Re: Exploring new vocabularies for HTML

> Semantic content isn't an end-goal of the HTML5 work.

There are differnet levels of semantics, and one can always draw the
line somewere but banning <semantics> because of teh (valid) concern
that someone may edit the object without editing teh annotation is
exactly the same  as banning the alt attribute on img because someone
may edit the dom form script and get the image and its description out of
sync. 

> Really? I was assured that presentational MathML was quite accessible.
It's quite accessibe for a prsentation format, with hooks tio
distinguish various invisible operators that are spoken differently for
example, but it won't tell you whetehr x^{-1} is an inverse function or
a reciprocal or a degree -1 cohomology group or....


> but how do people deal with that in LaTeX today? And if Content MathML
> is the solution here,  why do we need Presentational MathML?
LaTeX authors can (and often but not always do) define new commands so
that the document level markup for distinct functions is distinct,
even if their presentation is the same.

> And if Content MathML is the solution here, 
> why do we need Presentational MathML?

Because you need a language that can express mathematical layout so that
you can define the rendering of _any_ term, In a context of a piec of
softwar that isn't specifcally mathematical it is perhaps unreasonable
for it to build n knowledge of all of content mathml, but even if it
did, you would still need presentation mathml to specify rendeing of
more specialist forms not included in the core mathml set, or (commonly)
to force a partcular presentation form. The sematics of mathematics is
universal, the presentation of each semantic construct is subject to
wide variation.



> How so? Presentational MathML is as copy-pastable as content MathML
sorry I can't guess what you mean here.

> To most authors, it is, since most authors just want to show the equations 
> and be done with it.

so you say, disregarding decades of development of markup for
mathematical documents.



> There's over 80 different presentational MathML elements in MathML3. 

I can't say I've ever really counted, I'm not sure how you get 80
though, I just did this and got about half that:

$ pwd
/c/w3c/WWW/Math/RelaxNG/mathml3

$ grep element mathml3-presentation.rnc mathml3-common.rnc | wc -l
42


Douglas Adams would be pleased:-)


> Chapter 3 doesn't seem to list the actual content models.

we can possibly make it clearer (mathll3 is in WD stage) but in most
cases the only difference between the content models is the number of
arguments.The only thing you need to know aboyt mfac is that it takes
two children, you don't need to list which they are.
there are some special cases like table elements mtd that can only
appear in mtr etc, but they are called out separately.

David

________________________________________________________________________
The Numerical Algorithms Group Ltd is a company registered in England
and Wales with company number 1249803. The registered office is:
Wilkinson House, Jordan Hill Road, Oxford OX2 8DR, United Kingdom.

This e-mail has been scanned for all viruses by Star. The service is
powered by MessageLabs. 
________________________________________________________________________

Received on Sunday, 30 March 2008 09:55:12 UTC