if this shit keeps getting better like this man... you're really gonna want some sort of local, infrastructure grade language-aware codebase intelligence engine that builds deterministic indexes reproducably and incrementally, with snapshot diffs... someone should build that.

Sounds like a pretty good idea. Regret to inform you that you're currently at your "token limit" for cool PowerShell idea requests for the day. But gimme a follow and remind me this weekend / next week and I'll get on it. Already got something on the shelf that's not too far from this.

I'm ~10B tokens deep on this, started about a month ago. Got a huge refactor and a couple big features left, actual 0.1 should be landing this month

Good! / (slightly) oh dear. I feel like sometimes very efficient approaches become obvious with experience. Short version of my short approach: Localized data files with basic structure. Happy to dive into the nitty gritty (this weekend) 😉

just a bit past that... it's a hybrid semantic+lexical+graph index with some neat shit like cross-language call graphs, fast partial reindexing (which produces the same deterministic result as a full index of the same input), cross-file inferencing, honestly so much shit lmao

Here's my entry xD for(;;){$t=1.3*[datetime]::UtcNow.Ticks/1e7;[console]::Write("`e]4;264;rgb:{0:x2}/{1:x2}/{2:x2}`a"-f(0,2.09,4.18|%{[byte](127.5*(1+[math]::Sin($t+$_)))}));sleep -m 1}