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

David Carlisle said:
>
> Sorry but you appear to have completely missed the point as to why
> Mozilla uses special math fonts, that the fonts with math characters
> in unicode position simply don't (yet) exist.

Sorry but points i), ii), and iii) still hold. I do not remember to me
doing an interpretation of *why* Mozilla uses special fonts, just
reviewing the problems derived from a layout engine _based_ in special
fonts may be dowloaded and installed at the client side. If you want
another day we can talk about why Mozilla rendering engine is polluted
with CM metrics and why that is not good for a browser, but that will be
not today ;-)

> STIX (and other projects) may
> change that situation one day, but not today.

Do you read point iv)?

> Your examples don't go
> much further than a+b and an integral sign: yes you can find those
> symbols in standard text fonts but that is hardly enough to cover the
> range of mathematics that one would need.

First I would say is that examples are extracted (or based) from the w3c
MathML 2.0 specification. If they are sufficient for _illustration_ in a
MathML 2.0 spec also could be in a "HTML-Math" spec, no?

>From the source
[http://lists.w3.org/Archives/Public/www-math/2006Jul/0002.html]

<blockquote>
Of course, more complex formulae could be also rendered with CSS. For
example nested fractions are not limited to 2 levels. I have limited the
complexity of formulae simply because i am very busy those days.
</blockquote>

> Your screendumps show that your mozilla was incorrectly configured for
> mathml (no math fonts) so its unsurprising that it does not render
> mathml correctly.

I wonder in what part of the description I said that I had used the
specific fonts! Do not forget that the objective is we can compare both
systems (CSS and MathML) in the *same* conditions. I will repeat the
comparison with specific fonts installed only when I can use the
specific fonts in a CSS approach, apples and apples, ok?

If, however, you insist on repeating the comparison with special fonts
installed or special plugins at the MathML side then you would claim for
a non-balanced comparison, but then I could freely begin a more general
comparison, with differente requirements, using only HTML 4.0, or the
requirement of no-namespaces, or using Opera or Safari browsers for
certain comunities, or adding restriction to certain MIME types for
people who cannot access to server configurations and so on and then CSS
would be the *only* posibility to render mathematics online. Therefore,
you (as one of MathML authors) would be grateful that comparison I did
was being completely balanced.

Could you, please, say us what examples are rendered via CSS and what
are with MathML?

Moreover, note also that experiment was not for proving “mathml does not
render correctly” if not for proving that a CSS approach can be
indistinguishable.

Also would be interesting to know how the usage of specific fonts into
the Mozilla layout engine would affect to the rendering via CSS. Are you
claiming that the nested fraction or the root via CSS will look
incorrectly once specific fonts will be installed?

> Incidentally in your discussion on the whatwg list you
> gave a long list of pointers to the mathml test suite and implied that
> they didn't work well in firefox. You didn't say what you thought was
> wrong, but all the ones I looked at looked OK in firefox, perhaps your
> problem was again just an incorrect local installation?

Whereas large fences, roots, and others could be improved thanks to
usage of specific fonts (with problems of that in mind) I see difficult
that other failed tests I cited can be passed.

In a next round, I will submit screenshots from the MathML test suite on
examples are not correctly rendered via native MathML browser but are
when using a CSS approach.

> It's absolutely no surprise that if you have fonts with the symbols
> that you need, and don't have any complicated requirements such as
> large brackets or subscripts on large base expressions that you can go
> a long way with css and/or css aided by a bit of javascript.

Large brackets or subscripts on large base expressions could be done for
*any* font via CSS.

> For example there
> is my effort from 2001 (I think)
> http://www.dcarlisle.demon.co.uk/mathmlcss/
> and others have done similar things, either using MathML markup to
> base the CSS or using a different markup.  But in all cases while the
> result is "good enough" if you have very simple formulae, and your
> typesetting requirements are not too high, the mechanisms are really
> only useful as a fallback for cases where no mathematical rendering is
> possible.

Difficult to adivine that you mean by "good enough" and by "not too
high", but analizing a bit your css approach (e.g. the fourth rule on
mfrac) I suspect that you mean "I can render beatiful fractions via pure
CSS if _and only if_ numerator and denominators are *exactly* of the
same height" and further limitations of that kind.

Well, sorry but in this approach you can render beatiful mathematical
formulae that either your css approach cannot or it renders incorrectly.
I encourage you again to say us what of screnshoots were rendered using
CSS and what using p-MathML 2.0.

Could you cite examples of mathematical formulae can be rendered via
MathML cannot be via CSS, please?


Juan R.

Center for CANONICAL |SCIENCE)

Received on Wednesday, 5 July 2006 09:26:54 UTC