scikit_build_core.build build-backend issue: "ninja: build stopped: subcommand failed"

[Geremia]

When trying to uv build --wheel, I get the following error:

ninja: build stopped: subcommand failed.

*** CMake build failed
  × Failed to build `/tmp/core`
  ├─▶ The build backend returned an error
  ╰─▶ Call to `scikit_build_core.build.build_wheel` failed (exit status: 1)
      hint: This usually indicates a problem with the package or the build environment.

[burgholzer]

Could you paste the full output of uv build --wheel -v?
CI still worked just fine yesterday.

[Geremia]

@burgholzer

build_log.txt

[burgholzer]

Thanks. Definitely something I haven't seen before.
I suspect the change that triggered this is that we switched to building spdlog as a shared library instead of statically.
This, plus something that is different in your local environment compared to the CI environment we test in.
All of our CI passes (even the extensive one) and I also can't reproduce this locally.

Are there any particular changes that you made or is this on a fresh clone of mqt- ore?
Which OS are you using?
Any idea from your side what could be different than in our regular environments?

[Geremia]

@burgholzer

Are there any particular changes that you made or is this on a fresh clone of mqt- ore?

It's a fresh clone of master.

Which OS are you using?

Slackware64-current Linux, 6.14.11 kernel.

Any idea from your side what could be different than in our regular environments?

I've had issues with building it before. I'm not sure if it's a uv/pip incompatibility or something else.

Here's some info that might help (culled from pytorch's collect_env.py):

OS: Slackware Linux  (x86_64)
GCC version: (GCC) 15.2.0
Clang version: 21.1.8
CMake version: version 4.2.1
Libc version: glibc-2.42

Python version: 3.12.12 (main, Oct  9 2025, 14:09:57) [GCC 15.2.0] (64-bit runtime)
Python platform: Linux-6.14.11-x86_64-AMD_Ryzen_Threadripper_2990WX_32-Core_Processor-with-glibc2.42
UV version: uv 0.9.5