• XriTcW3cTag

1. About

This is the home page summarizing progress of the dialog between the XRI TC and the W3C TAG (Technical Architecture Group) regarding the TAG's concerns about XRI 2.0.

2. Links to Related Documents

• W3C TAG draft finding on URNs, Namespaces, and Registries (this version dated 2006-08-17)

• XriTcW3cTag/TagFindingReferences is a brief analysis of how the other projects cited in the TAG finding above do/do not relate to the XRI TC's requirements for abstract identifiers.

• AbstractIdentifierArchitecture is a page for collaborative drafting of a proposed XRI TC white paper explaining abstract identifier architecture.

• XriAsRelativeUri describes a new proposal arising from the W3C TAG discussions for how XRI architecture integrates with Web architecture.

3. Progress of the Discussions

• 31 May 2008: OASIS Standard vote on XRI 2.0 specifications fails approval by 1 percentage point.

• 2 June 2008: XRI TC meets and agrees with the comments provided by a number of OASIS members that the XRI TC should pursue a dialog with the TAG about their concerns.

• 10 June 2008: Emails are exchanged and both parties agree to schedule an initial joint telecon.

• 3 July 2008: Initial joint telecon.

  • See XriTcW3cTag/FirstTelecon for preparatory materials.

  • W3C TAG minutes of the meeting (note that they follow IRC log formats, so they can be somewhat challenging to read).

  • See also further discussion in the Draft Meeting Minutes TAG 2008-07-10 Telecon

• Ongoing: active email discussions on the public W3C TAG mailing list. Look for discussions with the tag [XRI] in the subject line.

• 22 August 2008: Joint telecon of XRI TC and W3C TAG coordination team (XRI TC: Peter Davis, Drummond Reed, John Bradley; TAG: Stuart Williams, Henry Thompson) – agreement that substantial progress is being made; action item to create and distribute remaining issues list and proposed next steps (below).

• 2 October 2008: XRI TC members Drummond Reed and John Bradley visit with W3C TAG co-chair Stuart Williams in Bristol, England to discuss an emerging proposal for better integration of XRIs with the Web.

• 14 October 2008: XriAsRelativeUri proposal posted on the XRI TC wiki for mailing list and telecon discussion among XRI TC members.

• 23 October 2008: XriAsRelativeUri proposal page updated with XRI TC member feedback and message sent to the public W3C TAG mailing list to invite feedback.

4. Primary Open Issue Areas and Proposed Next Steps in Each

On the 2008-08-20 coordination telecon between the XRI TC and W3C TAG, the XRI TC took the action item to:

1. Write up summaries of the remaining open issue areas.
2. Identify interrelationships and dependencies between them.
3. Propose next steps with respect to each issue.

The balance of this wiki page is this writeup. The XRI TC will endeavor to keep this list of issues/next steps current until the discussions are concluded.

5. #1: Abstract Identifier Architecture Issues

The first "bucket" of issues are related to abstract identifier architecture, of which the XRI specifications are an implementation. These issues are all related to each other, and all of them shape the requirements for the other two buckets of issues, so we list these first.

5.1. Abstract vs. Concrete Identifiers

The first issue is the need for a clear definition/model for: a) abstract identifiers for resources, and b) resource descriptors for resources, together a clear explanation of how these co-exist with concrete identifiers for resources and the target resources themselves.

5.2. Persistence and URN Requirements

One of the requirements for abstract identifiers (but not the only one) is the need for persistent identifiers that can serve as immutable keys when referencing distributed resources. This requirement is generally summarized in RFC 1737 – Functional Requirements for Uniform Resource Names, though there has been additional thinking about the requirements around persistence in the decade since that was published. The open issue is clearly mapping these requirements into the abstract identifier architecture defined above.

5.3. Determination By Inspection

One of the XRI TC's other requirements for abstract identifiers is that certain characteristics of the identifier, such as abstractness, persistence/reassignability, and authority, be determinable via inspection (i.e., not require resolution). The open issue is determining whether there is consensus between the XRI TC and the TAG on this requirement.

5.4. Other Requirements for Abstract Identifiers

There are other often more compelling requirements for abstract identifiers beyond persistence and inspection, for example structured identifiers and cross-references. The open issue is communicating these requirements clearly enough for the W3C TAG and other interested parties to understand them.

6. #2: URI Scheme Issues

The second bucket of issues are related to URI schemes. These issues have cross-dependencies with each other, and also to some extent depend on the conclusions regarding requirements above.

6.1. HTTP URI Profiles

The TAG's position (as summarized in Section 5.5 of their URNs, Namespaces, and Registries finding) is that the XRI requirements should be met by creating a "subclass" of HTTP URI rather than a separate scheme. It is proposed to call such an HTTP URI subclass an HTTP URI profile. The open issues regarding HTTP URI profiles are:

• Should they use a standardized mechanism (e.g., specified domain names or domain name patterns, specified path strings, specified query strings, etc.)?
• If standardization is recommended:
  • Where should this work be done?
  • How quickly can it be done (given that XRI would have a dependency on it)?
• If standardization is not recommended, are there still recommended best practices the XRI TC and others should follow?

6.2. Rationale for New URI Schemes

Even if some new types of identifiers can be implemented using an HTTP URI profile, the question remains: when does a new type of identifier justify a new URI scheme? This question is important for the XRI TC since it currently has implemented specifications that use the xri: scheme, yet due to procedural misunderstandings, has not yet applied to register this scheme with IANA. So the open issues are:

1. What are the criteria that justify registration of a new URI scheme?
2. Does XRI meet these criteria?

7. #3: Resolution/Description Issues

The last major bucket of issues regards resolution and resource descriptor architecture. These have some dependencies on the first bucket since resolution/description requirements flow from the requirements of mapping abstract identifiers to target resources. However are largely independent of the second bucket.

7.1. Resource Descriptor Documents as Resources

The first open issue is how resource descriptor documents (also called "discovery documents") are themselves treated as resources from an AWWW standpoint.

7.2. XRDS Discovery Protocol Methods

The second open issue is reviewing the protocol methods currently used (or proposed for the next revision) to discover an XRDS document from either an XRI or an HTTP(S) URI to see if there conform to AWWW recommendations and best practices, and whether the TAG recommends any improvements.