W3C home > Mailing lists > Public > www-math@w3.org > September 2010

Re: correct DOCTYPES to use on MathML 3.0 documents

From: William F Hammond <hammond@csc.albany.edu>
Date: Wed, 22 Sep 2010 17:57:57 -0400
To: W3C MathML Discussion <www-math@w3.org>
Message-ID: <i7eicl30gq.fsf@hilbert.math.albany.edu>
David Carlisle <davidc@nag.co.uk> writes:

>> Without a DOCTYPE how does the validator know what recommendation to
>> validate to?
>
> If you validate locally before putting the file on the server then you
> don't need to put a doctype on the file, which imposes a cost on every
> application reading the file, that it has to load the dtd.

Given that big search engines still ignore application/xhtml+xml
content, I expect that many math content providers will soon be
switching to the text/html serialization of html5 with math.

Some of the html5 folk seem to have strong aversions to (1) dtds and
(2) sgml.  They've gone out of their way to proclaim that the
text/html serialization of html5 "is not sgml".  At the very least, it
seems clear to me that, with a suitable sgml declaration, one will be
able to create an understandable case-sensitive sgml document type for
html5+mathml (with all html names lower space) whose document
instances will (1) pass for the text/html serialization of html5 and
(2) admit easy transformation to xhtml5+mathml.

That said, I should hasten to point out that opensp needs a little
maintenance in regard to the astral planes of unicode and, in
addition, I would very much like to see a command line switch for
onsgmls to request either downcasing or upcasing in the esis for a
case-insensitive document type.

>> The W3C validator does not YET correctly validate HTML5 documents that
>> use MathML.

My experience there is basically favorable.

> hmm w3c's validator didn't like many of the attributes for some
> reason: they look OK to me,

Specifically, I think there's some relaxng code that wants
all lengths in mathml attribute values to be integers.

> The HTML5 editor doesn't really believe in version numbers:-) So The
> HTML5 spec is a bit vague as to exactly which version of mathml can
> be used. However I believe the HTML5 spec will point to the "latest
> version" of mathml, do hopefully MathML3 in a week or so. Of course
> actual implementations may, as ever, take a while to catch up, and
> similarly the online validators may take a while to integrate the
> mathml3 schemas. (validator.nu is relaxng based)

It's not clearly stated, but I think the philosophy is that as time
evolves, validator.w3 will attempt to enforce what is OK with the
major browsers of the moment.  How else, for example, should one
understand the spec's misappropriated markup "<!DOCTYPE html>" ?
As long as backward compatibility remains a feature, it might
actually work.

                                    -- Bill
Received on Wednesday, 22 September 2010 21:58:51 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 22 September 2010 21:58:51 GMT