ENH: alleviate issues with inconsistent multi-level joining on a 0-column merge table

[prhbrt]

Feature Type

• Adding new functionality to pandas

• Changing existing functionality in pandas

• Removing existing functionality in pandas

Problem Description

FutureWarning: merging between different levels is deprecated and will be removed in a future version. (`some number` levels on the left, 1 on the right)

When running:

multilevel_columns = pd.DataFrame({('a', 'b'): np.random.randint(0,100, 10)})
multilevel_columns.index.name = 'alpha'

relations = pd.DataFrame({
    'alpha': np.random.randint(0,10, 10),
    'beta': np.random.randint(0,10, 10)
}).set_index(['alpha', 'beta'])

multilevel_columns.join(relations)

because the 0-column merge table relations only has 1 column-level, whereas multilevel_columns has two. However, it's irrelevant because relations has 0 columns anyway, and so concatenating them should be trivial.

Feature Description

Allow merging inconsistent levels if one frame has 0 columns.

Alternative Solutions

Change this check with the added condition and len(_left.columns) > 0 and len(_right.columns) > 0

Additional Context

No response

[phofl]

Hi, thanks for your report. I am -1 on this, we should avoid value dependent behavior. What are you trying to do here?

[mroeschke]

Thanks for the request but it appears there hasn't been much interest in this issue from the core team and community over the years so closing.