Just published Luma 🌙 — a small programming language designed to be kind. Explicit types, gentle errors, low cognitive load.
Built with AI assistance by someone who can't write code independently. That's kind of the point — tools should work for everyone.
github.com/dylanisaiahp/luma
#programming #opensource #accessibility #rust
@dylanisaiah Wow, I'm also interested in creating a programming language, now I'm in pre-development state (talking and brainstorming ideas with agents). I was thinking of making version 1 as a transpiler to C. So basically one would run my program, it is analyzed, transpiled to C then compiled with something like GCC or MSVC. However, I saw Luma is written in Rust. How much time did it take to you? What was your approach or, if it's 100% AI, approach of your agent? Do you have intermediate code, like C or LLVM? thank you so much for replying.
@menelion Luma is a tree-walking interpreter written in Rust — no intermediate representation, no LLVM, no transpilation yet. Source goes straight to tokens → AST → execution. The transpiler-to-C approach is solid for performance, it's how early languages like C++ started. Total time was about 3 days, built collaboratively with AI assistance. The approach was iterative — lexer first, then parser, then interpreter, one feature at a time.
@dylanisaiah wow, I wish you the best of luck!