Re: MathML vs HTML math, vs ???

Dear Dr. Landrum,

Even though it is a bit late in your discussion, I can't resist wading
in.  You wrote:

> Of course, the problem with all of these things, is that they are not
> part of the HTML standard, so plugins are required.  This inevitably
> limits the portability of whatever is used.  For example, I am sitting
> at an IBM workstation running AIX.  This means that I cannot use most
> plugins.  Even IBM's TechExplorer doesn't work on my machine.
> 
> I realize that this problem isn't even remotely the fault of the working
> group, but is more a consequence of the economic factors driving the 
> development of the HTML specification.  

To me, this is the heart of the issue.  It is indeed economic factors
that drive the development of browsers, and hence HTML.  We can debate
whether the scientific community is better served by a low-level
markup like MathML, or a high level markup like TeX, but the simple
economic fact is that the major browser vendors will *not* support for
something like TeX.  

If we had gone down that road, we would have inevitably marginalized
the scientific community yet again -- no doubt we could have had
wonderful math support through specialty browsers, plug-ins, applets,
etc, just as TeX is wonderful for scientific papers.  But just as the
rest of the world uses Word, etc. for word processing, and the math
people always have to go to great lengths to have TeX, the rest of
the world would use ordinary browsers, and the scientists would have
to wrangle with installing special Web browsing tools.

The key point about MathML, which I don't believe has been raised yet
in this discussion, is that it is an application of XML, and XML is
the HTML extension mechanism that the browser vendors have announced
they *will* support.  Obviously, they are not supporting XML on account
of math, but the point is, they are going to support it.  By casting
MathML as and XML application, we get to go along for the ride.

According to our best understanding, using XML (and XSL, the style
sheet language that goes along with it) it will be possible to have
most if not all MathML rendered natively in browsers in a year or two.
There are no guarantees about how this will play out, of course, but
according to the stated design goals of XML/XSL, early prototypes, and
the assurances of the people most involved in XML and XSL at W3C,
browser support *should* be adequate to render most common MathML
expressions directly in a browser.

So, your observation that MathML is about communicating with machines
is at least partially correct; we would very much like to communicate
with Web browsers in the future.

------------------------------

That said, let me add a postscript as an individual, rather than as a
member of the HTML-Math WG.  In your first message, you pointed out
that you like to edit by hand using emacs.  So do I.  But your analogy
was at the wrong level.  I can and have edited PostScript with emacs,
but I don't usually.  I don't think I have ever directly hacked a DVI
file.  When you edit LaTeX, you are then obliged to run a processor on
it, to put it in machine-readable form.  This is the model I
personally see for MathML as well.

We have developed a suite of MathML tools, WebEQ [1], at the Geometry
Center at the University of Minnesota.  One of its modes operations is
exactly that; authors prepare HTML code with embedded LaTeX-like
markup, and then run a processor which expands it into MathML (and/or
images, and/or applets, so you can actually view your document on
today's browsers, instead of waiting two years.)

[1] http://www.geom.umn.edu/locate/WebEQ/

The AMS, SIAM and the Geometry Center are also funding the development
of a latex2mathml translator, an enhancement of the current latex2html
translator.  Work is well underway.  

Design Science, makers of the MathType equation editor and the math
editor in Microsoft Word, will be export MathML in the next release of
their product later this winter.  Waterloo Maple is also working on
MathML support.  You have already heard about IBM TechExplorer, HP
EzMath and the development work underway at Wolfram Research.

I think we will soon have a "critical mass" of inter-operable tools
that enormously simplify creating and manipulating scientific
electronic documents based on HTML/MathML, for a very broad group of
users.  Add to that the probability that browsers will be able to
render MathML within a few years, and I believe that is adequate
justification for the design of MathML, imperfect as it may be.

Sincerely yours,

Robert Miner


--------------------------------------------------------------------
Robert Miner                               http://www.geom.umn.edu
The Geometry Center                        phone: (612) 626-8313
HTML-Math WG co-chair                      fax:   (612) 625-8083
--------------------------------------------------------------------

Received on Tuesday, 27 January 1998 10:47:40 UTC