typedef

From Wikipedia, the free encyclopedia
Jump to: navigation, search

typedef is a keyword in the C and C++ programming languages. The purpose of typedef is to form complex types from more-basic machine types[1] and assign simpler names to such combinations. They are most often used when a standard declaration is cumbersome, potentially confusing, or likely to vary from one implementation to another.[2]

Under C convention (such as in the C standard library or POSIX), types declared with typedef end with '_t' (e.g., size_t, time_t). Such type names are reserved by POSIX for future extensions and should generally be avoided for user defined types.

Usage examples[edit]

Indicating what a variable represents[edit]

A typedef can be used to indicate how a variable represents something, e.g., units of measurement or counts. This code sample uses no typedefs:

int current_speed ;
int high_score ;
...
 
void congratulate(int your_score) {
    if (your_score > high_score)
...

This code sample does use typedefs:

typedef int km_per_hour ;
typedef int points ;
 
km_per_hour current_speed ;  //"km_per_hour" is synonymous with "int" here,
points high_score ;          //and thus, the compiler treats our new variables as integers.
...
 
void congratulate(points your_score) {
    if (your_score > high_score)
...

Both sections of code do the same thing. However, the use of typedef in the second example makes it clear that the two variables, while represented by the same data type (int), represent different and incompatible things. The declaration in congratulate() of your_score indicates to the programmer that current_speed (or any other variable not declared as a points) should not be passed as an argument. This would not be as apparent if both were declared as ints. However, note that the indication is for the programmer only; the C/C++ compiler considers both variables to be ints and will not give any type mismatch warnings or errors for the "wrong" argument type for congratulate(points your_score) in the code snippet below:

void foo() {
    km_per_hour km100 = 100;
    congratulate(km100);
...

Although the compiler considers km_per_hour to be equivalent to int in the above code, the two cannot be used interchangeably when the type is changed via a prefix of unsigned, signed, or long.

void foo() {
    unsigned int a;         // Okay
    unsigned km_per_hour b; // Compiler complains
    long int c;             // Okay
    long km_per_hour d;     // Compiler complains
 
...

Simplifying a declaration[edit]

A typedef can be used to simplify the declaration for a compound type (struct, union) or pointer type.[3] For example,

struct MyStruct {
    int data1;
    char data2;
};

Here (above) a struct MyStruct data type has been defined. To declare a variable of this type in C (below) the struct key word is required (though it may be omitted in C++):

struct MyStruct a;

A typedef can be used to eliminate the need for the struct key word in C. For example, with:

typedef struct MyStruct newtype;

we can now create a variable of this type with:

newtype a;

Note that the structure definition and typedef can instead be combined into a single statement:

typedef struct MyStruct {
    int data1;
    char data2;
} newtype;

Or simply we can also use it as:

typedef struct {
    int data1;
    char data2;
} newtype;

In C++, in contrast to C, the struct, class, and enum key words are optional in variable declarations that are separate from the definitions, as long as there is no ambiguity to another identifier:

struct MyStruct x;       // This is legal
MyStruct y;              // This is also legal

As such, MyStruct can be used wherever newtype can be used. However, the reverse is not true; for instance, the constructor methods for MyStruct cannot be named newtype.

A notorious example where even C++ would need the struct keyword is POSIX' stat system call that uses a struct of the same name in its arguments:

int stat(const char *filename, struct stat *buf);

Here both C as well as C++ need the struct keyword in the parameter declaration.

Using typedef with pointers[edit]

Typedefs can also simplify declarations for pointer types. Consider this:

struct Node {
    int data;
    struct Node *nextptr;
};

Using typedef, the above code can be rewritten like this:

typedef struct Node Node;
struct Node {
    int data;
    Node *nextptr;
};

In C, one can declare multiple variables of the same type in a single statement, even mixing pointer and non-pointers. However, one would need to prefix an asterisk to each variable to designate it as a pointer. In the following, a programmer might assume that errptr was indeed a Node *, but a typographical error means that errptr is a Node. This can lead to subtle syntax errors.

struct Node *startptr, *endptr, *curptr, *prevptr, errptr, *refptr;

By defining a Node * typedef, it is assured that all the variables will be pointer types.

typedef struct Node *NodePtr;
...
NodePtr startptr, endptr, curptr, prevptr, errptr, refptr;

Using typedef with function pointers[edit]

Function pointers are somewhat different than all other types because the syntax does not follow the pattern typedef <old type name> <new alias>;. Instead, the new alias for the type appears in the middle between the return type (on the left) and the argument types (on the right). Consider the following code, which does not use a typedef:

int do_math(float arg1, int arg2) {
    return arg2;
}
 
int call_a_func(int (*call_this)(float, int)) {
    int output = call_this(5.5, 7);
    return output;
}
 
int final_result = call_a_func(&do_math);

This code can be rewritten with a typedef as follows:

typedef int (*MathFunc)(float, int);
 
int do_math(float arg1, int arg2) {
    return arg2;
}
 
int call_a_func(MathFunc call_this) {
    int output = call_this(5.5, 7);
    return output;
}
 
int final_result = call_a_func(&do_math);

Here, MathFunc is the new alias for the type. A MathFunc is a pointer to a function that returns an integer and takes as arguments a float followed by an integer.

When a function returns a function pointer, it can be even more confusing without typedef. The following is function prototype of signal(3) from FreeBSD:

void (*signal(int sig, void (*func)(int)))(int);

The function declaration above is cryptic as it does not clearly show what does the function accept as arguments, and what type does it return. A novice programmer may even assume that the function accepts a single int as its argument and returns nothing, but in reality it also needs a function pointer and returns another function pointer. It can be written more cleanly:

typedef void (*sighandler_t)(int)
sighandler_t signal(int sig, sighandler_t func);

Using typedef with type casts[edit]

A typedef is created using type declaration syntax but can be used as if it were created using type cast syntax. (Type casting changes a data type.) For instance, in each line after the first line of:

typedef int (*funcptr)(double);         // pointer to function taking a double returning int
funcptr x = (funcptr) NULL;             // C or C++
funcptr y = funcptr(NULL);              // C++ only
funcptr z = static_cast<funcptr>(NULL); // C++ only

funcptr is used on the left-hand side to declare a variable and is used on the right-hand side to cast a value. Thus, typedefs can be used by programmers who do not wish to figure out how to convert declaration syntax to type cast syntax.

Note that, without the typedef, it is generally not possible to use declaration syntax and cast syntax interchangeably. For example:

void *p = NULL;
int (*x)(double)  = (int (*)(double)) p; // This is legal
int (*)(double) y = (int (*)(double)) p; // Left-hand side is not legal
int (*z)(double)  = (int (*p)(double));  // Right-hand side is not legal

Usage concerns[edit]

Some people are opposed to the extensive use of typedefs. Most arguments center on the idea that typedefs simply hide the actual data type of a variable. For example, Greg Kroah-Hartman, a Linux kernel hacker and documenter, discourages their use for anything except function prototype declarations. He argues that this practice not only unnecessarily obfuscates code, it can also cause programmers to accidentally misuse large structures thinking them to be simple types.[4]

Others argue that the use of typedefs can make code easier to maintain. K&R states that there are two reasons for using a typedef. First, it provides a means to make a program more portable. Instead of having to change a type everywhere it appears throughout the program's source files, only a single typedef statement needs to be changed. Second, a typedef can make a complex declaration easier to understand.

Usage in C++[edit]

In C++ type names can be very complicated and typedef provides a mechanism to assign a simple name to the type. Consider:

std::vector<std::pair<std::string, int> > values;
for (std::vector<std::pair<std::string, int> >::const_iterator i = values.begin(); i != values.end(); ++i)
{
   std::pair<std::string, int> const & t = *i;
   // do something
}

and

typedef std::pair<std::string, int> value_t;
typedef std::vector<value_t> values_t;
 
values_t values;
for (values_t::const_iterator i = values.begin(); i != values.end(); ++i)
{
   value_t const & t = *i;
   // do something
}

Use with templates[edit]

There is no direct way to have templated typedefs in C++03. For instance, to have stringpair<T> represent std::pair<std::string, T> for every type T one cannot use:

template<typename T>
typedef std::pair<std::string, T> stringpair<T>; // Doesn't work

However, if one is willing to accept stringpair<T>::foo in lieu of stringpair<T> then it is possible to achieve the desired result via a typedef within an otherwise unused templated class or struct:

template<typename T>
class stringpair
{
private:
    // Prevent instantiation of stringpair<T>
    stringpair();
public:
    // Make stringpair<T>::type represent std::pair<std::string, T>
    typedef std::pair<std::string, T> type;
};
 
// Declare a variable of type std::pair<std::string, int>
stringpair<int>::type my_pair_of_string_and_int;

In C++11, templated typedefs are added with the following syntax, which requires the using keyword rather than the typedef keyword. (See alias templates.)

template <typename T>
using stringpair = std::pair<std::string, T>;
 
// Declare a variable of type std::pair<std::string, int>
stringpair<int> my_pair_of_string_and_int;

Other languages[edit]

In many statically typed functional languages, like Haskell, Miranda, OCaml, etc., one can define type synonyms, which are the same as typedefs in C. An example in Haskell:

type PairOfInts = (Int, Int)

This example has defined a type synonym PairOfInts which means the same as a pair of Ints.

In Swift, typedef is called typealias:

typealias PairOfInts = (Int, Int)

C# also contains a feature which is similar to the typedef of C.[5]

using newType = global::System.Runtime.Interop.Marshal;
using otherType = Enums.MyEnumType;
using StringListMap = System.Collections.Generic.Dictionary<string, System.Collections.Generic.List<string>>;

See also[edit]

References[edit]

  1. ^ Schildt, Herbert (1995). C : the complete reference (3rd ed. ed.). Berkeley, Calif.: Osborne McGraw-Hill. p. 198. ISBN 0-07-882101-0. Retrieved 12 September 2012. "C allows you to explicitly define new data type names by using the keyword typedef. You are not actually creating a new data type, but rather defining a new name for an existing type." 
  2. ^ Typedef as a Synonym, It allows you to introduce synonyms for types which could have been declared some other way.
  3. ^ Deitel, Paul J.; Deitel, H. M. (2007). C how to program (5th ed. ed.). Upper Saddle River, N.J.: Pearson Prentice Hall. ISBN 9780132404167. Retrieved 12 September 2012. "Names for structure types are often defined with typedef to create shorter type names." 
  4. ^ Kroah-Hartman, Greg (2002-07-01). "Proper Linux Kernel Coding Style". Linux Journal. Retrieved 2007-09-23. "Using a typedef only hides the real type of a variable." 
  5. ^ http://msdn.microsoft.com/en-us/library/aa664765(VS.71).aspx