7848 2009-10-08 17:54:42 +0000 MathML version used in HTML5 2010-10-04 14:56:02 +0000 1 1 1 Unclassified HTML WG pre-LC1 HTML5 spec (editor: Ian Hickson) unspecified PC Windows NT RESOLVED WORKSFORME NE P2 normal --- 1 pcotton ian davidc ian julian.reschke mike public-html-admin public-html-wg-issue-tracking public-html-bugzilla oldest_to_newest 28137 0 pcotton 2009-10-08 17:54:42 +0000 Section 4.8.15 MathML [1] defines the use of MathML in HTML5. Although this section does not directly use the [MathML] normative reference at least one other section does. The Normative References for HTML5 include a [MathML] reference to MathML 2.0 [2]. Although the [MathML] hyperlink is to the generic MathML location [2] I am assuming it references MathML 2.0 since the reference also includes the date "Oct 2003". The W3C is currently working on Version 3.0 of MathML and this specification is currently at Last Call [3]. Q1: Should HTML5 reference MathML Version 3.0 since it will likely to be a Recommendation before HTML5 becomes a Recommendation? I believe this question can be resolved as part of the HTML5 Last Call period. Note: If the HTML5 spec wants to reference the most recent edition of MathML 2.0 then I would recommend that the hyperlink be changed to: http://www.w3.org/TR/MathML2/ /paulc [1] http://www.w3.org/TR/html5/the-canvas-element.html#mathml [2] http://www.w3.org/TR/MathML/ [3] http://www.w3.org/TR/2009/WD-MathML3-20090924/ 28161 1 davidc 2009-10-09 11:18:59 +0000 As a personal comment, I think that it would be most desirable for html5 to reference mathml3 (or simply the latest version of mathml) mathml3 is being developed along a similar timeline as html5, and any issues that help integration will be addressed on the mathml side in mathml3, mathml2 being essentially frozen. To give one specific example; html5 draft currently includes all the "mathml" entity definitions. these are the definitions in mathml3 not mathml2. mathml3/html5 definitions take advantage of recent changes to unicode (and also fix some anomalies in the mathml2 ones). MathML3 also includes specific support for directionality (via an html-like dir attribute) which will be important for html+mathml documents in some locales. David Carlisle (MathML co-editor, but speaking for myself) 28559 2 ian 2009-10-20 10:05:37 +0000 The spec doesn't intentionally reference a particular version; the date is just the date of the spec at the time the references section was written. It doesn't mean anything. The HTML5 spec doesn't require any particular version of MathML be implemented. It just defines how you construct the DOM with nodes from the MathML namespace; whether those nodes do anything or not is not discussed. UAs can implement MathML or not, that's an issue between the UA and the MathML spec. So there doesn't seem to be anything to change here. I'll update the date of the reference to MathML at some point after the date of the document at the end of that URL changes, at the same time as I update many of the other references. (You'll notice the spec has references to drafts I edit myself, all of which have old dates that not longer match the latest versions.) 28568 3 davidc 2009-10-20 10:41:06 +0000 (In reply to comment #2) > The spec doesn't intentionally reference a particular version; the date is just > the date of the spec at the time the references section was written. That's fine (the main thing to avoid is that it should reference (or be seen to be referencing) _just_ mathml2, thus banning mathml3. I'd have thought that the thing to do given your stated policy is to reference the "latest version" uri that is http://www.w3.org/tr/mathml which is currently MathML2 (rather than the MathML3 draft) but will become mathml3 once that's a rec. I see that's what you do in the current HTML5 draft. So the only question is whether the following text should say "latest version of mathml" or whether it should say the version that you looked at. Currently it uses the latter style so lists the mathml2 editors and the date "october 2003". This is a general question about the style used in the html5 references section, so I'm happy to leave that to the discretion of the html5 chairs and editor. If the MathML reference uses the generic URI http://www.w3.org/TR/MathML/ and whatever wording convention used in the HTML5 references section for generic references, I don't think there is any MathML-specific issue. David 28578 4 ian 2009-10-20 11:42:19 +0000 The only place where MathML is explicitly and normatively referenced is where it says: # The semantics of MathML elements are defined by the MathML # specification and other relevant specifications. [MATHML] ...which seems pretty clearly version-agnostic to me. (I don't want to explicitly say "the latest version" because if, say, the MathML group gets overrun by crazy people and MathML4 is a big mess, it should be fine for UAs and authors to continue supporting MathML3 and ignore MathML4.)