🚀 go-pugleaf

In article <FISCHER.91Dec17174504@loke.iesd.auc.dk> fischer@iesd.auc.dk (Lars P. Fischer) writes:
>
>Given the fact that 1) TeX has got the linebreaking and kerning right
>2) The algorithms are published in the scientific literature 3) the
>implementation is freely available for all to see and use, why is that
>just about no one of the zillions of high end wordprocessors and DTP
>systems get it right? Why do they all produce this inferior mess when
>it is a matter of spending a small amount of time grapping the code
>from the hand of God, ehhr Knuth? So much for commercial software
>development, market drive, and competition for customers.
>
My answer: because a TeX document is a turing machine.

I wonder myself why there are no major word processors based on the TeX
typesetting architecture. One reason is that word processors have their
ancestry in PC hackers and not computer scientists. [for the same reason,
the most widely selling operating system in the world looks more like CP/M
than Unix, despite the availability of systems like OS/9]

But I think some good programmers have looked at the problem of dealing with
TeX documents interactively. The PC hackers have advanced the state of the
art to the point where users want to be able to see a formatted representation
of their document while they type it. The problem is to let the user edit
at the level of TeX source, and keep the dvi representation current and on-
screen.

From investigating several interchange formats for word processors,
(Microsoft Rich Text Format, and FrameMaker Interchange Format, for
example) I think that most PC word processor document languages are
regular, i.e. you only have to keep a finite amount of state around to
process any part of the document. None of the constructs nest to
arbitrary levels, and there are no loops, macros, or subroutines. Word
has a fully programmable macro language, but that just facilitates
typing documents -- not formatting them.

None of the commercial word processors "get it [line breaks and
kerning] right" because "it" is a Very Hard Problem. The problem cannot
be reduced to a finite state machine. Knuth took the time to research,
implement, and publish a turing machine that gets it right. Also, very
few of the "zillions of high end wordprocessors" were developed in
reasonable development environments: Microsoft Word has to run in a 640k
DOS partition, and MacWrite ran on a 128k Macintosh.

I am surrounded by the documentation development process, and my background
is in computer science. I certainly hope someone takes the time to leverage
Knuth's work and the capable modern workstation environment to develop a
rich and capable document development environment we all need. Someone
piped up about a GNU (or at least FSF) word processor a while ago. I'd sure
like to contribute. If it meshed with the goals of my employer a little better,
I'd be persuing it more actively.

Dan

p.s. I want to know more about the TeX DVI format. Can anyone point me
to readable, electronically available explanation of the format?