https://github.com/serprex/inliners

http://esolangs.org/wiki/Inline
https://github.com/serprex/inliners

esolang

Last synced: 3 days ago
JSON representation

http://esolangs.org/wiki/Inline

• Host: GitHub
• URL: https://github.com/serprex/inliners
• Owner: serprex
• Created: 2016-12-10T21:08:22.000Z (about 8 years ago)
• Default Branch: master
• Last Pushed: 2018-09-29T16:10:03.000Z (over 6 years ago)
• Last Synced: 2024-12-19T17:48:28.421Z (about 1 month ago)
• Topics: esolang
• Language: Rust
• Size: 16.6 KB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: readme

Awesome Lists containing this project

README

        
Somehow I came across this language on esolang, looked trivial to implement, started on it.. a couple weeks go by

I've taken 3 stabs at parsing. First I started with hand rolled recursive descent, but the structure felt simpler, tried to model it with nom, but I can never get working with nom right

But this language.. it has no scopes. It's a mash up assembler for a RISC architecture designed by someone who has no idea how to make a RISC architecture (exponent instruction.. seriously? I mean sure, we don't all need to be turing tar pits, but there's so much room for usability improvements & you invest in an exponentent instruction??)

Anyways the brainfuck interpreter example is broken, since it's missing 2 suffixes & $[200] needs an immediate. Then I found that the interpreter was having issues running BF interpreting Hello World because it'd hit [ & dirty 3:d & then continue checking it against character codes. Probably would've been better to model the instructions as gotos rather than calls

So now anyways parsing is that I use a big RegexSet to check for what the next instruction is, slowly stepping through. Conditions being able to include a single instruction instead of a label makes the grammar non regular since that means you can use pumping lemma with infinitely nested conditions. So specially handle that by having parse_instr be recursive there

Things I have yet to implement:

[100:h] is apparently valid, for now one has to do [1-0:h] (ie break up immediate memory addresses into byte components)

%. Still trying to decipher semantics

Compiling down to a bytecode and/or Rust

Fun thing about RegexSet::new, I hit the 'rust doesn't impl anything for [T; N] where N > 32' issue & had to move it to a static const which I then have be an &[&str] rather than &[&str; 40-something]