> On 14. Oct 2017, at 21:55, Xiaodi Wu via swift-evolution > <swift-evolution@swift.org> wrote: > > On Sat, Oct 14, 2017 at 14:36 Benjamin G <benjamin.garrig...@gmail.com > <mailto:benjamin.garrig...@gmail.com>> wrote: > I think what you're saying and what Kevin is saying are in way not > contradictory : > > You're saying the "SpongeBob" protocol functions make sense and are coherent, > Kevin is saying the consequence is that it creates weird functions for Sets, > and i think you're both right. > > an "UnorderedSequence" may very well have a "first" element. The problem is > that when you apply it to sets, you're not calling: > > mySet.generatedRandomOrderSequence().first() > > but > > mySet.first() > > That's where the issue stands, and that's IMHO, a symptom that the protocol > hierarchy is a bit misleading. > What you would like for sets isn't "first()", but "any()". > > 'first' is just a shorthand for 'let first; for element in sequence { first = > element; break }'. Any type that offers iterated access in a for...in loop > must, by construction, have a definable 'first'. > > Swift Sequences are deliberately allowed to be single-pass, infinite, or > unordered. A multi-pass, finite Sequence is a Collection. It's not an > oversight that 'first' is defined on Sequence and not Collection. It's not > 'weird' to ask what the first element you get on iteration must be; by > definition, iteration gives you elements sequentially and something must be > first.
I don't think this really changes the point you're making, but the 'first' property is in fact defined on Collection and not Sequence. Sequence has 'first(where:)', 'dropFirst', and 'prefix' methods, and essentially the same argument applies to them. One rationale I heard (I think from Ben Cohen) for why 'first' is not defined on Collection is that Sequences can be single-pass, and it would feel weird for a seemingly innocuous property like 'first' to have side effects.
_______________________________________________ swift-evolution mailing list swift-evolution@swift.org https://lists.swift.org/mailman/listinfo/swift-evolution
