Agenda for 4 February 2002 TAG teleconference

Next: 12 February face-face >>

architecture · administration

Nearby: TAG home · teleconference details · issues list

Any additions to the agenda? Problems with previous minutes (28 Jan)?

Architecture

First let us look at raising ome issues, suggestions belowin bold (Any others?)

Architecture threads

This is a reference summary of threads In chronological order oldest to newest. Ian has collected them, and Tim has marked some with proposed status. Today we go quickly (2 minutes per) to raise issues where we have consensus

1. Resource discovery: limits of URIs. Raised by Peter Pappamikail. Refer to TB's reply. "Why not have URNs?". See The Myth of Names and Addresses ...Needs arch document; not issue unless contentious.
2. Why does the Web use mime types and not URIs?. Raised by Aaron Swartz. propose raise issue, low urgency. Refer Don Eastlake's work - should be homewhor before we address it.
3. What's the proper terminology for distinguishing between the HTTP/HTML Web, things with URIs, the services Web, etc.? Alternately, what is the Web? . Raised by Aaron Schwartz.
4. Do Web Architecture principles continue to make sense in a world with decentralized systems like Freenet and Mojo Nation?. Raised by Aaron Schwartz.
5. URIs versus URI references. Raised by Mark Baker. See comments from Tim Bray about related issues:
   • relative or absolute?
   • fragments OK or not?
   • URNs sometimes always or never?
   • the trick that RDF does of conventionally using a trailing # and relying on concatenation for finer-grain identification (not quite the same issue as fragments?). How does it interact with XML?
6. HTTP is not well understood . Raised by Mark Baker.
7. System architecture document needed. Raised by David Orchard. A discussion between David and Roy follows. This also led to TAG home work on reading chapters of Roy's thesis. See comments from Graham Klyne on the thesis w.r.t. TAG work.
8. XML Packaging mechanism needed. Raised by David Orchard.
9. It would be beneficial to describe common profiles of XML processing. . Raised by David Orchard.
10. What is the XML Processing model?. Raised by David Orchard.
11. An XML Type Library would be useful. Raised by David Orchard.
12. Is simplicity a useful architectural constraint?. Raised by Tim Bray. A long thread follows.
13. Overlap of functionality between XSLT and XQuery. Raised by Graham Klyne. .
14. Comments about comments by James Clark. Raised by Stephen van Egmond. See comments from Norm, Chris, Tim Bray, and more.
15. Is testability a useful architectural constraint?. Raised by Chris Lilley. A long thread follows.
16. Design Issues: is GET the only idempotent method?. Raised by Mike Dierken. Sounds from comments from TBL that this may have been addressed. See followup from Massimo Marchiori: Foundational Web Model(s). Possible technical development: QUERY method in HTTP. propose raise this., to address oustanding need. Pseudothreads include:
    1. Background information on GET and XForms (was: GET should be enco uraged...) Micah Dubinko (Thu, Jan 24 2002)
    2. RE: Background information on GET and XForms (was: GET should be encouraged...) Martin Duerst (Sun, Jan 27 2002)
17. Design Issues: proactive distribution. Raised by Mark Dierken. No replies yet.
18. Architecture doc issues Raised by Tim Bray, who disagrees strongly with: "The namespace document (with the namespace URI) is a place for the language publisher to keep definitive material about a namespace. Schema languages are ideal for this." I think some discussion around this will be useful, and I woul hope to have a TAg statement at the end of it -TBL. Propose raise issue
19. Multiple namespaces. No issue raised by Paul Prescod. Just comments on mixing documents. However, I think there is a long-standing issue of validation for mixed namespaces. No issue
20. Language bindings for extending functionality. "Language binding was explicitly dropped from XSLT 2.0, in the recognition that a common approach was required across the W3C." Raised by Jim Fuller. propose raise issue.
21. TAG media type issues and XForms Last Call WDMark Nottingham (Thu, Jan 24 2002) Part of the iss . Mark needs TAG support IMHO.
22. [rdfmsQnameUriMapping-6] Algorithm for creating a URI from a QName in RDF Model? Ian B. Jacobs (Tue, Jan 29 2002)

Issue for discussion today

Continuing XML mime types and namespaces. Issue 3, what is the relationship between MXL namespaces and media types? See 18 above. I have proposed a relationship. Simon St L has made a draft suggestingthat all NSs be visible. I would go along with the top level NS being visible in the MIME type.

We also have to settle that the outermost element of the document is critical, and the meaning of a document hangs on that, or if not, what you are supposed to do with a multi-namespace document.

Requests for liaison

• No new. However, the new CEO of WAPforum offered to send 2 techies to talk about their architecture plans to w3C, and we could have some TAG members on it maybe? Discuss who would like to, could, etc. Probably a Thursday 9-10:30ET slot some week.

Requests for review

• No new. Well, didn't we say we don't take requests for review? This section will dissapear next week (by 2002/2/11)

Administrative

Review outstanding action items

Upcoming meetings

Nearby: meeting information

• Next meeting: 12 February. Meeting information 12 Feb ftf available.
• 27 Feb: Panel discussion. Refer to thread of panel suggestions.

How we work

• See a suggestion for TAG output from Tim Bray. This includes a strawman approach for issuing "Findings", more lightweight than W3C Rec process.
• Should we make formal models where appropriate? See DanC's work in Larch, on http etc @@link
• Paul Cotton has suggested that we consider how we will carry out some of our chartered roles:
  • very early review (prior to Last Call) of the deliverables of Working Groups chartered to produce Architectural Recommendations;
  • coordination with Working Groups that realize only after chartering that they are producing Architectural Recommendations.
  • establishing liaisons with groups outside of W3C involved in the development of Web architecture.
• On the "three-month" rule: Since the three-month rule is to ensure that a group doesn't spend too much time out of the public eye, the fact that TAG minutes are public satisfies the requirement.

Charter elaboration

Setting up a process for hearing appeal as described in the charter. We have to elaborate this.

Discuss briefly, possible subgroup. Volunteers?