W3C * TAG

Request for Proposed Recommendation Status: Architecture of the World Wide Web, 1st Edition

This is a transition request. Whereas

• the TAG is chartered to "document and build consensus around principles of Web architecture"
• we have engaged a substantial community in building a list of architectural issues
• we have developed an architecture document that addresses a useful subset of these issues
• this document has received thorough review, within the group and the community
• we have a significant body of experience using this document to negotiate and deploy principles of web architecture

the Technical Architecture Group has decided (25Oct) to request that you advance this specification to W3C Proposed Recommendation and call for its review and endorsement by the W3C membership.

Document

• Architecture of the World Wide Web, First Edition

  W3C Working Draft 16 August 2004

  editor's draft, changes page

Abstract

The World Wide Web uses relatively simple technologies with sufficient scalability, efficiency and utility that they have resulted in a remarkable information space of interrelated resources, growing across languages, cultures, and media. In an effort to preserve these properties of the information space as the technologies evolve, this architecture document discusses the core design components of the Web. They are identification of resources, representation of resource state, and the protocols that support the interaction between agents and resources in the space. We relate core design components, constraints, and good practices to the principles and properties they support.

The changes since last call include a number of editorial modifications, plus:

• refined URI ownership (section 2.2.2) to be more clearly a relationship between identifiers and resources
• softened extensibility (section 4.2)
• clarified "information resource" (section 2.2)
• relaxed constraints on representations (sections 3.2, 2.2, etc.)

Status (Proposed)

see also: First draft of status section for PR document

This is [not yet] a Proposed Rec. It was first released as a 30 August 2002 Working Draft and has been refined in light of comments on a best-effort basis until the 16 August 2004 Last Call Working Draft; since then, all issues raised in public-webarch-comments have been formally addressed.

Reviews are due 8 Dec 2004. provided we ship by 5Nov, according toplan.

hmm... can't find proposed rec boilerplate in process document pubrules, nor transition how-to.

Summary of Review

The TAG started meeting in January 2002; after a period of collecting architectural issues and drafting smaller findings, we release the first Working Draft of webarch, 30 Aug 2002. The working draft was revised a number of times in light of discussion and comments from www-tag.

We released a last call Working Draft 9 Dec 2003 whose scope includes some, but not all of the architectural issues. We decided that progressing this document through the W3C Recommendation Track at this time is of greater value to the community than waiting until all of the architectural issues have been covered, and the 231 issues raised in response to this draft confirm that the community finds it valuable, or at least interesting. We engaged the following peer groups in the last call process:

• The Internet Architecture Board: 21 Apr 2004
• The XML Schema WG 4 Mar 2004
• The Internationalization WG 18 Mar 2004
• The Device Independence WG 10 Sep 2004
• The HTML WG 29 Sep 2004

Addressing the comments on the Dec 2003 draft resulted in such pervasive changes to the document that we reached a point where it was no longer clear which comments remained relevant, so we issued another last call Working Draft, 6 August 2004. We received comments from 18 reviewers ( Amielh, Bray, Dubost, Hawke, Hayes, Hazaël-Massieux, Henderson/QA WG, Ishikawa, Jacobs, Klyne, Kopecky, Masinter, Meyer, Pemberton/HTML WG, Stickler, Uhl, Weitzner, drewangel ) in the last call review period from 6 August 2004 to 17 September 2004 or shortly thereafter. We reached a clear consensus with the reviewer in many cases (37 threads, including spam and off-topic comments). Addressing some comments involved several rounds of negotiation; for example, a joint meeting with the QA WG 27 Sep. In some cases (Weitzner, drewangel, Henderson/QA, Bray) , we recieved neither confirmation nor dissent to our ultimate reply.

Outstanding Formal Objections

There are 3 cases of outstanding dissent:

• Patrick Stickler sent a comment on "information resource". The TAG considered these comments, among others, in a discussion of 7 Oct, and decided on a revised definition (now available in section 2.2. URI/Resource Relationships of the editor's draft). This definition was seen as an improvement by some commentors (Hawke, Hayes), but when asked about it, Stickler replied, "I am not satisfied with this outcome, ..." 20 Oct.
• Jacek Kopecky sent a Representation of a secondary resource? comment. The TAG considered this comment in a discussion 7 Oct in Basel; after noting that the terms primary/secondary are imported from the URI Standard, considering diagrams to clarify, and considering other wording, we agreed to revise the text ala

  ... The terms primary and secondary in this context do not limit the nature of the resource; they are not classes. In this context, primary and secondary simply indicate that there is a relationship between the resources..

  After several more rounds of negoitation with the commentor, his most recent 28 Oct response says "I cannot live with the text as it stands. [...] Neither of these sections currently mentions the special handling necessary for URIs with fragment identifiers [...]".
• The HTML Working Group sent a comment:

  XLink was issued without reaching consensus, and did not follow due W3C process. This makes it an inappropriate specification for underpinning the Web architecture until such time as consensus has been achieved.

  The TAG considered this comment in a discussion of 7 Oct, and decided to "add that XLink is not the only linking design that has been proposed for XML, nor is it universally accepted as a good design. to section 4.5.2. Links in XML" (now available in section 4.5.2. Links in XML of the editor's draft) . When asked if this was acceptable, the reply said:

  I am afraid not. They responded very strongly that it is not acceptable to

  recommend a spec that has not reached consensus within W3C. They object in

  particular to the wording "[XLink] is an appropriate specification" and

  "Designers of XML-based formats should consider using XLink".

  20 Oct

While we regret that we were unable to achieve consensus with these reviewers, we suggest that the document is sufficiently valuable to the community in its present state to merit advancement to Proposed Recommendation.

Implementation Experience

The subject of the Web Architecture document is not quite the sort of technology that is deployed by direct implementation in software, and hence we do not have a suite of tests by which we test conformance. Rather, the principles in this document are deployed by people developing technologies, tools, content, etc. In particular, they are deployed by other W3C working groups. We have had a number of succesful interactions with other Working Groups regarding principles in the Web Architecture document:

• Earlier drafts of the Speech Recognition Grammar Specification Version 1.0 from the Voice Browser WG appeared to conflict with the HTTP protocol; the TAG worked with the Voice Browser Working Group on issuecontentTypeOverride-24; the TAG was pleased (7 July 2003) with the revisions to the SRGS specifications and condensed the discussion to a constraint, Agents MUST NOT ignore message metadata without the consent of the user derived from a principle, Agents that recover from error by making a choice without the user's consent are not acting on the user's behalf.
• After seeing these dicsussions, the SVG WG added a type attribute in a way that was consistent with the error recovery principle
• A comment on an XForms draft resulted in issue whenToUseGet-7; the TAG worked with the XForms WG to revise the form submission to maximize network effects of addressable resources
• the tag commented that its position on whenToUseGet-7 seemed to be in conflict with SOAP drafts; the XMLP WG revised the SOAP 1.2 specification to emphasize GET for safe operations
• discussion of the whenToUseGet-7 with the Web Services Description Working Group continues
• following request from the Web Services Description WG, the TAG adopted issue abstractComponentRefs-37; while the TAG has not yet resolved this issue, it was able to advise the WSD WG during the development of WSDL 2.0.
• following discussion of profiles of the XML specification, the TAG adopted issue xmlIDSemantics-32; while the Web Architecture discusses this issue (in section 4.5.6) the TAG has not resolved this issue but developed a finding collecting much of the history of the issue, which spans many efforts. The TAG asked that this issue be persued in the XML Activity. The XML Core WG has produced an xml:id requirements document and specification.

See also: some notes on implementation experience with each principle, constraint, and good practice

$Log: prreq.html,v $
Revision 1.16  2004/11/01 19:48:51  connolly
cite newer SOTD, more implemenation experience doc

Revision 1.15  2004/10/28 20:42:53  connolly
18 reviewers, not 17, including Bray
Bray has a thread in commentor-wait
37 closed thread, not 31
dissent += Kopecky on secondary resources