When passing a collection to Set.contains(_:), forward to Set.isSuperset(of:)

[marcomasser]

Motivation

There is a confusing behavior of Set.contains(_:) when passing a collection that contains more than one element:

let s: Set = [1, 2, 3]
s.contains([1]) // true
s.contains([1] as Set) // true
s.contains([1, 2]) // false 🤨
s.contains([1, 2] as Set) // false 🤨
s.contains([1, 2, 3]) // false 🤨
s.contains([1, 2, 3] as Set) // false 🤨

I think this confusion is introduced by a Collection extension in the _StringProcessing module:

extension Collection where Self.Element : Equatable {

    /// Returns a Boolean value indicating whether the collection contains the
    /// given sequence.
    /// - Parameter other: A sequence to search for within this collection.
    /// - Returns: `true` if the collection contains the specified sequence,
    /// otherwise `false`.
    @available(macOS 13.0, iOS 16.0, watchOS 9.0, tvOS 16.0, *)
    public func contains<C>(_ other: C) -> Bool where C : Collection, Self.Element == C.Element
}

I understand that this behavior is caused by the Set internally storing its elements in a random order (as viewed from outside), so checking whether “[…] the collection contains the given sequence” is not really a useful thing to do on a Set, making this method pretty much useless there.

Proposed solution

To make this method useful on Set, it could simply forward to Set.isSuperset(of:), which is what I would have expected in the first place:

s.isSuperset(of: [1, 2]) // true
s.isSuperset(of: [1, 2] as Set) // true
s.isSuperset(of: [1, 2, 3]) // true
s.isSuperset(of: [1, 2, 3] as Set) // true

While maybe not strictly correct in the sense of the documentation that “[…] the collection contains the given sequence” because the order of elements as passed is irrelevant, I think it is more useful because the behavior is at least deterministic.

Alternatives considered

None really. Do nothing and keep the behavior as-is, I guess.

Additional information

I posted on the Swift forum about this and received only one reply that agreed with the confusing behavior.

Here’s the thread: Passing a Collection to Set.contains(_:) leads to confusing results

[marcomasser]

I also want to mention that this caused a bug in our app because of a line like this:

canvas.tree.elements.contains(interactionState.canvasInteractionState.selection.nodeIDs)

This worked as expected for an empty selection or single selection, but evaluated to false unexpectedly for multi-selections.

IMHO, the issue is that elements is a Set and while an isSuperset(of:) check is the correct thing to do here, it’s really easy to reach for the wrong method (contains(_:)) because (1) that method exists and (2) it’s valid to think of this as “I want to check if set A contains set B”.

I’m not familiar with set theory at all, but a quick check of the Wikipedia entry for Subset shows that saying “set A contains set B” seems to be a valid statement. At least in set theory. But not in Swift.

[sveinhal]

let s: Set = [1, 2, 3]
s.contains([1]) // true
s.contains([1] as Set) // true
s.contains([1, 2]) // false 🤨
s.contains([1, 2] as Set) // false 🤨
s.contains([1, 2, 3]) // false 🤨
s.contains([1, 2, 3] as Set) // false 🤨

This is also non-deterministic and the output may change each time the program is run, depending on the seed used for hashing. s is treated as a collection, and its order is hash-dependent. So if Array(s) happens to return [1, 2, 3] (or some permutation with similarly-ordered subsets), the results will sometimes yield true, and other times false.

[marcomasser]

That’s absolutely true, it seems I forgot to point out the non-deterministic behavior explicitly. Thanks for mentioning it!

[AnthonyLatsis]

@hamishknight I think we need to forward this to the string processing folks.