Re: MathML-in-HTML5

> Well, MSIE does not deal with MathML in any form 

This isn't really the case. It's true that if you are using
IE+MathPlayer then the math rendering is being done by an application
produced by Design Science rather than Microsoft, but would you say that
"Opera doesn't deal with applets in any form" just because executing an
applet requires a JDK from sun (or some other Java virtual machine)?
In practice, what a user experiences as "the browser" might be any
number of applications from multiple companies.

IE, for all it's faults, has a rather sensible way of dealing with
extending HTML with XML languages (MathML, SVG, ...). Mozilla, leveraging
off its open source basis, requires the core engine to be extended to
support these languages. IE on the other hand exposes an API that
allows a particular rendering engine to register itself to render
specific XML namespaces. The actual implementation of the idea in IE
unfortunately has some flaws in that it requires explict COM ids being
declared in an object element in the page, and requires a non standard
namespace declaration syntax, However these flaws can be hidden from the
user as long as some guidelines are followed.

> and I am not against
> embededing MathML in environments other then XML (you can embed it in
> LaTeX if you want)

Yes, I once implemented an XML parser in TeX, with that in mind...
http://www.google.co.uk/search?q=xmltex

>  but I am against turning it into tagsoup which is
> different issue.

I agree, and this is one of the merits of the IE approach, that I hope
would be seriously considered for mozilla. It isn't necessary for
HTML <4+n> to specify "html-variants" of the various XML languages, _any_
_well formed_ XML fragments can be included, so long as you register the
namespace with the application to bind it to a rendering component.  In
IE that binding happens in the html page itself, but it would be better
done at the browser level.

I think that if a simpler linear input form without so much element
markup overhead is required, (and almost certainly it is required)
then something more like 
http://www1.chapman.edu/~jipsen/mathml/asciimath.html
is what is wanted (ie, no element markup at all). Asciimath as published
at the above address does the expansion to MathML on the client (so it
is the tex-like syntax that would be served) but an alternative would be
to do the expansions on the server, which is essentially the wiki
approach, allowing you to write 1+x^2 as shorthand for
<mn>1</mn><mo>+</mo>...
just as
 * zzz
is shorthand for <ul><li>zzz... in many wiki variants.

David

Received on Tuesday, 3 October 2006 15:30:06 UTC