[Axiom-developer] Axiom musings...

[Tim Daly]

I have been programming for 50 years. There is always the
personal challenge of "keeping up with the edge".

In the wire-board and punched card days it was the ability
to choose the optimal sort for your data (almost everything
involved sorting).

Then came the cpu optimizations... write self-modifying code
that fit into the 64 byte cache. Or pick specially chosen chebyshev
values optimized for your sin function. Or do function-maps using
the Translate-and-Test instruction. Or micro-coding your loops to
enhance the CPU microcode for your program.

Then it was CMOS ASICs and FPGAs so your 16-bit multiplies
could REALLY pipeline.

Then PASCAL showed up and you weren't leading-edge if
you didn't do types.

and so on...

Now you're not even in the game if you aren't at least doing
$F_\omega$. (see Lambda: The Ultimate Sublanguage)
https://dl.acm.org/ft_gateway.cfm?id=3342713&ftid=2076175&dwn=1&CFID=149744317&CFTOKEN=2f53f2232d5db617-85952791-F402-7A5C-37FE69F835BDD124

It's what all the leading-edge kids are learning in school.

Next week you're behind the times if you aren't proving your
programs correct. You think your code "works"? Prove it.

And in two weeks you need to be using $\lambda{}C$.

Oh, wait... I'm already behind the edge.

All of this learning is time consuming and painful. But when
the Boeing plane crashes and the Airbus plane loses its
fly-by-wire and the Arianne rocket blows up and the Therac-25
kills patients and the Uber runs down pedistrians and the
facial-recognition AI system sends you to jail...It is time
to get serious about correct programs.

Since Axiom is computational mathematics it seems to
be a good place to start.

Tim

"it takes all the running you can do, to keep in the same place.
If you want to get somewhere else, you must run at least twice as fast as that"
-- The Red Queen in Through the Looking Glass"