| 24 Feb 2026 |
crazychaoz | * what happened in between
• Updated input 'nixpkgs':
'github:NixOS/nixpkgs/41da1e3ea8e23e094e5e3eeb1e6b830468a7399e?narHash=sha256-jp0D4vzBcRKwNZwfY4BcWHemLGUs4JrS3X9w5k/JYDA%3D' (2025-06-14)
→ 'github:NixOS/nixpkgs/724cf38d99ba81fbb4a347081db93e2e3a9bc2ae?narHash=sha256-MpAKyXfJRDTgRU33Hja%2BG%2B3h9ywLAJJNRq4Pjbb4dQs%3D' (2026-02-19)
so that
error: Package ‘mingw_w64-pthreads-13.0.0’ in /nix/store/00k5v1i44s1dqjryvq6q414rnha8x6ld-source/pkgs/os-specific/windows/mingw-w64/headers.nix:35 is not available on the requested hostPlatform:
hostPlatform.system = "x86_64-linux"
package.meta.platforms = [
"aarch64-windows"
"x86_64-windows"
"i686-windows"
]
(i cannot cross compile for windows anymore)
error (gcc-15.2.0):
> The directory (BUILD_SYSTEM_HEADER_DIR) that should contain system headers does not exist:
> /usr/include
> make[2]: *** [Makefile:3641: stmp-fixinc] Error 1
> rm gcc.pod
> make[2]: Leaving directory '/build/build/gcc'
| 09:32:29 |
vcunat | It evaluated fine. I'm sorry for the false alarm. | 10:20:46 |
vcunat | Though the "removed jobs" might be worth investigating:
https://hydra.nixos.org/eval/1823123 | 10:20:53 |
vcunat | * The jobset evaluated fine. I'm sorry for the false alarm. | 10:21:51 |
alexfmpe | In reply to @crazychaoz:matrix.org
what happened in between
• Updated input 'nixpkgs':
'github:NixOS/nixpkgs/41da1e3ea8e23e094e5e3eeb1e6b830468a7399e?narHash=sha256-jp0D4vzBcRKwNZwfY4BcWHemLGUs4JrS3X9w5k/JYDA%3D' (2025-06-14)
→ 'github:NixOS/nixpkgs/724cf38d99ba81fbb4a347081db93e2e3a9bc2ae?narHash=sha256-MpAKyXfJRDTgRU33Hja%2BG%2B3h9ywLAJJNRq4Pjbb4dQs%3D' (2026-02-19)
so that
error: Package ‘mingw_w64-pthreads-13.0.0’ in /nix/store/00k5v1i44s1dqjryvq6q414rnha8x6ld-source/pkgs/os-specific/windows/mingw-w64/headers.nix:35 is not available on the requested hostPlatform:
hostPlatform.system = "x86_64-linux"
package.meta.platforms = [
"aarch64-windows"
"x86_64-windows"
"i686-windows"
]
(i cannot cross compile for windows anymore)
error (gcc-15.2.0):
> The directory (BUILD_SYSTEM_HEADER_DIR) that should contain system headers does not exist:
> /usr/include
> make[2]: *** [Makefile:3641: stmp-fixinc] Error 1
> rm gcc.pod
> make[2]: Leaving directory '/build/build/gcc'
That sounds very recent, I've been cross compiling to windows for months | 10:40:03 |
crazychaoz | how? for me it worked previously with:
TARGET_CC = "${pkgs.pkgsCross.mingwW64.stdenv.cc}/bin/${pkgs.pkgsCross.mingwW64.stdenv.cc.targetPrefix}cc";
depsBuildBuild = with pkgs; [
pkgsCross.mingwW64.stdenv.cc
pkgsCross.mingwW64.windows.pthreads
perl
];
| 11:22:33 |
alexfmpe | I've mostly used pkgsCross.ucrt64, not sure about that one | 11:34:48 |
alexfmpe | pkgsCross.ucrt64.windows.pthreads builds for me | 11:35:41 |
alexfmpe | actually pkgsCross.mingwW64.windows.pthreads also builds for me | 11:36:45 |
alexfmpe | on aarch64-darwin | 11:36:49 |
crazychaoz | i believe the error is with the cc | 11:38:27 |
crazychaoz | ┃ ┌─ ⏵ gcc-15.2.0 (buildPhase) ⏱ 5m0s
┃ ┌─ ⏸ gcc-wrapper-15.2.0
┃ ┌─ ⏸ stdenv-linux
┃ ┌─ ⏸ mingw_w64-pthreads-13.0.0
| 11:40:49 |
alexfmpe | nix-build -A pkgsCross.mingwW64.stdenv.cc also built for me | 12:01:27 |
alexfmpe | ┏━ Dependency Graph:
┃ ┌─ ✔ dlltool
┃ │ ┌─ ✔ 001-fix-rpath.patch
┃ ├─ ✔ meson-1.10.1 ⏱ 5m9s
┃ │ ┌─ ↓ ✔ gcc-15.2.0.tar.xz from [2]
┃ │ │ ┌─ ✔ mingw_w64-headers-13.0.0 ⏱ 7s
┃ │ │ ├─ ✔ x86_64-w64-mingw32-binutils-2.44 ⏱ 3m1s
┃ │ ├─ ✔ x86_64-w64-mingw32-binutils-wrapper-2.44 ⏱ 4s
┃ │ ┌─ ✔ x86_64-w64-mingw32-nolibc-gcc-15.2.0 ⏱ 6m26s
┃ │ ┌─ ✔ win-dll-hook.sh
┃ │ ┌─ ✔ x86_64-w64-mingw32-nolibc-gcc-wrapper-15.2.0
┃ │ ┌─ ✔ stdenv-darwin
┃ │ ┌─ ✔ mingw-w64-x86_64-w64-mingw32-13.0.0 ⏱ 3m53s
┃ │ ┌─ ✔ x86_64-w64-mingw32-binutils-wrapper-2.44
┃ │ ┌─ ✔ x86_64-w64-mingw32-nolibc-gcc-wrapper-15.2.0
┃ ├─ ✔ stdenv-darwin
┃ ┌─ ✔ mcfgthread-x86_64-w64-mingw32-2.3.1 ⏱ 13s
┃ ┌─ ✔ x86_64-w64-mingw32-gcc-15.2.0 ⏱ 7m52s
┃ ┌─ ✔ win-dll-hook.sh
┃ ✔ x86_64-w64-mingw32-gcc-wrapper-15.2.0
| 12:01:45 |
alexfmpe | I can't test on linux right now | 12:02:07 |
alexfmpe | * I can't test on x86_64-linux right now | 12:02:21 |
alexfmpe | this was on master | 12:03:49 |
alexfmpe | if I try in your 724... commit, pkgsCross.mingwW64.stdenv.cc is fetched from cache | 12:05:08 |
crazychaoz | just as a sanity check: please update & compile this: https://github.com/CrazyChaoz/backtor | 12:06:35 |
alexfmpe | "update" ? | 12:20:12 |
alexfmpe | wait a second | 12:21:49 |
crazychaoz | i currently committed the "working" state, so you'll need a nix flake update to get the broken state | 12:22:13 |
crazychaoz | or should i publish a broken branch ? | 12:23:03 |
alexfmpe | I mean, if you want me to reproduce, then yeah you should give me your exact flake.lock | 12:24:02 |
alexfmpe | or I might nix flake update into a different state | 12:24:23 |
alexfmpe | I really don't get why flakes want flake.nix to be non-reproducible | 12:24:45 |
crazychaoz | ack; https://github.com/CrazyChaoz/backtor/tree/broken-windows | 12:30:05 |
alexfmpe | what do I run?
nix build .\#packages.aarch64-darwin.windows | 12:38:17 |
alexfmpe | * what do I run?
nix build .\#packages.aarch64-darwin.windows ? | 12:38:20 |
crazychaoz | .#windows should suffice | 12:38:45 |