See also: IRC log
RESOLUTION: to approve minutes 3 Sep http://www.w3.org/2001/tag/2009/09/03-minutes
NM: 17 Sep telcon is canceled
NM: we seem to be set for 10am Thu during the ftf
NM: ftf regrets other than Raman?
JohnK: at risk [which day?]
NM: agenda sketch:
<noah> * Discuss HTML 5 working draft, and make plan for providing feedback to the working group
<noah> * Make progress on the Architecture of the Web of Applications
<noah> * Discuss issues relating to metadata on the Web
<noah> * Establish more clearly TAG priorities for coming year
<DanC_> johnk, which day of the ftf do you have a conflict?
NM: I think we're making good progress on 1 of our 3 priorities, HTML; I'm concerned we didn't do much on web apps nor metadata
... does anyone have any input in those areas?
AM: John and I are working on something around device APIs... I expect we'll have something...
<trackbot> ACTION-284 -- Jonathan Rees to flesh out the Web Application (http://www.w3.org/2001/tag/2009/06/webAppsTOC.html) outline with as many sentences as he can -- due 2009-08-25 -- OPEN
ACTION-284 due 15 Sep
<trackbot> ACTION-284 Flesh out the Web Application (http://www.w3.org/2001/tag/2009/06/webAppsTOC.html) outline with as many sentences as he can due date now 15 Sep
<trackbot> ACTION-282 -- Jonathan Rees to draft a finding on metadata architecture. -- due 2009-08-31 -- OPEN
JAR: I expect to do 284 tue
<trackbot> ACTION-282 Draft a finding on metadata architecture. closed
<trackbot> ACTION-283 -- Larry Masinter to update document on version identifiers w.r.t. Cambridge June discussion -- due 2009-08-31 -- PENDINGREVIEW
<Zakim> ht, you wanted to promise a new draft of Naming Schemes
<scribe> ACTION: John to prepare draft on device APIs [recorded in http://www.w3.org/2009/09/10-tagmem-irc]
<trackbot> Created ACTION-300 - Prepare draft on device APIs [on John Kemp - due 2009-09-17].
johnk, when did you say you expect to deliver that draft?
<johnk> DanC, by Thursday next week
action-300 due 15 Sep
<trackbot> ACTION-300 Prepare draft on device APIs due date now 15 Sep
<trackbot> ACTION-33 -- Henry S. Thompson to revise naming challenges story in response to Dec 2008 F2F discussion -- due 2009-09-18 -- OPEN
NM: on web of applications, while I see some low-level progress, I don't think we've made progress on a bigger picture.
... note updates to ftf logistics page
<DanC_> [oops; I haven't managed to link that from the TAG home page]
<Zakim> ht, you wanted to comment
HT: on the question of 'does this proposal merit a new URI scheme?' I think it does, since it defines a new protocol.
... the 'just use http:' approach doesn't apply.
... then w.r.t. Alan R's question, I don't have an informed opinion; hard to tell from the spec what it's for, among all the algorithms
<Zakim> noah, you wanted to ask do we have TAG consensus that new protocol ==> new scheme OK?
NM: it's not so clear to me that if you used the http protocol, the right things wouldn't happen...
... I think this is in the space of things that motivated Comet... simulate asynchrony, bidirectional communication, e.g. chat...
... this websockets protocol starts with bytes that look like an HTTP request... I think the request looks like an HTTP UPGRADE request...
<johnk> yup, HTTP Upgrade header
NM: how about using http:, and saying that servers should handle not only UPGRADE to websockets, but also GET, which would respond with some sort of description of the websocket
JAR: I had trouble finding motivation/context in/around the spec too... where do these URIs occur? in <a href="ws:...">?
<Zakim> jar, you wanted to ask how the URIs can be used
<Zakim> ht, you wanted to question noah's story
HT: my guess is they're used in xmlhttprequest...
JK: there's a new websocket api, part of the HTML 5 work
HT: I searched the HTML 5 draft for "websocket" and got no hit
JK: I think it was split out
DanC: NM, if they used http: , how would the client know to do websockets protocol?
NM: well, I assume it's somewhat like you know you're expecting an image: from context
<noah> API spec draft: http://dev.w3.org/html5/websockets/
<jar> compare with telnet:
HT: a problem is that webarch has a model of URIs appearing in documents... and being used to get other documents with MIME types... though that's not the case for all URI schemes
<noah> I agree jar, telnet is good comparison
<Zakim> noah, you wanted to say they need some ID scheme, and we usually encourage URIs, no?
HT: so there's a question of whether these things ever escape from APIs into documents
<johnk> noah: client would know to do websockets through the Upgrade header from the server
NM: yes, telnet: is an interesting point of comparison
[more discussion... exceed scribe's bandwidth]
<Zakim> ht, you wanted to question noah's story
NM: Ian Hickson has submitted the ws and wss URI scheme registrations to IETF. Documentation of APIs and Protocols is also out there in drafts somewhere, but I don't at the moment recall where.
<DanC_> I see a "Background" section http://tools.ietf.org/html/draft-hixie-thewebsocketprotocol-40#page-4 ... maybe that's got some motivation
<scribe> ACTION: John to review websocket protocol/api motivation and brief TAG at Sep ftf [recorded in http://www.w3.org/2009/09/10-tagmem-irc]
<trackbot> Created ACTION-301 - Review websocket protocol/api motivation and brief TAG at Sep ftf [on John Kemp - due 2009-09-17].
<jar> 1. Is a new protocol needed? 2. Does it need URIs at all? 3. Does it need a new URI scheme? ...
<noah> JK's note on HTML 5 issues: http://lists.w3.org/Archives/Public/www-tag/2009Sep/0012.html
<ht> I note that my issues email is not linked from the agenda: http://lists.w3.org/Archives/Public/www-tag/2009Sep/0015.html
JK: the HTML 5 intro has a section on relationship to other specs; it seems unclear
... on distributed extensibility, I saw several related symptoms; several of them seem to be HTML WG issues already, e.g. Media-type registration http://lists.w3.org/Archives/Public/public-html/2009Sep/0180.html
<Zakim> ht, you wanted to ask about HTML WG process wrt closing issues
HT: I see some issues closed in Aug... not clear what the resolutions were, nor whether raiser was satisfied
DanC: right; the clerical work of copying the issue resolutions into more visible places is ongoing; pick one for discussion?
(issue 5 got into the tracker before we really had a clear sense of how to use the tracker, I think)
ISSUE-28 http-mime-override http://www.w3.org/html/wg/tracker/issues/28
the proposal that carried on issue-28 was http://lists.w3.org/Archives/Public/public-html/2009Aug/0703.html
HT: I don't see a record that the raiser is satisfied
DanC: WG members silently agree after a week or so
NM: are raisers notified?
DanC: yes, the chairs are making an effort
HST: OK, thanks for clarifying, I understand better now
<noah> 3.2.1 (3.3.1) "Authors must not use elements, attributes, and attribute values for purposes other than their appropriate intended semantic purpose. Authors must not use elements, attributes, and attribute values that are not permitted by this specification or other applicable specifications." This is one of the most important sentences in the entire specification, but it's somewhat vague. If "other applicable specifications" means: any specification that an
<noah> turned into a hyperlink that sets down unambiguously the rules for determining whether another specification is "applicable".
NM: the bit about "or other applicable specifications" is potentially really cool... though it's not terribly clear.
... does it mean "specs from the HTML WG"? or "from W3C"? or "from WHATWG?" or "Joe in his basement"?
... some responses in blogs suggest the most liberal interpretation
danc: a request for clarification of "applicable" would be strengthened by inclusion of an example, such as "is spec X applicable?"
<noah> NM: ack noah
<Zakim> noah, you wanted to point to interesting text on extensibility
NM: ok, sounds like yes, the "other applicable specifications" bit is something to talk more about, w.r.t. distributed extensibility
<noah> ACTION: noah to raise (as individual issue) question of 3 words "other applicable specifictions" in 3.2.1 (3.3.1) of HTML 5 [recorded in http://www.w3.org/2009/09/10-tagmem-irc]
<trackbot> Created ACTION-302 - Raise (as individual issue) question of 3 words "other applicable specifictions" in 3.2.1 (3.3.1) of HTML 5 [on Noah Mendelsohn - due 2009-09-17].
JK: next concern of mine is "2. Relationship between the DOM and its HTML serializations" ...
<noah> So, my understanding is: the TAG has expressed a general sentiment that extensiblity MAY be something we'll eventually want to comment on, but with respect to the question raised by Noah (the 3 word phrase), he should start with ACTION 302, I.e. to ping the HTML WG as an individual
JK: seems that the DOM is a separate spec; these are too closely intertwined in the spec
<Zakim> ht, you wanted to agree with NM, as I understand
DC: The definition of HTML document is not nearly as clear as I'd like to be. Something about tree structure and serialization, and difference will be clear when necessary. Roy Fielding says this stuff is too implementation-specific. Henri Sivonen responded that DOM constructs weren't unduly constraining when he wanted to build a non-DOM based validator.
<noah> DC: As to namespaces in one serialization or another, that seems separate
<noah> DC: Mostly same DOMs serialize both ways.
<noah> DC: Also, W3C has tried separating DOM from HTML, didn't work well.
NM: the interesting thing to me about these layering are in document.write() and such that update -- well, you tell me -- is it the DOM or the serialization?
... the way the spec deals with that sort of thing is very fundamentally thru the DOM
... I expect if you wanted to focus on the non-scripting case things could be simplified considerably
... in some cases with scripts, there _is_ no piece of text that corresponds to the DOM
<noah> 3.5.3 (3.8.2): document.write: "If the method was invoked on an XML document, throw an INVALID_ACCESS_ERR exception and abort these steps."
NM: another part of this is that document.write() doesn't work in application/xhtml+xml
DanC: right; that's the design that's deployed
NM: if I want well-formedness to interface with XML databases but serve it as text/html, does that work?
DanC: yes, with a dozen or so caveats; these are known as "polyglot" documents
<DanC_> e.g. <blockquote />
JK: 3rd, "Lack of RFC2119 normative language"
<noah> 126.96.36.199 (188.8.131.52) Style attribute "Documents that use style attributes on any of their elements must still be comprehensible and usable if those attributes were removed." That appears to be a formal "must", but the criterion is informal and not testable. I think this one probably is better as a "should" than a "must".
DC: HTML WG history is that these are intended as RFC 2119, and it's known that some musts depend on author's intent.
<noah> > However, the very concept of semantic markup is useless unless markup
<noah> > is used according to the intended meaning. Therefore, I think it is
<noah> > even interop-sensitive (in the sense that receivers process the
<noah> > messages in ways compatible with sender expectations) that authors use
<noah> > markup according to the specified semantics.
DC: I sort of made my peace with that.
... ...(Dan reads other comments from his email)...
<jar> This doesn't make sense to me. The html has to be assumed to be the only communication between sender and receiver
<jar> .. the sender can't be held responsible for things it can't know
<johnk> yes, what in in Noah's quote does "comprehensible and usable" actually mean to an implementor?
<jar> Three senses of conformance needs to be kept distinct: conforming sender, conforming document, conforming receiver