W3C home > Mailing lists > Public > www-math@w3.org > October 2006

Re: MathML-in-HTML5

From: David Carlisle <davidc@nag.co.uk>
Date: Wed, 11 Oct 2006 17:08:40 +0100
Message-Id: <200610111608.k9BG8eCV013750@edinburgh.nag.co.uk>
To: hammond@math.albany.edu
Cc: www-math@w3.org

> In application/xhtml+xml there seems to be more than one understanding
> among current user agents on the question (given the use of namespace
> prefixing) of whether or not inside an "m:math" element unprefixed
> subelements (at all depths) are mathml.

As far as the specification goes (which is, I think what Mozilla does)
the situation is clear: Elements in the MathML namespace (whether
prefixed or not) are MathML, elements that are not in the MathML
namespace (whether prefixed or not) are not MathML.

Mozilla actually implements a compound document format where mathml, xhtml
(and probably svg) can be nested in ways that are not individually
sanctioned by the respective specs but in ways that could be sanctioned
by schema languages designed for the purpose, such as
ISO/IEC 19757-4 NVDL (Namespace-based Validation Dispatching Language)

This doesn't work in IE+MathPlayer (and the Math expression, once
extracted, would not work in any other MathML application, as it is
MathML+XHTML, not just MathML).

This difference in behaviour though is due to implementing (or not) a
compound document format that allows the various languages to be
extended, it isn't a difference in understanding which elements are

I think actually that that is the way it should be.

Rather than start adding html-like elements such as <b> <i> to <mtext>
which might look natural in mathml-in-xhtml but would look distinctly
odd in mathml-in-xslfo or mathml-in-docbook, I think we should keep the
leaf elements in "pure" mathml as more or less just pcdata, but
encourage larger fromats to use nvdl (or the W3C's CDI compound document
by Inclusion format's once that has been specified) to allow richer
markup at mtext and perhaps other token elements. This then makes it
clear that the richer markup is not "mathml" and there is no obligation
on CA systmems, etc to implement it, yet it does allow (at least the
specification of) documents with commutative diagrams marked up as 
xhtml (for the document) containing
svg  (for the diagram arcs) containing
mathml (for the diagram nodes) containing
xhtml (for marked up text in the formulae)

Received on Wednesday, 11 October 2006 16:09:06 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:27:38 UTC