🚀 go-pugleaf

In article <ROLFL.92Sep11084644@karl.uio.no>, rolfl@karl.uio.no (Rolf Lindgren) writes:
> 
> When Knuth wrote TeX, he knew what he was doing. This is a fundamental
> assumption, and all arguments pro or con should be checked against this.

Hey, my turn :-) I question this assumption.

A very good critique of Knuth's style of programming can be found in
David Gries' response to an Literate Programming example by DEK
(CACM, Vol. 29, No. 5, May 1986, p. 364-369).
    It's interesting to note that Gries did not criticise the concept
of Literate Programming, but the focus on a self-contained algorithm.
Also, since you mentinoned it, my rambling below should not be taken
as a point against Literate Programming. It's completely independent
to that issue.

Even though TeX was written very long ago, the concept of Abstract
Datatypes was already known (after all, Liskov & Snyder published
their paper in 1972, not to speak of Simula's classes). Already then,
it was _not_ state of the art to use global variables all over the
place for the communication between several parts of the program. If
pre- and postconditions were named, at least... The value of
procedural abstractions (ie, communication over formal parameters)
was well known. (Btw: a good introductionary text about this topic
may be found in Liskov & Guttag: Abstraction and Specification in
Program Development.)
    It remains an open questions for me, if one of the reasons behind
not changing TeX any more is the low maintainibility grade of this
code. ;-( (Hmm, at least it seems to be a low maintainibility grade
for mere mortals like me. :-) :-)

--
Joachim

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Joachim Schrod			Email: schrod@iti.informatik.th-darmstadt.de
Computer Science Department
Technical University of Darmstadt, Germany