Open utdemir opened 5 years ago
The generalised joins in discrimination
might be useful for this - probably not directly, but it should be possible to leverage the implementation there and take advantage of the fast Grouping
work.
@axman6 Thank you for the suggestion! I haven't used discrimination
before, but indeed it looks like it might be useful to distributed-dataset
on multiple future (joins, shuffles, sorts, ..).
I will look at it thoroughly when I get to implementing those; or you are always welcome to give it a go :)..
Implement a
join
function which joins two datasets based on a key/key function. Different join types (left/right/full outer, cartesian product) should be supported.