Files
George Barnett 042e1c4d9d Move platform requirements to availability annotations (#348)
* Move platform requirements to availability annotations

Adding or raising the deployment platforms in the package manifest is a
SemVer major breaking change as consumers must also add or raise their
deployment platforms. This is a known limitation of SwiftPM.

Unforunately this means that it's very difficult for non-leaf packages
to adopt packages which declare their platforms in the manifest. Doing
so puts the brakes on adoption and ecosystem growth. For 'core' packages
like this one availability constraints should be expressed on
declarations rather than in the manifest.

This patch adds equivalent availability annotations to declarations
across the package and removes platforms from the package manifest.

* Address naming feedback
2025-04-22 08:07:37 +01:00

120 lines
3.3 KiB
Swift

//===----------------------------------------------------------------------===//
//
// This source file is part of the Swift Async Algorithms open source project
//
// Copyright (c) 2022 Apple Inc. and the Swift project authors
// Licensed under Apache License v2.0 with Runtime Library Exception
//
// See https://swift.org/LICENSE.txt for license information
//
//===----------------------------------------------------------------------===//
import DequeModule
/// Creates an asynchronous sequence of elements from two underlying asynchronous sequences
@available(AsyncAlgorithms 1.0, *)
public func merge<
Base1: AsyncSequence,
Base2: AsyncSequence,
Base3: AsyncSequence
>(_ base1: Base1, _ base2: Base2, _ base3: Base3) -> AsyncMerge3Sequence<Base1, Base2, Base3>
where
Base1.Element == Base2.Element,
Base1.Element == Base3.Element,
Base1: Sendable,
Base2: Sendable,
Base3: Sendable,
Base1.Element: Sendable
{
return AsyncMerge3Sequence(base1, base2, base3)
}
/// An `AsyncSequence` that takes three upstream `AsyncSequence`s and combines their elements.
@available(AsyncAlgorithms 1.0, *)
public struct AsyncMerge3Sequence<
Base1: AsyncSequence,
Base2: AsyncSequence,
Base3: AsyncSequence
>: Sendable
where
Base1.Element == Base2.Element,
Base1.Element == Base3.Element,
Base1: Sendable,
Base2: Sendable,
Base3: Sendable,
Base1.Element: Sendable
{
public typealias Element = Base1.Element
private let base1: Base1
private let base2: Base2
private let base3: Base3
/// Initializes a new ``AsyncMerge2Sequence``.
///
/// - Parameters:
/// - base1: The first upstream ``Swift/AsyncSequence``.
/// - base2: The second upstream ``Swift/AsyncSequence``.
/// - base3: The third upstream ``Swift/AsyncSequence``.
init(
_ base1: Base1,
_ base2: Base2,
_ base3: Base3
) {
self.base1 = base1
self.base2 = base2
self.base3 = base3
}
}
@available(AsyncAlgorithms 1.0, *)
extension AsyncMerge3Sequence: AsyncSequence {
@available(AsyncAlgorithms 1.0, *)
public func makeAsyncIterator() -> Iterator {
let storage = MergeStorage(
base1: base1,
base2: base2,
base3: base3
)
return Iterator(storage: storage)
}
}
@available(AsyncAlgorithms 1.0, *)
extension AsyncMerge3Sequence {
@available(AsyncAlgorithms 1.0, *)
public struct Iterator: AsyncIteratorProtocol {
/// This class is needed to hook the deinit to observe once all references to the ``AsyncIterator`` are dropped.
///
/// If we get move-only types we should be able to drop this class and use the `deinit` of the ``AsyncIterator`` struct itself.
final class InternalClass: Sendable {
private let storage: MergeStorage<Base1, Base2, Base3>
fileprivate init(storage: MergeStorage<Base1, Base2, Base3>) {
self.storage = storage
}
deinit {
self.storage.iteratorDeinitialized()
}
func next() async rethrows -> Element? {
try await storage.next()
}
}
let internalClass: InternalClass
fileprivate init(storage: MergeStorage<Base1, Base2, Base3>) {
internalClass = InternalClass(storage: storage)
}
public mutating func next() async rethrows -> Element? {
try await internalClass.next()
}
}
}
@available(*, unavailable)
extension AsyncMerge3Sequence.Iterator: Sendable {}