May 9, 2026
Syntax and the City
Making your own programming language is easier than you think (but also harder)
Everyone wants a dream coding language—until the comments turn into a sandbox war
TLDR: A hobby developer built a custom programming language for a game-mod project and says it’s both easier and harder than people think. Commenters immediately split into teams: "just use Python," "use Lua," "libraries are the real problem," and "actually this is kind of awesome."
A game developer wrote a long, cheerful confession about building a custom programming language for fun, for learning, and maybe for game mods—and the crowd instantly turned it into a very online courtroom drama. The author is refreshingly self-aware: this is not the next world-changing software empire, just a passion project that already managed to produce a small working graphics demo. But in the comments, that humility barely slowed down the hot takes.
The biggest clash? Should you even make your own language at all? One camp was instantly relatable: why invent a new thing when, as one commenter joked, their ideal language would be basically Python... 100%. Another group hit the practical panic button, arguing that making a language is the easy part and that the real nightmare is building enough tools and add-ons so people can actually use it without suffering. In other words: congrats on the shiny new language—now where’s the stuff that makes it useful?
Then came the sandboxing fight, aka the "how do you keep player-made mods from becoming chaos" debate. One commenter side-eyed the whole plan and suggested Lua would be far easier for safe modding, while another swooped in with an alternate favorite, libriscv, basically saying, "Cool project, but have you considered this other rabbit hole?" Meanwhile, one brave soul chimed in with "actually, building language tools is surprisingly approachable," giving the thread its one wholesome "let them cook" moment.
Key Points
- •The author started building a custom programming language in mid-December 2025 and published a status reflection on May 6, 2026.
- •The language is not production-ready, but the author says it was capable of running a 1,000-line Monte Carlo path tracer.
- •The project was motivated by a long-standing interest in language design, the modding needs of the author's game, and Matt Godbolt's compiler-optimization series.
- •The author's game is described as simulation-heavy, with hundreds of thousands of entities handled by a custom ECS engine.
- •The article identifies three desired properties for a modding language: efficient low-level iteration over game data, easy sandboxing, and simple drop-in usability for mod authors.