>Recently the people from Wolfram Research gave a demonstration

>of the new Mathematica version in our town.

>They seem to have a system now which should be able,

>together with some macro package still to be written,

>to simulate an interactive WYSIWYG*system.

Unfortunately, you seem to believe things that you are told

by what are called in the computer industry "marketing slime".

Quote:>Already now one gets beautiful typesetting, interactive formatting of

>equations, on the fly reformatting when one edits, etc. And all is

>WYSIWIG.

A major difference between WYSIWYG editors and TeX-based systems is

that the latter are structure-based and thus one deals fairly

consistently with notions like "chapter heading" rather than "a bunch

of stuff that is in 14 point Helvetica bold". One can, by changing a

definition or a macro or a setting early in the text, change the

meaning of subsequent text. Many people take advantage of this. With

most WYSIWYG systems, what you see is ALL you get. Structure-based

editing, for serious large-scale work, is still interesting, and it is

not such a problem to "batch process", at a time when re-typesetting a

document is done at a rate of 10 pages/second e.g. on a 200MHz

Pentium. TeX still has many advocates.

Previous mixtures of TeX and WYSIWYG have been proposed and

implemented, including work by Bill Schelter (unfortunately only on a

Lisp Machine) called something like InFoR; something done at Berkeley

by Peehong Chen (VorTeX), and probably others too. They didn't catch

on, but perhaps that is unrelated to technical matters. A commercial

system combining WYSIWYG and TeX is Scientific Workplace. I don't

know how large their market is, but it does much of what you think

Mathematica might do in the future. This starts with a rather full

complement of text, data-base, graphics stuff and adds active

mathematics (your choice of Maple or Mathematica) from menus or

keyboard selection..

Now, perhaps one can introduce into Mathematica 3.0 structures that

correspond to (say) \begin{section} .... \end{section}. etc. In fact

one could, in principle, translate all of TeX into Mathematica's

language, etc. etc. So, sure you can do it. You lose some immediacy

when you do this, however.

Quote:>Is that the future of LaTeX? Will Mathematica replace LaTeX?

TeX is available free and is supported by the American Mathematics

Society (etc) as a standard. It is used by many people as is.

Quote:>Is anybody already working on this?

Wolfram believes all mathematicians should disambiguate

sin(x) by writing it as Sin[x] "standard form." I assume that

people at WRI would try to wean people away from old stuff to

new Mathematica stuff for TeX if they could. Just like any

commercial enterprise trying to sell their product.

(No harm in that if they provide fair value, of course!)

If you want to look at this ball of wax, you will also have to

look at other publishing systems, other computer algebra systems,

and other communication systems like HTML 3.0, SGML, OpenMath, as

well as the notebook interfaces for Maple, Macsyma, Axiom, Theorist,

MathCAD, etc.

Some people think that a big negative in using Mathematica for TeX is

that even using Mathematica for mathematics leads to bogus results.

Downright incorrect. Wrong branches of functions, wrong signs, wrong

precision of numerical stuff. Some people (Fortran-fans) just find

Mathematica about 1000 times too slow. For those people, paying big

bucks for a system that they would not use and could not use for their

kind of work, just so you could typeset, is a major disadvantage.

I encourage the use of better typesetting technology in computer

algebra systems, but history favors the primacy of the text,

with links to active mathematics, as a model. While it is possible

to begin stuffing any and all possible text into a math model,

it might not be comfortable, and it might be a mistake to insist

that everyone buy a particular expensive program in order to

run on your paper to read it.

--

Richard J. Fateman