[TUHS] Knuth and Pascal (was "Were cron and at ...")

[TUHS] Knuth and Pascal (was "Were cron and at ...")

[M Douglas McIlroy]

Off topic, but too much fun to pass up.

>> You wrote your algorithm in Pascal, debugged it, and then rewrote it in your favourite language (in my case, ALGOL-W).

> Now why didn't Don Knuth think of that for TeX?

I'm glad he didn't. He might have written it in Mix. Knuth once said
he didn't believe in higher-level languages. Of course he knew more
about them than anybody else and was CACM's associate editor for the
subject--like a minister who doesn't believe in God.

Doug

Re: [TUHS] Knuth and Pascal (was "Were cron and at ...")

[Toby Thain]

On 2020-12-15 6:45 a.m., M Douglas McIlroy wrote:
> Off topic, but too much fun to pass up.
> 
>>> You wrote your algorithm in Pascal, debugged it, and then rewrote it in your favourite language (in my case, ALGOL-W).
> 
>> Now why didn't Don Knuth think of that for TeX?

He did even more than that, inventing literate programming and tools for
it, so that he could write in a Pascal-like language enhanced according
to his preferences and beliefs about program exposition to humans.

Because the output of his tools was standard Pascal, Tex and METAFONT
were easily and widely ported.

--Toby



> 
> I'm glad he didn't. He might have written it in Mix. Knuth once said
> he didn't believe in higher-level languages. Of course he knew more
> about them than anybody else and was CACM's associate editor for the
> subject--like a minister who doesn't believe in God.
> 
> Doug
> 

Re: [TUHS] Knuth and Pascal (was "Were cron and at ...")

[Bakul Shah]

> On Dec 15, 2020, at 3:46 AM, M Douglas McIlroy <m.douglas.mcilroy@dartmouth.edu> wrote:
> 
> Off topic, but too much fun to pass up.
> 
>>> You wrote your algorithm in Pascal, debugged it, and then rewrote it in your favourite language (in my case, ALGOL-W).
> 
>> Now why didn't Don Knuth think of that for TeX?
> 
> I'm glad he didn't. He might have written it in Mix. Knuth once said
> he didn't believe in higher-level languages. Of course he knew more
> about them than anybody else and was CACM's associate editor for the
> subject--like a minister who doesn't believe in God.
> 
> Doug

Did he actually say that? In this delightful interview
    
https://conservancy.umn.edu/bitstream/handle/11299/107413/oh332dk.pdf

in response to a question about were there any giant steps
in CS, he says the idea of Structured Programming is a giant
step but those are few and far between. A number of times he
comments on structured programming (as opposed to HLLs). About
Go To and SP:

  "It is like the zero population growth movement: The goal isn’t
really to have zero population growth; the goal is to improve
people’s quality of life. But they substitute a quantitative goal
for the qualitative goal, and that’s like viewing structured
programming as only a non-‘Go To’ type of a program, which is
foolish, and I am sure Edsger didn’t mean it that way at all."

He also mentions he wrote an Algol compiler for Burroughs during
the summer of 1960.

Knuth finished the first draft of what became TAOCP in 1966.
Given that there were already many discussions about Algol X
but no clarity and that he really did want people to understand
what steps are needed and their cost, that is, get close to the
computer, MIX made more sense (I'm glad it wasn't Fortran!).
MIX wouldn't have made sense for TeX & METAFONT, which are in
essence *recipes* made out of algorithmic ingredients and in
effect he reimplemented these algorithms in Pascal.

Re: [TUHS] Knuth and Pascal (was "Were cron and at ...")

[Adam Thornton]

> On Dec 15, 2020, at 12:18 PM, Bakul Shah <bakul@iitbombay.org> wrote:
> 
>> On Dec 15, 2020, at 3:46 AM, M Douglas McIlroy <m.douglas.mcilroy@dartmouth.edu> wrote:
>> 
>> 
>> I'm glad he didn't. He might have written it in Mix. Knuth once said
>> he didn't believe in higher-level languages. Of course he knew more
>> about them than anybody else and was CACM's associate editor for the
>> subject--like a minister who doesn't believe in God.
>> 
> 
> Did he actually say that? In this delightful interview
> 

The strong impression I got from _Coders at Work_ is that Knuth, while perfectly capable in HLLs, prefers writing in MIX/MMIX simply because he finds it more fun than languages that provide you with prerolled abstractions.  It’s a charming book, although I think the Knuth chapter is the best.

Adam