Multi-key-column joins

[adamConnerSax]

It seems essential to me that the library be able to join using multiple columns as the join key. I don't know if the underlying Trie makes that simpler.

The simplest version might simply add a column holding a product (maybe [v]?) of the key-columns to each Frame, then join the new Frames on that new column, then remove that column in the result. That's how I'd do it from outside the library. And that doesn't seem horribly inefficient. But I'm imagining there's a better way?

[ocramz]

Trying to wrap my head around this; could you provide an example of the input and result?

[adamConnerSax]

I can give a more thorough example later but here's a sketch: Suppose I have rows of data, each recording, e.g., votes by people with certain demographic characteristics (age, sex, education) in certain geographic areas of the US. And suppose that the geography is specified by two keys: US State and Congressional district within the state. Now I want to join that data to similarly geographically specified data about, e.g., median income. The join is on the pair of geographic columns.

I can, of course, make a new column holding a tuple and then join on that. But it's convenient to skip those extra steps. I think a combinator could handle this fine.

The joins in Frames are already like this. The key is specified as a type-list of (Symbol, Type).

In the Heidi case, I guess k arguments would become Set k?