> o trade-off of complexity versus coverage
> o the impact of yet another standard for equations
> o the large numbers of math symbols in use
> o just how much code is needed for parsing/rendering?
> o what to do with line mode displays
My answer would be "whatever TeX can do, use that". The best way to do
that is to use TeX, either by displaying the output of an existing dvi
viewer, e.g. xdvi, or make WWW browsers into TeX viewers (maybe limited
to just mathematical TeX). What to do with char displays? Do what TeX
does (I don't *know* what it does, I don't want to know, I just know
that its users seem happy with it).
We use (amongst other things) DECWrite, which deals with equations by
embedding limited, math only, TeX code (actually stored in a separate
file), and translating it to DECwrite's internal format for final
display. It works well. DEC saw no reason to invent a new format, and
given that TeX already exists, I see no point in reinventing it,
either. Face it, the hard bit has been done - the browser author
doesn't have to know what the Tex code *does*, simply hand the input to
it and render the output. Yes, I know it's not trivial for browser
authors, but my guess is that reinventing maths symbol processing for
HTML would be worse. Do the TeX stuff in a separate process, if you
like. Ghostscript takes this approach of separating viewer and processor, I believe.
The same arguments apply to eqn, for sites which use that. We don't,
but the code exists as groff. Fine; lets use that. If WWW land goes for
eqn, I'm sure it's not too difficult to translate existing TeX math to
eqn math. And once we can embed eqn or Tex math, why not arbitrary
troff/groff or TeX? This has the distinct advantage of handling native
man pages and TeXinfo straight off.
Finally - a limited set of math symbols just wouldn't work. Authors
will stop using it as soon as they realised that bits of it failed to
come out. An HTML+ math format would need to be as complete as TeX/eqn
from day one, which is a tall order; otherwise its like only supporting
the 20 most frequently used letters of the alphabet on the basis that
they cover 98% of English text. True, but still not acceptable.
Peter Lister p.lister@cranfield.ac.uk
Computer Centre,
Cranfield Institute of Technology, Voice: +44 234 754200 ext 2828
Cranfield, Bedfordshire MK43 0AL UK Fax: +44 234 750875