarify the
affect of nested matching. I will think about this more. 


> 
> So, my answer to your question to the policy group [1] which appears 
> related to this is (a) No.
> 
> Tony: Sorry, I've spotted a problem with the doc in Example 3-4. The 
> wsam:AnonymousResponses and wsam:NonAnonymousResponses 
> elements should not 
> have wsp:Optional="true" attributes.
> 
> David
> 
> [1] http://lists.w3.org/Archives/Public/public-ws-policy/2007Jan/0069
> 
> David Illsley
> Web Services Development
> MP211, IBM Hursley Park, SO21 2JN
> +44 (0)1962 815049 (Int. 245049)
> david.illsley@uk.ibm.com
> 
> 

Cheers, 

--umit

> 
> From:
> "Rogers, Tony" <Tony.Rogers@ca.com>
> To:
> "Yalcinalp, Umit" <umit.yalcinalp@sap.com>, "Bob Freund" 
> <bob@freunds.com>, "[WS-A]" <public-ws-addressing@w3.org>
> Date:
> 15/01/2007 08:00
> Subject:
> RE: New draft of Metadata document - was RE: 2007-01-15 
> Teleconference 
> canceled due to low attendance <EOM>
> 
> 
> Hello Umit
> 
> you are quite right, the intersection will fail. 
> 
> If we interpret the client as MUST have anonymous response, 
> and the server 
> as NOT GUARANTEEING anonymous response, it is arguable that it SHOULD 
> fail.
> 
> That's my interpretation, anyway.
> 
> If the server wants to state support for both anon and 
> non-anon, it must 
> not provide the empty policy alone; it must provide empty, anon only, 
> non-anon only, and both.
> 
> Tony Rogers
> CA, Inc
> Senior Architect, Development
> tony.rogers@ca.com
> 
> From: Yalcinalp, Umit [mailto:umit.yalcinalp@sap.com]
> Sent: Mon 15-Jan-07 17:28
> To: Rogers, Tony; Bob Freund; [WS-A]
> Subject: RE: New draft of Metadata document - was RE: 2007-01-15 
> Teleconference canceled due to low attendance <EOM>
> 
> There is some discussion in the WS-Policy wg about the semantics of 
> intersection with empty policy alternatives and nesting. This 
> is why the 
> WS-A approach to using nesting is rather important. 
> 
> I read the document and the following statement is not very 
> clear. Could 
> the wg clarify what is intended: 
> 
> {Note also that the lack of either of these assertions 
> (AnonymousResponses 
> and NonAnonymousResponses) does not indicate lack of support. 
> So it is 
> suggested that a subject that does not have a strict compatibility 
> requirement that an interacting subject understands or is 
> concerned with 
> these assertions provides an alternative without either assertion. }
> 
> For example, your example 3.2 (with no statement on support 
> on supported 
> response EPRs) on a service will fail to intersect with a 
> clients policy 
> which would require anonymous responses. Is the statement 
> quoted above 
> trying to recommend use of alternatives that contain nested 
> assertions to 
> indicate explicit support for type of responses 
> (anonymous/non anonymous) 
> in one of the nested alternatives ? If that is the case, 
> Example 3.2 needs 
> to be positioned appropriately. Using example 3.2 alone as a policy 
> expression by a service will not allow the clients that 
> require a specific 
> type of responses to communicate with the service as the intersection 
> algorithm will fail, but that is not clear from the text. 
> Thus, example 
> 3.2 as "no-statement on supported responses" is misleading. 
> 
> Cheers, 
> 
> --umit
> 
> 
> 
> 
> From: public-ws-addressing-request@w3.org [
> mailto:public-ws-addressing-request@w3.org] On Behalf Of Rogers, Tony
> Sent: Sunday, Jan 14, 2007 12:56 PM
> To: Bob Freund; [WS-A]
> Subject: New draft of Metadata document - was RE: 2007-01-15 
> Teleconference canceled due to low attendance <EOM>
> 
> So everyone has a whole week to study the new Editor's Draft of the 
> Metadata document :-)  You will find it at: 
> http://dev.w3.org/cvsweb/~checkout~/2004/ws/addressing/ws-addr
> -wsdl.html?content-type=text/html;%20charset=utf-8
> 
> The main changes are the complete removal of UsingAddressing 
> and the SOAP 
> module as alternatives for indicating the use of 
> WS-Addressing (yes, I 
> have anticipated the WG slightly, but I can roll this back if 
> it is not 
> agreed - want you to see what it looks like without those) - the only 
> mechanism supported for indicating/requiring the use of 
> WS-Addressing is 
> the policy assertion.
> 
> Please e-mail the list with any omissions or mistakes. 
> 
> Tony Rogers
> CA, Inc
> Senior Architect, Development
> tony.rogers@ca.com
> 
> From: public-ws-addressing-request@w3.org on behalf of Bob Freund
> Sent: Mon 15-Jan-07 6:51
> To: [WS-A]
> Subject: 2007-01-15 Teleconference canceled due to low 
> attendance <EOM>
> 
> 
> 
> 
>