How you can use the consequence sort to deal with errors in Swift 5?

Error dealing with fundamentals in Swift

The way in which of dealing with errors modified loads because the first model of Swift. The primary huge milestone occurred in Swift 2, the place Apple fully revamped error administration. These days you need to use the do, strive, catch, throw, throws, rethrows key phrases as a substitute of coping with nasty NSError pointers, so this was a warmly welcomed addition for the language. Now in Swift 5 we take one other big leap ahead by introducing the Consequence sort as a built-in generic. First, let me present you all the most effective practices of error dealing with within the Swift programming language, subsequent I am going to present you some cool stuff through the use of outcomes to cope with errors. 🚧

Optionals as error indicators

For easy eventualities you possibly can at all times use elective values, to point that one thing dangerous occurred. Additionally the guard assertion is extraordinarily useful for conditions like this.

let zeroValue = Int("0")! 
let nilValue = Int("not a quantity") 

guard let quantity = Int("6") else {
    fatalError("Ooops... this could at all times work, so we crash.")
}
print(quantity)

For those who do not actually care in regards to the underlying sort of the error, this method is okay, however generally issues can get extra difficult, so that you may want some particulars about the issue. Anyway, you possibly can at all times cease the execution by calling the fatalError methodology, however should you accomplish that, properly… your app will crash. 💥

There are additionally a pair different methods of cease execution course of, however this might be a subject of a standalone submit, so right here is only a fast cheat sheet of obtainable strategies:

precondition(false, "ouch")
preconditionFailure("ouch")
assert(false, "ouch")
assertionFailure("ouch")
fatalError("ouch")
exit(-1)

The important thing distinction between precondition and assertion is that assert will work solely in debug builds, however precondition is evaluated at all times (even in launch builds). Each strategies will set off a deadly error if the situation fails aka. is fake. ⚠️

Throwing errors through the use of the Error protocol

You possibly can outline your individual error varieties by merely confirming to the built-in Error protocol. Often most builders use an enum with a view to outline totally different causes. You may also have a customized error message should you conform to the LocalizedError protocol. Now you are able to throw customized errors, simply use the throw key phrase if you would like to lift an error of your sort, however should you accomplish that in a operate, it’s a must to mark that operate as a throwing operate with the throws key phrases. 🤮

enum DivisionError: Error {
    case zeroDivisor
}

extension DivisionError: LocalizedError {
    public var errorDescription: String? {
        change self {
        case .zeroDivisor:
            return "Division by zero is sort of problematic. " +
                   "(https://en.wikipedia.org/wiki/Division_by_zero)"
        }
    }
}

func divide(_ x: Int, by y: Int) throws -> Int {
    guard y != 0 else {
        throw DivisionError.zeroDivisor
    }
    return x / y
}

Nice, so the divide operate above can generate a customized error message. If the divisor is zero it’s going to throw the zeroDivision error case. Now think about the next situation: you are attempting to learn the contents of a file from the disk. There might be a number of kinds of errors associated to permission or file existence, and so forth.

Rethrowing Capabilities and Strategies A operate or methodology will be declared with the rethrows key phrase to point that it throws an error provided that certainly one of it’s operate parameters throws an error. These capabilities and strategies are referred to as rethrowing capabilities and rethrowing strategies. Rethrowing capabilities and strategies should have not less than one throwing operate parameter.

Okay, so a throwing operate can emit totally different error varieties, additionally it may possibly propagate all of the parameter errors, however how will we deal with (or ought to I say: catch) these errors?

The do-try-catch syntax

You simply merely must attempt to execute do a throwing operate. So do not belief the grasp, there’s undoubtedly room for attempting out issues! Unhealthy joke, proper? 😅

do {
    let quantity = strive divide(10, by: 0)
    print(quantity)
}
catch let error as DivisionError {
    print("Division error handler block")
    print(error.localizedDescription)
}
catch {
    print("Generic error handler block")
    print(error.localizedDescription)
}

As you possibly can see the syntax is fairly easy, you could have a do block, the place you possibly can attempt to execute your throwing capabilities, if one thing goes mistaken, you possibly can deal with the errors in numerous catch blocks. By default an error property is obtainable inside each catch block, so you do not have to outline one your self by hand. You possibly can nevertheless have catch blocks for particular error varieties by casting them utilizing the let error as MyType sytnax proper subsequent to the catch key phrase. So at all times strive first, do not simply do! 🤪

Variations between strive, strive? and take a look at!

As we have seen earlier than you possibly can merely attempt to name a operate that throws an error inside a do-catch block. If the operate triggers some type of error, you possibly can put your error dealing with logic contained in the catch block. That is quite simple & simple.

Typically should you do not actually care in regards to the underlying error, you possibly can merely convert your throwing operate consequence into an elective through the use of strive?. With this method you may get a 0 consequence if one thing dangerous occurs, in any other case you may get again your common worth as it’s anticipated. Right here is the instance from above through the use of strive?:

guard let quantity = strive? divide(10, by: 2) else {
    fatalError("This could work!")
}
print(quantity) 

One other approach is to forestall error propagation through the use of strive!, however it’s a must to be extraordinarily cautious with this method, as a result of if the execution of the “tried operate” fails, your utility will merely crash. So use provided that you are completely certain that the operate will not throw an error. ⚠️

let quantity = strive! divide(10, by: 2) 
print(quantity)

There are a number of locations the place it is accepted to make use of drive strive, however in many of the circumstances it is best to go on an alternate path with correct error handlers.

Swift errors should not exceptions

The Swift compiler at all times requires you to catch all thrown errors, so a state of affairs of unhandled error won’t ever happen. I am not speaking about empty catch blocks, however unhandled throwing capabilities, so you possibly can’t strive with out the do-catch companions. That is one key distinction when evaluating to exceptions. Additionally when an error is raised, the execution will simply exit the present scope. Exceptions will often unwind the stack, that may result in reminiscence leaks, however that is not the case with Swift errors. 👍

Introducing the consequence sort

Swift 5 introduces a long-awaited generic consequence sort. Which means error dealing with will be much more easy, with out including your individual consequence implementation. Let me present you our earlier divide operate through the use of Consequence.

func divide(_ x: Int, by y: Int) -> Consequence<Int, DivisionError> {
    guard y != 0 else {
        return .failure(.zeroDivisor)
    }
    return .success(x / y)
}

let consequence = divide(10, by: 2)
change consequence {
case .success(let quantity):
    print(quantity)
case .failure(let error):
    print(error.localizedDescription)
}

The consequence sort in Swift is mainly a generic enum with a .success and a .failure case. You possibly can go a generic worth in case your name succeeds or an Error if it fails.

One main benefit right here is that the error given again by result’s sort secure. Throwing capabilities can throw any type of errors, however right here you possibly can see from the implementation {that a} DivisionError is coming again if one thing dangerous occurs. One other profit is that you need to use exhaustive change blocks to “iterate by means of” all of the potential error circumstances, even and not using a default case. So the compiler can hold you secure, e.g. if you will introduce a brand new error sort inside your enum declaration.

So through the use of the Consequence sort it is clear that we’re getting again both consequence information or a strongly typed error. It isn’t potential to get each or neither of them, however is that this higher than utilizing throwing capabilities? Properly, let’s get asynchrounous!

func divide(_ x: Int, by y: Int, completion: ((() throws -> Int) -> Void)) {
    guard y != 0 else {
        completion { throw DivisionError.zeroDivisor }
        return
    }
    completion { return x / y }
}

divide(10, by: 0) { calculate in
    do {
        let quantity = strive calculate()
        print(quantity)
    }
    catch {
        print(error.localizedDescription)
    }
}

Oh, my pricey… an internal closure! A completion handler that accepts a throwing operate, so we will propagate the error thrown to the outer handler? I am out! 🤬

Another choice is that we remove the throwing error fully and use an elective in consequence, however on this case we’re again to sq. one. No underlying error sort.

func divide(_ x: Int, by y: Int, completion: (Int?) -> Void) {
    guard y != 0 else {
        return completion(nil)
    }
    completion(x / y)
}

divide(10, by: 0) { consequence in
    guard let quantity = consequence else {
        fatalError("nil")
    }
    print(quantity)
}

Lastly we’re getting someplace right here, however this time let’s add our error as a closure parameter as properly. It is best to notice that each parameters should be optionals.

func divide(_ x: Int, by y: Int, completion: (Int?, Error?) -> Void) {
    guard y != 0 else {
        return completion(nil, DivisionError.zeroDivisor)
    }
    completion(x / y, nil)
}

divide(10, by: 0) { consequence, error in
    guard error == nil else {
        fatalError(error!.localizedDescription)
    }
    guard let quantity = consequence else {
        fatalError("Empty consequence.")
    }
    print(quantity)
}

Lastly let’s introduce consequence, so we will remove optionals from our earlier code.

func divide(_ x: Int, by y: Int, completion: (Consequence<Int, DivisionError>) -> Void) {
    guard y != 0 else {
        return completion(.failure(.zeroDivisor))
    }
    completion(.success(x / y))
}

divide(10, by: 0) { consequence in
    change consequence {
    case .success(let quantity):
        print(quantity)
    case .failure(let error):
        print(error.localizedDescription)
    }
}

See? Strongly typed errors, with out optionals. Dealing with errors in asynchronous operate is means higher through the use of the Consequence sort. For those who think about that many of the apps are doing a little type of networking, and the result’s often a JSON response, there you have already got to work with optionals (response, information, error) plus you could have a throwing JSONDecoder methodology… cannot wait the brand new APIs! ❤️

Working with the Consequence sort in Swift 5

We already know that the consequence sort is mainly an enum with a generic .succes(T) and a .failure(Error) circumstances, however there’s extra that I might like to indicate you right here. For instance you possibly can create a consequence sort with a throwing operate like this:

let consequence = Consequence {
    return strive divide(10, by: 2)
}

Additionally it is potential to transform again the consequence worth by invoking the get operate.

do {
    let quantity = strive consequence.get()
    print(quantity)
}
catch {
    print(error.localizedDescription)
}

Additionally there are map, flatMap for reworking success values plus you may also use the mapError or flatMapError strategies if you would like to rework failures. 😎

let consequence = divide(10, by: 2) 


let mapSuccess = consequence.map { divide($0, by: 2) } 


let flatMapSuccess = consequence.flatMap { divide($0, by: 2) } 
let mapFailure = consequence.mapError { 
    NSError(area: $0.localizedDescription, code: 0, userInfo: nil)
}

let flatMapFailure = consequence.flatMapError { 
    .failure(NSError(area: $0.localizedDescription, code: 0, userInfo: nil)) 
}

That is it in regards to the Consequence sort in Swift 5. As you possibly can see it is extraordinarily highly effective to have a generic implementation constructed immediately into the language. Now that we’ve got consequence, I simply want for increased kinded varieties or an async / await implementation. 👍