Far more failed (not sure if it was a fluke or not) with 1.3.3. The first two I think we can live with given they're natural as a result of FP / 32-bit rches, but the last one is more concerning. Dakon reported the same thing(?) in bug 809854 for sparc. Looks almost the same as https://github.com/pandas-dev/pandas/issues/36579. ---- FAILED pandas/tests/io/parser/test_c_parser_only.py::test_float_precision_options[c_high] - assert 243.164 == 243.16400000000002 FAILED pandas/tests/io/parser/test_c_parser_only.py::test_float_precision_options[c_low] - assert 243.164 == 243.16400000000002 FAILED pandas/tests/indexes/interval/test_astype.py::TestFloatSubtype::test_subtype_integer_errors - Failed: DID NOT RAISE <class 'TypeError'> ===================================================== 3 failed, 170235 passed, 9866 skipped, 873 xfailed, 6 xpassed, 866 warnings in 2675.34s (0:44:35) ====================================================== * pytest failed with python3.8 * ERROR: dev-python/pandas-1.3.4::gentoo failed (test phase): * Failed to run 'epytest pandas --skip-slow --skip-network --deselect pandas/tests/window/test_rolling.py::test_rolling_var_numerical_issues --deselect pandas/tests/base/test_misc.py::test_memory_usage[series-with-empty-index] --deselect pandas/tests/tools/test_to_datetime.py::TestToDatetime::test_to_datetime_tz_psycopg2 --deselect pandas/tests/io/xml/test_xml.py::test_wrong_url --deselect pandas/tests/io/test_fsspec.py::test_read_csv --deselect pandas/tests/io/test_fsspec.py::test_markdown_options -m not single -n 6' ---- ================================================================================================== FAILURES ================================================================================================== ____________________________________________________________________________________ test_float_precision_options[c_high] ____________________________________________________________________________________ [gw4] linux -- Python 3.8.12 /usr/bin/python3.8 c_parser_only = <pandas.tests.io.parser.conftest.CParserHighMemory object at 0xeb8c36b8> def test_float_precision_options(c_parser_only): # GH 17154, 36228 parser = c_parser_only s = "foo\n243.164\n" df = parser.read_csv(StringIO(s)) df2 = parser.read_csv(StringIO(s), float_precision="high") tm.assert_frame_equal(df, df2) df3 = parser.read_csv(StringIO(s), float_precision="legacy") if IS64: assert not df.iloc[0, 0] == df3.iloc[0, 0] else: > assert df.iloc[0, 0] == df3.iloc[0, 0] E assert 243.164 == 243.16400000000002 E +243.164 E -243.16400000000002 c_parser_only = <pandas.tests.io.parser.conftest.CParserHighMemory object at 0xeb8c36b8> df = foo 0 243.164 df2 = foo 0 243.164 df3 = foo 0 243.164 parser = <pandas.tests.io.parser.conftest.CParserHighMemory object at 0xeb8c36b8> s = 'foo\n243.164\n' pandas/tests/io/parser/test_c_parser_only.py:679: AssertionError ---- ____________________________________________________________________________________ test_float_precision_options[c_low] _____________________________________________________________________________________ [gw4] linux -- Python 3.8.12 /usr/bin/python3.8 c_parser_only = <pandas.tests.io.parser.conftest.CParserLowMemory object at 0xeb8c3700> def test_float_precision_options(c_parser_only): # GH 17154, 36228 parser = c_parser_only s = "foo\n243.164\n" df = parser.read_csv(StringIO(s)) df2 = parser.read_csv(StringIO(s), float_precision="high") tm.assert_frame_equal(df, df2) df3 = parser.read_csv(StringIO(s), float_precision="legacy") if IS64: assert not df.iloc[0, 0] == df3.iloc[0, 0] else: > assert df.iloc[0, 0] == df3.iloc[0, 0] E assert 243.164 == 243.16400000000002 E +243.164 E -243.16400000000002 c_parser_only = <pandas.tests.io.parser.conftest.CParserLowMemory object at 0xeb8c3700> df = foo 0 243.164 df2 = foo 0 243.164 df3 = foo 0 243.164 parser = <pandas.tests.io.parser.conftest.CParserLowMemory object at 0xeb8c3700> s = 'foo\n243.164\n' pandas/tests/io/parser/test_c_parser_only.py:679: AssertionError ---- pandas/tests/io/parser/test_c_parser_only.py:679: AssertionError ________________________________________________________________________________ TestFloatSubtype.test_subtype_integer_errors ________________________________________________________________________________ [gw3] linux -- Python 3.8.12 /usr/bin/python3.8 self = <pandas.tests.indexes.interval.test_astype.TestFloatSubtype object at 0xadc93208> @pytest.mark.xfail(is_platform_arm(), reason="GH 41740") def test_subtype_integer_errors(self): # float64 -> uint64 fails with negative values index = interval_range(-10.0, 10.0) dtype = IntervalDtype("uint64", "right") msg = re.escape( "Cannot convert interval[float64, right] to interval[uint64, right]; " "subtypes are incompatible" ) with pytest.raises(TypeError, match=msg): > index.astype(dtype) E Failed: DID NOT RAISE <class 'TypeError'> dtype = interval[uint64, right] index = IntervalIndex([(-10.0, -9.0], (-9.0, -8.0], (-8.0, -7.0], (-7.0, -6.0], (-6.0, -5.0] ... (5.0, 6.0], (6.0, 7.0], (7.0, 8.0], (8.0, 9.0], (9.0, 10.0]], dtype='interval[float64, right]') msg = ('Cannot\\ convert\\ interval\\[float64,\\ right\\]\\ to\\ ' 'interval\\[uint64,\\ right\\];\\ subtypes\\ are\\ incompatible') self = <pandas.tests.indexes.interval.test_astype.TestFloatSubtype object at 0xadc93208> pandas/tests/indexes/interval/test_astype.py:183: Failed
The build.log is too big to upload to BZ even (aggressively) compressed, so here it is: https://dev.gentoo.org/~sam/bugs/gentoo/pandas-1.3.4-bug818964-arm.log.bz2. But this is quite interesting. For the test I'm concerned about (test_subtype_integer_errors), it's meant to be skipped on arm: https://github.com/pandas-dev/pandas/blob/44237a5841a77e48e004e9fa1e705c51a88633e2/pandas/tests/indexes/interval/test_astype.py#L173. But at least in an arm32 chroot ("armv8l") on arm64, I get: >>> import platform >>> platform.machine() 'armv8l' And I'm not sure that the upstream check for 'arm64', 'aarch64' will cover real arm32 either. So I'll poke at that.
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=45d65df5773c2bcaf1241e5c0881c7308d9c6740 commit 45d65df5773c2bcaf1241e5c0881c7308d9c6740 Author: Sam James <sam@gentoo.org> AuthorDate: 2021-10-29 16:59:28 +0000 Commit: Sam James <sam@gentoo.org> CommitDate: 2021-10-29 16:59:28 +0000 dev-python/pandas: fix detecting 32-bit arm (for skipping some tests) Bug: https://bugs.gentoo.org/818964 Signed-off-by: Sam James <sam@gentoo.org> .../pandas/files/pandas-1.3.4-arm-tests.patch | 21 +++++++++++++++++++++ dev-python/pandas/pandas-1.3.4.ebuild | 5 +++++ 2 files changed, 26 insertions(+)
