> In the end, I really think Nix and NixOS are very strong tools for achieving reliable and reproducible system configurations and package management. Unfortunately, though, I don’t think the benefits I’ve gotten in this one month of using NixOS so far justified the cost I’ve initially spent and continue to spend learning Nix and NixOS.

This largely reflects my experience with NixOS as well. I installed it for a few months on my laptop (on two separate occasions) and there are parts of the OS I love. Being able to simply define your packages and user setup in a single configuration file is incredible, and even now if I wanted to switch back to Nix, I could dig up my configuration.nix and pick up where I left off. Gone too are the piles of mysterious packages installed on my system but I can't recall why they're there or if I still need them.

Unfortunately, once you start stepping off the beaten path, Nix often became a source of frustration. If there's a well working package available in its repositories, great. If not I found creating my own Nix packages to be an experience of arcane incantations, even for the simplest things. Compare this to writing Arch AUR PKGBUILDs, which are simple, glorified shell scripts, and it quickly led to me growing frustrated with the OS.

I think what eventually did me in was trying to get Neovim LSPs installed correctly. For some reason I never really understood, where ever my Nvim plugin was installing them was certainly not where Nix wanted them, and I never resolved the issue. Things like this eventually drove me back to the comforts of Arch, although every once in a while I look back over and wonder what could've been. I think it's the only distro I've tried in recent years that really tempted me to switch permanently.

The ideas of NixOS are fascinating, and I really think it can create a new paradigm in the Linux space. However, I'm unsure whether it'll be Nix itself that really gets there.