W3C home > Mailing lists > Public > public-ws-addressing@w3.org > November 2006

Re: Jury Instructions/ Preliminary Agenda for 2006-11-06 teleconference

From: David Hull <dmh@tibco.com>
Date: Mon, 06 Nov 2006 12:05:07 -0500
To: Katy Warr <katy_warr@uk.ibm.com>
Cc: bob@freunds.com, public-ws-addressing@w3.org
Message-id: <454F6B43.7090009@tibco.com>

I continue to be concerned that (1) below is a false dichotomy, as there
are two largely separate issues involved.

One issue is whether there should be some sort of extension point -- not
necessarily the one I proposed -- for describing what sort of addresses
are and are not acceptable beyond the ones WSA defines (i.e., anon and
none).  This is larger than WS-RX, though WS-RX is of course a driving
use case, and independent of the notion of "back-channel" or
"anonymous".  From the mailing list, there appears to be at least some
support for opening this issue, including (IIUC) the person who filed
CR33.  It is also in line with the findings of the async task force.

The other issue is whether we should define (or allude to, or allow
others to define, or mention but leave undefined ...) some notion of
"back-channel" beyond what we say in the SOAP binding about the anon
address.  If so, there is the separate issue of how to define this.  My
impression, perhaps incorrect, is that this proposal grew out of a
fairly tactical fix to the language in the WSDL binding, though one
could also argue that the notion of "back-channel" would be generally

I'm much more concerned about the first issue than the second.  I would
go along with a proposal for the second, as long as it is well-defined. 
I have, of course, expressed significant concern as to whether the
current such proposal is well-defined.

Framing the question as a choice between these two proposals for largely
distinct questions would tend to imply that accepting one means
rejecting the other, particularly given the length of time this whole
thing has dragged on and the universal desire to get on with it.  This
is why I would strongly prefer to see the issues split.  My expectation
-- and Lord knows I could be wrong -- is that we could get to a yes/no
decision on these separate issues more quickly, as the discussion would
be more focused.

Here is a more formal presentation of what I've already proposed in
other messages:

   1. Close CR 33 with no action.
   2. Go to the pub.

Oops ... sorry ... what I meant was

   1. Close CR 33 with no action, and concomitantly:
         1. Open an issue about a general extension point for describing
            what addresses are/are not allowed.  Use my previous
            proposal as a strawman.
         2. Open an issue about a general definition of "back-channel",
            with Anish & Paco's current proposal as a strawman.  FWIW, I
            would support a version along the lines of what Anish
            mentioned, tying "back-channel" to the SOAP request-response
         3. Open an issue about making our WSDL markings more
            policy-friendly.  I don't have a proposal here, as it's
            outside my comfort zone, but I believe others feel strongly
            enough about it to drive it.
   2. Bash the above issues to death
   3. Go to the pub.

I would propose covering these in the order given (except that (3) could
perhaps be interspersed), as the resolution of (a) might possibly shed
some light on (b), and (c) can probably be done better once the other
two are in place.

Katy Warr wrote:
> Hi Bob
> We have discussed CR33 for such a long time now - we must be in a
> position to make a decision.
> I propose that we formally vote in today's meeting in order to reach
> resolution.  
> I suggest the following staged approach:
> 1) (Assuming not close-with-no-action) Which proposal of those on
> table  (Paco's/Anish's vs DavidHull's)? Summarise proposals. Take a vote.
> 2) Does anyone need the solution to be reflected in WSDL, or will
> policy solution suffice? Take a vote.
> 3) Iron out details such as proposal syntax/names.  Namespace
> implications.
> Best regards
> Katy
> *"Bob Freund" <bob@freunds.com>*
> Sent by: public-ws-addressing-request@w3.org
> 04/11/2006 17:54
> To
> 	"Mark Little" <mark.little@jboss.com>
> cc
> 	"[WS-A]" <public-ws-addressing@w3.org>
> Subject
> 	RE: Jury Instructions/ Preliminary Agenda for 2006-11-06 teleconference
> More like Spencer Tracy (Inherit the wind)
> -bob
> ------------------------------------------------------------------------
> *From:* Mark Little [mailto:mark.little@jboss.com] *
> Sent:* Thursday, November 02, 2006 10:45 AM*
> To:* Bob Freund*
> Cc:* [WS-A]*
> Subject:* Re: Jury Instructions/ Preliminary Agenda for 2006-11-06
> teleconference
> Are you looking for a Henry Fonda person (12 Angry Men) ;-)?
> Mark.
> On 31 Oct 2006, at 23:52, Bob Freund wrote:
> Ladies and Gentlemen of the Jury
> All of the testimony has been given, and the evidence provided for
> your inspection.  The time has come to conclude your deliberations.
> You will be asked to decide the following questions with regard to the
> charges raised against WS-Addressing:
> First Charge:
> One count of flirting with anonymous addresses of unknown character
> without any intention of establishing a meaningful relationship.
> Soap binding 5.2.1 invites other anonymous addresses.
> "Note that other specifications MAY define special URIs that have
> other behaviors (similar to the anonymous URI)."
> If the Jury finds that the WG didn't really mean it, then the spec
> shall be found guilty of this charge.
> If found guilty of this charge, then the WG shall issue an errata
> removing the flirtatious prose and cr33 shall be closed with no action.
> If found innocent, then the WG is sentenced to accommodate such
> anonymous addresses without prejudice and to modify the WSDL binding
> and the policy assertions accordingly.
> How do you find?
> Second Charge:
> Core and Soap binding are inconsistent: The core spec is section 3.2.1
> says that anonymous is a recognizable uri detectable with simple
> string comparison for
> "_http://www.w3.org/2005/08/addressing/anonymous_".  If the Jury
> returns a guilty verdict to the first charge, then this charge is moot
> once the sentence has been served.  If the Jury returns an innocent
> verdict to the first charge, and a guilty verdict to the second
> charge, then the WG shall be sentenced to decide how to remove this
> inconsistency.
> How do you find?
> Third charge:
> One count of not being policy friendly
> Content in the element is not well matched with the policy framework
> that is forming into a specification.
> If found guilty, the mandatory sentence is that all markers are to be
> meaningful by their name alone which touches the WSDL binding as well
> as the policy assertion
> How do you find?
> Thanks
> -bob
Received on Monday, 6 November 2006 17:06:51 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:04:15 UTC