miniblog.

I've been experimenting with LLM autoresearch. I prompted Sol to make difftastic faster without changing output on the test suite, and log everything it tried: https://github.com/Wilfred/difftastic/blob/c6e9c6fed4c71276a8b0e377c92695c3d929dc7f/PERF_RESEARCH_LOG.md It found some interesting performance tweaks, although it's too tolerant of complexity.

I had a third attempt at porting difftastic from Dijkstra to A*, and I've eventually concluded that A* just isn't worth it for the implementation complexity. I had fun writing it up though, and I've even attempted ASCII art for building intuitions:

JSON is too small (no comments) and YAML is too big (many string syntaxes, relatively few implementations). TOML is in the sweet spot for complexity, but I agree this syntax is by far the most confusing part. TOML 1.1 improves it at least: