Reimplement TeX in Lisp? I think not...

Thomas Fischbacher tf@another.gun.de
Fri, 9 May 1997 17:15:33 +0200


Chris wrote:
            
lr> Absolutely.  However, I feel that if someone wants to spend their time 
lr> reimplmenting TeX, the only real harm is that their time might be useful 
lr> working on something that matters to me.

Certainly, a typesetter shouldn't be implemented for LISP for the sake of
itself. But what do we want? A LISP-based typesetter would essentially be
a package containing lots of typographic functions, e.g., for word
hyphenation, spell checking, line breaking. It wouldn't be a sealed
black box processing data. Since X font metrics are in many ways
similar to TeX font metrics (though a little bit more primitive), you
could use the paragraph-building functions from the TeX package for
pretty-formatting text in X, if you want. Just that's why I find
the idea of a LISP-based OS so interesting: the elements you work with
are programs in any today's os: sealed very-special-purpose
black-boxes. In a LISP environment, you can use the underlying functions.
We have much finer granularity here.
Imagine: you could just use the RSA and IDEA from your LISP-PGP within
your own programs whenever they seem useful.
(Not considering legal problems within the U.S., however...)

There have been attempts to make programs export some of their
functionality to other programs in the past. Maybe you know REXX.
But this is rather unsatisfactory when compared to what would be
possible with a "bag of functions" implementation.


But a LISP typesetter certainly shouldn't be something to start with
immediately. There are more important things to be done before.
One topic that needs special attention is security.
Who may call whose functions? Who may modify whose functions?
Who may export functions, to whom, and how?


I see no reason why a LISP system should not be able to run most Unix
stuff (since we need large part of Unix for X anyway). In the long run
however, quite many Unix utilities have to be replaced.

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