Re: political strategy for MathML

Hi.

> Thanks for your reply.
> 
> > Given your views on open source software (I did notice the
> > word 'share') I expect you will discount this, but one of the WebEQ
> > tools is designed to be used as a server module, doing the down
> > translation of documents on the fly. 
> 
> Please don't overstate my views about proprietary software.
> CAS is one area where I think that the concept of proprietary
> software sometimes makes sense.  I have used such products,
> and I even once wrote a pedagogical package for one.

Sorry.  I know you have thought a lot about these issues, and I didn't
mean to oversimply or overstate your views.  I just meant that in the 
past I have understood you to argue for the need for open source, or at
least freely available, MathML software, if it is to be a viable
standard in the long run.  I agree with this view, too, except until
recently I couldn't see a path to getting much freely available MathML
software built in the very short term.

However, I think the perception is growing that MathML is actually
going to be part of the Web infrastructure for a while.  As a
consequence, as I tried to suggest, I think it is inevitable that we
will start to see more and more free MathML software.  Indeed, it is
already beginning to happen, as I think/hope the MathML-in-Mozilla
effort is demonstrating.

> Question: Am I correct that the WebEQ input language is just for
> MathML segments?

It also accepts as input a syntax that is basically a subset of LaTeX
math, augmented with some interactivity commands like \highlight{},
etc.  We call this input syntax WebTeX.  Since we developed it
specifically to convert into MathML, we also devised new table layout
commands (using a Tex-like syntax, of course) that fit the MathML
model better.  Now I wish we hadn't, since otherwise it would be much
easier to tweak it a little to really make it a strcit subset of LaTeX
math.  As it is, it takes a moderately husky Perl script to get it to
'TeX'.  Ah well, live and learn; it mostly serves the purpose.

Another relevant comment here is that when we set up WebEQ, we put a
good deal of effort into modularizing the parser stuff.   It is quite
easy to write a new Java parser class, that makes API calls to build
up the internal WebEQ MathML data structre, and just drop it in
place.  Thus, a develop that wanted to teach WebEQ a new language only
has to write the language-specific parsing stuff.

--Robert

Received on Friday, 29 October 1999 13:12:03 UTC