========================================================= test session starts =========================================================
platform linux -- Python 3.13.12, pytest-9.0.2, pluggy-1.6.0
rootdir: /tmp/bitstring
configfile: pyproject.toml
plugins: hypothesis-6.151.9
collected 19 items
tests/test_fp8.py ..........F.......F [100%]
============================================================== FAILURES ===============================================================
______________________________________ TestConversionToFP8.test_compare_8bit_floats_with_gfloat _______________________________________
self = <tests.test_fp8.TestConversionToFP8 object at 0x7f20c7a19cd0>
def test_compare_8bit_floats_with_gfloat(self):
> for fi, lut in [(format_info_p3109(4), p4binary_fmt.lut_binary8_to_float),
^^^^^^^^^^^^^^^^^^^^
(format_info_p3109(3), p3binary_fmt.lut_binary8_to_float),
(format_info_ocp_e4m3, e4m3mxfp_saturate_fmt.lut_int_to_float),
(format_info_ocp_e5m2, e5m2mxfp_saturate_fmt.lut_int_to_float),
]:
E TypeError: format_info_p3109() missing 1 required positional argument: 'precision'
tests/test_fp8.py:235: TypeError
_________________________________________________ test_rounding_consistent_to_gfloat __________________________________________________
def test_rounding_consistent_to_gfloat():
> for fi, dt in [[format_info_p3109(4), Dtype('p4binary')],
^^^^^^^^^^^^^^^^^^^^
[format_info_p3109(3), Dtype('p3binary')]]:
E TypeError: format_info_p3109() missing 1 required positional argument: 'precision'
tests/test_fp8.py:325: TypeError
======================================================= short test summary info =======================================================
FAILED tests/test_fp8.py::TestConversionToFP8::test_compare_8bit_floats_with_gfloat - TypeError: format_info_p3109() missing 1 required positional argument: 'precision'
FAILED tests/test_fp8.py::test_rounding_consistent_to_gfloat - TypeError: format_info_p3109() missing 1 required positional argument: 'precision'
==================================================== 2 failed, 17 passed in 0.28s =====================================================
The test suite currently pins to
gfloat < 0.3which is pretty old. The0.4*versions work just fine, however there's a small regression with 0.5.2: