Installing from requirements-file fails because of numba version

[benatouba]

Numba version is not specified in requirements files and pyproject-file.

The installation fails (tested on python>=3.11) as pip solves version constraints by choosing numba 0.53.1

Numba added support for numpy2.0 in 0.60. I suggest using this version.

Stacktrace:

  × Failed to download and build `numba==0.53.1`
  ╰─▶ Build backend failed to determine requirements with `build_wheel()` (exit status: 1)

      [stderr]
      Traceback (most recent call last):
        File "<string>", line 14, in <module>
        File "/home/ben/.cache/uv/builds-v0/.tmpegLw1v/lib/python3.11/site-packages/setuptools/build_meta.py", line 334,
      in get_requires_for_build_wheel
          return self._get_build_requires(config_settings, requirements=[])
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
        File "/home/ben/.cache/uv/builds-v0/.tmpegLw1v/lib/python3.11/site-packages/setuptools/build_meta.py", line 304,
      in _get_build_requires
          self.run_setup()
        File "/home/ben/.cache/uv/builds-v0/.tmpegLw1v/lib/python3.11/site-packages/setuptools/build_meta.py", line 522,
      in run_setup
          super().run_setup(setup_script=setup_script)
        File "/home/ben/.cache/uv/builds-v0/.tmpegLw1v/lib/python3.11/site-packages/setuptools/build_meta.py", line 320,
      in run_setup
          exec(code, locals())
        File "<string>", line 50, in <module>
        File "<string>", line 47, in _guard_py_ver
      RuntimeError: Cannot install on Python version 3.11.6; only versions >=3.6,<3.10 are supported.

  help: `numba` (v0.53.1) was included because `cosipymodel` (v2.0.0) depends on `numba

[gampnico]

Thanks, did this occur with pip install cosipymodel or with pip install -r requirements.txt (or some other installation method)?

[benatouba]

Only when installing via any requirements file or editable mode (pip install -e .). The released version does not seem to be affected.

[gampnico]

Ref: #75 (comment)

I can't reproduce this issue, nor does it occur in the CI workflow (which installs COSIPY as an editable, and from the requirements file). Please may you run pip cache purge and reinstall in a clean environment? If you are reusing an environment, then pip install --upgrade -e . might also solve this issue.

If that doesn't work, please may you let me know:

1. Your OS (and distro if applicable)?
2. Are you installing using .venv, conda, or mamba?
3. If you are installing with conda/mamba, are you mixing pip and conda install commands? e.g. conda install gdal with pip install -e .?

[benatouba]

I used the pip interface of uv, because of performance. It has its own resolver. It seems the issue was/is upstream, with that resolver. Using uv sync works fine as well.

I reinstalled from a clean venv python3.12.7 environment with plain pip and it works. So it was just a special case because of the uncommon tooling that I use.