RE: The problems with namespaces in text/html

I second these concerns. MathML already contains markup for hyperlinks,
spacing, and bolding. They might not be ideal but, if not, it should be
taken up as part of MathML 3.0. And, as Jacques also notes, MathML must
live in other, non-HTML environments. Finally, math formatting has to
deal with spacing, fonts, character styles, etc. in ways that work for
math. While adding <b> inside <mtext>, for example, may seem to be
innocent enough, the potential complexity of interaction of such tags
with MathML's built-in formatting rules is scary. This is bad enough
already with CSS interacting with MathML.

Nesting languages such as HTML inside MathML inside HTML sounds
desirable, and is even robust from a language point of view if we
substitute XHTML for HTML. However, specifying how these languages are
to interact with each other is a nightmare. There are literally millions
of interactions that must be worked out if we are to both create
something useful and avoid yet another flavor of tag soup. 

BTW, this is how tag soup is made -- creating interesting and fun new
things by combining simple existing things without sorting out all the
details of how they interact.

Paul Topping

> -----Original Message-----
> From: 
> [] On Behalf 
> Of William F Hammond
> Sent: Monday, October 09, 2006 9:43 AM
> To:
> Cc:
> Subject: Re: The problems with namespaces in text/html
> Jacques Distler <> writes:
> > As to namespaces, looking at the output of itex2MML, there are two 
> > places where namespaces are used.
> >
> > 1) There's a
> >        <math xmlns=''>...</math>.
> >    The MathML elements themselves are un-prefixed.
> > 2) Hyperlinks in MathML are <mrow>s with XLink attributes:
> >        <mrow xmlns:xlink=""
> >            xlink:type="simple" xlink:href="..."> ... </mrow>
> >
> > Clearly, if presentational MathML elements are added to HTML5, the 
> > first can safely be ignored (as namespaces are currently ignored).
> Yes, the xmlns in math should be safely ignorable in HTML5.
> > The second is more of an issue. Substituting an (X)HTML <a> element 
> > seems likely to do nasty things to the layout.
> Regarding links from inside math: I think the most 
> interoperable approach is to allow HTML <a> elements inside 
> <mtext>.  Beyond that I would like to see the HTML elements 
> <a>, <em>, and <b> (or <strong>) incorporated in the 
> XHTML+MathML document type definition as elements that are 
> allowed in <mtext>.
> These things have actually been working in Mozilla when the 
> <mtext> element has an xmlns attribute flagging it for the 
> HTML namespace.
> But this namespace artifice has not been interoperable with 
> MathPlayer.  See 
>                                     -- Bill
> _______________________________________________
> dev-tech-mathml mailing list

Received on Monday, 9 October 2006 18:20:31 UTC