Re: MathML-in-HTML5


> But what I gather so far is that 
> IE+MathPlayer only supports prefixed tags, with the prefix declared in 
> the <html>. (If I understand it correctly, it doesn't even work with 
> <math xmlns="mathml-namespace">.

unfortunately this is true, although you can have everything _except_
the outer math element being prefixed
see the third example in
which probably had this use in mind...

The outer level math tag has to be prefixed as that is the hook that
allows the rendering behaviour to kick in.

> Hence, either I go off and add support for the plain <math>...</math>, 
> which will only work in Mozilla out-of-the-box and wouldn't be quite an 
> overture to IE+MathPlayer (without yet again content negotiation or some 
> JS trickeries by authors), or to get seamless interoperability, HTML5 
> may need a provision for prefixed tags (at least in the case of MathML), 
> and this is where we have got so far in the discussion.

> The starting patch that I indicated in the opening already supports 
> prefixed tags, i.e., <html xmlns:m="mathml-namespace">, with
> <m:math>...</m:math> in the document,
> as well as the typical
> <math xmlns="mathml-namespace">...</math>

If we're experimenting with mathml in html, I think that is what you
should do. It's true that  the second form doesn't currently work in IE
but it's not beyond possibility that it could be made to work in IE 7.x 
(or 8.x or whatever) and it _allows_ a form that works on both systems,
with the prefix, which is good.

As for the general discussion of whethe XML syntax would be enforced, I
think the most important point is that XML syntax (including namespaces)
be _allowed_  otherwise as others have pointed out, MathML authoring
systems have a real problem: DOM level compatibility of the parsed tree
doesn't really help if MathML is moving from MSWord to maple to
mathematica to the Web as _serialised_  MathML document fragments.

If implementation-wise MathML in HTML is going to be parsed by the HTML
parser rather than an XML one then it's presumably not feasible to
demand that all XML parsing rules are enforced. This is something that
personally I could live with especially if it were described as "error
recovery" rather than a different serialisation syntax. It would also be
good to have a specified algorithm for this (a la html5) rather than
just leaving it up to the browser (at least at the specification level,
even if implementations use existing parser stratgeies for the immediate


Received on Wednesday, 11 October 2006 15:50:53 UTC