The article conflates several concepts about what is and isn't expected from a reproducible build system.<p>This comment sums it up pretty well:
<a href="https://github.com/NorfairKing/nix-does-not-guarantee-reproducibility/issues/1#issue-1853130857" rel="nofollow">https://github.com/NorfairKing/nix-does-not-guarantee-reprod...</a> (<a href="https://archive.is/KKx0V" rel="nofollow">https://archive.is/KKx0V</a>)<p>Additionally, I find the author's response of "PRs welcome" somewhat unsettling. It comes across to me like a poor-taste joke: someone writes an opinion piece with misleading information, then expects those pointing out the errors to do the work of correcting them.<p>I would be remiss if I didn't mention Brandolini's law.<p>A few references:<p>- A more nuanced interpretation of reproducibility vs repeatability:
<a href="https://discourse.nixos.org/t/nixos-is-not-reproducible/42688/4" rel="nofollow">https://discourse.nixos.org/t/nixos-is-not-reproducible/4268...</a><p>- Reproducibility rates on Nixpkgs:
<a href="https://arxiv.org/abs/2501.15919" rel="nofollow">https://arxiv.org/abs/2501.15919</a>