Giter Site home page Giter Site logo

swift-style-guide's Introduction

Prolific Swift Style Guide

Table Of Contents

Overview

This is Prolific's style guide for writing code in Swift. The purpose of this guide is to develop a universal standard for Swift code that makes our codebases consistent and easy to read. This guide aims for consistent and clean code written in Swift in line with Apple and the general community.

The standards have been influenced by:

  • Apple's language design decisions -- specifically, its desire for Swift code to be written concisely and expressively
  • Xcode's defaults for code formatting
  • The general community

Contributing

If you wish to contribute, feel free to submit a pull request or file an issue on this repo. In the pull request, be sure to discuss what problem you are intending to solve and the rationale to do so. When submitting a pull request, consider:

  • Is my suggestion general enough to be considered a code standard for all code bases?
  • Does my suggestion fall in line with Apple's desire for the language?
  • Does your suggestion invalidate a language feature?

Make sure to consider the resources in the open-source Swift repo; specifically, read through the various proposals for new language features as well as the most-commonly rejected proposals in order to guide your design principals.

Standards

File structure

You should not define multiple public/internal types (ie class, struct, enum) in the same file; each type should have its own file.

The following list should be the standard organization of all your Swift files, in this specific order:

Before the type declaration:

  • Private Class/Struct/Enum

Inside the type declaration:

  • Override Properties
  • Properties
  • Static/Class Variables
  • Static/Class Functions
  • Init/Deinit
  • Override Functions
  • Instance Functions

After the type declaration:

  • Extensions

Each section above should be organized by accessibility:

  • Public
  • Internal
  • Private

When implementing a protocol you should create an extension of your class that lives in the same file to separate the core logic of your class and your protocol implementation.

Enum & Protocol

All enums should live in their own file, except in cases where the enum is declared as private. In cases where the enum is declared private, declare the enum at the top of the file, above the type declaration.

Rationale: With enum and protocol types Swift allows defining functions and extensions. Because of that these types can become complex which is why they should be defined in their own file.

Usage of MARK / TODO / FIXME

To help organize your files you may want to use pragma marks to clearly separate your functions, properties and extensions. For extensions, use one MARK per extension. For example, // MARK: UITableViewDelegate Functions instead of // MARK: Extensions.

Xcode is also able to display TODO and FIXME tags directly in the source navigator, you should consider using them to find your notes inside your files.

// TODO: implement this feature

// FIXME: fix it it's not working

Other conventional comment tags, such as NOTE are not recognized by Xcode.

Types

Prefer Swift native types over Objective-C types when possible. Because Swift types bridge to Objective-C, you should avoid types like NSString and NSNumber in favor of Int or String.

Rationale: Avoid subclassing NSObject or using the @objc flag unless it is required to implement an NSObjectProtocol type. Subclassing NSObject or using the @objc flag automatically creates an Objective-C object that uses dynamic dispatch over the preferred static of Swift which can impact the performance of the app.

Preferred:

class MyClass {
...
}

Not preferred:

@objc class MyClass {
...
}

If you need functionality from an Objective-C type that is not available in its corresponding Swift type (for instance, needing an NSString function that is not available on String), cast your Swift raw type to the corresponding Objective-C type instead of declaring the type as the Objective-C type.

Preferred:

let scale = 5.0
let scaleString = (5.0 as NSNumber).stringValue
let scaleInt = Int(scale)

Not preferred:

let scale: NSNumber = 5.0
let scaleString = scale.stringValue
let scaleInt = scale.integerValue

Statement Termination

Unlike Objective-C, omit the use of ; to terminate statements. Instead, simply use new lines to indicate the end of a statement.

let myVar = 0
doSomething(myVar)

return

Avoid multiple statements on a single line.

guard let obj = myObj else { print("Something went wrong"); return; } // Wrong! Instead, place each item on its own line.

Variable Declaration

For declaring variables, favor let instead of var unless you need a mutable object or container.

func formatDate(date: NSDate) -> String {
    let dateFormatter = NSDateFormatter() // In this case, use `let` since the variable `dateFormatter` never changes once set
    dateFormatter.dateStyle = .ShortStyle
    return dateFormatter.stringFromDate(date)
}

func arrays() {
    let array = ["Hello", "Ciao", "Aloha"] // use let here since this is an immutable container

    var mutableArray = [String]() // Use var here since this container is mutable
    mutableArray.append("Farewell")
    mutableArray.append("Arrivederci")
}

Self

Never use the self modifier except in cases where it is necessary for the compiler or to alleviate conflicts with other variable declarations.

class Object {
	private var name = ""
	
	func useName() {
		// Let self be implied when it can be understood.
		otherObject.doSomethingWithName(name)
		setName("Will Smith")
	}
	
	func setName(name: String) {
		// Use self here to prevent conflicts with the `name` parameter being passed.
		self.name = name
	}
	
	func setNameAsync(newName: String) {
		// Use implicit self outside closures...
		otherObject.doSomethingWithName(name, then: { 
			// .. but within, you must use self to ease the compiler.
			self.setName("Jason")
		})
	}
}

Rationale: The idea behind this is that implicit use of self makes the conditions where you must use self (for instance, within closures) much more apparent and will make you think more on the reasons why you are using it. In closures, think about: should self be weak instead of strong? Apple has even rejected a request to enforce use of self for this reason, among others.

Structs & Classes

In Swift, structs maintain value semantics which means their values are copied when assigned. Classes, on the other hand, act like pointers from C and Objective-C; they are called reference types and the internal data is shared amongst instances of assigning.

When composing your types, consider carefully what they're going to be used for before choosing what they should end up being. In general, consider structs for types that are:

  • Immutable
  • Stateless
  • Have a definition for equality

Swift structs also have other, tangible benefits as well:

  • Faster
  • Safer due to copying rather than referencing
  • Thread safe -- copies allow mutations to happen independently of other instances.

In general, you should favor structs and protocols over classes; even in cases where polymorphism would dictate the usage of a class, consider if you can achieve a similar result via protocols and extensions. This allows you to achieve polymorphism via composition rather than inheritance.

Bracket Syntax

For brackets, prefer the Xcode-default syntax of having the opening brace be on the same line as the statement opening it:

internal final class MyObject {
}

internal enum MyEnum {
}

func doSomething() {
}

if true == false {
}

let doSomething: () -> Void = { 
}

For type declarations, include a single space between the type declaration and the first item implemented within it:

internal final class MyObject {

	let value = 0

In addition, include a space before the type declarations closing bracket:

internal final class MyObject {
	
	let value = 0
	
	func doSomething() {
		value += 1
	}
	
}

This also applies to extension declarations:

extension MyObject {

	func doAnotherThing() {
		...
	}
	
}

Do not include this extra space in function declarations:

func doSomething() {
	let value = 0
}

Rationale: Simply put, this is the Xcode default and standard, and it's not worth fighting. This keeps things consistent across the board and makes our lives as developers considerably easier.

Force Unwrap

Unless there is a situation that absolutely calls for it, usage of the force-unwrap operator (!) should be minimized, if not eliminated all together. With the many and varied ways of unwrapping optionals, it is safer and simpler to declare variables as optional and unwrap them when needing their values than it is to force-unwrap them and potentially introduce runtime errors into the code base.

if let text = self.textLabel?.text {
    doSomethingWithText(text)
}

If you are interoping from Objective-C and the declaration automatically translates into force-unwrapped parameters, replace them with ? parameters instead.

The one exception to this rule are IBOutlets. @IBOutlets may be declared using ! so long as they are expected to exist for the lifetime of the view controller object:

@IBOutlet private weak var textLabel: UILabel!

Error Handling

The emergence of try / catch in Swift 2 has added powerful ways to define and return errors when something fails. The emergence of ErrorType as well for defining errors makes error definitions much more convenient over the cumbersome NSError. Because of this, for functions that can have multiple points of failure, you should always define it as throws and return a well-defined ErrorType.

Consider the following contrived example:

func multiplyEvensLessThan10(evenNumber: Int) -> Int? {
	guard evenNumber % 2 == 0 && evenNumber < 10 else {
		return nil
	}
	
	return evenNumber * 2
}

The function above fails because it only expects evens less than 10 and returns an optional if that is violated. While this works and is simple, it is more Objective-C than Swift in its composition. The caller may not know which parameter they violated. For Swift, instead consider refactoring it as follows:

internal enum NumberError: ErrorType {
	case NotEven
	case TooLarge
}

func multiplyEvens(evenNumber: Int) throws -> Int {
	guard evenNumber % 2 == 0 else {
		throw NumberError.NotEven
	}
	
	guard evenNumber < 10 else {
		throw NumberError.TooLarge
	}
	
	return evenNumber * 2
}

The above, while slightly more cumbersome, this has well-defined benefits:

  • The caller is able to explicitly determine why their call to the function failed and thus can take active steps to recover:
let result: Int
do {
    result = try multiplyEvens(3)
} catch NumberError.NotEven {
    return 0
} catch NumberError.TooLarge {
    print("The Number entered was too large! Try again.")
    return -1
} catch {
    fatalError("Unhandled error occurred.")
}

return result
  • Try/catch semantics allow the caller to still retain the old optional functionality if the error is not relevant and they only care about the outcome:
let result: Int? = try? multiplyEvens(1)
  • Or, if the caller knows that it will not violate any of the parameters for a valid input:
let result: Int = try! multiplyEvens(2)

So, even though we've now modified our API to use swift exceptions, we can still retain the old Objective-C functionality giving the caller the choice of how they wish to handle the result of this failable operation.

NSError

In general, you should avoid NSError in Swift in favor of defining your own ErrorType. However, in the event you do need to use NSError (for interop with Objective-C, for instance):

  • Define a proper domain for your NSError. This should be specific to your module and ideally would be reflective of your bundle identifier (i.e. com.prolificinteractive.MyApp)
  • Define a list of the various error codes and what they translate to. These should be some sort of easily readable constant or enum value so that way the caller is able to determine what exactly failed.
  • In the userInfo, include at least a localized description (NSLocalizedDescriptionKey) that accurately and concisely describes the nature of the error.

Access Modifiers

Always specify access modifiers to top-level definitions. Do not specify them otherwise unless the modifier is needed:

internal final class Object {
    var myInt: Int

    private func doWork() {
        ...
    }
}

Further, the access modifier should always be presented first in the list before any other modifiers:

// Good!
private unowned var obj: Object

internal func doSomething() {
}

// Wrong!
weak public var obj: Object?

Type Declarations

When declaring types, the colon should be placed immediately after the identifier followed by one space and the type name.

var intValue: Int

// Do NOT do this
var intValue : Int

In all use-cases, the colon should be associated with the left-most item with no spaces preceding and one space afterwards:

let myDictionary: [String: AnyObject] = ["String": 0]

Type Inference

Prefer letting the compiler infer the type instead of explcitly stating it, wherever possible:

var max = 0 		// Int
var name = "John" 	// String
var rect = CGRect()	// CGRect

// Do not do:

var max: Int = 0
var name: String = "John"
var rect: CGRect = CGRect()

// Ok since the inferred type is not what we wanted:

var max: Hashable = 0 // Compiler would infer Int, but we only want it to be hashable
var name: String? = "John" // Compiler would infer this not to be optional, but we may need to nil it out later.

Rationale The compiler is pretty smart, and we should utilize it where necessary. It is generally obvious what the type is going to be in the instances above, so unless we need to be more explicit (as in the last examples above), it is better to omit unneeded words.

Nil Checking

Favor if-let checking over direct nil checking in all cases except when the result of this check is required:

guard let item = myItem else {
	return
}

doSomethingWith(item)
if let _ = error { // Prefer this over `if error != nil`
	fatalError()
}
func isError(error: Error?) -> Bool {
	return (error != nil) // In this case, we need the result of the bool, and this is much cleaner than the other options.
}

Implicit Getters

When overriding only the getter of a property, omit the use of get:

var myInt: Int {
	return 0
}

// Do not do this:
var myInt: Int {
	get {
		return 0
	}
}

For all other cases, specify the modifier as needed (set, didSet, etc.). This is compiler enforced.

Rationale The getter is implied enough to make sense without having to make it explicitly. It also cuts down on unneccessary verbiage and spacing to make code clearer.

Enums

For enum declarations, declare each enum case on a new line with its own case statement instead of a comma-separated list.

internal enum State {
	case Open
	case Closed
	case Pending
	case Faulted
}

Prefer singular case for enum names instead of plural: State vs. States:

var currentState = State.Open
var previousState = States.Closed // Reads less clearly than the previous option.

For enums with raw values, declare the raw value on the same line as its declaration:

internal enum HTTPMethod: String {
	case Get = "GET"
	case Post = "POST"
}

For any other functions or properties associated with the enum, place them after the last case item in the enum list:

internal enum State {
	case Open
	case Closed
	case Pending
	case Faulted
	
	func nextState() -> State {
		...
	}
}

In cases where the enum's type name can be omitted, do so:

let state = State.Open

if state == .Closed { ... // Prefer .Closed instead of State.Closed

Use of final

Classes should always be marked as final unless they are being used as a base class for another type. In instances where a class can be subclassed, any function or variable that should not be overriden by a subclass should be diligently marked as final.

// Not built for inheritance.
internal final class Object { 

}

// Purposefully utilizable as a base class
internal class BaseClass { 
	
	func doSomething () {
	}
	
	// Properly marked as final so subclasses cannot override
	final func update() { 
	}
	
}

internal final class SubClass: BaseClass {
	
	override func doSomething() {
		update()
	}
	
}

Rationale Subclassing in instances where the original class was not built to support subclasses can be a common source of bugs. Marking classes as final indicates that it was developed under the assumption that it would act on its own without regard for subclasses.

swift-style-guide's People

Contributors

haud avatar htinlinn avatar jstnheo avatar klein-thibault avatar paulmiard avatar thibaultklein avatar yoyoyoseob avatar

Watchers

 avatar  avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.