numpy-dev-xfails.txt

# finfo(float32).eps returns float32 but should return float
array_api_tests/test_data_type_functions.py::test_finfo[float32]

# https://github.com/numpy/numpy/issues/21213
array_api_tests/test_special_cases.py::test_binary[__pow__(x1_i is -infinity and x2_i > 0 and not (x2_i.is_integer() and x2_i % 2 == 1)) -> +infinity]
array_api_tests/test_special_cases.py::test_binary[__pow__(x1_i is -0 and x2_i > 0 and not (x2_i.is_integer() and x2_i % 2 == 1)) -> +0]
array_api_tests/test_special_cases.py::test_iop[__ipow__(x1_i is -infinity and x2_i > 0 and not (x2_i.is_integer() and x2_i % 2 == 1)) -> +infinity]
array_api_tests/test_special_cases.py::test_iop[__ipow__(x1_i is -0 and x2_i > 0 and not (x2_i.is_integer() and x2_i % 2 == 1)) -> +0]

# The test suite is incorrectly checking sums that have loss of significance
# (https://github.com/data-apis/array-api-tests/issues/168)
array_api_tests/test_statistical_functions.py::test_sum
array_api_tests/test_statistical_functions.py::test_prod
array_api_tests/test_statistical_functions.py::test_cumulative_sum

# The test suite cannot properly get the signature for vecdot
# https://github.com/numpy/numpy/pull/26237
array_api_tests/test_signatures.py::test_func_signature[vecdot]
array_api_tests/test_signatures.py::test_extension_func_signature[linalg.vecdot]

# 2023.12 support
# Argument 'device' missing from signature
array_api_tests/test_signatures.py::test_func_signature[astype]
array_api_tests/test_signatures.py::test_func_signature[from_dlpack]
array_api_tests/test_signatures.py::test_array_method_signature[__dlpack__]
# uint64 repeats not supported
array_api_tests/test_manipulation_functions.py::test_repeat