Comment by drivers99 | Hacker News Reader

Comment by drivers99 | Hacker News Reader

> A single person should be able to understand a computing system in its entirety, so that they can change it to fit their needs

This idea combined with the notion that software written by someone else necessarily serves their purposes rather than yours has me gradually working in that direction. Really all I need to do is write (or run) a forth on some bare metal. It's fun to build your own hardware as well which is what I'm during learning by doing right now too. Once I realized it was possible after reading jonesforth, I've been interested in (obsessed with) the idea.

To quote jonesforth:

> You really can write a complete FORTH in, say, 2000 lines of code. I don't just mean a FORTH program, I mean a complete FORTH operating system, environment and language. You could boot such a FORTH on a bare PC and it would come up with a prompt where you could start doing useful work. The FORTH you have here isn't minimal and uses a Linux process as its 'base PC' (both for the purposes of making it a good tutorial). It's possible to completely understand the system. Who can say they completely understand how Linux works, or gcc?

> Secondly FORTH has a peculiar bootstrapping property. By that I mean that after writing a little bit of assembly to talk to the hardware and implement a few primitives, all the rest of the language and compiler is written in FORTH itself.

The part about the language being a little bootstrap and then mostly implemented in its own language...

Are there other languages like this? At least any mainstream ones?

Well, Norvig's (et. al.) JScheme - http://norvig.com/jscheme.html - is implemented in Java - partially, and then a bunch of core functions are written in JScheme itself.

Or, if you look at implementations of J... like this - https://code.jsoftware.com/wiki/Essays/Incunabulum - it's basically creates something closer to J in C and then writes the rest of the interpreter. If you squint hard enough...

Lisps, can be, and sometimes are, implemented in a similar way. Trying to figure out the exact number of Lisp primitives required to write the rest of Lisp in Lisp is seemingly a hobby for some. A compiler for a functional language usually transforms the program into another functional language, an intermediate representation, one much more friendly to machine-implementation. (The classic model has four pointers to four lists and has simple operations for the items in those lists, and that's about it.)

It's a relatively small leap to writing the language in that subset, and implementing a small virtual machine for that subset, and then executing the language. Many toy Scheme implementations have been written in this way. But most major implementations, while often written in Lisp (or Scheme), generally do not do that. They use the full language and are not able to bootstrap like that. Though the 1980s Lisp machines kind of did do it that way. The garbage collector and parts of the OS and runtime were written in a Lisp-like assembly language.

https://github.com/udem-dlteam/ribbit

And https://piumarta.com/software/maru/

Are two good examples. Maru is interesting in that it can implement a full-fat language/os/vm/compiler in under 2kloc

Bluebottle is another barbecues and up OS: https://github.com/btreut/a2

I remember and learned a little forth because of that console you could get with PowerPC Macs back in the day. I think I remember there was a key combination to boot in that embedded forth interpreter. It was pretty cool

https://en.wikipedia.org/wiki/Open_Firmware

(Hence the key combination cmd-opt-O-F)

Reply on news.ycombinator.com