Phantom types in Swift
Not every language with a static type system has so strong type-safety like Swift. Swift features like phantom types, generic type extensions, enums with associated types create an excellent foundation. This week we will learn how to use phantom types to build type-safe APIs.
The all-in-one macOS app that enhances App Store Connect, supercharging your app updates, localization, and ASO with AI-powered tools. Save 25% and try now!
Basics
A phantom type is a generic type that is declared but never used inside a type where it is declared. It is usually used as a generic constraint to build a more type-safe and robust API. Let’s take a look at the quick example.
struct Identifier<Holder> {
let value: Int
}
In the example above, we have the Identifier struct with a generic Holder type declared. As you can see, we don’t use the Holder type inside the Identifier type. That’s why it is called phantom type. Now let’s think about the benefits of using phantom types like this.
struct User {
let id: Identifier<Self>
}
struct Product {
let id: Identifier<Self>
}
let product = Product(id: .init(value: 1))
let user = User(id: .init(value: 1))
user.id == product.id
We create User and Product types and use the previously created Identifier struct. We set the value of the identifier to 1 for the newly created user and product. But when we try to compare them, the Swift compiler fails with the error:
Binary operator ‘==’ cannot be applied to operands of type ‘Identifier-User’ and ‘Identifier-Product’.
And that’s great because there is no reason to compare user and product identifiers. We can do it only accidentally. The Swift compiler doesn’t allow us to mix the identifiers between users and products because of phantom type and recognize them as entirely different types. Here is another example where the Swift compiler doesn’t allow us to mix identifiers.
func fetch(_ product: Identifier<Product>) -> Product? {
// return product by id
}
fetch(user.id)
To learn more about the benefits of using phantom types, look at my “Building type-safe networking in Swift” post.
Type safety in HealthKit
We learned the basics of phantom types. Now we can move to more advanced examples. I built a couple of health-oriented apps that use HealthKit to store and query health data from the Apple Watch. Let’s look at the typical code that I use to fetch data from the Apple Health app.
import HealthKit
let store = HKHealthStore()
let bodyMass = HKQuantityType.quantityType(
forIdentifier: HKQuantityTypeIdentifier.bodyMass
)!
let query = HKStatisticsQuery(
quantityType: bodyMass,
quantitySamplePredicate: nil,
options: .discreteAverage
) { _, statistics, _ in
let average = statistics?.averageQuantity()
let mass = average?.doubleValue(for: .meter())
}
store.execute(query)
In the example above, we create a query to fetch user’s weight from the Apple Health app. In the completion handler, we try to get the average and convert it to meters. As you can guess, it is impossible to convert body mass to meters, and the app will crash here. We will try to solve the issue by introducing phantom type to build more type-safe API.
enum Distance {
case mile
case meter
}
enum Mass {
case pound
case gram
case ounce
}
struct Statistics<Unit> {
let value: Double
}
extension Statistics where Unit == Mass {
func convert(to unit: Mass) -> Double {
}
}
extension Statistics where Unit == Distance {
func convert(to unit: Distance) -> Double {
}
}
let weight = Statistics<Mass>(value: 75)
weight.convert(to: Distance.meter)
Here is a possible solution for the HealthKit framework that uses phantom type to improve API safety. We introduce Mass and Distance enums to have distinct units. And as soon you try to convert mass into the distance, the Swift compiler stops you with a great error message:
Cannot convert the value of type ‘Distance’ to expected argument type ‘Mass’
Conclusion
Today we learned phantom types, one of my favorite features of the Swift language. It looks like there are a lot of possible applications for phantom types. Feel free to share with me how you make your API more type-safe by using phantom types. I hope you enjoy the post. Follow me on Twitter and ask your questions related to this article. Thanks for reading, and see you next week!