[API Review] Sorted Collections

[kylemacomber]

We also discussed this in the PriorityQueue review:

[API Review] PriorityQueue APIs

A concern I have with adding a closure/KeyPath comparator is that PriorityQueue is not the only data structure that would benefit from this kind of feature. I think Set, OrderedSet, and (soon!) SortedSet would similarly benefit from being able to compare/hash on a projection of their element. It seems undesirable to have an alternate variant of all of these data structures.

No comparator is still my preference for the sorted collections, as it's consistent with the existing types in the stdlib and the rest of the collections package.

@fbartho

Or I’d have to box them in a struct that compares the way I prefer?

This is what I'd recommend.

Part of my thinking is that if we keep everything consistent, it improves the likelihood we can identify a language feature that addresses this and applies equally well to all collections. For example, imagine something like a macro that synthesizes the boxes on your behalf:

Equal(person, by: \.name)
Compare(person, by: \.name)
Hash(person, by: \.name)