miniblog.

Doing another iteration on my diagnostics display. I'm reasonably happy with the bold highlighting within the error message. I'm not sure about the colour on Warning and Error though. It gives the output some visual structure, but arguably the message itself is more important.

I'm coming round to the view that a main function shouldn't take arguments. For example, in Rust you have to call std::env::args() to get CLI arguments. This makes Hello World less verbose, and gives you more flexibility in setting up CLI argument parsers.

One interesting consequence of the rise of LLMs: there's more demand for tools that handle untrusted input. Arbitrary HTML+JS can be safely run in a browser. Lean can check an arbitrary proof. These work really well with an LLM that can be wrong, but sometimes gives exactly what you want. Are there other tools in this family?