Update flake8 to 7.1.1.

[bdice]

Description

We need to update flake8 to fix a false-positive that appears with older flake8 versions on Python 3.12.

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[bdice]

The previous code assert TypeError was definitely wrong (it doesn't raise!). But now I'm seeing test failures:

___________________________ test_cudf_dataframe_init ___________________________
[gw7] linux -- Python 3.10.15 /pyenv/versions/3.10.15/bin/python

    def test_cudf_dataframe_init():
        df = cudf.DataFrame({"a": [1, 2, 3], "b": [4, 5, 6]})
        gdf = cuspatial.GeoDataFrame(df)
>       assert_eq_geo_df(gdf.to_pandas(), df.to_pandas())

tests/test_geodataframe.py:465: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

geo1 =    a  b
0  1  4
1  2  5
2  3  6, geo2 =    a  b
0  1  4
1  2  5
2  3  6

    def assert_eq_geo_df(geo1, geo2):
        if type(geo1) is not type(geo2):
>           raise TypeError
E           TypeError

tests/test_geodataframe.py:97: TypeError

[jameslamb]

I wanted to help so looked into this a bit.

Here's the specific error I now see in the one failing test:

FAILED tests/test_geodataframe.py::test_cudf_dataframe_init - AssertionError: assert <class 'geopandas.geodataframe.GeoDataFrame'> is <class 'pandas.core.frame.DataFrame'>
 +  where <class 'geopandas.geodataframe.GeoDataFrame'> = type(   a  b\n0  1  4\n1  2  5\n2  3  6)
 +  and   <class 'pandas.core.frame.DataFrame'> = type(   a  b\n0  1  4\n1  2  5\n2  3  6)

(build link)

Here's the source code of that test:

cuspatial/python/cuspatial/cuspatial/tests/test_geodataframe.py

Lines 462 to 465 in a2d21c9

	def test_cudf_dataframe_init():
	df = cudf.DataFrame({"a": [1, 2, 3], "b": [4, 5, 6]})
	gdf = cuspatial.GeoDataFrame(df)
	assert_eq_geo_df(gdf.to_pandas(), df.to_pandas())

It looks to me like it's intentional behavior that the cuspatial.GeoDataFrame.to_pandas() method actually returns a geopandas.GeoDataFrame, not a pandas.DataFrame.

cuspatial/python/cuspatial/cuspatial/core/geodataframe.py

Lines 70 to 80 in a2d21c9

	def to_pandas(self, nullable=False):
	"""
	Calls `self.to_geopandas`, converting GeoSeries columns into GeoPandas
	columns and cudf.Series columns into pandas.Series columns, and
	returning a pandas.DataFrame.
	Parameters
	----------
	nullable: matches the cudf `to_pandas` signature, not yet supported.
	"""
	return self.to_geopandas(nullable=nullable)

So I guess one of these has to happen:

• cuspatial.GeoDataFrame.to_pandas() needs to start returning a pandas.DataFrame instead (as the name implies)
• the have-the-same-type expectation needs to be removed from that test

[mroeschke]

From a cursory check, it appears the intention is for this check to assert that both arguments are geopandas.GeoDataFrame objects, so I think the usage here is incorrect.

I imagine the test should have been assert_eq_geo_df(gdf.to_pandas(), gpd.GeoDataFrame(df.to_pandas()))?

[bdice]

I'll try that locally. Thanks, that sounds like the right direction to me.

[bdice]

Yup, that worked on my local build. I pushed the change. Thanks @mroeschke!

[bdice]

/merge