逛奔的蜗牛

To get started as quickly as possible, let's look at some simple examples. The basic syntax for calling a method on an object is this:

Methods can return a value:

You can call methods on classes too, which is how you create objects. In the example below, we call the string method on the NSString class, which returns a new NSString object:

The id type means that the myObject variable can refer to any kind of object, so the actual class and the methods it implements aren't known when you compile the app. 

In this example, it's obvious the object type will be an NSString, so we can change the type:

This is now an NSString variable, so the compiler will warn us if we try to use a method on this object which NSString doesn't support. 

Notice that there's a asterisk to the right of the object type. All Objective-C object variables are pointers types. The id type is predefined as a pointer type, so there's no need to add the asterisk.

In many languages, nested method or function calls look like this:

The result of function2 is passed as input to function1. In Objective-C, nested messages look like this:

Avoid nested nesting more than two message calls on a single line, as it easily gets unreadable.

Some methods take multiple input values. In Objective-C, a method name can be split up into several segments. In the header, a multi-input method looks like this:

You call the method like this:

These are not just named arguments. The method name is actuallywriteToFile:atomically: in the runtime system.

All instance variables are private in Objective-C by default, so you should use accessors to get and set values in most cases. There are two syntaxes. This is the traditional 1.x syntax:

The code on the second line is not reading the instance variable directly. It's actually calling a method named caption. In most cases, you don't add the "get" prefix to getters in Objective-C. 

Whenever you see code inside square brackets, you are sending a message to an object or a class.

The dot syntax for getters and setters is new in Objective-C 2.0, which is part of Mac OS X 10.5:

You can use either style, but choose only one for each project. The dot syntax should only be used setters and getters, not for general purpose methods.

There are two main ways to create an object. The first is the one you saw before:

This is the more convenient automatic style. In this case, you are creating an autoreleased object, which we'll look at in more detail later. In many cases, though, you need to create an object using the manual style:

This is a nested method call. The first is the alloc method called on NSString itself. This is a relatively low-level call which reserves memory and instantiates an object. 

The second piece is a call to init on the new object. The init implementation usually does basic setup, such as creating instance variables. The details of that are unknown to you as a client of the class. 

In some cases, you may use a different version of init which takes input:

If you're writing an application for Mac OS X, you have the option to enable garbage collection. In general, this means that you don't have to think about memory management until you get to more complex cases. 

However, you may not always be working with an environment that supports garbage collection. In that case, you need to know a few basic concepts. 

If you create an object using the manual alloc style, you need to releasethe object later. You should not manually release an autoreleased object because your application will crash if you do. 

Here are two examples:

For this tutorial, you can assume that an automatic object will go away at the end of the current function. 

There's more to learn about memory management, but it will make more sense after we look at a few more concepts.

The Objective-C syntax for creating a class is very simple. It typically comes in two parts. 

The class interface is usually stored in the ClassName.h file, and defines instance variables and public methods. 

The implementation is in the ClassName.m file and contains the actual code for these methods. It also often defines private methods that aren't available to clients of the class. 

Here's what an interface file looks like. The class is called Photo, so the file is named Photo.h:

First, we import Cocoa.h, to pull in all of the basic classes for a Cocoa app. The #import directive automatically guards against including a single file multiple times. 

The @interface says that this is a declaration of the class Photo. The colon specifies the superclass, which is NSObject. 

Inside the curly brackets, there are two instance variables: caption andphotographer. Both are NSStrings, but they could be any object type, including id. 

Finally, the @end symbol ends the class declaration.

Let's add some getters for the instance variables:

Remember, Objective-C methods typically leave out the "get" prefix. A single dash before a method name means it's a instance method. A plus before a method name means it's a class method. 

By default, the compiler assumes a method returns an id object, and that all input values are id. The above code is technically correct, but it's unusual. Let's add specific types for the return values:

Now let's add setters:

Setters don't need to return a value, so we just specify them as void.

Let's create an implementation, starting with the getters:

This part of the code starts with @implementation and the class name, and has @end, just like the interface. All methods must appear between these two statements. 

The getters should look very familiar if you've ever written code, so let's move on to the setters, which need a bit more explanation:

Each setter deals with two variables. The first is a reference to the existing object, and the second is the new input object. In a garbage collected environment, we could just set the new value directly:

But if you can't use garbage collection, you need to release the old object, and retain the new one. 

There are actually two ways to free a reference to an object: release andautorelease. The standard release will remove the reference immediately. The autorelease method will release it sometime soon, but it will definitely stay around until the end of the current function (unless you add custom code to specifically change this). 

The autorelease method is safer inside a setter because the variables for the new and old values could point to the same object. You wouldn't want to immediately release an object which you're about to retain. 

This may seem confusing right now, but it will make more sense as you progress. You don't need to understand it all yet.

We can create an init method to set inital values for our instance variables:

This is fairly self-explanatory, though the second line may look a bit unusual. This is a single equals sign, which assigns the result of [super init] to self. 

This essentially just asks the superclass to do its own initialization. The ifstatement is verifying that the initialization was successful before trying to set default values.

The dealloc method is called on an object when it is being removed from memory. This is usually the best time to release references to all of your child instance variables:

On the first two lines, we just send release to each of the instance variables. We don't need to use autorelease here, and the standard release is a bit faster. 

The last line is very important. We have to send the message
[super dealloc] to ask the superclass to do its cleanup. If we don't do this, the object will not be removed, which is a memory leak. 

The dealloc method is not called on objects if garbage collection is enabled. Instead, you implement the finalize method.

Objective-C's memory management system is called reference counting. All you have to do is keep track of your references, and the runtime does the actual freeing of memory. 

In simplest terms, you alloc an object, maybe retain it at some point, then send one release for each alloc/retain you sent. So if you used alloc once and then retain once, you need to release twice.

That's the theory of reference counting. But in practice, there are usually only two reasons to create an object: 

1. To keep it as an instance variable
2. To use temporarily for single use inside a function 

In most cases, the setter for an instance variable should just autoreleasethe old object, and retain the new one. You then just make sure to release it in dealloc as well. 

So the only real work is managing local references inside a function. And there's only one rule: if you create an object with alloc or copy, send it arelease or autorelease message at the end of the function. If you create an object any other way, do nothing. 

Here's the first case, managing an instance variable:

Here's the other case, local references. We only need to release the object created with alloc:

And here's a combo: using a local reference to set an object as an instance variable:

Notice how the rules for managing local references are exactly the same, regardless of whether you're setting them as instance variables or not. You don't need to think about how the setters are implemented. 

If you understand this, you understand 90% of what you will ever need to know about Objective-C memory management.

Logging messages to the console in Objective-C is very simple. In fact, theNSLog() function is nearly identical to the C printf() function, except there's an additional %@ token for objects.

You can log an object to the console. The NSLog function calls thedescription method on the object, and prints the NSString which is returned. You can override the description method in your class to return a custom string.

When we wrote the accessor methods for caption and author earlier, you might have noticed that the code is straightforward, and could probably be generalized. 

Properties are a feature in Objective-C that allow us to automatically generate accessors, and also have some other side benefits. Let's convert the Photo class to use properties. 

Here's what it looked like before:

Here's what it looks like once converted to properties:

The @property is an Objective-C directive which declares the property. The "retain" in the parenthesis specifies that the setter should retain the input value, and the rest of the line simply specifies the type and the name of the property. 

Now let's take a look at the implementation of the class:

The @synthesize directive automatically generates the setters and getters for us, so all we have to implement for this class is the dealloc method. 

Accessors will only be generated if they don't already exist, so feel free to specify @synthesize for a property, then implement your custom getter or setter if you want. The compiler will fill in whichever method is missing. 

There are many other options for the property declarations, but those are outside of the scope of this tutorial.

In Objective-C, the nil object is the functional equivalent to the NULLpointer in many other languages. The difference is that you can call methods on nil without crashing or throwing an exception. 

This technique used by the frameworks in a number of different ways, but the main thing it means to you right now that you usually don't need to check for nil before calling a method on an object. If you call a method on nil that returns an object, you will get nil as a return value. 

We can also use this to improve our dealloc method slightly:

This works because when we set nil as an instance variable, the setter just retains nil (which does nothing) and releases the old value. This approach is often better for dealloc because there's no chance of the variable pointing at random data where an object used to be. 

Note that we're using the self.<var> syntax here, which means we're using the setter and picking up the memory management for free. If we just directly set the value like this, there would be a memory leak:

Categories are one of the most useful features of Objective-C. Essentially, a category allows you to add methods to an existing class without subclassing it or needing to know any of the details of how it's implemented. 

This is particularly useful because you can add methods to built-in objects. If you want to add a method to all instances of NSString in your application, you just add a category. There's no need to get everything to use a custom subclass. 

For example, if I wanted to add a method to NSString to determine if the contents is a URL, it would look like this:

This is very similar to a class declaration. The differences are that there is no super class listed, and there's a name for the category in parenthesis. The name can be whatever you want, though it should communicate what the methods inside do. 

Here's the implementation. Keep in mind this is not a good implementation of URL detection. We're just trying to get the concept of categories across:

Now you can use this method on any NSString. The following code will print "string1 is a URL" in the console:

Unlike subclasses, categories can't add instance variables. You can, however, use categories to override existing methods in classes, but you should do so very carefully. 

Remember, when you make changes to a class using a category, it affects all instances of that class throughout the application.