Today work has focused on a few things, most notably a transition from using software that may have potentially been coded with AI-assistance to no such projects.
Allow me to explain myself. I do not use AI for anything, and I feel downright xenophobic hate against AI itself (but quite definitely not against the backing research.) Of course, this makes me hate a lot of the software I use daily.
So after thinking it about for a few weeks, and deciding that I prefer struggle, I went down the ethical-foss/open-slopware route, and started looking for replacements to most software I use that may have had LLM involved in its maintenance.
This has so far been met with success on the side of getting rid of a crap ton stuff on my system. Thankfully, I can say both my shell and typesetting system of choice are both anti-AI. Unfortunately, Rust isn’t.
I plan on continuing my involvement with the Rust community, though. On the side of getting replacements for those tools, I’ve had moderate success in going back to using Vim instead of VSCode, and looking into SML/NJ instead of OCaml.
Yes, Vim is slop these days, but I’m temporarily using the Vim that comes pre-compiled in my system, which seems like a raw enough experience. Most software alternatives require GNU autotools, and those I’ve yet to compile.
I also got rid of my package manager, as it was riddled with AI stuff, and of my command launcher, which literally shipped an AI feature. The latter I can live with just fine, but the former is going to take a while getting used to.
Now, this was something I spent only half of today on, and may potentially spend some more time tomorrow. It is a tough transition, because a lot of nice stuff like ripgrep or lazygit are now slop to me. But I need it.
In terms of actual work (in the other half of the day,) today I worked on a bunch of stuff mostly concerned with PR reviews in rust-lang/libc. This required changing some stuff from old-time PRs like the AIX PR.
I also worked some on the compiler, which has finally seen the parser finished, and I’ve started working on resolving reduce/reduce conflicts from productions I can’t quite inline, but that I can quite definitely priority (through %prec.)
The no-AI in the software I use is going to potentially consume a bunch of hours in the next few weeks. I hope tomorrow to have the usual GNU tools for compilation set up, so that I may first install got, and then Vim Classic.
The rust-lang/libc work I expect to continue working tomorrow first thing in the morning, so as to take up the first half of the day. This is the same as what I’ve done today, because I can dedicate the best of myself to actual work.
In terms of compiler work, I don’t expect to continue working on the compiler in its current form tomorrow. I first would want to set up a working SML/NJ compiler, with which to start rethinking how I should address this project.