10biForthOS: A full 8086 OS in 46 bytes

https://git.sr.ht/~hocwp/10biForthOS

by mdp2021 • 1 year ago

8 3 1 year ago

git.sr.ht

anthk2• 1 year ago

More than Forth it looks like an ASM made into a REPL.

romforth• 1 year ago

By that metric, my implementation of romforth for x86 clocks in at just 12 bytes (see https://github.com/romforth/romforth/x86/forth.s)

The "bug" in the counting used here is that the overall Kolmogorov(?) complexity is not being accounted for because it is shunted elsewhere.

This is true for the other "tinyforth" implementations as well - such as sectorforth and milliforth, because the actual code ends up in the "input part of the Turing tape" if you want to think of it that way.

The right way to count it might be to measure the closure of all your dependencies (in this case, all of the input bytes that are needed plus the bytes in the BIOS that are needed).

anthk2• 1 year ago

That's what I tought. Even Miliforth lacks a proper complete integer based stack. From that and a few primitives, you can bootstrap a Forth.

On Lisp, Sectorlisp it's interesting; but the ones from https://t3x.org can do far more, even if they are not bootable per se. But T3XForth can, and it's highly usable.

EForth under SUBLEQ it's like that too.

Instead of "Boot sector languages", I'd pursuit a 386 compatbile language being able to fit in a floppy. T3XForth does, and tons more, because the T3X author made eiher standalone ones, or DOS and CP/M 2.2 ports.

10biForthOS: A full 8086 OS in 46 bytes

Related Stories