Yes, the latest response from Garnix side is this: https://matrix.to/#/!hQKBiJjjGQdiJBvMgK:matrix.org/$sVg4l7GJOvJCOVwUIBjaOr0RxLT-AU3iUjx-Ml93W54?via=matrix.org&via=t2bot.io&via=c-base.org

How do you mean? It’s just more people and systems that might be compromised and need to be trusted

$ ssh remote-linux-x86-builder 'cat /proc/cpuinfo | grep vendor_id | wc -l'
24

I agree. Nix tends to change option names a lot over time and the most reliable source tends to be the actual source of nixpkgs. The other part of nixos in my opinion is figuring out the syntax and the architecture of nixos itself. I find that people don’t tend to not learn when GenAI does everything for them. I’m aware GenAI is a very large and contentious topic right now, but I personally believe it shifts the job of writing and architecting code into code review (great when it works, disastrous when it doesn’t). MIT also did a study finding similar effects on learning and retaining information https://www.media.mit.edu/articles/a-i-s-effects-on-the-brain/.

Overall, very cool, and I think there’s potential, but I also think that the tool may need to help step people through what it’s doing so it can be a tool for learning not just click a button, pay 100$, and you have a “working” app or OS for about 3 weeks.