Ali Rizvi Frontear

Benchmarking Nix (and its derivatives)

The configuration used for testing was my own, and can be found at Frontear/dotfiles. To test in much the same way, the following command was used to run 10 isolated benchmarks:

# This was run within the root of my repo.
# To obtain working versions of the other nix* derivatives, use `nix shell` so that there is no overhead in processing the store.

for i in $(seq 1 10); do
 time nix eval '.#nixosConfigurations.LAPTOP-3DT4F02.config.system.build.toplevel.drvPath' --no-eval-cache

String interpolating a relative or absolute path value will cause Nix to copy the contents at that path to the store before constructing the string. (NixOS/nix#9428)

This has I/O implications for evaluation times, especially for large files/directories
This will error out if the path does not exist or if the user lacks permissions to view it
In a flake-controlled project (which is already copied as part of evaluation), this creates duplicate store objects

Solutions:

Avoid string interpolating path values. Use builtins.toString if you need the values to be a string.
Avoid using path values in general, and prefer using a generic string ("/var/empty" instead of /var/empty)

	{
	inputs = {
	nixpkgs.url = "github:NixOS/nixpkgs/nixos-unstable";
	};

	outputs = { self, ... } @ inputs:
	let
	# https://ayats.org/blog/no-flake-utils
	eachSystem = function: inputs.nixpkgs.lib.genAttrs [
	"x86_64-linux"

	#!/usr/bin/env python3

	import yaml
	import colorsys

	def hex2rgb(hex: str) -> tuple[int]:
	return tuple(int(hex[i:i+2], 16) for i in (0, 2, 4))

	def rgb2hex(rgb: tuple[int]):
	return "%x%x%x" % rgb