Message-Id: <9302101701.AA07555@sc.zib-berlin.dbp.de>
Reply-To: Mailing list for the LaTeX3 project <LATEX-L@vm.urz.Uni-Heidelberg.de>
Date:         Wed, 10 Feb 93 17:38:14 +0100
From: tarjeij@EXTERN.UIO.NO
Sender: Mailing list for the LaTeX3 project <LATEX-L@vm.urz.Uni-Heidelberg.de>
To: Multiple Recipients of <LATEX-L@vm.urz.Uni-Heidelberg.de>
Subject:      makeindex
Status: R


> From: Frank Mittelbach <MITTELBACH@MZDMZA.ZDV.UNI-MAINZ.DE>
>
> It may be true that, when we would try to re-implement the whole TeX
> system with all its utility programs (as NTS may one day might try to
> do, but please don't start an NTS discussion on this list) then one
> definitely should think about the base language first, eg WEB, CWEB,
> C, C++, smalltalk, .....who knows?

If I rember right web-to-C is something that is hardwired to TeX. It is a kit
with one mission only. Besides I think that someone in the Unix community has
reimplemtented TeX in C (CTeX or TeXC?). Last time I heard they had passed the
trip tests. I don't know if they have a version beyond 3.0. If this C
implementation is up to date and under active maintenance we should use it. That
will allow DEK to concentrate fully on "the art of computer programming". It
will also make it possible for others to work on the TeX kernel.


> But then who cares now since we are not trying to do this. And given,
> that on *any* plattform on which TeX runs, there must be a possibility
> to handle WEB, it helps a lot to make use of this fact. And enough
> people have outlined the advantages so that I don't need to repeat
> this. If you use different languages than you have additional
> maintenance or money problems.

As a person who have tried to figure out how to generate a big TeX from the
source I know a little about the maintainability of TeX. It is Pascal that
creates the maintenance problem. The C library is so ludicrously small that
portability for something like makeindex, dvips and TeX are relatively easy to
achieve (I'm not saying that C is the best language, because it isn't).

If there is one thing that is certain in this world for the foreseeable future
it is that there is a good free C compiler available for your workstation* and
or PC (from 386 PC and upwards). Pascal is if at all availble generally
something you have to _pay_ for. A conforming Posix environment has to have a
C compiler. Even Posix for VMS has one.

*) Intergraph computers currently excepted.

I doubt very much that it would be a very big job converting tex.web and friends
to cweb or fweb web (or any other actively maintained web). Anyway, it is a job
that would be done once. If Web is dead, why keep adding facilities that make
use of it? Will a Web'ed program fail to unpack with any of the other webs?

By doing these things one small step at a time we control change. We don't end
up with a big sweep and the problems that might bring. We end up with a system
in continous change, where problems can be dealt with as they arise.

If we insist on forcing everything into web we could end up with a situation
where it would not be possible to implement any NTS, because of the amount of
source code bound very tightly to the old system.

As a system manager I want the easiest system to administrate. I want a system
that I can _mend_ when something goes wrong. E.g. when users run out of memory
in TeX. I want to be able to fix these things in minutes, not in days and weeks
and I want to know as little as possible about TeX and Metafont (or Web for that
matter). As a system manager the best TeX I have is emTeX (requires least work).


[Trivia: According to an article in a recent issue of Computer Language magazine
Donald Knuth currently uses cweb.]

Finally: I don't advocate a rewrite of TeX or Metafont in C, I just argue
against moving programs that already exists in C to Pascal. And I recommend
that new programs are implemented in a modern web or in plain C.


greetings,

// Tarjei T. Jensen - if it ain't broken, fix it anyway!
//    tarjeij@ulrik.uio.no       || +47 4 563411
//   Support you local rescue centre: GET LOST!
// Working, but not speaking for the Norwegian Hydrographic Service.