Option to check column order when comparing polars dataframes

[physinet]

This adds check_column_order as a keyword argument to pl.testing.assert_frame_equal to allow comparison of DataFrames with columns in an arbitrary order.

[ritchie46]

How would this work? Which column is compared to which then?

Right, I now see the keyword in that function. Can we rename that to check_column_names?

[physinet]

I think check_column_order (as the variable was originally named) makes more sense. If False, the name of the column still has to match the values in that column, but the order of columns can be arbitrary.

check_column_names to me implies that as long as the values are in the correct positions, the column names do not matter. For example, does that imply these dataframes should be asserted as equal?

df1 = pl.DataFrame({"a": [1, 2], "b": [3, 4]})
df2 = pl.DataFrame({"b": [1, 2], "a": [3, 4]})

This is not the case that I'm hoping to cover with this keyword argument.

[physinet]

FWIW, the equivalent pandas argument is named check_like: https://github.com/pandas-dev/pandas/blob/v1.4.2/pandas/_testing/asserters.py#L1196

[ritchie46]

I think check_column_order (as the variable was originally named) makes more sense. If False, the name of the column still has to match the values in that column, but the order of columns can be arbitrary.

check_column_names to me implies that as long as the values are in the correct positions, the column names do not matter. For example, does that imply these dataframes should be asserted as equal?

df1 = pl.DataFrame({"a": [1, 2], "b": [3, 4]})
df2 = pl.DataFrame({"b": [1, 2], "a": [3, 4]})

This is not the case that I'm hoping to cover with this keyword argument.

To me it implies that the column names must be correct and in the same order. The other name confused me, because how can we check if the combination of columns is correct. It does not make much sense to me.

[physinet]

Changed to check_column_names - makes sense enough and the docstring should help clarify.

[ritchie46]

Thanks @physinet