CSC/ECE 517 Fall 2012/ch1b 1w47 sk

From PG_Wiki
Jump to: navigation, search

Generics

Contents

Introduction

Generic Programming is a programming paradigm for developing efficient and reusable software libraries. The Generic Programming process focuses on finding commonality among similar implementations of the same algorithm, then providing suitable abstractions so that a single, generic algorithm can cover many concrete implementations. This process is repeated until the generic algorithm has reached a suitable level of abstraction, where it provides maximal re-usability while still yielding efficient, concrete implementations.

In the simplest definition, generic programming is a style of computer programming in which algorithms are written in terms of to-be-specified-later types that are then instantiated when needed for specific types provided as parameters. This approach, pioneered by Ada in 1983, permits writing common functions or types that differ only in the set of types on which they operate on, thus reducing duplication. Such software entities are known as generics in Ada, Eiffel, Java and .NET; parametric polymorphism in ML, Scala and Haskell ; templates in C++; and parameterized types in the influential 1994 book, Design Patterns.

The term generic programming was originally coined by David Musser and Alexander Stepanov in a more specific sense than the above, to describe an approach to software decomposition whereby fundamental requirements on types are abstracted from across concrete examples of algorithms and data structures and formalized as concepts, analogously to the abstraction of algebraic theories in abstract algebra. Early examples of this programming approach were implemented in Scheme and Ada, although the best known example is the Standard Template Library (STL) in which is developed a theory of iterators which is used to decouple sequence data structures and the algorithms operating on them.

Generics in Object Oriented Languages

Generics .NET framework

Generics are classes, structures, interfaces, and methods that have placeholders (type parameters) for one or more of the types that they store or use. A generic collection class might use a type parameter as a placeholder for the type of objects that it stores; the type parameters appear as the types of its fields and the parameter types of its methods. A generic method might use its type parameter as the type of its return value or as the type of one of its formal parameters.

The following code illustrates a simple generic class definition in C#.

public class Generic<T>
{
    public T Field;
}

The following code illustrates a simple generic class definition in Visual Basic.

Public Class Generic(Of T)
    Public Field As T
End Class

When an instance of a generic class is created, one can specify the actual types to substitute for the type parameters. This establishes a new generic class, referred to as a constructed generic class, with the chosen types substituted everywhere that the type parameters appear. The result is a type-safe class that is tailored to the user's choice of types, which is illustrated in the code below in C# and VB.

//Code in C#
public static void Main()
{
    Generic<string> g = new Generic<string>();
    g.Field = "A string";
}
'Code in Visual Basic
Public Shared Sub Main()
    Dim g As New Generic(Of String)
    g.Field = "A string" 
End Sub

Generics Terminology in .NET

//Code in C#
T Generic<T>(T arg)
{
    T temp = arg;
    //... 
    return temp;
}
'Code in Visual Basic
Function Generic(Of T)(ByVal arg As T) As T
    Dim temp As T = arg
    '... 
    Return temp
End Function
//Code in C#
class A
{
    T G<T>(T arg)
    {
        T temp = arg;
        //... 
        return temp;
    }
}
class Generic<T>
{
    T M(T arg)
    {
        T temp = arg;
        //... 
        return temp;
    }
}
'Code in Visual Basic
Class A
    Function G(Of T)(ByVal arg As T) As T
        Dim temp As T = arg
        '... 
        Return temp
    End Function 
End Class 
Class Generic(Of T)
    Function M(ByVal arg As T) As T
        Dim temp As T = arg
        '... 
        Return temp
    End Function 
End Class

Nested Types and Generics

A type that is nested in a generic type can depend on the type parameters of the enclosing generic type. The common language runtime considers nested types to be generic, even if they do not have generic type parameters of their own. When you create an instance of a nested type, you must specify type arguments for all enclosing generic types.

Language Support

The .NET Framework provides a number of generic collection classes in the following namespaces:

Advantages and Disadvanatges

Generics in .NET 2.0 are very powerful. They allow developers to write code without committing to a particular type, yet your code can enjoy type safety. Generics are implemented in such a way as to provide good performance and avoid code bloat. While there is the drawback of constraints' inability to specify that a type must be a base type of another type, the constraints mechanism gives us the flexibility to write code with a greater degree of freedom than sticking with the least-common-denominator capability of all types.

Generics in Java

Generics were introduced in Java from JDK 1.5. Generics allows the abstraction over types. The most common examples are container types, such as those in the Collection hierarchy. So, if one could say that the code works with some unspecified type, rather than a specific interface or class, then the code uses Generics. Below is an example of a Genric:

List myIntList = new LinkedList(); 
myIntList.add(new Integer(0)); 
Integer x = (Integer) myIntList.iterator().next(); 

Typically, a programmer knows what kind of data has been placed into a particular list. However, the cast is essential. The compiler can only guarantee that an Object will be returned by the iterator. To ensure the assignment to a variable of type Integer is type safe, the cast is required. Of course, the cast not only introduces clutter, but also introduces the possibility of a run time error, since the programmer might be mistaken. It would be better if programmers could actually express their intent, and mark a list as being restricted to contain a particular data type. This is the core idea behind generics.

Here is a variation of the program fragment given above using generics:

List<Integer> myIntList = new LinkedList<Integer>();
myIntList.add(new Integer(0)); 
Integer x = myIntList.iterator().next();

The type declaration for the variable myIntList specifies that this is not just an arbitrary List, but a List of Integer, written List<Integer>. The List is a generic interface that takes a type parameter - in this case, Integer. One has to specify a type parameter when creating the list object. The compiler can now check the type correctness of the program at compile-time. Because myIntList is declared with type List<Integer>, this indicates that myIntList holds a list of integers, which holds true wherever and whenever it is used, and the compiler ensures to guarantee it. In contrast, the cast tells something the programmer thinks is true at a single point in the code. The net effect, especially in large programs, is improved readability and robustness.

Below is code for generic interfaces List and Iterator in package java.util.The content in angle brackets are the declarations of the formal type parameters of the interfaces List and Iterator.

public interface List<E> {
void add(E x);
Iterator<E> iterator();
}
public interface Iterator<E> {
E next();
boolean hasNext();
}

Generics and Subtyping

Below is an example showing code for Generics and Subtyping.

List<String> ls = new ArrayList<String>(); //1
List<Object> lo = ls; //2
lo.add(new Object()); //3
String s = ls.get(0); //4 attempts to assign an Object to a String

Line 1 is legal. Lines 2,3 and 4 aliased ls and lo. Accessing ls, a list of String, through the alias lo, one can insert arbitrary objects into it. As a result ls does not hold just Strings anymore. The Java compiler will prevent this from happening and Line 2 causes a compile time error.In general, if Foo is a subtype (subclass or subinterface) of Bar, and G is some generic type declaration, it is not the case that G<Foo> is a subtype of G<Bar>.

Wildcards

Consider the problem of writing a routine that prints out all the elements in a collection. In older version's of Java, it can be written as below:

void printCollection(Collection c) {
   Iterator i = c.iterator();
   for (k = 0; k < c.size(); k++) {
       System.out.println(i.next());
   }
}

In the later versions of Java, the same logic above can be written using generics using the syntax of enhanced for loop:

void printCollection(Collection<Object> c) {
   for (Object e : c) {
      System.out.println(e);
   }
}

The old code could be called with any kind of collection as a parameter, the new code only takes Collection<Object>, which is not a supertype of all kinds of collections. The supertype of all kinds of collections is written Collection<?>, that is, a collection whose element type matches anything. It’s called a wildcard type. Below is the code expressing the same functionality using a wildcard.

void printCollection(Collection<?> c) {
   for (Object e : c) {
      System.out.println(e);
   }
}

The above code can be called with any type of collection. Inside printCollection(), one can still read elements from c and give them type Object. This is always safe, since whatever the actual type of the collection, it does contain objects. However, It is not safe to add arbitrary objects to it.

Generic Methods

Generic methods are used more often than generic classes. A generic class is independent of whether you have a generic method. If a method is static, it has no access to the generic type parameters of the class, so if it needs to use genericity it must be a generic method. Unlike generic class one don’t usually have to specify the parameter types, because the compiler can figure that out.

public class Gen {
   public <T> void f(T x) {
       System.out.println(x.getClass().getName());
   }

   public static void main(String[] args) {
       Gen a = new Gen();
       a.f("");
       a.f(‘c’);
   }
}

Erasure

To implement generics, the Java compiler applies type erasure to:

Type erasure ensures that no new classes are created for parameterized types; consequently, generics incur no runtime overhead.

Consider the example below.

public class A {
   public static void main(String[] args) {
       System.out.println(new ArrayList<String>().getClass() == new ArrayList<Integer>().getClass());
   }
}

Here ArrayList<String> and ArrayList<Integer> seem to be of different types and so one expects them to behave differently. But the execution of the above program returns true which suggests they are of the same type.

class Fn {}
class Foo<Q> {}
class Boo<X,Y> {}

public class AbsentInformation {
   public static void main(String[] args) {
      Foo<Fn> foo = new Foo<Fn>();
      Boo<Long,Double> b = new Boo<Long,Double>();

      System.out.println(Arrays.toString(foo.getClass().getTypeParameters()));
      System.out.println(Arrays.toString(b.getClass().getTypeParameters()));
   }
} 
Output:
[Q]
[X,Y]

Class.getTypeParameters() suggest that one might be able to find out what the parameter types are. But in reality, the identifiers that are used as the parameter placeholders. This is one important feature where Java differs from C++. Java generics are implemented using erasure. So when one uses generics, the specific information is erased. The only thing one knows is that we are using an object. So both List<String> and List< Integer> have the same type at runtime.

Advantages and Disadvantages

With Java generics, we don't actually get any of the execution efficiency, because when one compiles a generic class in Java, the compiler takes away the type parameter and substitutes Object everywhere. So the compiled image for List<T> is like a List where it uses the type Object everywhere. Of course, if you now try to make a List<int>, you get boxing of all the ints. So there's a bunch of overhead there. Furthermore, to keep the VM happy, the compiler actually has to insert all of the type casts that were not written by the programmer. If it's a List of Object and we're trying to treat those Objects as Customers, at some point the Objects must be cast to Customers to keep the verifier happy. Java in their implementation does this by automatically inserting those type casts for you. Though this simplifies the programmer's job and avoids cluttering in code, there is no efficiency attained out of this optimization. Java's generics implementation relies on erasure of the type parameter, when we get to runtime, we don't actually have a faithful representation of what we had at compile time. When we apply reflection to a generic List in Java, we can't tell what the List is a List of


Generics in C++

C++ templates are a powerful mechanism for code reuse, as they enable the programmer to write code that behaves the same for data of any type.

Defining Templates

A Template Definition starts with the keyword template, followed by a list of Template Parameters. What follows is then either a class definition, or a function definition, defining a class template or a function template respectively. The template parameters introduce names into the scope of the definition, which can be types, values or templates. These names can be used just like any other name of the same kind. Then, when the template is instantiated, real types, values or templates are substituted in place of these names, and the code compiled.

Two template instantiations refer to the same template if their parameters are all the same, irrespective of any typedefs that may apply. Therefore vec1, vec2 and vec3 in the following example are all the same type.

typedef std::string MyString;
typedef std::vector<std::string> T1;
typedef std::vector<MyString> T2;
T1 vec1;
T2 vec2;
std::vector<std::string> vec3;

Multiple parameters may be specified, separated by commas in the template parameter list:

template<typename T1,typename T2>
class MyClass{};

MyClass is thus a class template with two template type parameters, T1 and T2. Every time a template is referenced with distinct template arguments, then the template is instantiated with the arguments. If the resultant code is not valid, then a compilation error will occur.

Templates can have one or more template parameters, which can be Type parameters, Non-Type parameters, or Template parameters.

Template Type Parameters

Template Type Parameters are template parameters that refer to a type; they are the most common form of template parameters. The syntax for Template Type Parameters is as below:

typename name
or
class name

Both alternatives are identical in meaning, and the choice is merely a matter of style. name is any valid C++ symbol name. Once name has been introduced in the template parameter list, then any reference to name within the body of the template automatically refers to the type of the corresponding template argument for each instantiation, and can be used anywhere a type can normally be used. e.g.

template<typename T>
void func(T value)
{
const T& ref=value;
T* p=new T;
T temp(23);
}

The code generated for func<int> is then identical to:

void func(int value)
{
const int& ref=value;
int* p=new int;
int temp(23);
}

If, however, a reference was made to func<std::string>, then a compilation error would result, as the statement std::string temp(23); is not valid.

Template Non-Type Parameters

Non-Type Template Parameters are template parameters that are values rather than types. They can be any value that is a compile-time constant. Their syntax is akin to a variable declaration, e.g.:

template<int i>
class A{};
template<double* dp>
class B{};
template<void (*func)(int)>
void c()
{}

Class template A has a non-type template parameter which is an integer, class template B has a non-type template parameter which is a pointer-to-double, and function template c has a non-type template parameter which is a pointer to a function returning void, with a single int parameter.

Template Template Parameters

Template Template Parameters enable a template to be parameterized by the name of another template. Say, for example, there is a class which contains a couple of collections of items, some strings, some integers, and one wants the users of the class to choose what type of collection to use (vector, list, stack, etc.). The natural thought is to make the collection type a template parameter. However, a collection of strings is a different type from a collection of integers, so the user would have to specify both individually if template type parameters are used. The solution is a Template Template Parameter which is shown in the code below:

template<template<typename T> class ContainerType>
class MyClass
{
ContainerType<int> intContainer;
ContainerType<std::string> stringContainer;
// rest of class
};

ContainerType is a Template Template Parameter that refers to a template with a single Template Type Parameter. One can thus say MyClass<vector> or MyClass<list> to have vectors or lists respectively.Within the template definition, the Template Template Parameters can be used just like any other template.

Using Templates

Class templates and function templates can be used anywhere normal classes and functions can be, as well as as Template Template Parameters to other templates. However, the compiler needs to know what to use for the template parameters. For class templates, the template name must be followed by a template argument list, specifying the parameters. This is a comma-separated list of expressions between angle brackets (<>). For Template Type Parameters, the corresponding expression must name a type, for Template Non-Type Parameters, the expression must evaluate to a compile-time constant of the appropriate type, and for Template Template Parameters, the expression must name a template with the correct signature. e.g.:

template<typename T,unsigned i>
struct FixedArray
{
T data[i];
};
FixedArray<int,3> a; // array of 3 integers
FixedArray<int,1+6/3> b; // array of 3 integers
template<template<typename T,typename Allocator> class Container>
struct ContainerPair
{
Container<int,std::allocator<int> > intContainer;
Container<std::string,std::allocator<std::string> > stringContainer;
};
ContainerPair<std::deque> deqCont; // two std::deques
ContainerPair<std::vector> vecCont; // two std::vectors

The second example demonstrates two things. Firstly, the template signature of the Template Template Parameter must exactly match the signature of the template passed as the argument, and standard containers have two template parameters. Secondly, if the last argument of a template parameter list is a template reference, as for Container<int,std::allocator<int> >, then the two closing angle brackets (>) must be separated by a space, to avoid being interpreted as the shift right operator (>>).

Templates requirements and concepts

Templates implicitly impose requirements on their parameters, particularly Template Type Parameters and Template Template Parameters. These requirements generally take the form of operations that must work on objects of the appropriate type, or class members that must exist and refer to objects or types or functions (with the implicit extra requirement that the type referred to is a class). A Concept is a set of requirements that describe a useful feature of a type. For example, the C++ Standard Library makes reference to things being Assignable or Copy-Constructible; these are Concepts that make the following requirements:

Template Specialization and Overloading

Template Specialization allows one to decide that for a specific set of template parameters, the code instantiated for a template should be different to the general case. Say, for example, one wish to use a template with a new class as a Template Type Parameter, and though it has the same semantic behaviour as the classes the template was designed for, the syntax is different, so the original template won’t compile with the new class as a parameter. The solution to this problem is specialization. Consider the following template function definition:

template<typename T>
void func(T value)
{
value.add(3);
}

This function assumes that the type substituted for the Template Type Parameter T is a class with a member function add that takes a single parameter of a type that can be constructed from an int. Thus the following classes would be OK:

class T1
{
public:
void add(int i,double d=0.0);
};
class T2
{
public:
std::string add(double d);
};

However, the class T3 below is not OK, because Add has a capital A.

class T3
{
public:
void Add(int i);
};

This means that one cannot use the original function template func for T3.

Advantages and Disadvanatges

Generics generate different classes based on the input type. An std::vector<int> is a completely different class than an std::vector<float>. However, since C++ templates create different classes for every variation of their template parameters, the size of the compiled executable is larger. Beyond that, compilation time increases greatly, since all template code must be included with each compilation unit and much more code must be generated. However, actual runtime memory footprint is typically smaller than the alternative (frees an extra void*) and performance is better, since the compiler can perform more aggressive optimizations with the known type. While a generic Java class compiles it's entire self, when using a C++ template, you only compile what you use. So, if you create an std::vector<int> and only use push_back and size, only those functions will be compiled into the object file. This eases the size of executable problem.

Conclusion

Generics in all languages have the common advantage of saving developer's effort on type casting of individual types and generalizes functionality irrespective of individual data types hence reducing code duplication, code cluttering and improving code re-usability. Compilation and generation of final executable follows a different path in C++ and Java. In Java, all the generic classes of different types are all considered the same with type Object, while in C++ each of the templates is a different class. As a result, in Java, there is not increase in the size of executable files, but there is no improvement in runtime efficiency as well. In C++, however, the size of the executable is much large, but there is improved efficiency.

References

1. http://www.artima.com/intv/genericsP.html
2. http://www.generic-programming.org
3. Generics in the Java Programming Language
4. http://www.cplusplus.com/doc/tutorial/templates
5. http://osl.iu.edu/publications/prints/2003/comparing_generic_programming03.pdf
6. http://msdn.microsoft.com/en-us/library/ms172192.aspx#Y1524
7. http://www.justsoftwaresolutions.co.uk/articles/intrototemplates.pdf
8. http://www.sgi.com/tech/stl/stl_introduction.html

Personal tools
Namespaces
Variants
Actions
Navigation
Toolbox