show missing factor levels on join report

[gmstanle]

It would be great to have a more-verbose option for joins (inner_join, left_join etc). This would show not just the number of rows filtered from each input df, but also the level combinations that were unique. So if you joined two datasets about cars on make, it might report that levels Ferrari and Mazda were only found in df_left and filtered out, and Ford was only found in df_right and filtered out. The issue is the output length will grow unbounded with the number of factor levels, so I think it would have to be optional and truncate the output past a certain number of lines. I am happy to help with development.

[elbersb]

Hi! At first glance I'm not sure whether this would be in scope for tidylog, as the join logic is already fairly complex, and this would add a lot of additional output. A few questions:

• what's the specific use case?
• if this is implemented, this should be opt in. How would one opt in?
• how to deal with high-cardinality values - as you mention, you might have 100 of levels
• why only factors? (could work as well for booleans, strings, even ints)