WIP: Setup GitHub Action Continuous Integration tests

[weiji14]

Minimal configuration to run unit tests on GitHub Actions CI. Using pytest to run the test suite.

Running matrix tests on Ubuntu-22.04 and Python 3.10/3.12.

TODO:

• Initial ci-tests.yml file for running GitHub Actions
• Simplify installation of optional test dependencies
  - [ ] Fix broken unit tests as much as possible (done in Fix broken doctest and tests on accessors #46)
• Setup GPU CI runner somehow...

Fixes #8

[weiji14]

Yeah, was probably too optimistic to think that we can test CuPy on standard GitHub Actions runners:

=================================== FAILURES ===================================
____________________________ test_data_set_accessor ____________________________

tutorial_ds_air = <xarray.Dataset> Size: 31MB
Dimensions:  (lat: 25, time: 2920, lon: 53)
Coordinates:
  * lat      (lat) float32 100B 7...).  These a...
    platform:     Model
    references:   http://www.esrl.noaa.gov/psd/data/gridded/data.ncep.reanaly...

    def test_data_set_accessor(tutorial_ds_air):
        ds = tutorial_ds_air
        assert hasattr(ds, "cupy")
        assert not ds.cupy.is_cupy
    
>       ds = ds.as_cupy()

cupy_xarray/tests/test_accessors.py:37: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
../../../micromamba/envs/cupy-xarray/lib/python3.10/site-packages/cupy_xarray/accessors.py:182: in as_cupy
    return ds.cupy.as_cupy(*args, **kwargs)
../../../micromamba/envs/cupy-xarray/lib/python3.10/site-packages/cupy_xarray/accessors.py:136: in as_cupy
    data_vars = {var: da.as_cupy() for var, da in self.ds.data_vars.items()}
../../../micromamba/envs/cupy-xarray/lib/python3.10/site-packages/cupy_xarray/accessors.py:136: in <dictcomp>
    data_vars = {var: da.as_cupy() for var, da in self.ds.data_vars.items()}
../../../micromamba/envs/cupy-xarray/lib/python3.10/site-packages/cupy_xarray/accessors.py:168: in as_cupy
    return da.cupy.as_cupy(*args, **kwargs)
../../../micromamba/envs/cupy-xarray/lib/python3.10/site-packages/cupy_xarray/accessors.py:70: in as_cupy
    data=cp.asarray(self.da.data),
../../../micromamba/envs/cupy-xarray/lib/python3.10/site-packages/cupy/_creation/from_data.py:88: in asarray
    return _core.array(a, dtype, False, order, blocking=blocking)
cupy/_core/core.pyx:2383: in cupy._core.core.array
    ???
cupy/_core/core.pyx:2410: in cupy._core.core.array
    ???
cupy/_core/core.pyx:2553: in cupy._core.core._array_default
    ???
cupy/_core/core.pyx:135: in cupy._core.core.ndarray.__new__
    ???
cupy/_core/core.pyx:223: in cupy._core.core._ndarray_base._init
    ???
cupy/cuda/memory.pyx:738: in cupy.cuda.memory.alloc
    ???
cupy/cuda/memory.pyx:1424: in cupy.cuda.memory.MemoryPool.malloc
    ???
cupy/cuda/memory.pyx:1444: in cupy.cuda.memory.MemoryPool.malloc
    ???
cupy/cuda/device.pyx:40: in cupy.cuda.device.get_device_id
    ???
cupy_backends/cuda/api/runtime.pyx:202: in cupy_backends.cuda.api.runtime.getDevice
    ??
?
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

>   ???
E   cupy_backends.cuda.api.runtime.CUDARuntimeError: cudaErrorInsufficientDriver: CUDA driver version is insufficient for CUDA runtime version

cupy_backends/cuda/api/runtime.pyx:146: CUDARuntimeError
___________________________ test_data_array_accessor ___________________________

tutorial_da_air = <xarray.DataArray 'air' (time: 2920, lat: 25, lon: 53)> Size: 31MB

Anyone want to sponsor some GPU runners? Or know of others ways to run GPU-based CI tests, e.g. https://docs.cirun.io/reference/yaml#gpu-gpu?

[weiji14]

Anyone want to sponsor some GPU runners? Or know of others ways to run GPU-based CI tests, e.g. https://docs.cirun.io/reference/yaml#gpu-gpu?

Blog post on how scikit-learn set up GPU CI - https://blog.scientific-python.org/scikit-learn/gpu-ci/ (see also discussion at scikit-learn/scikit-learn#24491). Need to ask if we could get this set up for cupy-xarray via NumFOCUS.