I’m building SOLONCODE, my own programming language

Why I’m creating it

I want to make data dependencies, permissions and checks explicit in the programs I use for my own products. SOLONCODE is where I’m turning those ideas into something I can run, inspect and improve.

One example that already runs

A Fibonacci sphere with a million possible points can be described by a 131-byte JSON rule. Storing all of its XYZ coordinates as Float64 values takes 24 MB. I keep the rule and calculate the point I ask for.

In the actual .solon run, the program returned point 42 and a block of eight points. I also checked 4,096 sampled points against a separate Python implementation. The largest coordinate difference was below 1e-12.

What those numbers mean

The 131 bytes cover the descriptor for this procedural sphere. Code, runtime and returned coordinates use additional memory. Equivalent lazy JavaScript uses the same 131-byte representation.

This example checks a representation and its output. It does not establish a general speed, memory or AI-cost advantage for SOLONCODE. I’ll judge those through matched tasks and the results I can verify.

The next builds

My next step is to develop the language through the products I’m building. I’ll share working examples, the decisions behind them and the checks that show what changed.