Note: What follows is another post in an on-going series for developers who are interested in learning to write applications for the iPhone. The entire series can be found here: iPhone Developer .

This post introduces messaging within Objective-C. Messaging is the terminology for invoking methods on an object. The format for a message expression is as follows (the brackets are required):

[object method]

or in Objective-C parlance

[receiver message]

Here’s a simple example:

  // Create an instance of SomeClass object
  // We'll cover this later...
  SomeClass *ptr = [[SomeClass alloc] init];
 
  // Send the message 'printInstanceVars' to the 'ptr' receiver
  [ptr printInstanceVars];

If we want to pass an argument as part of the message (that is, pass a parameter to the method), we send a message that looks like this:

[receiver message:argument]

For example, assume setStr and setX are two methods in the SomeClass object and ptr is a pointer an instance of SomeClass . Here is how we might pass arguments to each method.

 [ptr setStr:@"Testing"];
  [ptr setX:2008];

Side note: The @ symbol at the front of @”Testing” string is convenience method that converts the given string to an NSString object, which in the case of the setStr method, is the required type for the parameter.

Nest Messages
We can also nest messages, as shown below. In this example, we first pass a message to the NSDate object. This is a class object with a factory method ‘date’ for creating a new NSDate object. If that makes no sense, don’t worry about it. Think of it as nothing more than creating a new NSDate object that holds the current date and time. The return value of inner message (the NSDate object) is the argument for setDate message. What we’ve done here is to create a new date object and pass it as an argument to the setDate message of the receiver (ptr ). Sounds more confusing than it is.

[ptr setDate:[NSDate date]];

Multiple Arguments
Taking this another step further, we can pass multiple arguments along with our message. The message below takes three arguments, a string, date and integer. Sometimes it easier to read the method aloud to get this jist of what’s up. In this case, “pass a message to the ptr receiver that sets a string, and a date and an integer.”

  [ptr setStr:@"A new test..." andDate:[NSDate date] andInteger:99];

Here is how we define the message in the interface file:

-(void) setStr:(NSString *)str andDate:(NSDate *)date andInteger:(int)x;

And here is the implementation of the setStr() method which accepts three arguments:

-(void) setStr:(NSString *)strInput andDate:(NSDate *)dateInput
    andInteger:(int)xInput
{
  [self setStr:strInput];
  [self setDate:dateInput];
  [self setX:xInput];
}

Let me point out a few things about this method. First, notice how in the definition I used the names str , date and x . However, in the implementation, I used the names strInput , dateInput and xInput . The first thing to understand is that you can do this. The reason for doing so, is that in the class where setStr() is defined, I have instance variables with the names str , date and x . If I didn’t change the names in the actual implementation, I’d get a compiler warning that a local declaration hides an instance variable.

In this code example, the simplest thing to do is to change the variable names in the method to make it obvious what variables are referring to what. This may not be the best naming scheme for your applications, however, I wanted to point out the flexibility as to how you define and implement methods and their arguments. Obviously, do what makes the most sense for your specific needs and at the same time, results in the most readable code.

Also, note that this version of setStr() does nothing more than send a message as follows: the receiver is current object instance (self), the message is a setter method and the argument to each message is the appropriate parameter passed in to setStr() .

Using the format/syntax of Objective-C takes some getting used to. However, once you do, you’ll find that it’s easy to read code that has had some thought put into the names of message and the parameters.

Variable Number of Arguments
The last topic is creating messages with a variable number of arguments. Let’s start with the interface definition of a method that accepts a variable number of arguments:

-(void) printInstanceVars:(id)input, ...;

In this example, I declare on parameter that is of type ‘id’, which can represent any object. The implementation of the method looks as follows:

-(void) printInstanceVars:(id)input, ...
{
  id currentObject;
  va_list argList;
  int objectCount = 1;
 
  if (input)
  {
    NSLog(@"\n Object #%d is: %@\n", objectCount++, input);
 
    va_start(argList, input);
    while (currentObject = va_arg(argList, id))
      NSLog(@"\n Object #%d is: %@\n", objectCount++, currentObject);
    va_end(argList);
  }
}

This method will print out each argument using NSLog(). The assumption is made that the last argument will be nil. Here is how you might call the method, passing in two objects. What’s happening here is that I am sending getter messages, if you will, to the ‘ptr’ receiver. Each of these getters returns an object. Notice the nil value as the last parameter.

  [ptr printInstanceVars:[ptr str], [ptr date], nil];

Now, this example is a little contrived in that there is probably little value in passing in instance variables from an object, as I’ve done here. However, introducing more classes at this point may confuse more than help. If nothing else, hopefully you get the jist of how this works.

I haven’t come upon the case where this is necessary in an Objective-C application, however, when programming in C, this was quite handy when reading command line parameters. So,if you ever find the opportunity, now you’ll know how. And if you are into trivia, variable argument methods in Objective-C are referred to as variadic methods, go figure.

Many of the things that I’ve covered here are much better understood by looking at a real example. I’ve attached the Xcode project that I created to learn this stuff. You best bet is to download, open the project and mess around with the code.

Here is the main method of the example, so you can get an idea of how I tested each of the above examples.

int main(int argc, const char * argv[])
{
  NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init];
  SomeClass *ptr = [[SomeClass alloc] init];
 
  // Simple message
  [ptr printInstanceVars];
 
  // Passing a single argument
  [ptr setStr:@"Testing"];
  [ptr setX:2008];
  [ptr printInstanceVars];
 
  // Nesting of messages
  [ptr setDate:[NSDate date]];
  [ptr printInstanceVars];
 
  // Passing multiple arguments
  [ptr setString:@"A new test..." andDate:[NSDate date] andInteger:99];
  [ptr printInstanceVars];
 
  // Passing variable number of argument
  [ptr printInstanceVars:[ptr str], [ptr date], nil]; 
 
  // This won't work...
//    [ptr printInstanceVars:[ptr str], [ptr date], [ptr x], nil];
  [ptr printInstanceVars:[ptr str], [ptr date], [NSNumber numberWithInt:[ptr x]], nil];
 
  [ptr release];
  [pool drain];
  return 0;
}

Download the Xcode Project
Download the Xcode project and take some time to tinker.

Once you’ve got a good handle on the overall application, look at line 27 above. Make any sense? Here’s the deal: the variable length method is expecting objects to be passed in. The code in line 26 will generate a runtime error given the message [ptr x] returns an integer (not an object). The way around this is to create an NSNumber object by sending a message to the NSNumber receiver, with the message ‘numberWithInt’ passing in as the parameter the integer returned from the getter message sent to the ‘ptr’ receiver. If all that makes sense, you’re well on your way :)

For completeness, the entire code listing and a screenshot are shown below:

// ===========================
// = SomeClass.h =
// ===========================
#import <Foundation/Foundation.h>
 
@interface SomeClass : NSObject
{
  NSString *str;
  NSDate *date;
  int x;
} 
 
// Getters
-(int) x;
-(NSString *) str;
-(NSDate *) date;
 
// Setters
-(void) setStr:(NSString *)input;
-(void) setDate:(NSDate *)input;
-(void) setX:(int)input;
-(void) setStr:(NSString *)str andDate:(NSDate *)date andInteger:(int)x;
 
// Other
-(void) printInstanceVars;
-(void) printInstanceVars:(id)input, ...;
-(void) dealloc; 
 
@end

#import "SomeClass.h"
#import <stdio.h>
 
// ================================
// = SomeClass.m =
// ================================
 
@implementation SomeClass
// =================
// = Getter methods =
// =================
-(int) x
{
  return x;
}
 
-(NSString *) str
{
  return str;
}
 
-(NSDate *) date
{
  return date;
}
 
// =================
// = Setter Methods =
// =================
-(void) setStr:(NSString *)foo
{
  [foo retain];
  [str release];
  str = foo;
}
 
-(void) setDate:(NSDate *)input
{
  [input retain];
  [date release];
  date = input;
}
 
-(void) setX:(int)input
{
  x = input;
}
 
-(void) setStr:(NSString *)strInput andDate:(NSDate *)dateInput andInteger:(int)xInput
{
  [self setStr:strInput];
  [self setDate:dateInput];
  [self setX:xInput];
}
 
// ================================
// = Print the instance vars =
// ================================
-(void) printInstanceVars
{
  // Use the getter method of the ’self’ object to print object instance variables
  //  NSLog(@"\n x: %d\n str: %@\n date: %@\n", [self x], [self str], [self date]);
 
  // The class can directly access the instance variables (versus calling message as above)
  NSLog(@"\n x: %d\n str: %@\n date: %@\n", x, str, date);
}
 
-(void) printInstanceVars:(id)input, ...
{
  id currentObject;
  va_list argList;
  int objectCount = 1;
 
  if (input)
  {
    NSLog(@"\n Object #%d is: %@\n", objectCount++, input);
 
    va_start(argList, input);
    while (currentObject = va_arg(argList, id))
      NSLog(@"\n Object #%d is: %@\n", objectCount++, currentObject);
    va_end(argList);
  }
}
 
// ====================================
// = Dealloc all object instance vars =
// ====================================
 
-(void) dealloc
{
  // No release needed of the integer instance variable ‘x’
  [str release];
  [date release];
  [super dealloc];
}
@end

// ===========================
// = Messaging.m =
// ===========================
#import <Foundation/Foundation.h>
#import "SomeClass.h"
 
int main(int argc, const char * argv[])
{
  NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init];
  SomeClass *ptr = [[SomeClass alloc] init];
 
  // Simple message
  [ptr printInstanceVars];
 
  // Passing a single argument
  [ptr setStr:@"Testing"];
  [ptr setX:2008];
  [ptr printInstanceVars];
 
  // Nesting of messages
  [ptr setDate:[NSDate date]];
  [ptr printInstanceVars];
 
  // Passing multiple arguments
  [ptr setStr:@"A new test..." andDate:[NSDate date] andInteger:99];
  [ptr printInstanceVars];
 
  // Passing variable number of argument
  [ptr printInstanceVars:[ptr str], [ptr date], nil]; 
 
  // This won't work...
  //    [ptr printInstanceVars:[ptr str], [ptr date], [ptr x], nil];
  [ptr printInstanceVars:[ptr str], [ptr date], [NSNumber numberWithInt:[ptr x]], nil];
 
  [ptr release];
  [pool drain];
  return 0;
}

Note: This post is the start of an on-going series for developers who are interested in learning to write applications for the iPhone. The entire series can be found here: iPhone Developer .

One of the first topics to cover when learning to develop native iPhone applications is how to code in Objective-C. Apple offers the Objective C Reference , a good resource, however, the best way to learn is by writing code. I took to Xcode to write a few simple examples, you’ll find the code below. At the end of this post I also include a link to download the Xcode project I was working with.

There are two aspects to a class, the interface and the implementation, both of which I recommend you store in separate files (although this is not a requirement).

The interface looks as follows:

@interface NameOfClass : NameOfSuperclass
{
  instance variables here...
}
class methods
instance methods
@end

The interface for my example:

// ===========================
// = Interface for SomeClass =
// ===========================
 
@interface SomeClass : NSObject
{
  NSString *str;
  NSDate *date;
  int x;
} 
 
// Getters
-(int) x;
-(NSString *) str;
-(NSDate *) date;
 
// Setters
-(void )setX:(int) input;
-(void) setStr:(NSString *)input;
-(void) setDate:(NSDate *)input;
 
// Other
-(void) printInstanceVars;
-(void) dealloc;
@end

A good coding practice is to save the implementation definition in a file with a name that matches the class name, with an extension of .h (for exampe: SomeClass.h).

This class is inherited from NSObject, the uber object. The class has three instance variables, two that point to other objects, one that references an integer variable. Take note of the getter methods: in Objective-C there is typically no ‘get’ in the front of the method name (in Java this might look like getX or getStr). Second, it should be obvious, that an instance variable can have the same name as a method, as it generally does with a getter. The ‘-’ in the front of the definition, signifies that the method is an instance method. We use a ‘+’ to define a class method (more on class methods in a future post).

One important thing to point out is the format used when declaring methods. For example, setStr() is defined as -(void)setStr: (NSString *) input ; This is translated to, the method setStr is an instance method (given the ‘-’) that returns a void type. The method takes one argument, that is a pointer to an NSString object, the name assigned to the parameter is ‘input’. The reason for the name will become more apparent when you see the implementation of the method below.

The format for the implementation of a class looks as follows:

@implementation NameOfClass : NameOfSuperclass
{
  instance variables here...
}
class methods
instance methods
@end

Here is how the implementation for the above class looks:

#import "SomeClass.h"
#import <stdio.h>
 
// ================================
// = Implementation for SomeClass =
// ================================
 
@implementation SomeClass
// =================
// = Getter methods =
// =================
- (int) x
{
  return x;
}
 
- (NSString *) str
{
  return str;
}
 
- (NSDate *) date
{
  return date;
}
 
// =================
// = Setter Methods =
// =================
- (void) setX:(int)input
{
  x = input;
}
 
- (void) setStr:(NSString *)input
{
  [input retain];
  [str release];
  str = input;
}
 
- (void) setDate:(NSDate *)input
{
  [input retain];
  [date release];
  date = input;
}
 
// ================================
// = Print the instance vars =
// ================================
 
-(void) printInstanceVars
{
  // Use the getter method of the 'self' object to print object instance variables
//  NSLog(@"\n x: %d\n str: %@\n date: %@\n", [self x], [self str], [self date]);
 
  // The class can directly access the instance variables (versus calling message as above)
  NSLog(@"\n x: %d\n str: %@\n date: %@\n", x, str, date);
}
 
// ====================================
// = Dealloc all object instance vars =
// ====================================
 
-(void) dealloc
{
  // No release needed of the integer instance variable 'x'
  [str release];
  [date release];
  [super dealloc];
}
@end

Other than learning the syntax of Objective-C, if you are familiar with OO development, most of this should be pretty clear.

A couple of things to point out:

• The preferred file name for the implementation is the class name with a .m extension, in this example: SomeClass.m
• Notice how this file imports “SomeClass.h” to read the class definition. If you are familiar with C, this is analgous to the #include directive. The benefit of #import is that the compiler will do the work for you to verify that the include file is only read once. If you’ve done any amount of coding in C, you’ll appreciate this convenience, if not, you won’t understand how nice a feature this is.
• Within an instance method, all instance variables are within scope. For example, notice how the getter and setter methods refer to the instance variables.
• Notice in printInstanceVars() method that there are two means to access the instance variables. You can use the ‘self’ object an send a message to the getter method (more on objects and messages in the next post), or you can directly access the instance variables.
• If instance variables are pointers to objects, as are ‘str’ and ‘date’, it’s your responsibility as the developer to free the memory for those objects. The dealloc method is where you do this work. More on that to come…

To complete the example, the code that follows declares an instance of the SomeClass object, and uses the setter/getter methods to print the instance variables to the console.

#import <Foundation/Foundation.h>
#import "SomeClass.h"
 
int main(int argc, const char * argv[])
{
  NSAutoreleasePool * pool = [[NSAutoreleasePool alloc] init];
  SomeClass *ptr = [[SomeClass alloc] init];
 
  [ptr setX:99];
  [ptr printInstanceVars];
 
  [ptr setStr:@"Testing"];
  [ptr printInstanceVars];
 
  [ptr setDate:[NSDate date]];
  [ptr printInstanceVars];
 
  [ptr release];
  [pool drain];
  return 0;
 
}

A few comments on the above code:

• Notice this file imports the SomeClass.h interface file.
• Like working with C, main() is the function that gets everything started.
• ‘ptr’ is a reference to an object of the SomeClass class.
• Calling instance methods of an object follows this form: [object message:parameters]

A screenshot of the output from within Xcode of this example is below:

I recommend you download the Xcode project and give it a go.

Let’s go with that for today. In the next post I’ll talk further about this simple example, including instantiation of classes, sending messages to methods and freeing memory of the instance variables.

Having spent the past 8 years working with mobile application development (beginning with the book Core J2ME that I wrote in 2000), it’s odd that its taken me so long to dive into developing for the iPhone. The good news is, I’m now fully engaged and plan to spend a fair amount of time writing about this new endeavor…

This is the first in what I anticipate to be a long series of posts as I ramp up on all things iPhone. The intention is to share how I am going about learning to develop iPhone applications, in the hopes it can help you.

To begin, my plan is to learn the following, in the order shown below (with Xcode, Apple’s development environment, thrown in as needed):

• Objective-C
• Cocoa
• iPhone SDK (API’s)

So let me begin by pointing you to a few resources:

• iPhone Developer Center
  (you can create a new Apple ID if you don’t already have one)
• Learning Objective-C a Primer
• Objective-C Language Reference
• Cocoa Fundamentals Guide
• iPhone OS Programming Guide
• Xcode Overview

As you’ll see once you download a few of the docs, there is huge amount of information here. To keep things in perspective, my intention is not to revisit all the material in the documents, as much as point out the nuances that I think will be important in becoming a proficient iPhone developer.

A good example is the next post in this series where I’ll point out how to work around one of the sore spots for many who come to Objective-C from an object-oriented language, lack of support for private methods. With Objective-C there are a few tricks to "hide" methods, however, it’s really just a slight-of-hand, so to speak. I’ll explain more later.

There is a fair amount of information to digest above (assuming you are new to Objective-C, Cocoa and/or Xcode). Once you are ready for more, here is a list of additional documents to further immerse yourself:

• iPhone Guides (with related technology)
• Development References
• Sample Code

I hope you’ll join me as learn the ropes for developing iPhone applications.

I’ve been spending some time with the iPhone SDK to get a perspective on the architecture, tools and overall landscape. In addition, I’ve started to develop a few applications with Xcode (Cocoa/Objective-C) to exercise the tools for building iPhone applications. More on application development next week…

So far, I’m impressed. Here’s how I see things…

The Platform:
The iPhone architecture consists of the Core OS (kernel level resources), Core Services (system services), Media (audio, video and graphics) and Cocoa Touch (classes for managing graphics and event-driven applications).

From my perspective, there are two key areas that are exposed. First, the system services in Core OS, accessible through a C library (LibSystem) where one can work with wrappers for low-level features such as threading, networking, file system, Bonjour, among others.

Second, and even more compelling from a user interaction perspective, is the Media layer. Quartz is a 2-D drawing engine. Through the C-based API one can work with vector graphics, lines and shapes, etc. Core Animation is an Objective-C API providing animation and is also part-and-parcel to providing dynamic feedback from the UI. Many of the standard animations that make the iPhone so compelling (screens sliding, flipping over, etc) are included in View classes in the UIKit. OpenGL ES is a mobile version of the OpenGL standard and is the engine behind 3-D graphics. When high-frame rates are in order (think games), OpenGL is the answer.

From what I can tell, the iPhone SDK looks quite comprehensive, providing access across all areas of the device.

Distribution:
One of the challenges since day one for mobile developers has been getting their applications in front of potential users. With the iPhone, applications will be available through Apple’s App Store, which is accessible on the phone as well as through a desktop/laptop system. My understanding is that Apple will offer developers 70% of the revenue. Not unreasonable given Apple will provide the store front and manage all that goes with it.

Overall, an impressive start to what I think could be a significant, and welcome change, to the mobile development landscape.