HTML WG Teleconference

13 Nov 2008


See also: IRC log


MikeSmith, DanC, Josh, Cynthia, MurrayMaloney, AdrianBatemen
ChrisWilson, LauraCarlson, Julian


Markup Language Spec

MikeS: I posted a draft of the spec that I have been working on. It would be helpful as a starting point.

<MikeSmith> http://www.w3.org/html/wg/markup-spec/

<pimpbot> Title: HTML: The Markup Language (at www.w3.org)

MikeS: I realize very few have looked at it. Has anyone initial comments?
... Gives outline of abstract
... The doc defines authors, producers and consumers differently.
... Gives further details. No normative criteria, web browsers are not defined in terms of how they parse HTML, Is not intended to be an authoring guide.
... HTML Syntax is described. Various Mime Types are discussed. Its the same prose as defined in the current draft, pretty much. Optional BOM are mentioned etc.
... DOC Type, character encoding etc are defined. The remaining part of spec is a list of HTML elements and their content models, attributes and values etc.

<takkaria> I had a brief look, looked reasonable, but I would be worried people take it for normative

MikeS: In addition there is a section on common content models, phrase and prose content matches block and inline content. Then definitions of sets of common attributes. Similar to HTML 4 draft and other markup specs.
... Last part deals with ARIA markup, attribute sets, enumerated values for ARIA attributes. Semantics undefined as they are in the ARIA spec. Then exhaustive list of name character references.

MM: Test kit being build.

MikeS: Not a schema?

MM: Its a grammar to build a parser.

MikeS: Interesting

MM: I will ask him to join WG.

MS: Will you have more info next week?

MM: Yes

Adrian: Do you have a view as to how having this doc changes what the HTML 5 spec is/does?

MS: Right now as far as content models and syntax description. This matches what is in the HTML 5 draft. We want to keep things that way. We need to decide that the current part of semantics, content models etc should be kept there. We need to keep them in sync. As different docs have diff editors there may not always be agreement.
... We want this to be normative. If we were to go forward with a separate normative markup-language spec, there can only be one spec so it would necessarily need to supersede anything else.

Adrian: This looks like a good start. In terms of a descriptive doc that talks about the language and not its use. However, how practical is this? How much of the text has been taken from the HTML 5 draft?

MS: This spec should not have a lot of non-normative content.
... It should not describe rendering behaviors normatively, or have too much description of rendering behavior etc. many say the current draft conflates authoring and rendering domains. These are separate so there is confusion. I like to have the markup spec not do this anymore. Separate some of the under the hood stuff from the user manual aspect. Want to see the spec defined as an abstract language without processing assumptions.

Adrian: That is a good goal.

<DanC> (trying to construct a proof in my head that the language defined in Mike's draft is smaller than the language in Hixie's draft; hmm... don't think there is one... I think it's not actually a theorem. I think there are counter-examples)

Joshue thinks this may make it easier to understand for all concerned.

DanC: Its not smaller than the language Hixie defines as conformant.

DanC: In that docs conforming to his spec is conforming to yours.

MS: It is.

DanC: I don't think so.

<DanC> (other way around)

MS: You are right.

<DanC> DanC: e.g. documents that misuse headings, cite, etc. are prohibited by the HTML 5 spec

MS: Discusses schemas, parsing of schemas, attribute model and pattern definitions. RelaxNG etc.
... Programmatic extracts/additions of certain content via Schematron. Josh unable to parse some statements.

Cynthia: I am curious why this is done that way?

<DanC> (I think having feedback between validation tools and the spec is good... though this is something of an extreme approach)

MS: It is circular. Not ideal. Changes to the spec will go other way, or not be one way from validator to the spec. If changes are made the assertions that validator.nu are making will have to be changed to match the spec. At his point they are one way.

Cynthia: It is reasonable to do this in order to get the spec out.

MS: Its about having a formal description of the language. Formalisms are currently prose descriptions in order to not lock people who write a conformance checker. High level language used in order to design a tool around it loosely.

<DanC> (publishing the schema as a note is an interesting idea.)

MS: Hixie feels there should not a normative schema for the language. Other builders have a disincentive to build anything. All of these normative schemas for the language seemed to stop others from developing their own. We want to avoid this, having only one tool.

Cynthia: Yes, some need this behind a firewall.

MS: This can be done and works well.

Cynthia: We don’t want to give advantage to one set of schemas.

MM: When developing a formalism for HTML, we can build a grammar, define constraints etc. It depends on what you are trying to do.
... Grammar needs to be correct. Stuff taken from different namespaces can be dealt with. Others have more rigorous purposes, may not be public facing. The grammar needs to be examined to be a more liberal version that conformance checkers want to use, then good stuff.

MS: Existing validators, and HTML 4. XHTML 1.0 and 1.1 (DTD based validation tools)

MM: When you produce a DTD, the doc that accompanies it is produced alongside it. There are better formalisms to do this etc

MS: I understand. Validator.nu is doing a lot more that just conformance checking.

MM: You claim that it does that is false.

<takkaria> you have to be very very careful that people don't start trying to consume HTML via a grammar rather than an implementation of the parsing algorithm

MS: I concede that, however when a decent schema is available, validation against a schema etc there are more sophisticated tools. But the problem is that many see that passing the validator is perceived as meaning their content is fit for purpose. Schema checking alone does not always mean your doc can be processed the way you want it to be.

MM: Again this is false.

MS: I hear what you are saying. Other comments?

Cynthia: This is a good idea. It will be helpful.

MS: I think to have the Authoring guide as a way to make it clear to help them have their docs work on the web. It also needs to cover the DOM interface for scripting purposes. Real world use cases etc. This will keep the spec minimal. remove informative stuff into the authoring guide etc

MM: Then call it something else.

MS: No

Cynthia: It could have subtitle?

MS: We have talked to developers and they want this.


MM: How about a browsers guide, developers guide etc?

MS: We need a normative guide for browsers..

MM: You can't have that.
... I am not understanding this.

DanC: You said this was a spec for how UAs behave.

MM: Strong objection


<MikeSmith> Joshue: some document that is specifically for authors, that cuts out a lot of the under-the-hood stuff is in principle a good idea

MM: I am going to make this an issue.

<DanC> issue-61?

<trackbot> ISSUE-61 -- Conformance depends on author's intent -- RAISED

<trackbot> http://www.w3.org/html/wg/tracker/issues/61

<pimpbot> Title: ISSUE-61 - HTML Issue Tracking Tracker (at www.w3.org)

<DanC> maybe that's not so close to what Murray wanted on the issues list after all

<DanC> action-77?

<trackbot> ACTION-77 -- Michael(tm) Smith to lead HTML WG to response to TAG discussion and report back to TAG -- due 2008-10-30 -- OPEN

<trackbot> http://www.w3.org/html/wg/tracker/actions/77

<pimpbot> Title: ACTION-77 - HTML Issue Tracking Tracker (at www.w3.org)

MS: I did want to talk to the TAG list about this. Let them know we have followed up on the discussion. I have an item to do this. This should take place on the public HTML list.

MM: I don't follow

MS: The action item is complete.

<DanC> ISSUE-59?

<trackbot> ISSUE-59 -- Should the HTML WG produce a separate document that is a normative language reference and if so what are the requirements -- RAISED

<trackbot> http://www.w3.org/html/wg/tracker/issues/59

<pimpbot> Title: ISSUE-59 - HTML Issue Tracking Tracker (at www.w3.org)

<DanC> (maybe that's closer)

MS: Lets take the rest of the discussion to public HTML.



<pimpbot> Joshue: Huh?

<DanC> (just briefly, who has the ball on headers?)

<DanC> (the actions listed in http://www.w3.org/html/wg/tracker/issues/20 seem stale. )

<pimpbot> Title: ISSUE-20 - HTML Issue Tracking Tracker (at www.w3.org)

<MikeSmith> Joshue: we are talking with PF about @headers and discussing how to move this along a little farther

<DanC> (hmm... so it sounds like anybody/somebody/nobody has the ball.)

<MikeSmith> ACTION: Joshue to prepare status report on @headers discussion by next week [recorded in http://www.w3.org/2008/11/13-html-wg-minutes.html#action01]

<trackbot> Created ACTION-84 - Prepare status report on @headers discussion by next week [on Joshue O Connor - due 2008-11-20].

waves bye

Next meeting

<MikeSmith> we will have the telcon at the regular time next week, probably with ChrisWilson chairing

<MikeSmith> [adjourned]

Summary of Action Items

[NEW] ACTION: Joshue to prepare status report on @headers discussion by next week [recorded in http://www.w3.org/2008/11/13-html-wg-minutes.html#action01]