Following are the minutes of the unofficial telecon of the XRI TC at:

Date: Thursday, 18 September 2008 USA Time: 8:00AM - 9:00PM Pacific Time

ATTENDING

John Bradley Nika Jones Peter Davis Nat Sakimura Nick Nicholas Drummond Reed Les Chasen Markus Sabadello Eran Hammer-Lahav

GUEST

David Recordon

AGENDA

1) INTRODUCTION - NICK NICHOLAS

Nick is a business analyst working with Kerry Blinco on the PILIN project in Australia (https://www.pilin.net.au). They have been analyzing and preparing to deploy infrastructure for persistent identifiers and linking.

2) XRI AND OPENID AUTHENTICATION 2.1

David Recordon, one of the OpenID Authentication 2.0 editors, is preparing a charter for members of the OpenID community to begin work on OpenID Authentication 2.1. One the potential items of that charter is how to deal with XRI discovery. David joined us to solicit our feedback about that option.

David Recordon introduced himself as an employee of SixApart and one of the editors of the OpenID Authentication 2.0. He explained that the OpenID Authentication 2.0 spec was finalized in December of 2007, representing a convergence of four communities. It has steadily grown in adoption, and now large RPs (relying parties) are starting to adopt.

The 2.0 spec reflects somewhat of a "design by committee", and now 2.1 is being proposed as a way to deal with fixing some areas and a way to simplify implementations.

We discussed XRI support. Many OpenID libraries are not implementing it, and implementations are uneven (for example, they do not always use CanonicalID verification, or use it correctly). David's proposal is to turn XRI support into an OpenID discovery extension.

XDI TC members provided the following key points of feedback.

- No form of OpenID identifer discovery should be optional. OpenID users should know that if they adopt an OpenID identifier, it should work everywhere OpenID is supported.

- OpenID discovery should be a consistent framework for dealing with discovery, security, privacy, and usability of all forms of OpenID identifiers -- including email addresses if they are supported.

- XRIs are the only abstract identifiers currently supported by OpenID Authentication 2.0. They have special security and usability properties that are only available if they are implemented properly. Since OpenID discovery really requires any identifier to be treated as abstract for purposes of discovery, this is the underlying issue that needs to be fixed for all OpenID identifiers -- including http: and https: URIs, not just XRIs.

- The potential introduction of email addresses will make this aspect of OpenID architecture even more important.

David appreciated the feedback. He plans to wrap up preliminary work on the charter this week and send it out next week. It will then proceed through a discussion period. All XRI TC members interested in OpenID are urged to comment. The OpenID mailing lists are publicly available at:

http://openid.net

3) XRDS DISCOVERY UPDATE

Eran Hammer-Lahav gave us a brief update on discussions taking place with regard to how the http(s): URI version of XRDS discovery should evolve. He wrote an analysis of the key issues in a recent blog post:

http://www.hueniverse.com/hueniverse/2008/09/discovery-and-h.html

Eran said the new proposed XRDS spec needs a simple, uniform, consistent discovery protocol. He has talked with Mark Cunningham, who is working with IETF on a LINK header and a site mapping mechanism, and with Phil Archer from POWDER. Eran says there is broad agreement that there should be a common discovery mechanism. He is working on a dynamic mapping mechanism that will tie into XRDS as a discovery document format.

John Bradley described how this might fit with the "http: subscheme" or "http: profile" mechanism currently under discussion with the W3C TAG.

4) W3C TAG DISCUSSION AND XRI TC ABSTRACT IDENTIFIER ARCHITECTURE PAGE

The TAG has placed the "URNsAndRegistries-50" issue, under which they have classified XRI, at the start of the agenda of their F2F meeting.

http://lists.w3.org/Archives/Public/www-tag/2008Sep/0077.html

Drummond explained that he had noted to the discussion coordination group that the "URNsAndRegistries-50" issue -- which deals primarily with URNs and URN namespaces -- is only tangential to XRI architecture, and that we feel the real underlying issues are:

a) Abstract identifier architecture and its relationship to concrete identifier architecture. b) Standardizing metadata discovery, since abstract identifier architecture (among other things) depends on this.

The TAG agenda references the wiki page we have started on the definition of abstract identifier architecture:

http://wiki.oasis-open.org/xri/AbstractIdentifierArchitecture

We agree the action item is to update this with as much additional information as we can before the TAG meeting begins next Tuesday.

# DRUMMOND to coordinate the addition of other material to this page.