flake.nix declares three inputs and uses two:
inputs = {
flake-utils.url = "github:numtide/flake-utils";
rainix.url = "github:rainlanguage/rainix";
rain.url = "github:rainlanguage/rain.cli";
};
outputs = { flake-utils, rainix, ... }: ...
rain is never referenced. outputs destructures flake-utils and rainix, and the ... swallows the rest.
It is not inert
The unused input is the sole reason flake.lock carries a second, much older rainix. rain.cli has its own rainix dependency, so the lock holds two nodes:
| node |
rev |
lastModified |
reached via |
rainix_2 |
f22d4dc |
1780287289 (2026-06-01) |
root.inputs.rainix — the one actually used |
rainix |
f3bdb28 |
1714764843 (2024-05-03) |
root.inputs.rain → rain.cli |
That second node is a live trap for anyone reading the lock: grepping "rainix": { finds the 2024 one first, and concluding from it that this repo's toolchain is two years stale is wrong by about two years. That already happened once during review of #26.
Ask
Remove rain from inputs and regenerate flake.lock. Nothing in outputs changes, and the two orphaned nodes drop out with it.
Scope is the unused input only. The drift between rainix_2 and CI's RAINIX_SHA is a separate question, tracked in #28 — this issue should not bump the pin.
flake.nixdeclares three inputs and uses two:rainis never referenced.outputsdestructuresflake-utilsandrainix, and the...swallows the rest.It is not inert
The unused input is the sole reason
flake.lockcarries a second, much older rainix.rain.clihas its own rainix dependency, so the lock holds two nodes:rainix_2f22d4dcroot.inputs.rainix— the one actually usedrainixf3bdb28root.inputs.rain→ rain.cliThat second node is a live trap for anyone reading the lock: grepping
"rainix": {finds the 2024 one first, and concluding from it that this repo's toolchain is two years stale is wrong by about two years. That already happened once during review of #26.Ask
Remove
rainfrominputsand regenerateflake.lock. Nothing inoutputschanges, and the two orphaned nodes drop out with it.Scope is the unused input only. The drift between
rainix_2and CI'sRAINIX_SHAis a separate question, tracked in #28 — this issue should not bump the pin.