W3C home > Mailing lists > Public > www-math@w3.org > May 2004

Re: MathML and CSS

From: White Lynx <whitelynx@operamail.com>
Date: Sat, 29 May 2004 16:38:55 +0400
To: "Paul Libbrecht" <paul@activemath.org>
Cc: www-math@w3.org
Message-Id: <20040529123855.8B03423EE52@ws5-4.us4.outblaze.com>

> > We gone slightly offtopic, lets turn back to CSS3 math module.
> Here's my preference about the existance of CSS3-Math:
> I understand there may be more chances or it may be better for some 
> applications to realize mathematical presentation in (boosted) CSS than 
> in such a specialized language as MathML.
Yep, some browsers like Opera and Safari has good enough CSS
support to afford rendering of math expressions, so it will be
reasonable to reuse existing abilities of CSS rendering engine 
instead of reinventing wheel and creating 'ad hoc' MathML rendering engine.
Note that Mozilla - browser with native MathML support, actually
transforms presentational MathML into sequence of CSS frames 
and then transfers it to CSS rendering engine (Gecko) for further rendering.
So as far as CSS is the only part of web technology responsible for
formatting issues and MathML claims to be 
'the only correct way of putting math on web' I think it is time to think
about merging MathML into XML + CSS rendering framework.

> It would be "yet another luxury", and we would welcome it with pleasure!
I don't think it is luxury. It is just correct way of doing things.
Markup must focus on structure, while presentation is better controlled
via style language. In HTML transition from presentational tag soup
to separation of structure and formatting in HTML + CSS framework
happened long time ago, Tex to LateX transmission is also similar process
when formatting related junk was moved to external macro packages 
and LateX focused on structure only. Another examples are XML + XSL and
SGML + DSSSL. This way allows to use power of rendering engine in very
flexible way for multiple purposes. 

> But now, I also understand that a CSS3-Math module looks to be a way by 
> which browsers who don't implement MathML-presentation would not need 
> to do so "because they implement the CSS3-Math module".

Exactly. Even without having CSS3 math module one can see how it works.
See for instance this demo page:
http://geocities.com/csssite/mathml2.xml
MathML supporting browsers like Mozilla can just rely on native support
why Opera 7 renders exactly the same page exploiting its CSS2.1 implementation.
Page covers essential part of presentational MathML
(msub, msup, msubsup, munder, munderover, mfrac, mtable) but style sheet 
used there is not comprehensive as the purpose was just to show possibility
rather then to create comprehensive style sheet
(there is another version of the same page 
http://geocities.com/csssite/mathml2.xml
that does not trigger MathML support in browser and does not hide style sheet from Mozilla).
To make this approach work
better we lack ability to render radicals, lack some border styles, 
need better control over vertical alignment, white space processing,
content rearranging etc. So there are some technical issues that I think
can be addressed by forthcoming CSS3 in general and math module in particular.

> But to be able to declare such, at least in my opinion, means that such 
> a browser will have been working with a converter from a set MathML 
> samples into XHTML+CSS3(-Math). So that they should, probably with low 
> effort, be able to actually receive MathML-p.
This is one point. Another point is that browser with native MathML 
support should know how to apply CSS style sheets to MathML markup.
For instance should position of indices like msub, msup be changed
using CSS vertical-align property, relative positioning? 
MathML in any case should be integrated with existing web technology i.e. XML, CSS and DOM. Otherwise I see no point in using XML based format for mathematics, we could just have LateX and require browsers to implement its subset 
(if you don't reuse existing possibilities of CSS/DOM rendering engine then does it matter for browser whether it will be MathML, LateX or whatever else, as parcing markup is not a big issue). 

-- 
_____________________________________________________________
Web-based SMS services available at http://www.operamail.com.
From your mailbox to local or overseas cell phones.

Powered by Outblaze
Received on Saturday, 29 May 2004 08:55:21 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Saturday, 20 February 2010 06:12:56 GMT