npc: init at 1.0.0 by me-and · Pull Request #543008 · NixOS/nixpkgs

me-and · 2026-07-17T17:55:18Z

npc is an MIT-licensed tool for working with Nix channels and channel histories. In particular, it provides an npc bisect command that provides similar function to git bisect but only considering commits that have actually been pointed to by a relevant Nixpkgs release branch.

@samestep I've added you as a maintainer, since you're the author, but you obviously get to opt out!

Things done

samestep

@me-and it makes me so happy to see this, thank you! Unfortunately there's a blocker; see my comments below. Would love to hear your thoughts if you have a suggestion to unblock this!

samestep · 2026-07-17T18:01:19Z

+
+  env = {
+    NPC_REV = finalAttrs.src.rev;
+    GIT_BIN = lib.getExe git;


Unfortunately this is wrong; see samestep/npc#6. What do you think is the best solution here?

Try sending another email on the Git mailing list to see if people respond this time?

Add Jonathan Tan's one-line patch to the Git package definition in Nixpkgs, and target staging.

Use that patch but only for the Git that gets fed to npc itself.

Something else?

I spotted that; I'd assumed (perhaps naively) that the bug was at least rare, so it'd be fine to just live with it until it gets fixed in the Git source code.

That said, I rather expect submitting Jonathan Tan's one-line patch as a properly formatted patch to the Git mailing list would get it accepted. And once there's a patch on the mailing list, regardless of whether it's accepted or not, I think it'd be reasonable to add it to the Nixpkgs Git package.

I'd assumed (perhaps naively) that the bug was at least rare, so it'd be fine to just live with it until it gets fixed in the Git source code.

It's been several months since I've tried running npd with Git 2.48+, but if I remember correctly, I was hitting the bug pretty consistently before I pinned the older Git version. If it's not too much to ask, would you be able to try it out on your end to see if you hit it?

That said, I rather expect submitting Jonathan Tan's one-line patch as a properly formatted patch to the Git mailing list would get it accepted.

Yeah I think you're right 😅 to be honest, I've had this on my TODO list for months now. Obviously the patch itself is super easy, but then I'd also have to include a regression test, and I just haven't gotten around to understanding the Git codebase's testing setup well enough to feel confident in writing and sending it...

FWIW, I just ran the following and didn't hit the problem in any of the ten test clones

( set -xeuo pipefail; for (( n=0; n<10; n++ )); do d="$(mktemp -d)"; git clone --mirror --filter=tree:0 https://github.com/NixOS/nixpkgs.git "$d"; rm -rf "$d"; done )

Right, the initial clone works fine; it's the git fetch that gives the error. Although, I did also recently change how the fetching works in samestep/npc@8d5e25b, so maybe it avoids the bug now somehow? I'd be surprised though.

Yep, that was 50 runs without reproducing the bug with git v2.54.0! I'll test with v2.49 now, just to be sure it's not an environmental issue...

Oh wow! And I assume this is after the more recent change to how npc fetches? If you want, it may also be useful to try with an npc commit prior to that change.

I'll look into sending that patch on the Git mailing list later today.

Okay, I'm not managing to reproduce the problem with git v2.49.0 either. Specifically, I've just run the following and didn't see any issues:

( set -euo pipefail; git () { /nix/store/lchj5sfm6yyhv5jhahh5xr2fg239218r-git-2.49.0/bin/git "$@"; }; d="$(mktemp -d)"; git clone --mirror --filter=tree:0 https://github.com/NixOS/nixpkgs.git "$d"; cd "$d"; for (( n=0; n<20; n++ )); do git fetch --no-show-forced-updates; done )

So I don't think we can conclude anything from my testing other than there's some environmental factor at play :(

Thank you for testing! I'll do some more testing on my end later today.

And as a slightly different datapoint: I just rebuilt a couple of my machines using npc from your flake, but with both nixpkgs and nixpkgs-git pinned to a recent nixpkgs-unstable. Twenty runs of npc fetch on two different systems and I didn't see anything untoward.

me-and · 2026-07-17T18:13:28Z

`nixpkgs-review` result

Generated using nixpkgs-review.

Command: nixpkgs-review pr 543008
Commit: 05c990188cc3ac8e1781e14fab5482a295e51a83

`x86_64-linux`

✅ 1 package built:

npc

`aarch64-linux`

✅ 1 package built:

npc

samestep

Welp. I tried a bunch yesterday and today to reproduce the git fetch issue, and I couldn't 🙃

So I'd say ship it! Thanks again for opening this PR 😄

samestep · 2026-07-18T15:59:28Z

Oh wait hang on I tried a bit more and I think I actually reproduced it this time; please hold...

samestep · 2026-07-18T16:24:08Z

OK yep, turns out we're good 🥳

The bug was fixed by git/git@7a4bd1b, which was released in Git 2.53, so the Git 2.54 currently in Nixpkgs works great! I also did samestep/npc#44 upstream.

samestep · 2026-07-18T17:15:58Z

`npb` · `38ced06` → `a495202`

Expand this for a reproducible command.

npb --base 38ced06161c11eb06ba8791240f06ce2a7221052 --head 59682e0069f0ed0a452e2179a7f4c1f247027b9e --patch 59682e0069f0ed0a452e2179a7f4c1f247027b9e...a4952026d240b18ffc5503c9b481aac861fbe35b -s x86_64-linux -s aarch64-linux -s aarch64-darwin

Expand this for a legend of all symbols below.

✅ = successfully built
❌ = failed to build
🚫 = dependency failed to build
⏩ = didn't try to build
❔ = couldn't try to build
➖ = doesn't exist

`x86_64-linux`

➖ → ✅ · 1 package

npc

`aarch64-linux`

➖ → ✅ · 1 package

npc

`aarch64-darwin`

➖ → ✅ · 1 package

npc

nixos-discourse · 2026-08-04T15:05:11Z

This pull request has been mentioned on NixOS Discourse. There might be relevant details there:

https://discourse.nixos.org/t/prs-already-reviewed/2617/3213

samestep suggested changes Jul 17, 2026

View reviewed changes

nixpkgs-ci Bot removed the 9.needs: reviewer This PR currently has no reviewers requested and needs attention. label Jul 17, 2026

me-and force-pushed the npc branch from 05c9901 to 48f1bcc Compare July 17, 2026 18:38

npc: init at 1.0.0

a495202

me-and force-pushed the npc branch from 48f1bcc to a495202 Compare July 17, 2026 22:26

me-and changed the title ~~npc: init at 0-unstable-2026-07-11~~ npc: init at 1.0.0 Jul 17, 2026

samestep approved these changes Jul 18, 2026

View reviewed changes

nixpkgs-ci Bot added the 12.approvals: 1 This PR was reviewed and approved by one person. label Jul 18, 2026

K900 added this pull request to the merge queue Aug 9, 2026

Merged via the queue into NixOS:master with commit 9462939 Aug 9, 2026
29 checks passed

samestep mentioned this pull request Aug 23, 2026

{lixPackageSets.*.,}npb: init at 0-unstable-2026-07-27 #546792

Closed

13 tasks

Uh oh!

Conversation

me-and commented Jul 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Things done

Uh oh!

samestep left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

samestep Jul 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

me-and Jul 17, 2026

Choose a reason for hiding this comment

Uh oh!

samestep Jul 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

me-and Jul 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

samestep Jul 17, 2026

Choose a reason for hiding this comment

Uh oh!

me-and Jul 17, 2026

Choose a reason for hiding this comment

Uh oh!

samestep Jul 17, 2026

Choose a reason for hiding this comment

Uh oh!

me-and Jul 17, 2026

Choose a reason for hiding this comment

Uh oh!

samestep Jul 17, 2026

Choose a reason for hiding this comment

Uh oh!

me-and Jul 17, 2026

Choose a reason for hiding this comment

Uh oh!

me-and commented Jul 17, 2026

nixpkgs-review result

x86_64-linux

aarch64-linux

Uh oh!

samestep left a comment

Choose a reason for hiding this comment

Uh oh!

samestep commented Jul 18, 2026

Uh oh!

samestep commented Jul 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

samestep commented Jul 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

npb · 38ced06 → a495202

x86_64-linux

aarch64-linux

aarch64-darwin

Uh oh!

nixos-discourse commented Aug 4, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

me-and commented Jul 17, 2026 •

edited

Loading

samestep Jul 17, 2026 •

edited

Loading

samestep Jul 17, 2026 •

edited

Loading

me-and Jul 17, 2026 •

edited

Loading

`nixpkgs-review` result

`x86_64-linux`

`aarch64-linux`

samestep commented Jul 18, 2026 •

edited

Loading

samestep commented Jul 18, 2026 •

edited

Loading

`npb` · `38ced06` → `a495202`

`x86_64-linux`

`aarch64-linux`

`aarch64-darwin`