Repository files navigation

This is not meant to be a beginner's guide or a detailed discussion about Objective-C; it is meant to be a quick reference to common, high level topics.

• Read mySwift cheatsheet as well (Swift will replace Objective-C for iOS apps).
• Support this project via Gratipay
• To download a PDF version of this, useGitPrint.com

Note: I wrote this over a stretch of 3 days, so it could probably use some editing and clarifications where necessary. Please feel free to edit this document to update or improve upon it, making sure to keep with the general formatting of the document. The list of contributors can befound here.

For advanced Objective-C topics, these two sites are updated weekly:

• http://www.objc.io
• http://nshipster.com

If something isn't mentioned here, it's probably covered in detail in one of these:

• Swift, Apple's new language to replace Objective-C
• NSHipster's "Fake" book aka concise usage examples of common Objective-C tasks
• The Objective-C Programming Language
• Coding Guidelines for Cocoa
• iOS App Programming Guide
• Cocoa Encyclopedia

• Commenting
• Data Types
• Constants
• Operators
• Declaring Classes
• Preprocessor Directives
• Compiler Directives
• Literals
• Methods
• Properties and Variables
• Naming Conventions
• Blocks
• Control Statements
• Enumeration
• Extending Classes
• Error Handling
• Passing Information
• User Defaults
• Common Patterns

Comments should be used to organize code and to provide extra information for future refactoring or for other developers who might be reading your code. Comments are ignored by the compiler so they do not increase the compiled program size.

Two ways of commenting:

// This is an inline comment/* This is a block comment   and it can span multiple lines.*/// You can also use it to comment out code/*- (SomeOtherClass *)doWork{    // Implement this}*/

Usingpragma to organize your code:

#pragma mark - Use pragma mark to logically organize your code// Declare some methods or variables here#pragma mark - They also show up nicely in the properties/methods list in Xcode// Declare some more methods or variables here

Back to top

Permissible sizes of data types are determined by how many bytes of memory are allocated for that specific type and whether it's a 32-bit or 64-bit environment. In a 32-bit environment,long is given 4 bytes, which equates to a total range of2^(4*8) (with 8 bits in a byte) or4294967295. In a 64-bit environment,long is given 8 bytes, which equates to2^(8*8) or1.84467440737096e19.

For a complete guide to 64-bit changes, pleasesee the transition document.

Note: Objective-C inherits all of the C language primitive types and then adds a few extras.

void is C's empty data type. It is most commonly used to specify the return type for functions that don't return anything.

Integers can be signed or unsigned. When signed, they can be either positive or negative and when unsigned, they can only be positive. Example: When declaring anunsigned int, the range of allowable integer values for a 32-bit compiler will shift from -2147483648 to +2147483647 to instead be 0 to +4294967295.

Integer types with their accompanying byte sizes:

// Char (1 byte for both 32-bit and 64-bit)unsignedchar anUnsignedChar =255;NSLog(@"char size:%zu",sizeof(char));// Short (2 bytes for both 32-bit and 64-bit)short aShort = -32768;unsignedshort anUnsignedShort =65535;NSLog(@"short size:%zu",sizeof(short));// Integer (4 bytes for both 32-bit and 64-bit)int anInt = -2147483648;unsignedint anUnsignedInt =4294967295;NSLog(@"int size:%zu",sizeof(int));// Long (4 bytes for 32-bit, 8 bytes for 64-bit)long aLong = -9223372036854775808;// 64-bitunsignedlong anUnsignedLong =18446744073709551615;// 64-bitNSLog(@"long size:%zu",sizeof(long));// Long Long (8 bytes for both 32-bit and 64-bit)longlong aLongLong = -9223372036854775808;unsignedlonglong anUnsignedLongLong =18446744073709551615;NSLog(@"long long size:%zu",sizeof(longlong));

Fixed width integer types with their accompanying byte sizes as the variable names:

// Exact integer typesint8_t aOneByteInt =127;uint8_t aOneByteUnsignedInt =255;int16_t aTwoByteInt =32767;uint16_t aTwoByteUnsignedInt =65535;int32_t aFourByteInt =2147483647;uint32_t aFourByteUnsignedInt =4294967295;int64_t anEightByteInt =9223372036854775807;uint64_t anEightByteUnsignedInt =18446744073709551615;// Minimum integer typesint_least8_t aTinyInt =127;uint_least8_t aTinyUnsignedInt =255;int_least16_t aMediumInt =32767;uint_least16_t aMediumUnsignedInt =65535;int_least32_t aNormalInt =2147483647;uint_least32_t aNormalUnsignedInt =4294967295;int_least64_t aBigInt =9223372036854775807;uint_least64_t aBigUnsignedInt =18446744073709551615;// The largest supported integer typeintmax_t theBiggestInt =9223372036854775807;uintmax_t theBiggestUnsignedInt =18446744073709551615;

Floats cannot be signed or unsigned.

// Single precision floating-point (4 bytes for both 32-bit and 64-bit)float aFloat =72.0345f;NSLog(@"float size:%zu",sizeof(float));// Double precision floating-point (8 bytes for both 32-bit and 64-bit)double aDouble = -72.0345f;NSLog(@"double size:%zu",sizeof(double));// Extended precision floating-point (16 bytes for both 32-bit and 64-bit)longdouble aLongDouble =72.0345e7L;NSLog(@"long double size:%zu",sizeof(longdouble));

id : Known as the anonymous or dynamic object type, it can store a reference to any type ofobject with no need to specify a pointer symbol.

id delegate = self.delegate;

Class : Used to denote an object's class and can be used for introspection of objects.

Class aClass = [UIViewclass];

Method : Used to denote a method and can be used for swizzling methods.

Method aMethod =class_getInstanceMethod(aClass, aSelector);

SEL : Used to specify a selector which is compiler-assigned code that identifies a method name.

SEL someSelector =@selector(someMethodName);

IMP : Used to point to the memory address of the start of a method. You will probably never need to use this.

IMP theImplementation = [selfmethodForSelector:someSelector];

BOOL : Used to specify a boolean type where0 is consideredNO (false) and any non-zero value is consideredYES (true). Anynil object is also considered to beNO so there is no need to perform an equality check withnil (e.g. just writeif (someObject) notif (someObject != nil)).

// BooleanBOOL isBool =YES;// Or NO

nil : Used to specify a null object pointer. When classes are first initialized, all properties of the class are set to0 which means they point tonil.

Objective-C also has a number of other types such asNSInteger,NSUInteger,CGRect,CGFloat,CGSize,CGPoint, etc.

Enumeration types can be defined a number of different ways:

// Specifying a typed enum with a name (recommended way)typedefNS_ENUM(NSInteger, UITableViewCellStyle) {    UITableViewCellStyleDefault,    UITableViewCellStyleValue1,    UITableViewCellStyleValue2,    UITableViewCellStyleSubtitle};// Specify a bitmask with a name (recommended way)typedefNS_OPTIONS(NSUInteger, RPBitMask) {    RPOptionNone      =0,    RPOptionRight     =1 <<0,    RPOptionBottom    =1 <<1,    RPOptionLeft      =1 <<2,    RPOptionTop       =1 <<3};// Other methods:// Untypedenum {    UITableViewCellStyleDefault,    UITableViewCellStyleValue1,    UITableViewCellStyleValue2,    UITableViewCellStyleSubtitle};// Untyped with a nametypedefenum {    UITableViewCellStyleDefault,    UITableViewCellStyleValue1,    UITableViewCellStyleValue2,    UITableViewCellStyleSubtitle} UITableViewCellStyle;

// Set bits (only valid if it makes sense that your status may have many of the bitmask values)RPBitMask status = RPOptionNone;status |= RPOptionBottom;status != RPOptionTop;// Toggle bitstatus ^= RPOptionTop;// Set single bit to zerostatus &= ! RPOptionBottom;// Check if it matches a certain bitif (status & RPOptionTop) {     [selfdoSometing]; }

Sometimes it is necessary to cast anid or different type into a specific class or data type. Examples of this would be casting from afloat to anint or from aUITableViewCell to a subclass such asRPTableViewCell.

Casting non-object data types:

// Format: nonObjectType variableName = (nonObjectType)variableNameToCastFrom;int anInt = (int)anAnonymouslyTypedNonObjectOrDifferentDataType;

Casting object data types:

// Format: ClassNameOrObjectType *variableName = (ClassNameOrObjectType *)variableNameToCastFrom;UIViewController *aViewController = (UIViewController *)anAnonymouslyTypedObjectOrDifferentDataType;

Back to top

Usingconst is usually the better approach sinceconst points to the same memory address for any references to that item in code.#define defines a macro which replaces all references with the actual constant value contents before compilation starts, instead of being a memory pointer to that constant value.

Constants can be defined as:

// Format: type const constantName = value;NSString *constkRPShortDateFormat =@"MM/dd/yyyy";// Format: #define constantName value#definekRPShortDateFormat@"MM/dd/yyyy"

To make the constant available to external classes, you must also add it to the header.h file:

externNSString *constkRPShortDateFormat;

If you know that a constant will only be available within it's containing implementation.m file, specify it as:

staticNSString *constkRPShortDateFormat =@"MM/dd/yyyy";

A static variable declared within a method retains its value between invocations. This can be useful when declaring asingleton or creating custom setters/getters for a property.

Back to top

Back to top

Classes are declared using two files: a header (.h) file and an implementation (.m) file.

The header file should contain (in this order):

• Any needed#import statements or forward@class declarations
• Any protocol declarations
• An@interface declaration specifying which class you're inheriting from
• All publicly accessible variables, properties and methods

The implementation file should contain (in this order):

• Any needed#import statements
• An anonymous category, or class extension, for any private variables or properties
• An@implementation declaration specifying the class
• All public and private methods

Example:

MyClass.h

#import"SomeClass.h"// Used instead of #import to forward declare a class in property return types, etc.@class SomeOtherClass;// Place all global constants at the topexternNSString *constkRPErrorDomain;// Format: YourClassName : ClassThatYouAreInheritingFrom@interfaceMyClass :SomeClass// Public properties first@property (readonly,nonatomic,strong) SomeClass *someProperty;// Then class methods+ (id)someClassMethod;// Then instance methods- (SomeOtherClass *)doWork;@end

MyClass.m

#import"MyClass.h"#import"SomeOtherClass.h"// Declare any constants at the topNSString *constkRPErrorDomain =@"com.myIncredibleApp.errors";staticNSString *constkRPShortDateFormat =@"MM/dd/yyyy";// Class extensions for private variables / properties@interfaceMyClass (){int somePrivateInt;}// Re-declare as a private read-write version of the public read-only property@property (readwrite,nonatomic,strong) SomeClass *someProperty;@end@implementationMyClass// Use #pragma mark - statements to logically organize your code#pragma mark - Class Methods+ (id)someClassMethod{return [[MyClassalloc]init];}#pragma mark - Init & Dealloc methods- (id)init{if (self = [superinit]) {// Initialize any properties or setup code here    }return self;}// Dealloc method should always follow init method- (void)dealloc{// Remove any observers or free any necessary cache, etc.    [superdealloc];}#pragma mark - Instance Methods- (SomeOtherClass *)doWork{// Implement this}@end

When you want to create a new instance of a class, you use the syntax:

MyClass *myClass = [[MyClassalloc]init];

Thealloc class method returns a pointer to a newly allocated block of memory large enough to store an instance of the class. The allocated memory contains zeros except for one instance variable,isa, that all Objective-C objects are required to have. Theisa variable is automatically initialized to point to the class object that allocated the memory and enables the instance to receive messages such asinit that are used to complete initialization.

Back to top

This section needs a lot of love, so please feel free to improve upon it!

The special operatordefined is used in#if and#elif expressions to test whether a certain name is defined as a macro.

defined is useful when you wish to test more than one macro for existence at once. For example,#if defined(__IPHONE_8_0) || defined(__MAC_10_9)would succeed if either of the names__IPHONE_8_0 or__MAC_10_9 is defined as a macro.

Back to top

Also see theliterals section.

The default is@protected, so there is no need to explicitly specify this.

Back to top

Literals are compiler directives which provide a shorthand notation for creating common objects.

NSArray *example = @[@"hi",@"there", @23, @YES ];NSLog(@"item at index 0:%@", example[0]);

NSDictionary *example = @{@"hi" :@"there",@"iOS" :@"people" };NSLog(@"hi%@", example[@"hi"]);

Similar toNSString literals, collection objects made via literal arrays and dictionaries are immutable. Instead, you will have to make a mutable copy after making the immutable dictionary or array. Additionally, you cannot have static initializers like you can withNSString.

Back to top

For methods without a return type, usevoid:

// Does not return anything or take any arguments- (void)someMethod;

+ precedes declarations of class methods:

// Call on a class (e.g. [MyClass someClassMethod]);+ (void)someClassMethod;

- precedes declarations of class instance methods:

// Called on an instance of a class (e.g. [[NSString alloc] init]);- (void)someClassInstanceMethod;

Method arguments are declared after colons: and the method signature should describe the argument type:

// Does something with an NSObject argument- (void)doWorkWithObject:(NSObject *)object;

Argument and return types are declared using type casting syntax:

// Returns an NSString object for the given NSObject arguments- (NSString *)stringFromObject:(NSObject *)object andSomeOtherObject:(NSObject *)otherObject;

Methods are called using bracket syntax:[self someMethod]; or[self sometMethodWithObject:object];

self is a reference to the method's containing class. Theself variable is present in all Objective-C methods and it is one of two hidden arguments passed to code that implements a method, the other being_cmd, which identifies the received message.

At times, it is necessary to call a method in the superclass using[super someMethod];.

Under the hood, methods are implemented via message sending and they are turned into a variation of one of these two C functions:

idobjc_msgSend(id self,SEL op, ...);idobjc_msgSendSuper(struct objc_super *super,SEL op, ...);

If you'd like to test if a class responds to a certain selector before you send it (and possibly crash), you can do so with:

if ([someClassOrInstancerespondsToSelector:@selector(someMethodName)]){// Call the selector or do something here}

This pattern is common when you have a delegate implemented and need to test for methods declared as@optional before calling them on the delegate object.

Back to top

Declaring a property allows you to maintain a reference to an object within a class or to pass objects between classes.

Public properties are declared in the header (.h) file:

@interfaceMyClass :NSObject@property (readonly,nonatomic,strong)NSString *fullName;@end

Private properties are declared in an anonymous category, or class extension, in the implementation (.m) file:

#import"MyClass.h"// Class extension for private variables / properties@interfaceMyClass (){// Instance variableint somePrivateInteger;}// Private properties@property (nonatomic,strong)NSString *firstName;@property (nonatomic,strong)NSString *lastName;@property (readwrite,nonatomic,strong)NSString *fullName;@end@implementationMyClass// Class implementation goes here@end

The LLVM compiler automatically synthesizes all properties so there is no longer a need to explicitly write@synthesize statements for properties anymore. When a property is synthesized, accessors are created which allow you to set and get the value of a property.

Even though you may not see them since they are created at build time, a getter/setter pair can be shown as:

- (BOOL)finished{return _finished;}- (void)setFinished:(BOOL)aValue{    _finished = aValue;}

You can overrride the getter and setter of a property to create customized behavior, or even use this pattern to create transient properties such as:

- (NSString *)fullName{return [NSStringstringWithFormat:@"%@%@",self.firstName,self.lastName];}

Properties are typically backed by an instance variable with a leading underscore, so creating a property calledfirstName would have a backing instance variable with the name_firstName. You should only access that private instance variable if you override the getter/setter or if you need to setup the ivar in the classinit method.

When a property is specified, it is given the syntax:

@property SomeClass *someProperty;// Or@property (xxx) SomeClass *someProperty;

wherexxx can be a combination of:

Properties can be accessed using either bracket syntax or dot notation, with dot notation being cleaner to read.

[selfmyProperty];// Orself.myProperty

Local variables exist only within the scope of a method.

- (void)doWork{NSString *localStringVariable =@"Some local string variable.";   [selfdoSomethingWithString:localStringVariable];}

Back to top

The general rule of thumb: Clarity and brevity are both important, but clarity should never be sacrificed for brevity.

These both usecamelCase where the first letter of the first word is lowercase and the first letter of each additional word is capitalized.

These both useCapitalCase where the first letter of every word is capitalized.

These should use verbs if they perform some action (e.g.performInBackground). You should be able to infer what is happening, what arguments a method takes, or what is being returned just by reading a method signature.

Example:

// Correct- (UITableViewCell *)tableView:(UITableView *)tableView cellForRowAtIndexPath:(NSIndexPath *)indexPath{// Code}// Incorrect (not expressive enough)- (UITableViewCell *)table:(UITableView *)tableView cell:(NSIndexPath *)indexPath{// Code}

When using properties, instance variables are internally created with a preceeding underscore, somyVariableName is created as_myVariableName. However, Objective-C now synthesizes these properties for you so you should never have to access that underscored instance variable directly except in a custom setter.

Instead, instance variables should always be accessed and mutated usingself.

Local variables should not contain underscores.

These should start withk andXXX whereXXX is a prefix, possibly your initials, to avoid naming conflicts. You should not be afraid to be expressive with your constant naming, especially if it's a global constant. UsingkRPNavigationFadeOutAnimationDuration is much better thanfadeOutTiming.

Back to top

Blocks are essentially anonymous functions that are used to pass arbitrary code between methods or to execute code as a callback within a method. Since blocks are implemented as closures, the surrounding state is also captured (which can sometimes lead to retain cycles).

// As a local variablereturnType (^blockName)(parameterTypes) = ^returnType(parameters) {// Block code here};// As a property@property (nonatomic, copy) returnType (^blockName)(parameterTypes);// As a method parameter- (void)someMethodThatTakesABlock:(returnType (^)(parameterTypes))blockName {// Block code here};// As an argument to a method call[someObjectsomeMethodThatTakesABlock: ^returnType (parameters) {// Block code here}];// As a typedeftypedefreturnType (^TypeName)(parameterTypes);TypeName blockName = ^(parameters) {// Block code here};

Since variables within a block are just snapshots of what they were outside of the block scope, you must preceed any variables that you want to mutate within the block with__block, such as:

__blockint someIncrementer =0;[someObjectsomeMethodThatTakesABlock:^{    someIncrementer++;}];

Since blocks strongly capture all variables within the scope of the block, you have to be careful how you setup your blocks code. Here are two examples of retain cycles:

[someObjectsomeMethodThatTakesABlock:^{    [someObjectperformSomeAction];// Will raise a warning}];[selfsomeMethodThatTakesABlock:^{    [selfperformSomeAction];// Will raise a warning}];

In both of these cases, the object which performs the block owns the block, which also owns the object. This creates a loop, or a retain cycle, which means the memory is eventually leaked.

To get around this warning you can either refactor the code to be:

[selfsomeMethodThatTakesABlock:^{    [objectperformSomeAction];// No retain cycle here}];

Or you can use a__weak object:

__weaktypeof(self) weakSelf = self;[selfsomeMethodThatTakesABlock:^{    [weakSelfperformSomeAction];// No retain cycle here}];

Back to top

Objective-C uses all of the same control statements that other languages have:

if (someTestCondition) {// Code to execute if the condition is true}elseif (someOtherTestCondition) {// Code to execute if the other test condition is true}else {// Code to execute if the prior conditions are false}

The shorthand notation for anif-else statement is a ternary operator of the form:someTestCondition ? doIfTrue : doIfFalse;

Example:

- (NSString *)stringForTrueOrFalse:(BOOL)trueOrFalse{return trueOrFalse ?@"True" :@"False";}

There is also another lesser known form:A ?: B; which basically returnsA ifA isYES or non-nil, otherwise it returnsB.

for (int i =0; i < totalCount; i++) {// Code to execute while i < totalCount}

for (Person *person in arrayOfPeople) {// Code to execute each time}

wherearrayOfPeople can be any object that conforms to theNSFastEnumeration protocol.NSArray andNSSet enumerate over their objects,NSDictionary enumerates over keys, andNSManagedObjectModel enumerates over entities.

while (someTextCondition) {// Code to execute while the condition is true}

do {// Code to execute while the condition is true}while (someTestCondition);

Switch statements are often used in place ofif statements if there is a need to test if a certain variable matches the value of another constant or variable. For example, you may want to test if an error code integer you received matches an existing constant value or if it's a new error code.

switch (errorStatusCode){casekRPServerErrorCode:// Code to execute if it matchesbreak;casekRPNetworkErrorCode:casekRPWifiErrorCode:casekRPSystemErrorCode:// Code to execute if it matchesbreak;default:// Code to execute if nothing else matchedbreak;}

Note:switch statements are fallthrough: when control reaches the matchedcase (ordefault block if nothing matches), it continues execution of the next statements in the source code (includingdefault) until thebreak statement or the end ofswitch is reached. This also allows multiple values to match the same point without any special syntax; they are just listed with empty bodies.

• return : Stops execution and returns to the calling function. It can also be used to return a value from a method.
• break : Used to stop execution of a loop.

Newer enumeration methods now have specialBOOL variables (e.g.BOOL *stop) that are used to stop loop execution. Setting that variable toYES within the loop is similar to callingbreak.

Back to top

Fast enumeration was already mentioned in thecontrol statements section, but many collection classes also have their own block-based methods for enumerating over a collection.

Block-based methods perform almost as well as using fast enumeration, but they just provide extra options for enumerating over collections. An example of block-based enumeration over anNSArray would be:

NSArray *people = @[@"Bob",@"Joe",@"Penelope",@"Jane" ];[peopleenumerateObjectsUsingBlock:^(NSString *nameOfPerson,NSUInteger idx,BOOL *stop) {NSLog(@"Person's name is:%@", nameOfPerson);}];

Back to top

There are a few different ways to extend a class in Objective-C, with some approaches being much easier than others.

Inheritance essentially allows you to create concretesubclasses, which typically have specialized behavior, while inheriting all methods and properties from a superclass which are specified as@public or@protected within the superclass' header file.

Looking through any framework or open source project, you can see the use of inheritance to not only get behavior for free, but to also consolidate code and allow it to be easily reused. Examples of this approach can be seen with any of the mutable framework classes such asNSMutableString which is a subclass ofNSString.

In Objective-C, all objects have much of their behavior defined by theNSObject class through the act of class inheritance. Without inheritance, you would have to implement common methods like object or class equality checks on your own and you'd end up with a lot of duplicate code across classes.

// MyClass inherits all behavior from the NSObject class@interfaceMyClass :NSObject// Class implementation@end

Inheritance inherently creates coupling between classes, so be sure to take this into consideration.

Categories are a very useful and easy way of extending classes, especially when you don't have access to the original source code (such as for Cocoa Touch frameworks and classes). A category can be declared for any class and any methods that you declare in a category will be available to all instances of the original class,as well as any subclasses of the original class. At runtime, there's no difference between a method added by a category and one that is implemented by the original class.

Categories are also useful to:

• Declare informal protocols
• Group related methods similar to having multiple classes
• Break up a large class implementation into multiple categories, which helps with incremental compilation
• Easily configure a class differently for different applications

Categories are named with the format:ClassYouAreExtending +DescriptorForWhatYouAreAdding

As an example, let's say that we need to add a new method to theUIImage class (and all subclasses) that allows us to easily resize or crop instances of that class. You'd then need to create a header/implementation file pair namedUIImage+ResizeCrop with the following implementation:

UIImage+ResizeCrop.h

@interfaceUIImage (ResizeCrop)- (UIImage *)croppedImageWithSize:(CGSize)size;- (UIImage *)resizedImageWithSize:(CGSize)size;@end

UIImage+ResizeCrop.m

#import"UIImage+ResizeCrop.h"@implementationUIImage (ResizeCrop)- (UIImage *)croppedImageWithSize:(CGSize)size{// Implementation code here}- (UIImage *)resizedImageWithSize:(CGSize)size{// Implementation code here}

You could then call these methods on any instances ofUIImage or it's subclasses such as:

UIImage *croppedImage = [userPhotocroppedImageWithSize:photoSize];

Unless you have access to the source code for a class at compile time, it is not possible to add instance variables and properties to a class by using a category. Instead, you have to essentially fake this by using a feature of the Objective-C runtime calledassociative references.

For example, let's say that we want to add a public property to theUIScrollView class to store a reference to aUIView object, but we don't have access toUIScrollView's source code. We would have to create a category onUIScrollView and then create a pair of getter/setter methods for this new property to store a reference like this:

UIScrollView+UIViewAdditions.h

#import<UIKit/UIKit.h>@interfaceUIScrollView (UIViewAdditions)@property (nonatomic,strong) UIView *myCustomView;@end

UIScrollView+UIViewAdditions.m

#import"UIScrollView+UIViewAdditions.h"#import<objc/runtime.h>staticchar UIScrollViewMyCustomView;@implementationUIScrollView (UIViewAdditions)@dynamic myCustomView;- (void)setMyCustomView:(UIView *)customView{    [selfwillChangeValueForKey:@"myCustomView"];objc_setAssociatedObject(self, &UIScrollViewMyCustomView, customView,OBJC_ASSOCIATION_RETAIN_NONATOMIC);    [selfdidChangeValueForKey:@"myCustomView"];}- (UIView *)myCustomView {returnobjc_getAssociatedObject(self, &UIScrollViewMyCustomView);}@end

Let's explain a bit about what's happening here:

• We create a static key calledUIScrollViewMyCustomView that we can use to get and set the associated object. Declaring it asstatic ensures that it is unique since it always points to the same memory address.
• Next, we declare the property we are adding as@dynamic which tells the compiler that the getter/setter is not implemented by theUIScrollView class itself.
• Within the setter, we usewillChangeValueForKey followed bydidChangeValueForKey to ensure that we notify any key-value observers of a change to this property.
• Within the setter, we useobjc_setAssociatedObject to store a reference to the object that we really care about,customView under the static key that we created.& is used to denote that it is a pointer to a pointer toUIScrollViewMyCustomView
• Within the getter, we retrieve the object reference usingobjc_getAssociatedObject and a pointer to the static key

We could then use the property just like we would any other property:

UIScrollView *scrollView = [[UIScrollViewalloc]init];scrollView.myCustomView = [[UIViewalloc]init];NSLog(@"Custom view is%@", scrollView.myCustomView);// Custom view is <UIView: 0x8e4dfb0; frame = (0 0; 0 0); layer = <CALayer: 0x8e4e010>>

More info from the docs:

The [objc_setAssociatedObject] function takes four parameters: the source object, a key, the value, and an association policy constant. The key is a void pointer.

The key for each association must be unique. A typical pattern is to use a static variable. The policy specifies whether the associated object is assigned, retained, or copied, and whether the association is be made atomically or non-atomically. This pattern is similar to that of the attributes of a declared property

The possible property declaration attributes are:

OBJC_ASSOCIATION_RETAIN_NONATOMIC, OBJC_ASSOCIATION_ASSIGN, OBJC_ASSOCIATION_COPY_NONATOMIC, OBJC_ASSOCIATION_RETAIN, OBJC_ASSOCIATION_COPY

In the section aboutdeclaring classes, it shows how the private instance variables and properties within a class are actually added to the class by using an anonymous (unnamed) category, also known as a class extension.

// Class extensions for private variables / properties@interfaceMyClass (){int somePrivateInt;}// Re-declare as a private read-write version of the public read-only property@property (readwrite,nonatomic,strong) SomeClass *someProperty;@end

Class extensions are the only way that you can add variables and properties using a category, so you must have access to the source code at compile time in order to use this method.

Categories are very useful when you are working with Core Data models and want to add additional methods to anNSManagedObject subclass without worrying about Xcode writing over the model class each time you migrate a model.

Model classes should be kept to a minimum, containing only the model properties and Core Data generated accessor methods. All others, such as transient methods, should be implemented in a category on the model class.

Great advice fromApple's docs:

Because the methods declared in a category are added to an existing class, you need to be very careful about method names.

If the name of a method declared in a category is the same as a method in the original class, or a method in another category on the same class (or even a superclass), the behavior is undefined as to which method implementation is used at runtime. This is less likely to be an issue if you’re using categories with your own classes, but can cause problems when using categories to add methods to standard Cocoa or Cocoa Touch classes.

Delegation is basically a way to allow one class to react to changes made in another class or to influence the behavior of another class while minimizing coupling between the two.

The most commonly known example of the delegate pattern in iOS is withUITableViewDelegate andUITableViewDataSource. When you tell the compiler that your class conforms to these protocols, you are essentially agreeing to implement certain methods within your class that are required for aUITableView to properly function.

To conform to an existing protocol, import the header file that contains the protocol declaration (not necessary for framework classes). Then, insert the protocol name within< > symbols and separate multiple protocols by a comma. Both options shown below will work just fine, but I prefer to keep them in the header file since it feels cleaner to me.

Option 1: In your.h file:

#import"RPLocationManager.h"@interfaceMyViewController :UIViewController <RPLocationManagerDelegate, UITableViewDelegate, UITableViewDataSource>@end

Option 2: In the.m file:

#import"MyViewController.h"@interfaceMyViewController () <RPLocationManagerDelegate, UITableViewDelegate, UITableViewDataSource>{// Class extension implementation}@end

To create your own protocol for other classes to conform to, follow this syntax:

RPLocationManager.h

#import<SomeFrameworksYouNeed.h>// Declare your protocol and decide which methods are required/optional// for the delegate class to implement@protocolRPLocationManagerDelegate <NSObject>- (void)didAcquireLocation:(CLLocation *)location;- (void)didFailToAcquireLocationWithError:(NSError *)error;@optional- (void)didFindLocationName:(NSString *)locationName;@end@interfaceRPLocationManager :NSObject// Create a weak, anonymous object to store a reference to the delegate class@property (nonatomic,weak)id <RPLocationManagerDelegate>delegate;// Implement any other methods here@end

When we declare the@protocol namedRPLocationManagerDelegate, all methods are defaulted to being@required so it's not necessary to explicitly state this. However, if you want certain methods to be@optional for conforming classes to implement, you must state this.

Additionally, it is necessary to weakly declare an anonymously typed property calleddelegate which also references theRPLocationManagerDelegate protocol.

In the example above,RPLocationManager.h declares some methods that the class which is acting as the delegate must implement. WithinRPLocationManager.m itself, you could implement these a few different ways, but we'll just show two cases: a) required methods; b) optional methods.

Required Methods

- (void)updateLocation{// Perform some work// When ready, notify the delegate method    [self.delegatedidAcquireLocation:locationObject];}

Optional Methods

- (void)reverseGeoCode{// Perform some work// When ready, notify the delegate methodif ([self.delegaterespondsToSelector:@selector(didFindLocationName:)]) {        [self.delegatedidFindLocationName:locationName];    }}

The only difference between@required and@optional methods is that you should always check if the referenceddelegate implemented an optional method before calling it on that class.

To implement a delegate method, just conform to the protocol like was discussed earlier, and then write it as if it was a normal method:

MyViewController.m

- (void)didFindLocationName:(NSString *)locationName{NSLog(@"We found a location with the name%@", locationName);}

Subclassing is essentially the same asinheritance, but you would typically create the subclass to either

• Override a method or property implementation in the superclass; or
• Create specialized behavior for the subclass (e.g.Toyota is a subclass ofCar... it still has tires, an engine, etc, but it has additional custom behavior that uniquely makes it aToyota)

Many design patterns, such ascategories anddelegation, exist so that you don't have to subclass another class. For example, theUITableViewDelegate protocol was created to allow you to provide the implementation of methods liketableView:didSelectRowAtIndexPath: in your own class instead of having to subclassUITableView to override that method.

Other times, classes likeNSManagedObject are designed to be easily subclassed. The general rule of thumb is to subclass another class only if you can satisfy theLiskov substitution principle:

If S is a subtype of T, then objects of type T in a program may be replaced with objects of type S without altering any of the desirable properties of that program.

Let's assume that we want to model cars. All cars have similar behavior and characteristics, so let's put some of that in a superclass calledCar.

Car.h

#import<Foundation/Foundation.h>@interfaceCar :NSObject@property (nonatomic,strong)NSString *make;@property (nonatomic,strong)NSString *model;@property (nonatomic,assign)NSInteger year;- (void)startEngine;- (void)pressGasPedal;- (void)pressBrakePedal;@end

Car.m

#import"Car.h"@implementationCar- (void)startEngine{NSLog(@"Starting the engine.");}- (void)pressGasPedal{NSLog(@"Accelerating...");}- (void)pressBrakePedal{NSLog(@"Decelerating...");}@end

Now when we want to spawn off new car makes and models with unique characteristics, we use theCar superclass as a starting point and then add custom behavior in the subclass.

Toyota.h

#import"Car.h"@interfaceToyota :Car- (void)preventAccident;@end

Toyota.m

#import"Toyota.h"@implementationToyota- (void)startEngine{// Perform custom start sequence, different from the superclassNSLog(@"Starting the engine.");}- (void)preventAccident{    [selfpressBrakePedal];    [selfdeployAirbags];}- (void)deployAirbags{NSLog(@"Deploying the airbags.");}@end

Even thoughpressBrakePedal is declared in theCar class, that method is still accessible in theToyota class due to inheritance.

Often times, classes implement designated class initializers to allow for easy instantiation. If you override the main designated initializer for a class or provide a new designated initializer, it's important to ensure that you also override all other designated initializers so they use your new implementation instead of the superclass version. If you forget to do so and someone calls one of the secondary designated initializers on your subclass, they will get behavior from the superclass instead.

// The new designated initializer for this class- (instancetype)initWithFullName:(NSString *)fullName{if (self = [superinit]) {        _fullName = fullName;        [selfcommonSetup];    }return self;}// Provide a sensible default for other initializers- (instancetype)init{return [selfinitWithFullName:@"Default User"];}

If you'd rather someone not use a default initializer for some rare case, you should throw an exception and provide them with an alternative solution:

- (instancetype)init {        [NSExceptionraise:NSInvalidArgumentExceptionformat:@"%s Using the%@ initializer directly is not supported. Use%@ instead.", __PRETTY_FUNCTION__,NSStringFromSelector(@selector(init)),NSStringFromSelector(@selector(initWithFrame:))];returnnil;}

If you're subclassing another class to override a method within that class, you must be a little cautious. If you want to maintain the same behavior as the superclass, but just modify it slightly, you can callsuper within the override like this:

- (void)myMethod{    [supermyMethod];// Provide your additional custom behavior here}

If you don't want any of the superclass's behavior for the overridden method, simply leave out that call tosuper, but be careful that there aren't any memory or object lifecycle consequences for doing so.

Additionally, if the superclass has primitive methods upon which other derived methods are implemented, you must ensure that you override all necessary primitive methods necessary for the derived methods to work properly.

Certain classes don't lend themselves well to being easily subclassed and therefore subclassing is discouraged in those cases. An example of this is when trying to subclass a class cluster such asNSString orNSNumber. Class clusters have quite a few private classes within them so it's difficult to ensure that you have overridden all of the primitive methods and designated initializers within the class cluster properly.

As is often the case, clarity is better than cleverness. As a general rule, it's typically better to work around a bug in a method implementation than it is to replace the method by using method swizzling. The reason being that other people using your code might not realize that you replaced the method implementation and then they are stuck wondering why a method isn't responding with the default behavior.

For this reason, we don't discuss swizzling here but you are welcome toread up on it here.

Back to top

Errors are typically handled three different ways: assertions, exceptions, and recoverable errors. In the case of assertions and exceptions, they should only be used in rare cases since crashing your app is obviously not a great user experience.

Assertions are used when you want to ensure that a value is what it is supposed to be. If it's not the correct value, you force exit the app.

NSAssert(someCondition,@"The condition was false, so we are exiting the app.");

Important: Do not call functions with side effects in the condition parameter of this macro. The condition parameter is not evaluated when assertions are disabled, so if you call functions with side effects, those functions may never get called when you build the project in a non-debug configuration.

Exceptions are only used for programming or unexpected runtime errors. Examples: attempting to access the 6th element of an array with 5 elements (out-of-bounds access), attempting to mutate immutable objects, or sending an invalid message to an object. You usually take care of these sorts of errors with exceptions when an application is being created rather than at runtime.

An example of this might be if you have a library which requires an API key to use.

// Check for an empty API key- (void)checkForAPIKey{if (!self.apiKey || !self.apiKey.length) {        [NSExceptionraise:@"Forecastr"format:@"Your Forecast.io API key must be populated before you can access the API.",nil];    }}

If you're worried that a block of code is going to throw an exception, you can wrap it in a try-catch block but keep in mind that this has slight performance implications

@try {// The code to try here}@catch (NSException *exception) {// Handle the caught exception}@finally {// Execute code here that would run after both the @try and @catch blocks}

Many times, methods will return anNSError object in a failure block or as a pointer to a pointer (in the case ofNSFileManager). These are typically returned for recoverable errors and offer a much more pleasant user experience since they can clue the user into what just went wrong.

[forecastrgetForecastForLocation:locationsuccess:^(id JSON) {NSLog(@"JSON response was:%@", JSON);}failure:^(NSError *error,id response) {NSLog(@"Error while retrieving forecast:%@", error.localizedDescription);}];

It's also possible to create your ownNSError objects to return in methods.

// Error domain & enumsNSString *constkFCErrorDomain =@"com.forecastr.errors";typedefNS_ENUM(NSInteger, ForecastErrorType) {kFCCachedItemNotFound,kFCCacheNotEnabled};@implementationForecastr- (void)checkForecastCacheForURLString:(NSString *)urlStringsuccess:(void (^)(id cachedForecast))successfailure:(void (^)(NSError *error))failure{// Check cache for a forecastid cachedItem = [forecastCacheobjectForKey:urlString];if (cachedItem) {success(cachedItem);    }else {// Return an error since it wasn't foundfailure([NSErrorerrorWithDomain:kFCErrorDomaincode:kFCCachedItemNotFounduserInfo:nil]);    }}@end

Back to top

We have already discussed many ways of passing information between classes, such as through methods or delegates, but we'll discuss a few more here and also give one more example ofdelegation.

A very common way to pass data from one view controller to another is to use a delegate method. An example of this would be if you had a modal view with a table that showed over top of your view controller and you needed to know which table cell the user pressed.

AddPersonViewController.h (the modal view)

#import<UIKit/UIKit.h>#import"Person.h"@protocolAddPersonTableViewControllerDelegate <NSObject>- (void)didSelectPerson:(Person *)person;@end@interfaceAddPersonTableViewController :UITableViewController@property (nonatomic,weak)id <AddPersonTableViewControllerDelegate>delegate;@end

AddPersonViewController.m

// Other implementation details left out- (void)tableView:(UITableView *)tableView didSelectRowAtIndexPath:(NSIndexPath *)indexPath{    Person *person = [peopleobjectAtIndex:indexPath.row];    [self.delegatedidSelectPerson:person];}

GroupViewController.m (the normal view)

// Other implementation details left out, such as showing the modal view// and setting the delegate to self#pragma mark - AddPersonTableViewControllerDelegate- (void)didSelectPerson:(Person *)person{    [selfdismissViewControllerAnimated:YEScompletion:nil];NSLog(@"Selected person:%@", person.fullName);}

We left out a few implementation details, such as conforming to theAddPersonTableViewControllerDelegate, but you are welcome to read thedelegation section for those.

Also, notice that we dismiss the modal view controller (AddPersonViewController) in the same class that originally showed it. This is the recommended approach by Apple.

Notifications are broadcast messages that are used to decouple classes and establish anonymous communication between objects at runtime. Notifications may be posted by any number of objects and received by any number of objects thus enabling one-to-many and many-to-many relationships between objects.

Note: Notifications are sent synchronously so if your observer method takes a long time to return, you are essentially prevently the notification from being sent to other observing objects.

You can register to be notified when a certain event has happened, including system notifications, such as aUITextField which has begun editing.

[[NSNotificationCenterdefaultCenter]addObserver:selfselector:@selector(textFieldDidBeginEditing:)name:UITextFieldTextDidBeginEditingNotificationobject:self];

When theUITextFieldTextDidBeginEditingNotification notification is broadcast by the OS framework, thetextFieldDidBeginEditing: will be called byNSNotificationCenter and an object will be sent along with the notification that could contain data.

A possible implementation of thetextFieldDidBeginEditing: method could be:

#pragma mark - Text Field Observers- (void)textFieldDidBeginEditing:(NSNotification *)notification{// Optional check to make sure the method was called from the notificationif ([notification.nameisEqualToString:UITextFieldTextDidBeginEditingNotification])    {// Do something    }}

It's important to remove yourself as an observer when the class is deallocated, otherwiseNSNotificationCenter will attempt to call a method on a deallocated class and a crash will ensue.

- (void)dealloc{    [[NSNotificationCenterdefaultCenter]removeObserver:self];}

You can also create and post your own notifications. It's best practice to keep notification names in aconstants file so that you don't accidentally misspell one of the notification names and sit there trying to figure out why the notification wasn't sent/received.

Naming notifications:

Notifications are identified byNSString objects whose names are composed in this way:

[Name of associated class] + [Did | Will] + [UniquePartOfName] + Notification

Declare a string constant, using the notification name as the string's value:

// Remember to put the extern of this in the header fileNSString *constkRPAppDidResumeFromBackgroundNotification =@"RPAppDidResumeFromBackgroundNotification";

Post notification:

[[NSNotificationCenterdefaultCenter]postNotificationName:kRPAppDidResumeFromBackgroundNotificationobject:self];

When you are preparing to display a new view controller, you can assign data to one of it's properties prior to display:

MyViewController *myVC = [[MyViewControlleralloc]init];myVC.someProperty = self.someProperty;[selfpresentViewController:myVCanimated:YEScompletion:nil];

When you are transitioning from one view controller to another in a storyboard, there is an easy way to pass data between the two by implementing theprepareForSegue:sender: method:

#pragma mark - Segue Handler- (void)prepareForSegue:(UIStoryboardSegue *)segue sender:(id)sender{if ([segue.identifierisEqualToString:@"showLocationSearch"] {        [segue.destinationViewControllersetDelegate:self];    }elseif ([segue.identifierisEqualToString:@"showDetailView"]) {        DetailViewController *detailView = segue.destinationViewController;        detailView.location = self.location;    }}

User defaults are basically a way of storing simple preference values which can be saved and restored across app launches. It is not meant to be used as a data storage layer, like Core Data or sqlite.

NSUserDefaults *userDefaults = [NSUserDefaultsstandardUserDefaults];[userDefaultssetValue:@"Some value"forKey:@"RPSomeUserPreference"];

NSUserDefaults *userDefaults = [NSUserDefaultsstandardUserDefaults];id someValue = [userDefaultsvalueForKey:@"RPSomeUserPreference"];

There are also other convenience methods onNSUserDefaults instances such asboolForKey:,stringForKey:, etc.

Back to top

Singleton's are a special kind of class where only one instance of the class exists for the current process. They are a convenient way to share data between different parts of an app without creating global variables or having to pass the data around manually, but they should be used sparingly since they often create tighter coupling between classes.

To turn a class into a singleton, you place the following method into the implementation (.m) file, where the method name is prefixed withshared plus another word which best describes your class. For example, if the class is a network or location manager, you would name the methodsharedManager instead ofsharedInstance.

+ (instancetype)sharedInstance{staticid sharedInstance =nil;staticdispatch_once_t onceToken;dispatch_once(&onceToken, ^{      sharedInstance = [[selfalloc]init];   });return sharedInstance;}

The use ofdispatch_once ensures that this method is only ever executed once, even if it's called multiple times across many classes or different threads.

If the above code were placed withinMyClass, then you would get a reference to that singleton class in another class with the following code:

MyClass *myClass = [MyClasssharedInstance];[myClassdoSomething];NSLog(@"Property value is%@", myClass.someProperty);

Back to top