A successful lisp machine?

[Thomas Fischbacher]

Tim and Mike wrote:

> >Port TeX.  It may surprise you to learn that TeX was not developed
> >on Unix, and is not wedded to the Unix paradigm.  (Or it may not
> >surprise you.  But people seem to forget that this is an
> >applications question, one which I think is largely tangential to
> >the important questions surrounding the design of a new Lisp
> >system.)
>
>   Better yet, write a TeX to your favorite lisp based document system.
> What? You don't have one? Then volunteer to write one. Something called
> DSSSL is a scheme(???) based thing. Could be a good idea to start
> there.

This is not quite as simple as one initially might think.

TeX is a typographic system, it was written to do
typesetting on a computer, so TeX has little in common
with SGML or the like.


Hacking up a program like TeX is the minor problem, I think.
(Indeed, virtex is an extremely small program. And stuff like
implementation of hashing, print functions and so on
take up a large part.)
The problem is to get all the details right.
TeX is guaranteed to produce 100% identical output on a huge
variety of very different systems. (Most people don't think
of that initially, but it's quite clear that this is an
absolute must for a sensible typesetting system.
Imagine different machines producing different page breaks
in an important article. You couldn't cite from such a .tex
document, for example.)
This leads to consequences such as that floatingpoint arith
can't be used, since different machines round differently.
TeX is a rather static thing by now. (You can also see this
from the fact that TeX's version number seems to converge
against a famous transcendental number. :) )
There would be much room for improvements of TeX, but they
are deliberately not implemented.

A TeX port to LISP would have to guarantee that its output
is 100% identical to what TeX might produce, in every case.
You see, we get in trouble here.


On the other hand, if not porting TeX, but developing
something new and better (I hope), the question is
what the status of such a typesetting system should be.
Maybe it could exist besides TeX.


I think I could help in building a LISP-based typesetting
system (though I should pay a little bit more attention to
my studies :) ), but some things are too important to be
decided by a single individual or a small group.
If we really want a good LISP typesetting system,
not to get in contact with TUG would be a big mistake.

Anyone else interested in "LISP-TeX"?


--
regards,           Thomas Fischbacher - tf@another.gun.de
                                        fischbac@informatik.tu-muenchen.de