Why I love NixOS

Original Article ↗ Hacker News Discussion ↗

NixOS’ declarative, fully reproducible approach to configuring an operating system attracts power users who value being able to define an entire machine in version-controlled files and roll back any change with confidence. Commenters highlight how this model shines for development environments, CI, homelabs, and even AI-driven “agentic” configuration, but also point to a steep learning curve, poor and fragmented documentation, and the idiosyncratic Nix language and filesystem as major barriers to wider adoption. Comparisons with alternatives like Fedora’s immutable variants, Guix, and traditional distros underscore a trade‑off: exceptional composability and reproducibility in exchange for complexity and niche tooling.

Adoption and Learning Curve

  • Many describe Nix/NixOS as “all or nothing”: people either bounce off quickly or commit long-term.
  • Steep learning curve and complex mental model are recurrent themes; several tried multiple times across years before it “clicked.”
  • Some use Nix only as a package manager or for Home Manager on top of another OS as a middle ground.

Perceived Strengths

  • Declarative, reproducible system configuration is widely praised: rollback, easy rebuilds, and “one source of truth” for machines.
  • Strong fit for CI and homelabs: deterministic builds, caching, and easy cloning of servers or desktops.
  • Good at mixing package versions, pinning old nixpkgs commits, and patching packages (from small app tweaks to kernel patches).
  • Containers, NixOS containers, and microVMs are used to isolate workloads; some want per-app isolation derived from Nix configs.

Dev Environments

  • Tools like devenv.sh are appreciated as friendlier abstractions over raw Nix shells, adding services, tasks, profiles, LSP and git‑hook setup.
  • Some insist project tooling should not depend on Nix (for cross‑platform team compatibility), preferring language‑specific managers.

AI + NixOS

  • Many argue NixOS is uniquely well‑suited to LLM/agent configuration: changes are just config diffs, easy to audit, and safe to roll back.
  • People report success having agents refactor flakes, fix nagging issues, or package complex software.
  • Others note AI still struggles with Nix’s abstractions, option discovery, and can hallucinate packages/options; human oversight remains necessary.

Nix Language, Alternatives, and UX

  • Strong split: some grow to like Nix’s functional language; others call it “awful,” “insane,” or wish for a simpler syntax.
  • Guix is repeatedly mentioned as a similar but Scheme‑based alternative, with its own trade‑offs (e.g., S‑expression readability, slower builds, stricter firmware stance).

Disk Space, Performance, and Auto‑Updating Apps

  • Nix can consume more disk as a cache, but users say garbage collection and optimization tame it; some even report lower usage than other distros.
  • Complaints about slowness (builds, upgrades) and complexity are common.
  • Auto‑updating apps (Discord, Slack, Docker Desktop, JetBrains Toolbox) don’t fit well with Nix’s purity; people fall back to Flatpak, Homebrew, or impure installs.

Documentation and Fragmentation

  • Documentation is widely criticized as scattered, outdated, and unfriendly to beginners, especially around flakes and best practices.
  • Community often compensates by reading nixpkgs source, wikis, forum posts, and increasingly, by relying on LLMs to “stitch it together.”