Erik Erlandson

In that case, the map would map each key to a unique parition, and `numPartitions` would just return the size of that map

Maybe a formalism built around RDDs that specifically map each key to unique partition, using a partitioner built around the ideas above, combined with interesting abstractions for manipulating they keys...

The math3 implementation works on doubles, which I'd assume handles most practical cases. We could consider adding support for generalizations, although I'm not aware of any particular applications at the...

LGTM we aren't using the s2i scripts?

I think if you are using the trees as lookup tables, then an alternative representation is going to be superior, maybe `map`. It's hard to avoid that kind of overhead...

Sometimes people will use a tree as a way to map a key to a value (at the leaves), in which case something like map is likely to be more...

@macdew thanks for pointing this out, that is a bit unintuitive. I'm not immediately sure what to do in terms of backward compatibility - I'd hate to break people's code...

IIRC, both were to satisfy some version of the compiler that wanted a return value. It was 10 years ago :grinning: Line 723 is probably because there is a `throw`...

Hi Mark! My memory of some design decisions is a bit rusty (I wrote this 10 years ago) but IIRC this is a consequence of some assymetries from trying to...

I see this has been open nearly a year :grimacing: However, I'm going to leave this open in the event that either I have the bandwidth to work on it,...