Sets do not have keys, they have values

tc39 / proposal-collection-normalization

MIT License

41 stars 8 forks source link

Sets do not have keys, they have values #8

Closed domenic closed 5 years ago

domenic commented 5 years ago

Related to #6: although set values are unique like map keys are, they are still generally referred to as values, not keys.

In the spec so far, this only manifests unobservably:

keys() and [@@iterator]() are defined in terms of values(), instead of the other way around.
Arguments to the set methods, and intermediate variable names used in algorithms, are all "value" and not "key"

But I still think it's good to align with common understanding of how sets work.

bmeck commented 5 years ago

I think the counter example is that .has(_) works on keys for Maps and has similarity across Sets. Either way, ideally Set and Map should have some level to do this in a compatible way. The naming conventions of the spec are interesting here. Would given https://github.com/tc39/proposal-richer-keys/issues/6 it might make sense to provide both, merely because of this issue. I actually think we should move discussion back to https://github.com/tc39/proposal-richer-keys/issues/6 as it relates to multiple collection types.