W3C home > Mailing lists > Public > www-math@w3.org > July 2006

Re: Math on the web without MathML (CSS 2.1 rendering for HTML and XML)

From: <juanrgonzaleza@canonicalscience.com>
Date: Fri, 14 Jul 2006 08:03:21 -0700 (PDT)
Message-ID: <3091.>
To: <www-math@w3.org>

Neil Soiffer said:
> The difference is that Prof. Hutchinson adds the remark that some of
> these are part of the standard, and some are not.

And contradicted your previous

> It is quite likely that whatever problems
> Prof. Hutchinson pointed out are those of the renderer he was showing,
> not MathML.

> I also note that you choose to ignore his comment that the
> rendering problems are minor.

The fact is that there exist problems, as noted by most people using
MathML (even if w3c propaganda, for example recent one at the 2006 Charter
list, omits say us). The importance of each problem is a subjective point,
and I am not really interested in that kind of debate. What can be minor
for ones is major for others.

> All renderers have problems, as do all browsers, all CSS
> implementations, etc.  That doesn't mean the standards on which they are
> based are necessarily flawed, nor does it mean that problems with the
> current standard mean the standard is bad.

Sometimes does not mean, others it mean. For instance, unfriendly CSS and
DOM features of p-MathML are one of causes of the lack of interest by
browser developers. The first mathematical markup HTML3 was fatally
flawed, with lot of errors, the second attempt (Wolfram) was also
rejected, the first MathML spec contained a number of very important
flaws, and MathML 2 corrected several mistakes, e.g. trouble with binding
and all that deprecated stuff...

The use of presentation markup for online documents is one of biggest
mistakes I saw in my life. Whereas rest of community did an effort to
abandon <font>, <b>, <i>, <center> and all that, the MathML WG introduced
redundant “<font>” tags for mathematics.

> However, you stand pretty much alone in failing to recognize that MathML
> has a number of strong points and that CSS, as currently defined, is not
> well-suited to rendering math.

Just some days ago David Carlisle recognized that via absolute positioning
in CSS you can render math with high quality. An advanced CSS approach or
an mixed CSS-SVG approach can render mathematics you cannot render with

In canonical science today, I see that CSS renders equals or even better
than MathML for examples I tried. For instance, the simple fraction
renders nice in CSS and distort in MathML.

> You often cite XML Maiden as an example of what can be done.  However,
> it's rendering (which only works in Opera), would not be considered
> acceptable by most people who author math.

The comment about Opera is false (did you notice that in the comparative
CSS MathML I am using Mozilla Firefox?).

Curiously I offered examples of math rendered with MathML and with CSS and
most of people choosed the visual CSS rendering. I will provide more and
more examples.

Moreover, most of people chose CSS because prefer aditional advantages of
CSS: not requirement for special tools, not special DTDs, not special
MIMES, not extensions, not special browers, no special fonts, no special
plugins, no special DOM, no special style markup, no redundancies...

And do not forget the limitations on accesibility, wrong structural
markup, and other limitations of MathML + tools.

> Because it doesn't tag
> variables, numbers, or operators, it doesn't render variables in italics
> when appropriate, nor can it use appropriate spacing between operators.
> Furthermore, despite your assertion that CSS (and XML-Maiden?) can draw
> large curved "brackets", XML Maiden only draws vertical or horizontal
> lines, so square roots look poor, parens are drawn as brackets,
> under/over braces are drawn like rotated brackets, integrals don't
> stretch, etc (at least based upon the examples linked from
> http://www.geocities.com/csssite/index.xml.

It is nice to see your claim on spliting spec from implementations whereas
being unable to do the same CSS work. In short, all of above is false or

> These are more than minor
> problems to most people.

Instead claiming that people will like, dislike. I prefer offer to people
alternative approach not using MathML. And people will chose the option
they prefer.

> Despite your claims that CSS can do a good job of rendering math, you've
> yet show an example that it can.  In the US, there is a saying, "put up
> or shut up".

It is trivial to see that CSS _can_ do an excellent work, even Carlisle
already recognized that. Moreover, I already provided simple examples
where rendering via MathML is poor than with CSS in Mozilla Firefox. More
and more examples will be added in future. Most of people would wait a
completely inadequate rendering via CSS and perfect via MathML, but they
are nearly equaled. Some stuff is better done with CSS and other stuff is
better done with MathML.

> David Carlisle provided a simple example that he feels can
> not be rendered well in CSS. Please show that you know what you are
> taking about by showing how to do this in CSS so that we too can
> understand how it is done.

The reply to his question had been addressed many time ago. And I cited
here the link to examples, specs, annotated css...

Juan R.

Received on Friday, 14 July 2006 15:03:32 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:27:37 UTC