chore: refactor compare_dicts

[EdAbati]

What type of PR is this? (check all applicable)

• 💾 Refactor
• ✨ Feature
• 🐛 Bug Fix
• 🔧 Optimization
• 📝 Documentation
• ✅ Test
• 🐳 Other

Related issues

• Related issue #
• Closes #

Checklist

• Code follows style guide (ruff)
• Tests added
• Documented the changes

If you have comments or can explain your changes, please do so below.

I propose a couple of changes:

• I'd like to have a more verbose assertion error message when a unitest fail. Taking inspiration from pandas, I'm printing all the values of the columns that differ and the index where the difference was first noticed. This is particularly helpful for dataframes libraries that don't keep row order (* cough cough * pyspark * cough cough *)

def test_select(constructor: Constructor) -> None:
    data = {"a": [1, 3, 2], "b": [4, 4, 6], "z": [7.0, 8, 9]}
    df = nw.from_native(constructor(data))
    result = df.select("a")
    expected = {"a": [1, 3, 3]}
    compare_dicts(result, expected)

Before:

After:

• I think we should rename compare_dicts because we are not actually comparing 2 dictionaries. 😅 I propose something like assert_equal_data, any better name? This makes the diff of the PR a bit large, happy to revert if you think we should keep the previous name

[FBruzzesi]

Thanks @EdAbati I like this a lot more! I left a comment but can also be thought as a follow up.

For the naming, I think this is the closest we have to what in pandas and polars is assert_frame_equal, but we are not there quite yet, so I am not sure 🙈

[FBruzzesi]

Should we record the full diff instead of stopping at the first unequal encounter?