nixVersions.nix_2_26: fix cross by roberth · Pull Request #392832 · NixOS/nixpkgs

roberth · 2025-03-24T19:29:10Z

This contains a formatting commit, so best to review the commits individually.

Things done

Add a 👍 reaction to pull requests you find important.

roberth · 2025-03-24T19:32:11Z

@ofborg build pkgsCross.armv7l-hf-multiplatform.nixVersions.nix_2_26
@ofborg build pkgsStatic.nixVersions.nix_2_26
@ofborg build nixVersions.nix_2_26

roberth · 2025-03-24T19:44:55Z

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

Exposing the components as a first-class scope is required for cross compilation.

nixpkgs-ci · 2025-03-26T12:47:23Z

Backport failed for release-24.11, because it was unable to cherry-pick the commit(s).

Please cherry-pick the changes locally and resolve any conflicts.

git fetch origin release-24.11
git worktree add -d .worktree/backport-392832-to-release-24.11 origin/release-24.11
cd .worktree/backport-392832-to-release-24.11
git switch --create backport-392832-to-release-24.11
git cherry-pick -x c0df86f0ab7460fd64f5789f1b92d189eb16305d e9494dd2232526279f6229d3f1ce3b6da1609cb3 ac543ea035a0d6b3b8dc5187d82a04fc06c78d20

K900 · 2025-03-26T17:02:27Z

This broke eval. Reverting for now.

❯ nom-build pkgs/top-level/release.nix -A tarball
error:
       … while evaluating an expression to select 'drvPath' on it
         at «internal»:1:552:
       … while calling the 'derivationStrict' builtin
         at «internal»:1:208:
       … while evaluating derivation 'nixpkgs-tarball-25.05pre1234.abcdef'
         whose name attribute is located at /home/k900/gh/NixOS/nixpkgs/pkgs/stdenv/generic/make-derivation.nix:375:7

       … while evaluating attribute 'buildInputs' of derivation 'nixpkgs-tarball-25.05pre1234.abcdef'
         at /home/k900/gh/NixOS/nixpkgs/pkgs/stdenv/generic/make-derivation.nix:422:7:
          421|       depsHostHost                = elemAt (elemAt dependencies 1) 0;
          422|       buildInputs                 = elemAt (elemAt dependencies 1) 1;
             |       ^
          423|       depsTargetTarget            = elemAt (elemAt dependencies 2) 0;

       … while calling the 'getAttr' builtin
         at «internal»:1:500:
       … while calling the 'derivationStrict' builtin
         at «internal»:1:208:
       … while evaluating derivation 'nixpkgs-lib-tests'
         whose name attribute is located at /home/k900/gh/NixOS/nixpkgs/pkgs/stdenv/generic/make-derivation.nix:375:7

       … while evaluating attribute 'paths' of derivation 'nixpkgs-lib-tests'
         at /home/k900/gh/NixOS/nixpkgs/pkgs/build-support/trivial-builders/default.nix:542:9:
          541|         inherit preferLocalBuild allowSubstitutes;
          542|         paths = mapPaths (path: "${path}${stripPrefix}") paths;
             |         ^
          543|         passAsFile = [ "paths" ];

       … while calling anonymous lambda
         at /home/k900/gh/NixOS/nixpkgs/pkgs/build-support/trivial-builders/default.nix:534:33:
          533|     let
          534|       mapPaths = f: paths: map (path:
             |                                 ^
          535|         if path == null then null

       … from call site
         at /home/k900/gh/NixOS/nixpkgs/pkgs/build-support/trivial-builders/default.nix:537:14:
          536|         else if isList path then mapPaths f path
          537|         else f path
             |              ^
          538|       ) paths;

       … while calling anonymous lambda
         at /home/k900/gh/NixOS/nixpkgs/pkgs/build-support/trivial-builders/default.nix:542:27:
          541|         inherit preferLocalBuild allowSubstitutes;
          542|         paths = mapPaths (path: "${path}${stripPrefix}") paths;
             |                           ^
          543|         passAsFile = [ "paths" ];

       … while calling the 'getAttr' builtin
         at «internal»:1:500:
       … while calling the 'derivationStrict' builtin
         at «internal»:1:208:
       … while evaluating the derivation attribute 'name'
         at /home/k900/gh/NixOS/nixpkgs/pkgs/stdenv/generic/make-derivation.nix:375:7:
          374|     // (optionalAttrs (attrs ? name || (attrs ? pname && attrs ? version)) {
          375|       name =
             |       ^
          376|         let

       … while evaluating the `name` attribute passed to builtins.derivationStrict

       … from call site
         at /home/k900/gh/NixOS/nixpkgs/pkgs/stdenv/generic/make-derivation.nix:393:9:
          392|         in
          393|         lib.strings.sanitizeDerivationName (
             |         ^
          394|           if attrs ? name

       … while calling anonymous lambda
         at /home/k900/gh/NixOS/nixpkgs/lib/strings.nix:2704:3:
         2703|   in
         2704|   string:
             |   ^
         2705|   # First detect the common case of already valid strings, to speed those up

       … while evaluating a branch condition
         at /home/k900/gh/NixOS/nixpkgs/lib/strings.nix:2706:3:
         2705|   # First detect the common case of already valid strings, to speed those up
         2706|   if stringLength string <= 207 && okRegex string != null
             |   ^
         2707|   then unsafeDiscardStringContext string

       … in the left operand of the AND (&&) operator
         at /home/k900/gh/NixOS/nixpkgs/lib/strings.nix:2706:33:
         2705|   # First detect the common case of already valid strings, to speed those up
         2706|   if stringLength string <= 207 && okRegex string != null
             |                                 ^
         2707|   then unsafeDiscardStringContext string

       … in the argument of the not operator
         at /home/k900/gh/NixOS/nixpkgs/lib/strings.nix:2706:26:
         2705|   # First detect the common case of already valid strings, to speed those up
         2706|   if stringLength string <= 207 && okRegex string != null
             |                          ^
         2707|   then unsafeDiscardStringContext string

       … while calling the 'lessThan' builtin
         at /home/k900/gh/NixOS/nixpkgs/lib/strings.nix:2706:26:
         2705|   # First detect the common case of already valid strings, to speed those up
         2706|   if stringLength string <= 207 && okRegex string != null
             |                          ^
         2707|   then unsafeDiscardStringContext string

       … while calling the 'stringLength' builtin
         at /home/k900/gh/NixOS/nixpkgs/lib/strings.nix:2706:6:
         2705|   # First detect the common case of already valid strings, to speed those up
         2706|   if stringLength string <= 207 && okRegex string != null
             |      ^
         2707|   then unsafeDiscardStringContext string

       … while evaluating the attribute 'name'
         at /home/k900/gh/NixOS/nixpkgs/pkgs/build-support/trivial-builders/default.nix:63:30:
           62|         enableParallelBuilding = true;
           63|         inherit buildCommand name;
             |                              ^
           64|         passAsFile = [ "buildCommand" ]

       … while evaluating an expression to select 'version' on it
         at /home/k900/gh/NixOS/nixpkgs/lib/tests/test-with-nix.nix:17:42:
           16|
           17| pkgs.runCommand "nixpkgs-lib-tests-nix-${nix.version}"
             |                                          ^
           18|   {

       … from call site
         at /home/k900/gh/NixOS/nixpkgs/lib/tests/release.nix:8:65:
            7|   nix ? pkgs-nixVersions.stable,
            8|   nixVersions ? [ pkgs-nixVersions.minimum nix pkgs-nixVersions.latest ],
             |                                                                 ^
            9|   pkgs-nixVersions ? import ./nix-for-tests.nix { pkgs = pkgsBB; },

       … while calling anonymous lambda
         at /home/k900/gh/NixOS/nixpkgs/lib/tests/nix-for-tests.nix:15:6:
           14| builtins.mapAttrs (
           15|   _: pkg: if builtins.isAttrs pkg then pkg.override { withAWS = false; } else pkg
             |      ^
           16| ) pkgs.nixVersions

       … while calling a functor (an attribute set with a '__functor' attribute)
         at /home/k900/gh/NixOS/nixpkgs/lib/tests/nix-for-tests.nix:15:40:
           14| builtins.mapAttrs (
           15|   _: pkg: if builtins.isAttrs pkg then pkg.override { withAWS = false; } else pkg
             |                                        ^
           16| ) pkgs.nixVersions

       … from call site
         at /home/k900/gh/NixOS/nixpkgs/lib/trivial.nix:1000:7:
          999|     { # TODO: Should we add call-time "type" checking like built in?
         1000|       __functor = self: f;
             |       ^
         1001|       __functionArgs = args;

       … while calling anonymous lambda
         at /home/k900/gh/NixOS/nixpkgs/lib/customisation.nix:169:36:
          168|         # Re-call the function but with different arguments
          169|         overrideArgs = mirrorArgs (newArgs: makeOverridable f (overrideWith newArgs));
             |                                    ^
          170|         # Change the result of the function call by applying g to it

       … while calling a functor (an attribute set with a '__functor' attribute)
         at /home/k900/gh/NixOS/nixpkgs/lib/customisation.nix:169:45:
          168|         # Re-call the function but with different arguments
          169|         overrideArgs = mirrorArgs (newArgs: makeOverridable f (overrideWith newArgs));
             |                                             ^
          170|         # Change the result of the function call by applying g to it

       … from call site
         at /home/k900/gh/NixOS/nixpkgs/lib/trivial.nix:1000:7:
          999|     { # TODO: Should we add call-time "type" checking like built in?
         1000|       __functor = self: f;
             |       ^
         1001|       __functionArgs = args;

       … while calling anonymous lambda
         at /home/k900/gh/NixOS/nixpkgs/lib/customisation.nix:161:7:
          160|     mirrorArgs (
          161|       origArgs:
             |       ^
          162|       let

       … while evaluating a branch condition
         at /home/k900/gh/NixOS/nixpkgs/lib/customisation.nix:173:7:
          172|       in
          173|       if isAttrs result then
             |       ^
          174|         result

       … while calling the 'isAttrs' builtin
         at /home/k900/gh/NixOS/nixpkgs/lib/customisation.nix:173:10:
          172|       in
          173|       if isAttrs result then
             |          ^
          174|         result

       … from call site
         at /home/k900/gh/NixOS/nixpkgs/lib/customisation.nix:163:18:
          162|       let
          163|         result = f origArgs;
             |                  ^
          164|

       error: function 'anonymous lambda' called with unexpected argument 'withAWS'
       at /home/k900/gh/NixOS/nixpkgs/pkgs/tools/package-management/nix/vendor/2_26/packaging/everything.nix:1:1:
            1| {
             | ^
            2|   lib,

emilazy · 2025-03-26T17:06:53Z

Can we please just have a simple derivation for 2.26 that works rather than trying to vendor a bunch of code from the Nix flake that keeps causing packaging issues…? I don’t know if we want to ship 25.05 with nixVersions.latest having this many issues.

Ericson2314 · 2025-03-26T17:36:47Z

I guess the test failure came from lib/tests/nix-for-tests.nix?

Can we please just have a simple derivation for 2.26 that works rather than trying to vendor a bunch of code from the Nix flake that keeps causing packaging issues…?

IMO that is not a totally fair way to look at this. We've been writing the code with the goal of including things in Nixpkgs the whole time --- it's not a "flake first" design at all. We're going from a single "big chungus" package to a package set. Sketchy free-floating overrides like the above are bound to cause problems.

The vast majority of the code changing here is from formatting, too.

emilazy · 2025-03-26T17:53:52Z

It really doesn’t feel like idiomatic Nixpkgs code to me, even for a large package set. There’s parts of the packaging that look load‐bearing but are pure dead code in Nixpkgs, and I think the complexity of packaging/components.nix would have generated substantial review feedback if it wasn’t just being vendored in from the Nix repository.

But I’m more concerned about the fact that the freeze is on the horizon and stuff is still broken because of it. The package wasn’t split up prior to 2.26, and I think splitting it up currently doesn’t give a practical benefit, so having a working nixVersions.latest in the next stable release should take precedence over having it split up, surely?

emilazy · 2025-03-26T17:57:29Z

(And even if the override is sketchy, we probably should not change the public flag API of a package as central as Nix without a deprecation cycle or at least an explicit error message.)

Ericson2314 · 2025-03-26T18:10:26Z

I think splitting it up currently doesn’t give a practical benefit.

IMO there are huge practical benefits, and I wrote an RFC about better layering to spell them out, but it takes a while for these things to bake. For example, we wouldn't need any nixForLinking if everything was using the right stuff downstream.

(And even if the override is sketchy, we probably should not change the public flag API of a package as central as Nix without a deprecation cycle or at least an explicit error message.)

The "nix everything" was supposed to be the compat shim. IMO once we're fully transitioned over, nothing should be using it.

I think the complexity of packaging/components.nix would have generated substantial review feedback

I would like to do that. We're pioneering a lot of stuff, including file sets, so there are a lot of idioms to be shaken out.

But I’m more concerned about the fact that the freeze is on the horizon and stuff is still broken because of it.

Yeah, that is the frustrating part. It's been 6 months since we "just" delayed Meson for 24.11, and somehow it still feels like we are running out out of time.

Mic92 · 2025-03-26T18:59:14Z

I think splitting it up currently doesn’t give a practical benefit.

IMO there are huge practical benefits, and I wrote an RFC about better layering to spell them out, but it takes a while for these things to bake. For example, we wouldn't need any nixForLinking if everything was using the right stuff downstream.

Getting rid of one alias doesn't seem a good enough reasons to have this much complexity in nixpkgs.

(And even if the override is sketchy, we probably should not change the public flag API of a package as central as Nix without a deprecation cycle or at least an explicit error message.)

The "nix everything" was supposed to be the compat shim. IMO once we're fully transitioned over, nothing should be using it.

I am voting for a simpler meson package until this transition is actually being finished.

I think the complexity of packaging/components.nix would have generated substantial review feedback

I would like to do that. We're pioneering a lot of stuff, including file sets, so there are a lot of idioms to be shaken out.

But I’m more concerned about the fact that the freeze is on the horizon and stuff is still broken because of it.

Yeah, that is the frustrating part. It's been 6 months since we "just" delayed Meson for 24.11, and somehow it still feels like we are running out out of time.

I also feel like I no longer have a good overview what differences between the nixpkgs and the nix repo is. There subtle different.

This was done in NixOS#392832, but then that was reverted. Want to land the formatting right away before making the next version we hopefully don't need to revert.

This was done in #392832, but then that was reverted. Want to land the formatting right away before making the next version we hopefully don't need to revert. (cherry picked from commit ab043a8)

Mic92 · 2025-03-26T21:14:57Z

#393509

This was done in #392832, but then that was reverted. Want to land the formatting right away before making the next version we hopefully don't need to revert. (cherry picked from commit ab043a8)

roberth added 2 commits March 24, 2025 19:19

nix/default.nix: Unused parameter

c0df86f

nixVersions.nix_2_26: Fix cross and static

e9494dd

ofborg bot added the 6.topic: cross-compilation Building packages on a different platform than they will be used on label Mar 24, 2025

nix-owners bot requested review from Ericson2314, Mic92, RaitoBezarius, edolstra and tomberek March 24, 2025 19:31

github-actions bot added 10.rebuild-darwin: 0 This PR does not cause any packages to rebuild on Darwin. 10.rebuild-linux: 0 This PR does not cause any packages to rebuild on Linux. labels Mar 24, 2025

roberth mentioned this pull request Mar 24, 2025

nixVersions.nix_2_26: cross-compiled nix fails to evaluate #381952

Closed

3 tasks

roberth commented Mar 24, 2025

View reviewed changes

package-management/nix: nixfmt

ac543ea

Mic92 force-pushed the nix-cross branch from 61a3639 to ac543ea Compare March 25, 2025 08:47

Ericson2314 approved these changes Mar 25, 2025

View reviewed changes

wegank added 12.approvals: 1 This PR was reviewed and approved by one person. 12.approved-by: package-maintainer This PR was reviewed and approved by a maintainer listed in any of the changed packages. labels Mar 25, 2025

Mic92 merged commit f7a715c into NixOS:master Mar 26, 2025
31 of 41 checks passed

roberth added the backport release-24.11 label Mar 26, 2025

K900 mentioned this pull request Mar 26, 2025

Revert "nixVersions.nix_2_26: fix cross" #393428

Merged

emilazy mentioned this pull request Mar 26, 2025

GitHub Actions eval still doesn’t check every channel‐blocking eval issue #393430

Closed

Ericson2314 mentioned this pull request Mar 26, 2025

pkgs/tools/package-management/nix/default.nix: Format #393498

Merged

13 tasks

Ericson2314 mentioned this pull request Mar 26, 2025

nixVersions.nix_2_26: fix cross #393530

Merged

13 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

nixVersions.nix_2_26: fix cross#392832

nixVersions.nix_2_26: fix cross#392832
Mic92 merged 3 commits intoNixOS:masterfrom
roberth:nix-cross

roberth commented Mar 24, 2025 •

edited

Loading

Uh oh!

roberth commented Mar 24, 2025 •

edited

Loading

Uh oh!

roberth Mar 24, 2025

Uh oh!

Uh oh!

nixpkgs-ci bot commented Mar 26, 2025

Uh oh!

K900 commented Mar 26, 2025

Uh oh!

emilazy commented Mar 26, 2025

Uh oh!

Ericson2314 commented Mar 26, 2025 •

edited

Loading

Uh oh!

emilazy commented Mar 26, 2025

Uh oh!

emilazy commented Mar 26, 2025

Uh oh!

Ericson2314 commented Mar 26, 2025

Uh oh!

Mic92 commented Mar 26, 2025

Uh oh!

Mic92 commented Mar 26, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

Uh oh!

Conversation

roberth commented Mar 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Things done

Uh oh!

roberth commented Mar 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

roberth Mar 24, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

nixpkgs-ci bot commented Mar 26, 2025

Uh oh!

K900 commented Mar 26, 2025

Uh oh!

emilazy commented Mar 26, 2025

Uh oh!

Ericson2314 commented Mar 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

emilazy commented Mar 26, 2025

Uh oh!

emilazy commented Mar 26, 2025

Uh oh!

Ericson2314 commented Mar 26, 2025

Uh oh!

Mic92 commented Mar 26, 2025

Uh oh!

Mic92 commented Mar 26, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

roberth commented Mar 24, 2025 •

edited

Loading

roberth commented Mar 24, 2025 •

edited

Loading

Ericson2314 commented Mar 26, 2025 •

edited

Loading