3 ms·

Lisp with GC in 436 Bytes

173 points by behnamoh 2y ago

IncreasePosts 2y ago
Why does brainfuck not count as a real language?
- IshKebab 2y ago
  It's a toy language or esoteric language, designed to just have fun designing weird languages. The point of Brainfuck is to make it difficult to write programs so writing them becomes a "fun" challenge. In contrast Lisp is a real language designed to make writing useful programs easier, and has been used for decades to write real useful programs.
  - dixie_land 2y ago
    But brainfuck is Turing complete so you can bootstrap a LISP interpreter from brainfuck, thus making it "real"
    - IshKebab 2y ago
      That's really missing the point of this challenge.
      - odo1242 2y ago
        I mean, the Lisp here technically uses the same approach - it runs just enough Lisp to boostrap the remaining functionality on top of what is provided.
  - akira2501 2y ago
    > Lisp is a real language designed to make writing useful programs easier This implementation is the opposite of this goal. They explicitly eschew this in favor of making something small. So, no error messages, no printer, no macros, none of the things that make lisp "real." To the extend that BF is not real then this implementation of lisp isn't real either.
    - IshKebab 2y ago
      I don't know anything about Lisp really but they claim it can run "real" Lisp, and have a demo. Are you saying this is a lie?
      - akira2501 2y ago
        I challenge the definition of "real" as applied to _this_ implementation here. It can run some programs made only of exceptionally limited forms. You can, of course, build the components like integer addition and subtraction yourself in the least efficient way possible; however, how is this any different from the situation in BF? They themselves also say this: "The code above is a LISP within a LISP within a LISP: three levels. You can use this technique to implement missing features like macros."
        jart 2y ago
        I demonstrated in the blog post that SectorLISP can run real programs that John McCarthy and his crew wrote back in the 60's for his IBM 703 LISP 1.5 system. See https://justine.lol/sectorlisp2/proof.html https://justine.lol/sectorlisp2/proof.html where, with only light modifications to the original source code, I got his theorem prover working on SectorLISP, which uses Wang's algorithm. The original source code is here for comparison: https://justine.lol/sectorlisp2/wang.job.txt https://justine.lol/sectorlisp2/wang.job.txt
        akira2501 2y ago
        It was written as an example for the LISP I Programmers Manual. The algorithm itself is not particularly powerful and this implementation can only return a singular true or false value. I wouldn't necessarily call this a "real" program as McCarthy was trying to demonstrate how to translate logical forms into s-expressions more than anything. I'm not saying any of this to be critical of this team's implementation, more so to defend the notion that brainfuck is just as "real." Or, if brainfuck is "not real" then this particular implementation isn't for more or less the same reasons.
        2y ago
  - xvedejas 2y ago
    Brainfuck isn't maximizing difficulty, it's just a turing tarpit, maximizing a narrow definition of simplicity. If you want to see a language that maximizes difficulty, take a look at INTERCAL or Malbolge.
  - Onavo 2y ago
    It's excellent for teaching Turing completeness and Turing machines
- deciduously 2y ago
  This is a hilarious takeaway from this writeup.
- behnamoh 2y ago
  Define "real"
dang 2y ago
Related: Show HN: Lisp with GC in 436 Bytes - https://news.ycombinator.com/item?id=29630293 https://news.ycombinator.com/item?id=29630293 - Dec 2021 (131 comments)
sim7c00 2y ago
anything on justine.lol makes me feel like a single celled organism. weirdly it feels good. there's so many things in each project, which spirit can be applied to many things. also love the fact always previous projects and code are either used or referenced to explain things. i am only in awe each time something is posted.
anthk 2y ago
Justne.lol would love "The Computational Beauty of Nature". It has a github repo, with references to Mandelbrot, more fractals, atractors, and, of course, a Lisp and building blocks.
bankcust08385 2y ago
It uses AT&T syntax. Yuck!
dloss 2y ago
How small could this be if it was implemented on the SectorLambda VM (a sister project)? http://justine.lol/lambda/ http://justine.lol/lambda/
- jart 2y ago
  Check out https://github.com/woodrush/lambdalisp https://github.com/woodrush/lambdalisp which was written for my Blc VM.

173 points by behnamoh 2y ago

IncreasePosts 2y ago

Why does brainfuck not count as a real language?

IshKebab 2y ago
It's a toy language or esoteric language, designed to just have fun designing weird languages. The point of Brainfuck is to make it difficult to write programs so writing them becomes a "fun" challenge. In contrast Lisp is a real language designed to make writing useful programs easier, and has been used for decades to write real useful programs.
- dixie_land 2y ago
  But brainfuck is Turing complete so you can bootstrap a LISP interpreter from brainfuck, thus making it "real"
  - IshKebab 2y ago
    That's really missing the point of this challenge.
    - odo1242 2y ago
      I mean, the Lisp here technically uses the same approach - it runs just enough Lisp to boostrap the remaining functionality on top of what is provided.
- akira2501 2y ago
  > Lisp is a real language designed to make writing useful programs easier This implementation is the opposite of this goal. They explicitly eschew this in favor of making something small. So, no error messages, no printer, no macros, none of the things that make lisp "real." To the extend that BF is not real then this implementation of lisp isn't real either.
  - IshKebab 2y ago
    I don't know anything about Lisp really but they claim it can run "real" Lisp, and have a demo. Are you saying this is a lie?
    - akira2501 2y ago
      I challenge the definition of "real" as applied to _this_ implementation here. It can run some programs made only of exceptionally limited forms. You can, of course, build the components like integer addition and subtraction yourself in the least efficient way possible; however, how is this any different from the situation in BF? They themselves also say this: "The code above is a LISP within a LISP within a LISP: three levels. You can use this technique to implement missing features like macros."
      - jart 2y ago
        I demonstrated in the blog post that SectorLISP can run real programs that John McCarthy and his crew wrote back in the 60's for his IBM 703 LISP 1.5 system. See https://justine.lol/sectorlisp2/proof.html https://justine.lol/sectorlisp2/proof.html where, with only light modifications to the original source code, I got his theorem prover working on SectorLISP, which uses Wang's algorithm. The original source code is here for comparison: https://justine.lol/sectorlisp2/wang.job.txt https://justine.lol/sectorlisp2/wang.job.txt
        akira2501 2y ago
        It was written as an example for the LISP I Programmers Manual. The algorithm itself is not particularly powerful and this implementation can only return a singular true or false value. I wouldn't necessarily call this a "real" program as McCarthy was trying to demonstrate how to translate logical forms into s-expressions more than anything. I'm not saying any of this to be critical of this team's implementation, more so to defend the notion that brainfuck is just as "real." Or, if brainfuck is "not real" then this particular implementation isn't for more or less the same reasons.
        2y ago
- xvedejas 2y ago
  Brainfuck isn't maximizing difficulty, it's just a turing tarpit, maximizing a narrow definition of simplicity. If you want to see a language that maximizes difficulty, take a look at INTERCAL or Malbolge.
- Onavo 2y ago
  It's excellent for teaching Turing completeness and Turing machines
deciduously 2y ago
This is a hilarious takeaway from this writeup.
behnamoh 2y ago
Define "real"

dang 2y ago

Related: Show HN: Lisp with GC in 436 Bytes - https://news.ycombinator.com/item?id=29630293 https://news.ycombinator.com/item?id=29630293 - Dec 2021 (131 comments)

sim7c00 2y ago

anything on justine.lol makes me feel like a single celled organism. weirdly it feels good. there's so many things in each project, which spirit can be applied to many things. also love the fact always previous projects and code are either used or referenced to explain things. i am only in awe each time something is posted.

anthk 2y ago

Justne.lol would love "The Computational Beauty of Nature". It has a github repo, with references to Mandelbrot, more fractals, atractors, and, of course, a Lisp and building blocks.

bankcust08385 2y ago

It uses AT&T syntax. Yuck!

dloss 2y ago

How small could this be if it was implemented on the SectorLambda VM (a sister project)? http://justine.lol/lambda/ http://justine.lol/lambda/

jart 2y ago
Check out https://github.com/woodrush/lambdalisp https://github.com/woodrush/lambdalisp which was written for my Blc VM.