lixPackageSets.{lix_2_93,git}: init by alois31 · Pull Request #398141 · NixOS/nixpkgs

alois31 · 2025-04-12T13:35:33Z

Things done

Add a 👍 reaction to pull requests you find important.

alois31 · 2025-04-12T13:51:38Z

The checks performed are as follows:

built on x86_64-linux up to: lixPackageSets.git.lix.passthru.tests, lixPackageSets.git.nix-fast-build, pkgsStatic.lixPackageSets.git.lix pkgsCross.aarch64-multiplatform.lixPackageSets.git.nix-fast-build
evaluated my entire configuration with it

Updates roughly once a week with the same testing are planned. Please tell me if I missed some things I should test and I can add it to the list. (Unfortunately anything Darwin is not possible for me since I do not have appropriate hardware.)

lf-

overall lgtm, I've pointed to a few tiny nits but we can get this going pretty soon!

pkgs/tools/package-management/lix/common-lix.nix

lf- · 2025-04-26T17:20:54Z

pkgs/tools/package-management/lix/common-lix.nix

huh, can you remind me why we depend on yq? i actually don't remember how that got in there.

The dependency was introduced in https://gerrit.lix.systems/c/lix/+/2190 for some parser tests so that the output is more readable.

ah that's a little bit silly of us since we probably should just use PyYAML instead which we already depend on but oh well.

lf- · 2025-04-26T17:22:25Z

pkgs/tools/package-management/lix/common-nix-eval-jobs.nix

hm, i suspect we will need a lix branded capnproto package really soon if not now, due to patches we're doing to it. need to check details though.

The library is already used by Lix itself. This only adds the dependency to nix-eval-jobs which started depending on canproto headers due to the async changes. If a special capnproto build is needed that could be handled similar to boehmgc, aws-sdk-cpp and editline so that it is picked up by both.

ah okay! that should be fine then. we could potentially address this need via propagated build inputs as well; the only reason i was a little shaky on wanting to do it as propagated is that it would also affect derivations using lix only for the cli.

either approach is fine.

Correct, I think it might even have been me who suggested making it propagated. I decided to add it to nix-eval-jobs for now at least since that's also done in the vendored packaging.

pkgs/tools/package-management/lix/default.nix

lf-

looks good to me, and derisks 2.93 packaging issues. wins all around.

getchoo

Would be awesome to get this merged for 25.05. Maybe add a release note for it too?

`nixpkgs-review` result

Generated using nixpkgs-review.

Command: nixpkgs-review pr 398141

`x86_64-linux`

✅ 30 packages built:

lixPackageSets.git.aws-sdk-cpp (lixPackageSets.lix_2_93.aws-sdk-cpp)
lixPackageSets.git.aws-sdk-cpp.dev (lixPackageSets.lix_2_93.aws-sdk-cpp.dev)
lixPackageSets.git.boehmgc (lixPackageSets.lix_2_93.boehmgc)
lixPackageSets.git.boehmgc.debug (lixPackageSets.lix_2_93.boehmgc.debug)
lixPackageSets.git.boehmgc.dev (lixPackageSets.lix_2_93.boehmgc.dev)
lixPackageSets.git.boehmgc.doc (lixPackageSets.lix_2_93.boehmgc.doc)
lixPackageSets.git.editline (lixPackageSets.lix_2_93.editline)
lixPackageSets.git.editline.dev (lixPackageSets.lix_2_93.editline.dev)
lixPackageSets.git.editline.doc (lixPackageSets.lix_2_93.editline.doc)
lixPackageSets.git.editline.man (lixPackageSets.lix_2_93.editline.man)
lixPackageSets.git.lix
lixPackageSets.git.lix.debug
lixPackageSets.git.lix.dev
lixPackageSets.git.lix.devdoc
lixPackageSets.git.lix.doc
lixPackageSets.git.lix.man
lixPackageSets.git.nix-direnv
lixPackageSets.git.nix-eval-jobs
lixPackageSets.git.nix-fast-build
lixPackageSets.git.nix-fast-build.dist
lixPackageSets.latest.lix (lixPackageSets.lix_2_93.lix)
lixPackageSets.latest.lix.debug (lixPackageSets.lix_2_93.lix.debug)
lixPackageSets.latest.lix.dev (lixPackageSets.lix_2_93.lix.dev)
lixPackageSets.latest.lix.devdoc (lixPackageSets.lix_2_93.lix.devdoc)
lixPackageSets.latest.lix.doc (lixPackageSets.lix_2_93.lix.doc)
lixPackageSets.latest.lix.man (lixPackageSets.lix_2_93.lix.man)
lixPackageSets.latest.nix-direnv (lixPackageSets.lix_2_93.nix-direnv)
lixPackageSets.latest.nix-eval-jobs (lixPackageSets.lix_2_93.nix-eval-jobs)
lixPackageSets.latest.nix-fast-build (lixPackageSets.lix_2_93.nix-fast-build)
lixPackageSets.latest.nix-fast-build.dist (lixPackageSets.lix_2_93.nix-fast-build.dist)

getchoo · 2025-05-15T19:25:38Z

pkgs/tools/package-management/lix/default.nix

Both aarch64/x86_64-darwin are failing here and in the git package, but only the former is a regression from 2.92

It seems this is a repeat of #353576, as applying this diff fixed it:

diff --git a/pkgs/tools/package-management/lix/common-lix.nix b/pkgs/tools/package-management/lix/common-lix.nix index 13f6d88828e6..635f52082140 100644 --- a/pkgs/tools/package-management/lix/common-lix.nix +++ b/pkgs/tools/package-management/lix/common-lix.nix @@ -254,7 +254,8 @@ stdenv.mkDerivation (finalAttrs: { # Enable LTO, since it improves eval performance a fair amount # LTO is disabled on: # - static builds (strange linkage errors) - (lib.mesonBool "b_lto" (!stdenv.hostPlatform.isStatic && (isLLVMOnly || stdenv.cc.isGNU))) + # - darwin builds (install test failures) + (lib.mesonBool "b_lto" (!stdenv.hostPlatform.isStatic && !stdenv.hostPlatform.isDarwin && (isLLVMOnly || stdenv.cc.isGNU))) (lib.mesonEnable "gc" true) (lib.mesonBool "enable-tests" true) (lib.mesonBool "enable-docs" enableDocumentation)

It might be best to keep this, as I'm unsure if it's UB from the test's code (from a quick look this seems unlikely) or the LTO-ized Lix binary. The only side effect it would have is retroactively disabling LTO on Lix 2.92 for Darwin (as that's the only place it's enabled currently) where tests do pass; though this might still be an issue for 2.92 and it just isn't being caught

Edit: reported upstream in https://git.lix.systems/lix-project/lix/issues/832

getchoo · 2025-05-15T19:29:24Z

pkgs/tools/package-management/lix/default.nix

Should we set this as stable too? AFAIK there are no upstream or third party blockers like 2.92, and @RaitoBezarius signaled towards skipping over that release anyways here

I do think we have one major regression, which is the SSH wrapper stuff, atm, that I will be trying to fix in https://gerrit.lix.systems/c/lix/+/3159 and backport it. cc @lf-

yuyuyureka · 2025-05-16T13:19:35Z

Can we go ahead with bumping lixVersions.latest to this while keeping lixVersions.stable at 2.91? Then we can sort out the remaining issues asynchronously.

RaitoBezarius · 2025-05-16T15:41:37Z

Can we go ahead with bumping lixVersions.latest to this while keeping lixVersions.stable at 2.91? Then we can sort out the remaining issues asynchronously.

In agreement with this.

Previously always the top-level lix package would be used, which is obviously not what we want.

github-actions bot added 6.topic: teams Relating to team creation, updates, other management actions 10.rebuild-darwin: 1-10 This PR causes between 1 and 10 packages to rebuild on Darwin. 10.rebuild-linux: 11-100 This PR causes between 11 and 100 packages to rebuild on Linux. labels Apr 12, 2025

alois31 marked this pull request as ready for review April 12, 2025 13:51

Kranzes requested review from 9999years and lf- and removed request for lf- April 18, 2025 11:46

alois31 force-pushed the push-qwwlwyylnqpt branch from e2d8e3e to b5c0cb6 Compare April 26, 2025 14:55

lf- reviewed Apr 26, 2025

View reviewed changes

alois31 force-pushed the push-qwwlwyylnqpt branch from b5c0cb6 to 016506a Compare April 26, 2025 17:45

github-actions bot added 10.rebuild-linux: 1-10 This PR causes between 1 and 10 packages to rebuild on Linux. and removed 10.rebuild-linux: 11-100 This PR causes between 11 and 100 packages to rebuild on Linux. labels Apr 26, 2025

lf- approved these changes Apr 26, 2025

View reviewed changes

wegank added the 12.approvals: 1 This PR was reviewed and approved by one person. label Apr 26, 2025

alois31 force-pushed the push-qwwlwyylnqpt branch 2 times, most recently from b24672a to ca5fd40 Compare May 7, 2025 04:04

alois31 force-pushed the push-qwwlwyylnqpt branch from ca5fd40 to c889250 Compare May 10, 2025 09:34

github-actions bot added the 6.topic: nixos Issues or PRs affecting NixOS modules, or package usability issues specific to NixOS label May 10, 2025

alois31 changed the title ~~lixPackageSets.git: init at 2.93.0-pre-20250410_81d476984901~~ lixPackageSets.{lix_2_93,git}: init May 10, 2025

nix-owners bot requested review from Qyriad and RaitoBezarius May 10, 2025 09:42

getchoo reviewed May 15, 2025

View reviewed changes

lixPackageSets: support splicing

2ad2194

alois31 added 3 commits May 16, 2025 17:43

lixPackageSets.*.lix: test the correct Lix in passthru.tests

8966395

Previously always the top-level lix package would be used, which is obviously not what we want.

lixPackageSets.lix_2_93: init at 2.93.0

cf78bf6

lixPackageSets.git: init at 2.94.0-pre-20250509_dcb0a97000d5

39529f9

yuyuyureka force-pushed the push-qwwlwyylnqpt branch from c889250 to 39529f9 Compare May 16, 2025 15:44

ipetkov mentioned this pull request May 16, 2025

lixPackageSets.stable.lix: 2.91 -> 2.92 #405718

Closed

13 tasks

RaitoBezarius approved these changes May 16, 2025

View reviewed changes

yuyuyureka merged commit 62536dc into NixOS:master May 16, 2025
17 of 21 checks passed

alois31 deleted the push-qwwlwyylnqpt branch August 2, 2025 12:04

Uh oh!

Conversation

alois31 commented Apr 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Things done

Uh oh!

alois31 commented Apr 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

lf- left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

lf- left a comment

Choose a reason for hiding this comment

Uh oh!

getchoo left a comment

Choose a reason for hiding this comment

nixpkgs-review result

x86_64-linux

Uh oh!

getchoo May 15, 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!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

yuyuyureka commented May 16, 2025

Uh oh!

RaitoBezarius commented May 16, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

alois31 commented Apr 12, 2025 •

edited

Loading

alois31 commented Apr 12, 2025 •

edited

Loading

`nixpkgs-review` result

`x86_64-linux`

getchoo May 15, 2025 •

edited

Loading