h-vetinari opened a new issue, #45229: URL: https://github.com/apache/arrow/issues/45229
### Describe the enhancement requested The test suite in https://github.com/conda-forge/arrow-cpp-feedstock/pull/1664 has a single test failure ``` =================================== FAILURES =================================== ____________ test_sparse_coo_tensor_scipy_roundtrip[f2-arrow_type8] ____________ dtype_str = 'f2', arrow_type = DataType(halffloat) @pytest.mark.skipif(not coo_matrix, reason="requires scipy") @pytest.mark.parametrize('dtype_str,arrow_type', tensor_type_pairs) def test_sparse_coo_tensor_scipy_roundtrip(dtype_str, arrow_type): dtype = np.dtype(dtype_str) data = np.array([1, 2, 3, 4, 5, 6]).astype(dtype) row = np.array([0, 0, 2, 3, 1, 3]) col = np.array([0, 2, 0, 4, 5, 5]) shape = (4, 6) dim_names = ('x', 'y') # non-canonical sparse coo matrix > scipy_matrix = coo_matrix((data, (row, col)), shape=shape) pyarrow/tests/test_sparse_tensor.py:408: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ scipy/sparse/_coo.py:62: in __init__ self.data = getdata(obj, copy=copy, dtype=dtype) scipy/sparse/_sputils.py:150: in getdata getdtype(data.dtype) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ dtype = dtype('float16'), a = None, default = None def getdtype(dtype, a=None, default=None): """Form a supported numpy dtype based on input arguments. Returns a valid ``numpy.dtype`` from `dtype` if not None, or else ``a.dtype`` if possible, or else the given `default` if not None, or else raise a ``TypeError``. The resulting ``dtype`` must be in ``supported_dtypes``: bool_, int8, uint8, int16, uint16, int32, uint32, int64, uint64, longlong, ulonglong, float32, float64, longdouble, complex64, complex128, clongdouble """ if dtype is None: try: newdtype = a.dtype except AttributeError as e: if default is not None: newdtype = np.dtype(default) else: raise TypeError("could not interpret data type") from e else: newdtype = np.dtype(dtype) if newdtype not in supported_dtypes: supported_dtypes_fmt = ", ".join(t.__name__ for t in supported_dtypes) > raise ValueError(f"scipy.sparse does not support dtype {newdtype.name}. " f"The only supported types are: {supported_dtypes_fmt}.") E ValueError: scipy.sparse does not support dtype float16. The only supported types are: bool, int8, uint8, int16, uint16, int32, uint32, int64, uint64, longlong, ulonglong, float32, float64, longdouble, complex64, complex128, clongdouble. scipy/sparse/_sputils.py:137: ValueError =============================== warnings summary =============================== ``` As far as I can [tell](https://github.com/scipy/scipy/blob/v1.14.0/scipy/sparse/_sputils.py#L16), SciPy hasn't supported `float16` in sparse code for the last several years, so what happened most likely is simply that the checks got stricter in the API. If I'm wrong about that, please raise an issue in scipy. Sidenote: compare also the [migration guide](https://scipy.github.io/devdocs/reference/sparse.migration_to_sparray.html) for getting from `scipy.sparse.*_matrix` to `scipy.sparse.*_array`, though that probably warrants a separate issue. ### Component(s) Python -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@arrow.apache.org.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
