Monday, December 3, 2007

Objective-C, often referred to as ObjC and sometimes as Objective C or Obj-C, is a reflective, object-oriented programming language which adds Smalltalk-style messaging to C.
Today it is used primarily on Mac OS X and GNUstep, two environments based on the OpenStep standard (although Mac OS X is OpenStep incompatible), and is the primary language used for the NeXTSTEP, OPENSTEP, and Cocoa application frameworks. Generic Objective-C programs that do not make use of these libraries can also be compiled for any system supported by gcc, which includes an Objective-C compiler.

In 1988, NeXT, the next company started by Steve Jobs after Apple, licensed Objective-C from StepStone (the owner of the Objective-C trademark) and released their own Objective-C compiler and libraries on which the NeXTstep user interface and interface builder were based. The success of the tools and quality of the resultant operating system helped NeXT become a fairly popular niche workstation provider.
After acquiring NeXT in 1996, Apple used OpenStep as the basis for its new main operating system, Mac OS X. This included Objective-C and NeXT's Objective-C based developer tool, Project Builder (later replaced by Xcode), as well as its interface design tool, Interface Builder. Most of Apple's present-day Cocoa API is based on OpenStep interface objects, and is the most significant Objective-C environment being used for active development.

Popularization through NeXT
Objective-C is a very "thin" layer on top of C. Objective-C is a strict superset of C. That is, it is possible to compile any C program with an Objective-C compiler. Objective-C derives its syntax from both C and Smalltalk. Most of the syntax (including preprocessing, expressions, function declarations, and function calls) is inherited from C, while the syntax for object-oriented features was created to enable Smalltalk-style message passing.

The added syntax is for built-in support of object-oriented programming. The Objective-C model of object-oriented programming is based on sending messages to objects, similar to the model of Smalltalk. This is unlike the Simula programming model, which is used by C++ among other programming languages. This distinction is semantically important. The basic difference is that in Objective-C, one does not call a method; one sends a message.
An object called obj whose class has a method doSomething implemented is said to respond to the message doSomething. If we wish to send a doSomething message to obj, we write
This mechanism allows messages to be sent to an object even if the object is not able to respond to them. This differs from statically typed languages such as C++ and Java in which all method calls to objects must be predefined. (See the dynamic typing section below.)

Objective-C requires the interface and implementation of a class to be in separate specially declared code blocks. By convention, the interface is put in a header file and the implementation in a code file; the header files, suffixed .h, are similar to C header files.

Objective-C Interfaces and implementations
The interface of the class is usually defined in a header file. Convention is usually to create the name of the header file based on the name of the class. So if we have the class Thing, Thing's interface goes in the file Thing.h.
The interface declaration is in this form:
Hyphens mark instance methods and plus signs mark class methods (like static member functions in C++). This is different from the meaning of a preceding – and + in UML diagrams which mean private and public method, respectively.

The interface only declares the prototypes for the methods, and not the methods themselves, which go in the implementation. The implementation is usually stored in a main file, for example, Thing.m. The implementation is written
Methods are written in a different way from C-style functions. For example, a function in both C and Objective-C follows this general form:
with int do_something(int) as the prototype.
When this is implemented as a method, this becomes:
A more canonical way of writing the above method would be like this, by naming the first argument in the selector name:
This syntax may appear to be more troublesome but it allows the naming of parameters, for example
which can be invoked thus:
Internal representations of this method vary between different implementations of Objective-C. If myColor is of the class Color, internally, instance method -changeColorWithRed:green:blue: might be labeled _i_Color_changeColorWithRed_green_blue. The i is to refer to an instance method, with the class and then method names appended, colons translated to underscores.
However, internal names of the function are rarely used directly, and generally even message-sends are converted to a call to a function defined in a run-time library rather than directly accessing the internal name. This is partially because it is rarely known at compile-time which method will actually be called, because the class of the receiver (i.e., the object being sent the message) is rarely known until runtime.

Objective-C was extended at NeXT to introduce the concept of multiple inheritance of specification, but not implementation, through the introduction of protocols. This is a pattern achievable either as an abstract multiply inherited base class in C++, or else, more popularly, adopted (e.g., in Java or C#) as an "interface". Objective-C makes use of both ad-hoc protocols, called informal protocols, and compiler enforced protocols called formal protocols.
An informal protocol is a list of methods that a class can implement. It is specified in the documentation, since it has no presence in the language. Informal protocols often include optional methods, where implementing the method can change the behavior of a class. For example, a text field class might have a delegate that should implement an informal protocol with an optional autocomplete method. The text field discovers whether the delegate implements that method (via reflection), and, if so, calls it to support autocomplete.
A formal protocol is similar to an interface in Java or C#. It is a list of methods that any class can declare itself to implement. The compiler will emit an error if the class does not implement every method of its declared protocols. The Objective-C concept of protocols is different from the Java or C# concept of interfaces in that a class may implement a protocol without being declared to implement that protocol. The difference is not detectable from outside code. Formal protocols cannot provide any implementations, they simply assure callers that classes that conform to the protocol will provide implementations. In the NeXT/Apple library, protocols are frequently used by the Distributed Objects system to represent the capabilities of an object executing on a remote system.
The syntax
denotes that there is the abstract idea of locking that is useful and when stated in a class definition
denotes that instances of SomeClass will provide an implementation for the two instance methods using whatever means they want. This abstract specification is particularly useful to describe the desired behaviors of plug-ins for example, without constraining at all what the implementation hierarchy should be.

Objective-C, like Smalltalk, can use dynamic typing; we can send an object a message not specified in its interface. This can allow for increased flexibility — in Objective-C an object can "capture" this message, and depending on the object, can send the message off again to a different object (who can respond to the message correctly and appropriately, or likewise send the message on again). This behavior is known as message forwarding or delegation (see below). Alternatively, an error handler can be used instead, in case the message cannot be forwarded. If the object does not forward the message, handle the error, or respond to it, a runtime error occurs.
Static typing information may also optionally be added to variables. This information is then checked at compile time. In the following statements, increasingly specific type information is provided. The statements are equivalent at runtime, but the additional information allows the compiler to warn the programmer if the passed argument does not match the type specified. In the first statement, the object must conform to the aProtocol protocol, and in the second, it must be a member of the NSNumber class.
Dynamic typing can be a powerful feature. When implementing container classes using statically-typed languages without generics like pre-1.5 Java, the programmer is forced to write a container class for a generic type of object, and then cast back and forth between the abstract generic type and the real type. Casting however breaks the discipline of static typing – if you put in an Integer and read out a String, you get an error. One way of alleviating the problem is to resort to generic programming, but then container classes must be homogeneous in type. This need not be the case with dynamic typing.

Dynamic typing
Since Objective-C permits the sending of a message to an object that might not respond to it, the object has a number of things it can do with the message. One of these things could be to forward the message on to an object that can respond to it. Forwarding can be used to implement certain design patterns, such as the Observer pattern or the Proxy pattern very simply.
The Objective-C runtime specifies a pair of methods in Object
and as such an object wishing to implement forwarding needs only to override the forwarding method to define the forwarding behaviour. The action methods performv:: need not be overridden as this method merely performs the method based on the selector and arguments.

forwarding methods:
action methods: Forwarding
Here is an example of a program that demonstrates the basics of forwarding.

If we were to compile the program, the compiler would report that
The compiler is reporting the point that was made earlier, Forwarder does not respond to hello messages. In certain circumstances, such a warning can help us find errors, but in this circumstance, we can safely ignore this warning, since we have implemented forwarding. If we were to run the program

Cox's main concern was the maintainability of large code bases. Experience from the structured programming world had shown that one of the main ways to improve code was to break it down into smaller pieces. Objective-C added the concept of Categories to help with this process.
A category collects method implementations into separate files. The programmer can place groups of related methods into a category to make them more readable. For instance, one could create a "SpellChecking" category "on" the String object, collecting all of the methods related to spell checking into a single place.
Furthermore, the methods within a category are added to a class at runtime. Thus, categories permit the programmer to add methods to an existing class without the need to recompile that class or even have access to its source code. For example, if the system you are supplied with does not contain a spell checker in its String implementation, you can add it without modifying the String source code.
Methods within categories become indistinguishable from the methods in a class when the program is run. A category has full access to all of the instance variables within the class, including private variables.
Categories provide an elegant solution to the fragile base class problem for methods.
If you declare a method in a category with the same method signature as an existing method in a class, the category's method is adopted. Thus categories can not only add methods to a class, but also replace existing methods. This feature can be used to fix bugs in other classes by rewriting their methods, or to cause a global change to a class' behavior within a program. If two categories have methods with the same method signature, it is undefined which category's method is adopted.
Other languages have attempted to add this feature in a variety of ways. TOM took the Objective-C system a step further and allowed for the addition of variables as well. Other languages have instead used prototype oriented solutions, the most notable being Self.

This example builds up an Integer class, by defining first a basic class with only accessor methods implemented, and adding two categories, Arithmetic and Display that extend the basic class. Whilst categories can access the base class' private data members, it is often good practice to access these private data members through the accessor methods, which helps keep categories more independent from the base class. This is one typical usage of categories—the other is to use categories to add or replace certain methods in the base class (however it is not regarded as good practice to use categories for subclass overriding).

Example usage of categories
Compilation is performed, for example, by
One can experiment by omitting the #import "Arithmetic.h" and [num1 add:num2] lines and omit Arithmetic.m in compilation. The program will still run. This means that it is possible to "mix-and-match" added categories if necessary – if one does not need to have some capability provided in a category, one can simply not compile it in.

Objective-C permits a class to wholly replace another class within a program. The replacing class is said to "pose as" the target class. All messages sent to the target class are then instead received by the posing class. There are several restrictions on which classes can pose:
Posing, similarly to categories, allows globally augmenting existing classes. Posing permits two features absent from categories:
For example,
This intercepts every invocation of setMainMenu to NSApplication.

A class may only pose as one of its direct or indirect superclasses
The posing class must not define any new instance variables that are absent from the target class (though it may define or override methods).
No messages must have been sent to the target class prior to the posing.
A posing class can call overridden methods through super, thus incorporating the implementation of the target class.
A posing class can override methods defined in categories. Posing
In the C language, the #include pre-compile directive allows for the insertion of entire files before any compilation actually begins. Objective-C adds the #import directive, which does the same thing, except that it knows not to insert a file that has already been inserted.
For example, if file A includes files X and Y, but X and Y each include the file Q, then Q will be inserted twice into the resultant file, causing "duplicate definition" compile errors. But if file Q is included using the #import directive, only the first inclusion of Q will occur—all others will be ignored.
A few compilers, including GCC, support #import for C programs too; its use is discouraged on the basis that the user of the header file has to distinguish headers that should be included only once, from headers designed to be used multiple times. It is argued that this burden should be placed on the implementor; to this end, the implementor may place the directive #pragma once in the header file, or use the traditional include guard technique:
If a header file uses guards or #pragma once, it makes no difference whether it is #included or #imported. The same objection to #import actually applies to Objective-C as well, and many Objective-C programs will also adopt guards in their headers.

Objective-C in fact included a laundry-list of features that are still being added to other languages, and some that are unique to it. These led from Cox's (and later, NeXT's) realization that there is considerably more to programming than the language. The system has to be usable and flexible as a whole in order to work in a real-world setting.

Delegating methods to other objects at run-time is trivial. Simply add a category that changes the "second chance" method to forward the invocation to the delegate.
Remote invocation is trivial. Simply add a category that changes the "second chance" method to serialize the invocation and forward it off.
Swizzling of the isa pointer allows for classes to change at runtime. Typically used for debugging where freed objects are swizzled into zombie objects, whose only purpose is to report an error when someone calls them. Swizzling was also used in EOF to create database faults. Swizzling is used today by Apple's Foundation Framework to implement Key-Value Observing.
Archiving. An object can be archived into a stream, such as a file, and can be read and restored on demand. Other features
Objective-C++ is a front-end to the GNU Compiler Collection that can compile source files that use a combination of C++ and Objective-C syntax. Objective-C++ adds to C++ the extensions Objective-C adds to C. As nothing is done to unify the semantics behind the various language features, certain restrictions apply:

A C++ class cannot derive from an Objective-C class and vice versa.
C++ namespaces cannot be declared inside an Objective-C declaration.
Objective-C classes cannot have instance variables of C++ classes that do not have a default constructor, or that have one or more virtual methods, but pointers to C++ objects can be used as instance variables without restriction (allocate them with new in the -init method).
C++ "by value" semantics cannot be applied to Objective-C objects, which are only accessible through pointers.
An Objective-C declaration cannot be within a C++ template declaration and vice versa. Objective-C types, (e.g., Classname *) can be used as C++ template parameters, however.
Objective-C and C++ exception handling is distinct; the handlers of each cannot handle exceptions of the other type.
Care must be taken since the destructor calling conventions of Objective-C and C++'s exception run-time models do not match (i.e., a C++ destructor will not be called when an Objective-C exception exits the C++ object's scope). Objective-C++
Objective-C today is often used in tandem with a fixed library of standard objects (often known as a "kit" or "framework"), such as Cocoa or GNUstep. These libraries often come with the operating system: the GNUstep libraries often come with Linux distributions and Cocoa comes with Mac OS X. The programmer is not forced to inherit functionality from the existing base class (NSObject). Objective-C allows for the declaration of new root classes that do not inherit any existing functionality. Originally, Objective-C based programming environments typically offered an Object class as the base class from which almost all other classes inherited. With the introduction of OpenStep, NeXT created a new base class named NSObject that offered additional features over Object (an emphasis on using object references and reference counting instead of raw pointers, for example). Almost all classes in Cocoa inherit from NSObject.
Not only did the renaming serve to differentiate the new default behavior of classes within the OpenStep API, but it allowed code that used Object — the original base class used on NeXTSTEP (and, more or less, other Objective-C class libraries) — to co-exist in the same runtime with code that used NSObject (with some limitations). As well, the introduction of the two letter prefix became a sort of simplistic form of namespaces, which Objective-C lacks. Using a prefix to create an informal packaging identifier became an informal coding standard in the Objective-C community, and continues to this day.

Objective-C 2.0
Objective-C 2.0 allows for garbage collection, but it is an opt-in system. One may use garbage collection in a backwards compatible way, such that code written for previous versions will continue to work.

Garbage collection
Whereas instance variables previously required the creation of methods to get and set (getters and setters) those variables, Objective-C 2.0 introduces the property syntax:
Once added to the interface, properties can be accessed using dot notation (example given an instance aPerson of the above Person class):
The compiler translates property dot notation into accessor method calls. The above statement is equivalent to:

Instead of using an Enumerator object to iterate through a collection, Objective-C 2.0 offers the foreach syntax (given an array thePeople of objects of the above defined Person):

Fast iteration
Besides the GCC/NeXT/Apple Computer implementation, which added several extensions to the original Stepstone implementation, there exists another free open-source Objective-C implementation, which implements a slightly different set of extensions: The Portable Object Compiler [5] implements, among other things, also Smalltalk-like blocks for Objective-C.

Portable Object Compiler
Objective-C implementations use a thin runtime written in C that adds little to the size of the application. In contrast, most OO systems at the time that it was created (and Java even today) used large VM runtimes that took over the entire system. Programs written in Objective-C tend to be not much larger than the size of their code and that of the libraries (which generally do not need to be included in the software distribution), in contrast to Smalltalk systems where a large amount of memory was used just to open a window.
Likewise, the language can be implemented on top of existing C compilers (in GCC, first as a preprocessor, then as a module) rather than as a new compiler. This allows Objective-C to leverage the huge existing collection of C code, libraries, tools, and mindshare. Existing C libraries — even in object code libraries — can be wrapped in Objective-C wrappers to provide an OO-style interface. All of these practical changes lowered the barrier to entry, likely the biggest problem for the widespread acceptance of Smalltalk in the 1980s.
The first versions of Objective-C did not support garbage collection. At the time this decision was a matter of some debate, and many people considered long "dead times" (when Smalltalk did collection) to render the entire system unusable. Although some 3rd party implementations have added this feature (most notably GNUstep), Apple implemented it as of Mac OS X v10.5. But it is unavailable to applications targeting older versions of the Mac OS
Another common criticism is that Objective-C does not have language support for namespaces. Instead programmers are forced to add prefixes to their class names, which can cause collisions. As of 2007, all Mac OS X classes and functions in the Cocoa programming environment are prefixed with "NS" (as in NSObject or NSButton) to clearly identify them as belonging to the Mac OS X core; the "NS" derives from the names of the classes as defined during the development of NeXTSTEP.
Since Objective-C is a strict superset of C, it does not treat C primitive types as first-class objects either.
Unlike C++, Objective-C does not support operator overloading. Also unlike C++, Objective-C allows an object only to directly inherit from one class (forbidding multiple inheritance). As Java was influenced by the design of Objective-C, the decision to use single inheritance was carried into Java. Categories and protocols may be used as alternative functionality to multiple inheritance; Java however lacks categories.

Analysis of the language
The design and implementation of C++ and Objective-C represent different approaches to extending C.
In addition to C's style of procedural programming, C++ directly supports object-oriented programming, generic programming, and metaprogramming. C++ also comes with a large standard library which includes several container classes. Objective-C, on the other hand, adds only object-oriented features to C. Objective-C in its purest fashion does not contain the same number of standard library features, but in most places where Objective-C is used, it is used with an OpenStep-like library such as OPENSTEP, Cocoa, or GNUstep which provide similar functionality to some of C++'s standard library.
One notable difference is that Objective-C provides runtime support for some reflective features, whereas C++ adds only a small amount of runtime support to C. In Objective-C, an object can be queried about its own properties, for example whether it will respond to a certain message. In C++ this is not possible without the use of external libraries; however, it is possible to query whether two objects are of the same type (including built-in types) and whether an object is an instance of a given class (or superclass).
The use of reflection is part of the wider distinction between dynamic (run-time) features versus static (compile-time) features of a language. Although Objective-C and C++ each employ a mix of both features, Objective-C is decidedly geared toward run-time decisions while C++ is geared toward compile-time decisions. The tension between dynamic and static programming involves many of the classic trade-offs in computer science.

See also

No comments: