flake: add remaining drvs to ci.buildbot #2060

0xda157 · 2025-12-03T21:47:14Z

Closes: https://github.com/nix-community/stylix/issues/2056

I certify that I have the right to submit this contribution under the MIT license
Commit messages adhere to Stylix commit conventions
Theming changes adhere to the Stylix style guide
Changes have been tested locally
Changes have been tested in testbeds
Each commit in this PR is suitable for backport to the current stable branch

trueNAHO

Although this is not a critical test, it would be great to also evaluate formatter.${system} in ci.buildbot.

flake/dev/packages.nix

flake/dev/pre-commit.nix

flake/dev/public-and-dev-version-consistency.nix

MattSturgeon

A couple prompts/questions about grouping up some of the related/trivial checks for the purpose of reducing the number of workers/jobs buildbot-nix needs to spin up:

flake/dev/public-and-dev-version-consistency.nix

MattSturgeon · 2025-12-06T23:37:30Z

flake/dev/treefmt.nix

-          indentWidth = 2;
-          lineWidth = 80;
+      treefmt = {
+        projectRootFile = "flake.nix";


Off-topic: this projectRootFile is no longer correct, now that you have two flake.nix files in-tree.

You could use "flake/dev/flake.nix" or any other path that can only be resolved when at the root of the flake.

We don't really need a flakeRootFile now, since treefmt knows to ask git rev-parse --show-toplevel if it's not defined. However, I don't think treefmt-nix is updated to know that, so it will always define it...

flake/dev/pre-commit.nix

flake/dev/public-and-dev-version-consistency.nix

.github/workflows/pre-commit.yml

flake/dev/public-and-dev-version-consistency.nix

MattSturgeon · 2025-12-07T01:17:41Z

Testing locally:

$ nix repl --option extra-experimental-features pipe-operators

nix-repl> :lf .
Added 26 variables.

nix-repl> :a inputs.nixpkgs.lib
Added 465 variables.

nix-repl> ci.buildbot |> foldlAttrs (acc: system: checks: acc ++ (checks |> attrNames |> map (name: "${system}.${name}"))) []
[
  "aarch64-darwin.devShells"
  "aarch64-darwin.packages"
  "aarch64-darwin.public-and-dev-version-consistency"
  "aarch64-darwin.testbeds"
  "aarch64-linux.devShells"
  "aarch64-linux.packages"
  "aarch64-linux.public-and-dev-version-consistency"
  "aarch64-linux.testbeds"
  "x86_64-darwin.devShells"
  "x86_64-darwin.packages"
  "x86_64-darwin.public-and-dev-version-consistency"
  "x86_64-darwin.testbeds"
  "x86_64-linux.devShells"
  "x86_64-linux.packages"
  "x86_64-linux.public-and-dev-version-consistency"
  "x86_64-linux.testbeds"
]

However, I can't see any of the new attrs being built here: https://buildbot.nix-community.org/#/builders/7620/builds/158

Does buildbot-nix get the attrs to be built from the base branch instead of the head of the PR?

0xda157 · 2025-12-07T04:30:41Z

Does buildbot-nix get the attrs to be built from the base branch instead of the head of the PR?

it shouldn't. cc @zowoq

zowoq · 2025-12-07T05:08:09Z

https://buildbot.nix-community.org/api/v2/logs/1848491/raw_inline

Looks like they are skipped because they already built and cached?

trueNAHO

What about evaluating formatter.${system} in ci.buildbot? I assume this is a negligible performance hit and ensures everything is tested.

MattSturgeon · 2025-12-10T16:18:34Z

What about evaluating formatter.${system} in ci.buildbot?

The formatter derivation (treefmt) is a dependency of your default devShell, which is built by buildbot-nix:

      ci.buildbot = { inherit (config) devShells; };

(Incidentally, that should probably be a linkFarm, for the same reasons as elsewhere:

      ci.buildbot.devShells = pkgs.linkFarm "devShells" config.devShells;

0xda157 · 2025-12-10T17:08:02Z

What about evaluating formatter.${system} in ci.buildbot?

The formatter derivation (treefmt) is a dependency of your default devShell, which is built by buildbot-nix:
      ci.buildbot = { inherit (config) devShells; };
(Incidentally, that should probably be a linkFarm, for the same reasons as elsewhere:
      ci.buildbot.devShells = pkgs.linkFarm "devShells" config.devShells;

treefmt is also built as part of the pre-commit drv

Closes: nix-community#2056

MattSturgeon · 2025-12-23T01:53:39Z

.github/workflows/check.yml

If removing this workflow, you'll need to update your ruleset just before you merge; currently this PR is blocked waiting on required checks that'll never run.

If removing this workflow, you'll need to update your ruleset just before you merge

Will this not automatically be updated once this workflow is no longer on the target master branch? In that case, pending PRs that are not rebased on top of this patchset will implicitly use an incomplete Buildbot CI, meaning pending PRs should be rebased before merging.

currently this PR is blocked waiting on required checks that'll never run.

I force merged this PR for now.

Will this not automatically be updated once this workflow is no longer on the target master branch? In that case, pending PRs that are not rebased on top of this patchset will implicitly use an incomplete Buildbot CI, meaning pending PRs should be rebased before merging.

No. Required status checks relates to the name of the status check, it has nothing to do with the workflow. For workflows, the status check name is the same as the job name.

Currently your ruleset will have something like:

Required status checks:

"x86_64-linux"

"aarch64-linux"

"x86_64-darwin"

"aarch64-darwin"

I.e. it expects status checks with each of the above names to be reported as "successful" or "skipped". It doesn't know or care that these come from jobs in the check.yml workflow.

You need to update your ruleset to instead expect the nix-build status check, as used by buildbot. Or, for belts and braces, nix-eval and nix-build.

In that case, pending PRs that are not rebased on top of this patchset will implicitly use an incomplete Buildbot CI

That's already the case, and is unrelated to your ruleset. Buildbot will run on all branches, using that branch's configuration and flake outputs.

Other PRs not rebased on this should still run the GHA workflow, since it is still present in their branch. You'll have to manually pay attention to this during the transition period once it is not a "required" status check.

Even without "required status checks", you will still notice the big red ❌ on the PR indicating that at least some CI thing failed.

Will this not automatically be updated once this workflow is no longer on the target master branch? In that case, pending PRs that are not rebased on top of this patchset will implicitly use an incomplete Buildbot CI, meaning pending PRs should be rebased before merging.

No. Required status checks relates to the name of the status check, it has nothing to do with the workflow. For workflows, the status check name is the same as the job name.

Currently your ruleset will have something like:

Required status checks:

"x86_64-linux"

"aarch64-linux"

"x86_64-darwin"

"aarch64-darwin"

I.e. it expects status checks with each of the above names to be reported as "successful" or "skipped". It doesn't know or care that these come from jobs in the check.yml workflow.

You need to update your ruleset to instead expect the nix-build status check, as used by buildbot. Or, for belts and braces, nix-eval and nix-build.

Thanks as always for the thorough explanation.

Based on the CI status from commit 551df12 ("flake: add remaining drvs to ci.buildbot (#2060)"), it seems the buildbot/nix-build, buildbot/nix-effects, and buildbot/nix-eval checks are already included. According to nix-community/infra#2027 (comment), this happened automatically. In that case, only the following assumed entries have to be removed:

"x86_64-linux"

"aarch64-linux"

"x86_64-darwin"

"aarch64-darwin"

AFAICT, I do not have permission to view or modify the rulesets. IIUC, @nix-community/stylix-committers seems to indicate that @danth has the necessary permissions. IIRC, #655 (comment) is the last time @danth changed the rulesets.

In that case, pending PRs that are not rebased on top of this patchset will implicitly use an incomplete Buildbot CI

That's already the case, and is unrelated to your ruleset. Buildbot will run on all branches, using that branch's configuration and flake outputs.

Other PRs not rebased on this should still run the GHA workflow, since it is still present in their branch. You'll have to manually pay attention to this during the transition period once it is not a "required" status check.

Even without "required status checks", you will still notice the big red ❌ on the PR indicating that at least some CI thing failed.

I suppose unless we force merge, the broken CI will be obvious enough that no extra precautions are needed.

The point of required status checks is to make "force" merging necessary when the specified check failed (or hasn't run).

I.e. you can't normally merge a PR until all required status checks have finished successfully (or been skipped).

The required status checks ruleset rule doesn't affect which checks actually run.

Any repo admin or owner, and any org owner, can manage rulesets. ~~Cc @zowoq.~~

I've updated the ruleset to require the new Buildbot checks, and removed the old GitHub Actions ones. Existing PRs will show the Buildbot checks as "expected" until a new commit is pushed, triggering a CI run.

trueNAHO

LGTM.

stylix-automation · 2025-12-24T19:04:00Z

Successfully created backport PR for release-25.11:

[release-25.11] flake: add remaining drvs to ci.buildbot #2105

stylix-automation bot added the topic: flake /flake.nix, /flake.lock, and /flake/ subsystems label Dec 3, 2025

0xda157 force-pushed the builtbot-full branch 2 times, most recently from dc1e859 to 39615fa Compare December 4, 2025 00:50

trueNAHO requested changes Dec 4, 2025

View reviewed changes

flake/dev/packages.nix Show resolved Hide resolved

flake/dev/pre-commit.nix Show resolved Hide resolved

flake/dev/public-and-dev-version-consistency.nix Outdated Show resolved Hide resolved

trueNAHO added the backport: release-25.11 To be backported to the release-25.11 stable branch label Dec 4, 2025

0xda157 force-pushed the builtbot-full branch from 39615fa to 380ec70 Compare December 5, 2025 23:54

stylix-automation bot added the status: merge conflict Merge conflict label Dec 6, 2025

0xda157 force-pushed the builtbot-full branch from 380ec70 to c49a504 Compare December 6, 2025 00:11

0xda157 removed the status: merge conflict Merge conflict label Dec 6, 2025

0xda157 force-pushed the builtbot-full branch from c49a504 to e013472 Compare December 6, 2025 23:13

MattSturgeon reviewed Dec 6, 2025

View reviewed changes

0xda157 force-pushed the builtbot-full branch from e013472 to 50fa0a2 Compare December 6, 2025 23:56

stylix-automation bot added the topic: ci /.github/ subsystem label Dec 6, 2025

0xda157 force-pushed the builtbot-full branch from 50fa0a2 to 653f10e Compare December 7, 2025 00:00

MattSturgeon reviewed Dec 7, 2025

View reviewed changes

flake/dev/public-and-dev-version-consistency.nix Outdated Show resolved Hide resolved

MattSturgeon reviewed Dec 7, 2025

View reviewed changes

.github/workflows/pre-commit.yml Outdated Show resolved Hide resolved

0xda157 force-pushed the builtbot-full branch 2 times, most recently from 4632081 to 7d28206 Compare December 7, 2025 00:40

MattSturgeon reviewed Dec 7, 2025

View reviewed changes

flake/dev/public-and-dev-version-consistency.nix Outdated Show resolved Hide resolved

0xda157 requested a review from trueNAHO December 9, 2025 06:22

trueNAHO requested changes Dec 10, 2025

View reviewed changes

0xda157 force-pushed the builtbot-full branch from 7d28206 to 72dc809 Compare December 22, 2025 21:02

0xda157 requested a review from trueNAHO December 22, 2025 21:02

0xda157 added 2 commits December 22, 2025 13:03

flake: add remaining drvs to ci.buildbot

7420f20

Closes: nix-community#2056

ci: don't run checks on github actions

69f1522

0xda157 force-pushed the builtbot-full branch from 72dc809 to 69f1522 Compare December 22, 2025 21:03

MattSturgeon approved these changes Dec 23, 2025

View reviewed changes

trueNAHO approved these changes Dec 24, 2025

View reviewed changes

trueNAHO merged commit 551df12 into nix-community:master Dec 24, 2025
3 checks passed

github-actions bot added the has: port to stable This has been backported to the latest stable branch label Dec 24, 2025

stylix-automation bot mentioned this pull request Dec 24, 2025

[release-25.11] flake: add remaining drvs to ci.buildbot #2105

Merged

0xda157 deleted the builtbot-full branch December 26, 2025 02:11

Uh oh!

flake: add remaining drvs to ci.buildbot #2060

flake: add remaining drvs to ci.buildbot #2060

Uh oh!

Conversation

0xda157 commented Dec 3, 2025

Uh oh!

trueNAHO left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

MattSturgeon left a comment • 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!

Uh oh!

MattSturgeon Dec 6, 2025 • 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!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

MattSturgeon commented Dec 7, 2025

Uh oh!

0xda157 commented Dec 7, 2025

Uh oh!

zowoq commented Dec 7, 2025

Uh oh!

trueNAHO left a comment • 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!

MattSturgeon commented Dec 10, 2025 • edited by trueNAHO Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

0xda157 commented Dec 10, 2025

Uh oh!

MattSturgeon Dec 23, 2025

Choose a reason for hiding this comment

Uh oh!

trueNAHO Dec 24, 2025

Choose a reason for hiding this comment

Uh oh!

MattSturgeon Dec 24, 2025

Choose a reason for hiding this comment

Uh oh!

trueNAHO Dec 25, 2025

Choose a reason for hiding this comment

Uh oh!

MattSturgeon Dec 25, 2025 • 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!

danth Dec 25, 2025 • 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!

trueNAHO left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

stylix-automation bot commented Dec 24, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

MattSturgeon left a comment •

edited

Loading

MattSturgeon Dec 6, 2025 •

edited

Loading

trueNAHO left a comment •

edited

Loading

MattSturgeon commented Dec 10, 2025 •

edited by trueNAHO

Loading

MattSturgeon Dec 25, 2025 •

edited

Loading

danth Dec 25, 2025 •

edited

Loading