RE: How to extend MathML?; was RE: LeftRightArrow equivalence operator?

Hello David,

Thanks for your reply.
Your comments steered me toward reading again about the fn element.
I see now that I did not have to edit the mathml.dtd,
as long as I'm willing to repeat fn elements.
You are right, I was after the convenience of a short form, empty element,
which I understand DOES require an edit,
but punts on any possibility of a standard rendering.
I see your point that XSL will fix this last (rendering) issue.

Thanks again,

> -----Original Message-----
> From: David Carlisle [mailto:davidc@nag.co.uk]
> Sent: Friday, October 15, 1999 5:59 PM
> To: devlinh@nwu.edu
> Cc: www-math@w3.org
> Subject: Re: How to extend MathML?; was RE: LeftRightArrow equivalence
> operator?
> If you just add a new element by editing the DTD then that may or may
> not be sufficient depending what you are doing with the markup.
> You have then made a mathml-like markup extended with the new element
> and if you are using it to pass mathematical information between two
> systems that understand that, then that may be all that is required.
> However a MathML system won't know what to do with your new element, it
> won't know what it means mathematically, and won't know how to render
> it. If it reads the DTD it will know that the element is supposed to be
> there, but that is all.
> One possible solution to this is to specify an XSL transform from your
> new markup to presentation MathML, then (once browsers have sufficiently
> good XSL and MathML support) the browser will know how to render the new
> element by converting it on the fly to presenttaion MathML. If your
> transform also includes semantic as well as presentation information
> (eg by mapping to a suitable semantics element construction) then 
> you should also be able to pass the semantics of the new element to
> a native MathMl system as well.
> > How do I sneak an operator in there without an edit?
> basically by using a ci to get the operator name and a suitable
> definition element or definitionURL attribute to give the semantics.
> If you want the convenience of an empty element short form as in the
> existing <sin/> etc, then something like the above is needed.
> The latest mozilla test code claims to have support for xsl and
> mathml so I'm hoping to be able to build that and try this out in the
> near future....
> Will let you know if it works....
> David