zig: fix darwin SDK lookup #376427
zig: fix darwin SDK lookup #376427
Conversation
niklaskorz
force-pushed
the
zig-sdk-darwin
branch
from
0e37410 to
f92762e
Compare
github-actions
bot
added
6.topic: nixos
niklaskorz
force-pushed
the
zig-sdk-darwin
branch
from
f92762e to
cfadb65
Compare
|
Zig 0.11 still fails to build in the sandbox, looking into it Edit: this was fixed by #320374 but the PR only addresses 0.12 and 0.13, so I'm adding this change to 0.11 as well |
niklaskorz
force-pushed
the
zig-sdk-darwin
branch
from
cfadb65 to
cb30108
Compare
|
Edit: This is finalized by #331011 |
|
|
|
niklaskorz
force-pushed
the
zig-sdk-darwin
branch
from
cb30108 to
1bab826
Compare
niklaskorz
force-pushed
the
zig-sdk-darwin
branch
from
1bab826 to
a4f58b1
Compare
github-actions
bot
removed
the
6.topic: nixos
github-actions
bot
removed
8.has: documentation
niklaskorz
changed the title
github-actions
bot
added
10.rebuild-linux: 0
|
This fixes the ghostty on darwin incident see here for context: #369788 |
niklaskorz
force-pushed
the
zig-sdk-darwin
branch
from
a4f58b1 to
d3a64dc
Compare
|
Looks like the CI is broken |
This has absolutely nothing to do with that. |
| substituteInPlace lib/std/zig/system/darwin.zig \ | ||
| --replace-fail /usr/bin/xcrun xcrun \ | ||
| --replace-fail /usr/bin/xcode-select '${lib.getExe' xcbuild "xcode-select"}' | ||
| ''; |
There was a problem hiding this comment.
I don't like that this is duplicated three times over the span of three compilers. This is something unlikely to change in the future compilers as well, so this will continue to be duplicated in 0.14, 0.15, ...
There was a problem hiding this comment.
This also applies to code above, but I don't know how you can select a range of lines in GitHub PR reviews
There was a problem hiding this comment.
This is already addressed by #331011
There was a problem hiding this comment.
Amazing, I didn't know about that PR, sorry :(
Zig 0.11 for Linux is broken on master as well, so it's not caused by the changes of this PR: |
Building Zig packages on darwin can currently fail because of the hardcoded
/usr/bin/xcruncall in Zig's macOS SDK lookup method (ziglang/zig#22600).https://github.com/ziglang/zig/blob/f3c29dcb247be5e50d2d05f9fb2e814ebf115a86/lib/std/zig/system/darwin.zig#L51
This is fixed by replacing it with just
xcrun, which will then use thexcrunprovided by the nixpkgs darwin stdenv. I decided against using an absolute nix store path forxcrunas replacement as @reckenrode raised concerns about that causing issues in Nix development shells.Zig also calls
/usr/bin/xcode-select, which I have replaced byxcode-selectin thexcbuildpackage. In this case, the absolute path to the nix store is acceptable even for devShells as Zig discards the output ofxcode-selectand just checks if the command is successful:https://github.com/ziglang/zig/blob/f77e1b86225cd49c2d04dfa6ca4a7ede315dc0b1/lib/std/zig/system/darwin.zig#L18
I also removed the by now rather old Zig versions 0.9 and 0.10 (not completely arbitrarily, they are the only releases marked as "pre-release" on https://github.com/ziglang/zig/releases and had either no or only one dependant in nixpkgs). ForThis part has been moved to a separate PR: #376540zig_0_9, there was a single dependant colorstorm which already is Zig 0.13 compatible in its HEAD state. I opened an issue requesting a new tagged release for colorstorm as the last one is three years old, and in the meanwhile bumped colorstorm to the commit bringing Zig 0.13 compatibility.Things done
nix.conf? (See Nix manual)sandbox = relaxedsandbox = truenix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage./result/bin/)Add a 👍 reaction to pull requests you find important.