- From: Tom Rutt <tom@coastin.com>
- Date: Mon, 05 Mar 2007 12:38:35 -0500
- To: WS-Addressing <public-ws-addressing@w3.org>, ws policy <public-ws-policy@w3.org>
I thought it would be good to close my action Item from WS addressing WG with this summary of my 4 proposed solutions. (I now prefer alterntiave D) Alternatives A through C would rely on multiple policy alternatives to indicate that a response sender can use either Anonymous or Non-anonymous Replies. Alternative A: http://lists.w3.org/Archives/Public/public-ws-addressing/2007Mar/0017.html Summary: Define AnonymousResponses and NonAnonymous responses as requirements for sending responses. NonAnonymous EPR address is anything other than wsa:Anonymous URI. Missing nested assertion within an Anonymous assertion implies prohibition for that alternative. (one problem with Alternative A is that missing nested assertion has same meaning as presence of the other nested assertion, since NonAnonymous URI is a uri which is NOT Anonymous) Alternative B: http://lists.w3.org/Archives/Public/public-ws-addressing/2007Feb/0016.html Summary: Define AnonymousResponses and NonAnonymous responses as requirements for sending responses. NonAnonymous EPR address is defined as any "connectable" URI.. Missing nested assertion within an Anonymous assertion implies prohibition for that alternative. (one problem with Alternative B is difficulty in defining "connectable" URI in a transport independent manner) Alternative C: http://lists.w3.org/Archives/Public/public-ws-addressing/2007Mar/0016.html Summary: Define AnonymousResponses and NonAnonymous responses as requirements for sending responses. NonAnonymous EPR address is defined as anything other than wsa:Anonynous URI.. Missing nested assertion within an Anonymous assertion has no meaning with respect to use of response URIs. Alternative D: http://lists.w3.org/Archives/Public/public-ws-addressing/2007Mar/0018.html Summary: Remove nested policy assertions for wsa:Addressing assertion. I personally could live with either alternative D or C, but prefer Alternative D. If we pick Alternative C, I see the only practical thing for a response sender to put in its policy is three alternatives (one for addressing with nonAnon repolies, another for addressing with anon replies, and another with addressing showing no nested policy assertion). However this is no better than saying that addressing is supported. If we select Alternatives A , B, or C, we should probable add text stating that the policy attached to a response sender subject pertains to individual instances of responses. Also clarify that if alternatives exist for either non anon or anon, that different response EPRs in the same requiest can obey any of the allowed alternatives. -- ---------------------------------------------------- Tom Rutt email: tom@coastin.com; trutt@us.fujitsu.com Tel: +1 732 801 5744 Fax: +1 732 774 5133
Received on Monday, 5 March 2007 17:40:13 UTC