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

David Carlisle said:
>
>
>
>> CSS techniques to render mathematics can be used for HTML, XML,
>> XHTML...
>
> yes but to what extent you can style any particular expression depends
> on the markup used. If you go <span>a+b</span> you can't really style
> the a b and + separately with css, nor can you ensure that the spacing
> is adequate.In order to do that you need a more mathml like markup with
> <span><span class="i">a</span><span class="o">+</span><span
> class="i">b</span></span>

Well that depend of the approach taken, the complexity of markup desired,
the quality waited, the target (educative K-12 site or an academic journal
on mathematics?)...

You can use

<span>
  <span class="i">a</span>
  <span class="o">+</span>
  <span class="i">b</span>
</span>

or

<span>a+b</span>

or

<span>a + b</span>

or

<span>
  <span class="omi" name=”a”/>
  <span class="omo" name=”plus”/>
  <span class="omi" name=”b”/>
</span>

...

Of course, something like <span>a+b</span> cannot be directly styled and
spaced, but you can do it, in an indirect way, via code. That is similar
to Itex $a+b$ or ASCIMath `a+b` inputs for MathML.

Something as <span>a + b</span> is good enough for mathematics on the web;
it is lighweight, can be authored by hand and machine, can be directly
styled with CSS (or XSL-FO) and can be styled as the full token case you
cited with asistance from XSLT or JS (or, of course, via browser
extensions).

> Which is one reason MathML is the way it is, to allow per-character
> styling and font changes as required in math typesetting.

This was adressed in the past. Let me introduce a bit of presentational
HTML for illustration

<p>
  <span class=”roman”>MathML</span>
  <span class=”roman”>is</span>
  <span class=”italic”>not</span>
  <span class=”roman”>popular</span>
  <span class=”roman”>on</span>
  <span class=”roman”>the</span>
  <span class=”roman”>web</span>
</p>

vs.

<p>MathML is <span class=”italic”>not</span> popular on the web</p>

> Just saying
> "use CSS" is meaningless without some indication of the target markup
> that CSS is styling.
>
>> However, prof. Ian Hutchinson (the father of one of the MathML tools)
>> cited as one of anyonances of MathML rendering at the beginning of
>> this year in a IAP talk about mathematics on the web
>
> So Ian is saying his personal preference differs from the usual
> conventions of mathematical typesetting. That's fine (so do some of
> mine) MathML allows those differences but the defaults follow the usual
> (in certain cultures) conventions.

I do not believe that he was talking in a personal way. Any case this
would be replied better by him.

Talking of usual conventions, how would you render (italic, roman)

F = F_0 exp(- delta t) sin(omega t)?

And what about Re, Sr, or Nu?

>
>> > Choice of fonts is again largely a feature of the renderer, not of
>> the markup and would obviously apply equally to any markup system.
>
>> And are annoyances or not as said in the talk?
>
> It's quite funny that you highlight non-use of TeX fonts as an annoyance
> since your earlier complaint about mozilla was that it (had the option
> of using) TeX fonts. In any case the solution is the same whether you
> use MathML or html for markup or anything else: use a mathematical font
> set designed to work better in a web browser than the cm fonts (which
> are very light) The mathematica fonts or (one day) stix fonts are
> possibilities as is the new math font that comes with the ms word 2007
> beta (if that's allowed by the licence it's eventually distributed
> with).

This is rather misleading and mixed several points in so any messy way I
see not need to repeat I (and others) said, whereas waiting to be
misunderstood again.

> So whether or not this is an annoyance has nothing to do with
> mathml.

I do not see the point to cite again; files, links, and data are there.

>
>> Sorry but again I would say you that one can render matrices in CSS 2
>> without joining fragments not rely on special fonts. Please visit some
>> of links I cited.
>
> I've visited some of your links and seen some examples with a+b and the
> occasional fraction, but nothng that would require any real mathematical
> typesetting notation. If your CSS solution is just scaling up single
> characters it's not really a solution, if it's not doing that, what is
> it doing?

No Carlisle. I cannot repeat the same again and again. Search the link i
cited to CSS techniques in this list days ago. Look also in the internet
for other approaches. The only CSS approach rendering brackets from
fragments I know is jsMath (and use images as complement). A way to render
fences in matrices is via CSS rounded borders technique.

Ah and examples in canonical science today were extracted from the MathML
2.0 specification. Therefore, the examples contained at the MathML 2.0
specification contain (in your own words ;-)
> nothing that would require any real mathematical
> typesetting notation.


>> Of course, you choosed a simple prototipical example where aligment is
>> good. Somewhat as fraction 3/2 renders nice in your CSS approach but
>> {x^2}/5 would render a bit distorted, true?
>
> Not really, alignment of the = sign with the surrounding text is
> independent of the rest of the expression, as is the fact that alignment
> of the matrices is unaffected by the surrounding brackets So these
> features are a feature of MathMl rendering independent of the simplicty
> of the expressions.

I was refering to fact that 1 2 3 all are of same high. Try again now with
1^3 in first matrix and 3_44 in second and you can see as the whole
rendering distorts. E.g. the 2 at the left side of the equation is not
longer aligned with the 2 at the right. The 2 at the left is not aligned
with the xxx before, looking really ugly for many TeXers.

Do you know I asked three new persons about the comparative rendering of
CSS with MathML and they found childish (sorry but was the word) your
comment on the position of the + sign, whereas were atonished of the ugly
rendering of the single fraction provided by MathML (that you omitted to
analize)?

> David


Juan R.

Center for CANONICAL |SCIENCE)

Received on Friday, 14 July 2006 14:17:08 UTC