RE: [xdi] Agenda:Joint XRI & XDI TC Telecon 10AM PT Thursday 2007-06-14

["Steven Churchill" <steven.churchill@xdi.org>]

Drummond, 

I will place here my comments regarding the proposal at
<http://wiki.oasis-open.org/xri/XriCd02/CanonicalIdVerification>.

Thanks for "pushing" the model here to encompass URI. I think that in the
end this will be of benefit.

Here are a couple of observations about the proposal above.

The proposal does not mention whether the resolving agent returns (A) the
XRDS for the original URI, or (B) the XRD obtained by resolving the
canonical ID. If the answer is not (B) then your new definition of Canonical
ID verification is not consistent with that currently specified in section
11 [Resolution Specification.] Recall that that section defines Canonical ID
verification as a promise from the resolving agent (in this case the XRI
Resolver) that if the XRD contains a Canonical ID then that XRD has been
produced by the authority chain making up the canonical identifier path.

So, in order to obtain consistency with the Canonical ID Verification
semantics of section 11, the resolving agent needs to return (B). 

More importantly, it turns out the new proposal in its current form is not
"model consistent" with that currently specified in section 11. [This is
irregardless of my proposal to add the graph drawings and formalisms. "The
existing model is the existing model" whether we document the graph or not.]
The section 11 model clearly says that two authority nodes cannot have the
same Canonical ID. But in the new proposal, they can. Consider the following
graph drawing representing the proposed model. (The curly braces denote
Canonical ID. The square braces denote that "the enclosed identifier
resolves to this node".)

                           O  [https://]
             
                       /         \

                    O              O  
            [https://A.com]    [https://B.com]
           [{https://B.com}]  [{https://B.com}]

As shown here, new model allows both nodes to have the same Canonical ID. 

In order to achieve consistency with the current model for Canonical ID
verification, we can simply change the proposal from the "pointer" living in
a <CanonicalID> to the pointer living in a <Ref>. 

Now, when you make this Ref from the left-most leaf node above to the
right-most leaf node, then you are in effect "creating" a polyarchical edge
from the root node "https://"; to its right-most child. This is precisely the
same thing that happens in figure 11-3 [in my proposed text.] The presence
of the Ref effectively makes the left-most leaf node disappear from the
graph.

With this approach, now the "verification" behavior is precisely consistent
with that in the current section 11--the resolving agent simply "follows the
Ref" and returns the XRDS of the ref target. It thus fulfills the promise
that the XRDS is produced by the authority chain in the returned canonical
identifier path. (Note that, in this case, this is simply the root
"https://"; producing the XRD for the right-most child node above.)

So this approach completely eliminates the notion of "verifying backrefs".
This is a good thing because that clearly involves a different
identity/synonymity model than the one currently used for Canonical ID
verification.

~ Steve



-----Original Message-----
From: =drummond.reed [mailto:drummond.reed@cordance.net] 
Sent: Wednesday, June 13, 2007 7:28 PM
To: xri@lists.oasis-open.org; xdi@lists.oasis-open.org
Subject: [xdi] Agenda:Joint XRI & XDI TC Telecon 10AM PT Thursday 2007-06-14

Following is the agenda for a joint unofficial telecon of the XRI and XDI
TCs at:

Date:  Thursday, 14 June 2007 USA
Time:  10:00AM - 12:00PM Pacific Time

Event Description:
Weekly unofficial joint call of the XRI and XDI Technical Committees.

TO ACCESS THE AUDIO CONFERENCE:
    Dial In Number: 571-434-5750
    Conference ID: 5474


AGENDA

1) EXTENDING CANONICAL ID VERIFICATION TO HTTP(S) URIS

After the special telecon on this subject held on Monday (see minutes at
http://lists.oasis-open.org/archives/xri/200706/msg00050.html), a formal
writeup was posted on the XRI TC wiki at:

	http://wiki.oasis-open.org/xri/XriCd02/CanonicalIdVerification

Feedback on this proposal has been requested from the OpenID Specs mailing
list. We will review this proposal in detail and discuss how it, 

2) REVIEW OF STEVE'S TEXT ON THE CANONICAL ID VERIFICATION GRAPH

We will follow up on the extensive list discussion this past week, and
decide how this text should best be integrated into the next draft. Steve's
document, which was an attachment to an email to the list, is available at:

	http://lists.oasis-open.org/archives/xri/200706/doc00003.doc 


3) OTHER ISSUES IN XRI RESOLUTION 2.O WORKING DRAFT 11

As of this writing, not enough progress has been made to post Editor's Draft
03, so we are still working off the second editor's draft posted last week:
	
http://www.oasis-open.org/committees/download.php/24286/xri-resolution-v2.0-
wd-11-ed-02.doc

Drummond will prepare a list of open questions, and also proposed
assignments for completing ED03.