Many languages, such as Kotlin, Go, JavaScript, Python, Rust, C#, C++ and others, already have coroutines support that makes the async/await pattern implementation possible. This feature is not yet supported in Swift, but this can be improved by a framework without the need to change the language.
- It is the first implementation of coroutines for Swift with iOS, macOS and Linux support.
- It includes futures and channels that supplement coroutines for more flexibility.
- It is completely lock-free and uses only atomic primitives for synchronizations.
Asynchronous programming is usually associated with callbacks. It is quite convenient until there are too many of them and they start nesting. Then it's called a pyramid of doom or even callback hell.
Another problem of asynchronous programming is error handling, because Swift's natural error handling mechanism cannot be used.
There are many other frameworks that make it easy to use asynchronous code, such as Combine, RxSwift, PromiseKit and so on. They use other approaches that have some drawbacks:
- Similar to callbacks, you also need to create chained calls, that’s why you can’t normally use loops, exception handling, etc.
- Usually you need to learn a complex new API with hundreds of methods.
- Instead of working with the actual data, you need to operate with some wrappers all the time.
- Chaining of errors can be really complicated to handle.
The async/await pattern is an alternative that allows an asynchronous, non-blocking function to be structured in a way similar to an ordinary synchronous function.
It is already well-established in other programming languages and is an evolution in asynchronous programming. The implementation of this pattern is possible thanks to coroutines.
Let’s have a look at the example with coroutine inside of which await() suspends it and resumes when the result is available without blocking the thread.
//executes coroutine on the main thread
DispatchQueue.main.startCoroutine {
//extension that returns CoFuture<(data: Data, response: URLResponse)>
let dataFuture = URLSession.shared.dataTaskFuture(for: imageURL)
//await CoFuture result that suspends coroutine and doesn't block the thread
let data: Data = try dataFuture.await().data
//create UIImage from the data
guard let image = UIImage(data: data) else { return }
//execute heavy task on global queue and await the result without blocking the thread
let thumbnail: UIImage = try DispatchQueue.global().await { image.makeThumbnail() }
//set image in UIImageView on the main thread
self.imageView.image = thumbnail
}- Support only 64-bit architectures
- iOS 10+ / macOS 10.12+ / Ubuntu
- Xcode 10.4+
- Swift 5.2+
- It's available through the Swift Package Manager for iOS, macOS and Linux.
- It's available through CocoaPods for iOS and macOS.
A coroutine is a computation that can be suspended and resumed at a later time without blocking a thread. Coroutines build upon regular functions and can be executed on any scheduler with a possibility to switch among them during execution.
- Suspend instead of block. The main advantage of coroutines is the ability to suspend their execution at some point without blocking a thread and resuming later on.
- Fast context switching. Switching between coroutines is much faster than switching between threads as it does not require the involvement of operating system.
- Asynchronous code in synchronous manner. The use of coroutines allows an asynchronous, non-blocking function to be structured in a manner similar to an ordinary synchronous function. And even though coroutines can run in multiple threads, your code will still look consistent and therefore easy to understand.
The coroutines API design is as minimalistic as possible. It consists of the CoroutineScheduler protocol that describes how to schedule coroutines (DispatchQueue already conforms it), and the Coroutine structure with utility methods. This API is enough to do amazing things.
The following example shows the usage of await() inside a coroutine to wrap asynchronous calls.
//execute coroutine on the main thread
DispatchQueue.main.startCoroutine {
//await URLSessionDataTask response without blocking the thread
let (data, response, error) = try Coroutine.await { callback in
URLSession.shared.dataTask(with: url, completionHandler: callback).resume()
}
. . . use response on the main thread . . .
}Here's how we can conform NSManagedObjectContext to CoroutineScheduler for launching coroutines on it.
extension NSManagedObjectContext: CoroutineScheduler {
func scheduleTask(_ task: @escaping () -> Void) {
perform(task)
}
}
//execute coroutine on the main thread
DispatchQueue.main.startCoroutine {
let context: NSManagedObjectContext //context with privateQueueConcurrencyType
let request: NSFetchRequest<NSDictionary> //some complex request
//execute request on the context without blocking the main thread
let result: [NSDictionary] = try context.await { try context.fetch(request) }
}A future is a read-only holder for a result that will be provided later and the promise is the provider of this result. They represent the eventual completion or failure of an asynchronous operation.
The futures and promises approach itself has become an industry standart. It is a convenient mechanism to synchronize asynchronous code. But together with coroutines, it takes the usage of asynchronous code to the next level and has become a part of the async/await pattern. If coroutines are a skeleton, then futures and promises are its muscles.
- Performance. It is much faster than most of other futures and promises implementations.
- Awaitable. You can await the result inside the coroutine.
- Cancellable. You can cancel the whole chain as well as handle it and complete the related actions.
Futures and promises are represented by the corresponding CoFuture class and its CoPromise subclass.
//wraps some async func with CoFuture
func makeIntFuture() -> CoFuture<Int> {
let promise = CoPromise<Int>()
someAsyncFunc { int in
promise.success(int)
}
return promise
}It allows to start multiple tasks in parallel and synchronize them later with await().
//create CoFuture<Int> that takes 2 sec. from the example above
let future1: CoFuture<Int> = makeIntFuture()
//execute coroutine on the global queue and returns CoFuture<Int> with future result
let future2: CoFuture<Int> = DispatchQueue.global().coroutineFuture {
try Coroutine.delay(.seconds(3)) //some work that takes 3 sec.
return 6
}
//execute coroutine on the main thread
DispatchQueue.main.startCoroutine {
let sum: Int = try future1.await() + future2.await() //will await for 3 sec.
self.label.text = "Sum is \(sum)"
}It's very easy to transform or compose CoFutures into a new one.
let array: [CoFuture<Int>]
//create new CoFuture<Int> with sum of future results
let sum = CoFuture { try array.reduce(0) { try $0 + $1.await() } }Futures and promises provide a convenient way to transfer a single value between coroutines. Channels provide a way to transfer a stream of values. Conceptually, a channel is similar to a queue that allows to suspend a coroutine on receive if it is empty, or on send if it is full.
This non-blocking primitive is widely used in such languages as Go and Kotlin, and it is another instrument that improves working with coroutines.
To create channels, use the CoChannel class.
//create a channel with a buffer which can store only one element
let channel = CoChannel<Int>(capacity: 1)
DispatchQueue.global().startCoroutine {
for i in 0..<100 {
//imitate some work
try Coroutine.delay(.seconds(1))
//sends a value to the channel and suspends coroutine if its buffer is full
try channel.awaitSend(i)
}
//close channel when all values are sent
channel.close()
}
DispatchQueue.global().startCoroutine {
//receives values until closed and suspends a coroutine if it's empty
for i in channel.makeIterator() {
print("Receive", i)
}
print("Done")
}All launched coroutines, CoFutures and CoChannels, usually do not need to be referenced. They are deinited after their execution. But often there is a need to complete them earlier, when they are no longer needed. For this, CoFuture and CoChannel have methods for canceling.
CoScope makes it easier to manage the life cycle of these objects. It allows you to keep weak references to them and cancel if necessary or on deinit.
You can add coroutines, CoFutures, CoChannels and other CoCancellable to CoScope to cancel them when they are no longer needed or on deinit.
class ViewController: UIViewController {
let scope = CoScope() //will cancel all objects on `cancel()` or deinit
func performSomeWork() {
//create new `CoChannel` and add to `CoScope`
let channel = makeSomeChannel().added(to: scope)
//execute coroutine and add to `CoScope`
DispatchQueue.main.startCoroutine(in: scope) { [weak self] in
for item in channel.makeIterator() {
try self?.performSomeWork(with: item)
}
}
}
func performSomeWork(with item: Item) throws {
//create new `CoFuture` and add to `CoScope`
let future = makeSomeFuture(item).added(to: scope)
let result = try future.await()
. . . do some work using result . . .
}
}