How to achieve function overloading in C?

Question

Is there any way to achieve function overloading in C? I am looking at simple functions to be overloaded like

foo (int a)  
foo (char b)  
foo (float c , int d)

I think there is no straight forward way; I'm looking for workarounds if any exist.

Answer 1

Yes!

In the time since this question was asked, standard C (no extensions) has effectively gained support for function overloading (not operators), thanks to the addition of the _Generic keyword in C11. (supported in GCC since version 4.9)

(Overloading isn't truly "built-in" in the fashion shown in the question, but it's dead easy to implement something that works like that.)

_Generic is a compile-time operator in the same family as sizeof and _Alignof. It is described in standard section 6.5.1.1. It accepts two main parameters: an expression (which will not be evaluated at runtime), and a type/expression association list that looks a bit like a switch block. _Generic gets the overall type of the expression and then "switches" on it to select the end result expression in the list for its type:

_Generic(1, float: 2.0,
            char *: "2",
            int: 2,
            default: get_two_object());

The above expression evaluates to 2 - the type of the controlling expression is int, so it chooses the expression associated with int as the value. Nothing of this remains at runtime. (The default clause is optional: if you leave it off and the type doesn't match, it will cause a compilation error.)

The way this is useful for function overloading is that it can be inserted by the C preprocessor and choose a result expression based on the type of the arguments passed to the controlling macro. So (example from the C standard):

#define cbrt(X) _Generic((X),                \
                         long double: cbrtl, \
                         default: cbrt,      \
                         float: cbrtf        \
                         )(X)

This macro implements an overloaded cbrt operation, by dispatching on the type of the argument to the macro, choosing an appropriate implementation function, and then passing the original macro argument to that function.

So to implement your original example, we could do this:

foo_int (int a)  
foo_char (char b)  
foo_float_int (float c , int d)

#define foo(_1, ...) _Generic((_1),                                  \
                              int: foo_int,                          \
                              char: foo_char,                        \
                              float: _Generic((FIRST(__VA_ARGS__,)), \
                                     int: foo_float_int))(_1, __VA_ARGS__)
#define FIRST(A, ...) A

In this case we could have used a default: association for the third case, but that doesn't demonstrate how to extend the principle to multiple arguments. The end result is that you can use foo(...) in your code without worrying (much[1]) about the type of its arguments.

---

EDIT Cosinus has a much more elegant solution for multi-argument overloads that works with C23 or GNU extensions, the technique below was written against C11 (which didn't really want to let you do this).

For more complicated situations, e.g. functions overloading larger numbers of arguments, or varying numbers, you can use utility macros to automatically generate static dispatch structures:

void print_ii(int a, int b) { printf("int, int\n"); }
void print_di(double a, int b) { printf("double, int\n"); }
void print_iii(int a, int b, int c) { printf("int, int, int\n"); }
void print_default(void) { printf("unknown arguments\n"); }

#define print(...) OVERLOAD(print, (__VA_ARGS__), \
    (print_ii, (int, int)), \
    (print_di, (double, int)), \
    (print_iii, (int, int, int)) \
)

#define OVERLOAD_ARG_TYPES (int, double)
#define OVERLOAD_FUNCTIONS (print)
#include "activate-overloads.h"

int main(void) {
    print(44, 47);   // prints "int, int"
    print(4.4, 47);  // prints "double, int"
    print(1, 2, 3);  // prints "int, int, int"
    print("");       // prints "unknown arguments"
}

(implementation here) So with some effort, you can reduce the amount of boilerplate to looking pretty much like a language with native support for overloading.

As an aside, it was already possible to overload on the number of arguments (not the type) in C99.

---

[1] note that the way C evaluates types might trip you up though. This will choose foo_int if you try to pass it a character literal, for instance, and you need to mess about a bit if you want your overloads to support string literals. Still overall pretty cool though.

Answer 2

There are a few possibilities:

1. printf style functions (type as an argument);
2. OpenGL style functions (type in function name);
3. C subset of C++ (if You can use a C++ compiler).

Answer 3

As already stated, overloading in the sense that you mean isn't supported by C. A common idiom to solve the problem is making the function accept a tagged union. This is implemented by a struct parameter, where the struct itself consists of some sort of type indicator, such as an enum, and a union of the different types of values. Example:

#include <stdio.h>

typedef enum {
    T_INT,
    T_FLOAT,
    T_CHAR,
} my_type;

typedef struct {
    my_type type;
    union {
        int a; 
        float b; 
        char c;
    } my_union;
} my_struct;

void set_overload (my_struct *whatever) 
{
    switch (whatever->type) 
    {
        case T_INT:
            whatever->my_union.a = 1;
            break;
        case T_FLOAT:
            whatever->my_union.b = 2.0;
            break;
        case T_CHAR:
            whatever->my_union.c = '3';
    }
}

void printf_overload (my_struct *whatever) {
    switch (whatever->type) 
    {
        case T_INT:
            printf("%d\n", whatever->my_union.a);
            break;
        case T_FLOAT:
            printf("%f\n", whatever->my_union.b);
            break;
        case T_CHAR:
            printf("%c\n", whatever->my_union.c);
            break;
    }

}

int main (int argc, char* argv[])
{
    my_struct s;

    s.type=T_INT;
    set_overload(&s);
    printf_overload(&s);

    s.type=T_FLOAT;
    set_overload(&s);
    printf_overload(&s);

    s.type=T_CHAR;
    set_overload(&s);
    printf_overload(&s); 
}

Answer 4

Here is the clearest and most concise example I've found demonstrating function overloading in C:

#include <stdio.h>
#include <stdlib.h>
#include <string.h>

int addi(int a, int b) {
    return a + b;
}

char *adds(char *a, char *b) {
    char *res = malloc(strlen(a) + strlen(b) + 1);
    strcpy(res, a);
    strcat(res, b);
    return res;
}

#define add(a, b) _Generic(a, int: addi, char*: adds)(a, b)

int main(void) {
    int a = 1, b = 2;
    printf("%d\n", add(a, b)); // 3

    char *c = "hello ", *d = "world";
    printf("%s\n", add(c, d)); // hello world

    return 0;
}

https://gist.github.com/barosl/e0af4a92b2b8cabd05a7

Answer 5

If your compiler is gcc and you don't mind doing hand updates every time you add a new overload you can do some macro magic and get the result you want in terms of callers, it's not as nice to write... but it's possible

look at __builtin_types_compatible_p, then use it to define a macro that does something like

#define foo(a) \
((__builtin_types_compatible_p(int, a)?foo(a):(__builtin_types_compatible_p(float, a)?foo(a):)

but yea nasty, just don't

EDIT: C1X will be getting support for type generic expressions they look like this:

#define cbrt(X) _Generic((X), long double: cbrtl, \
                              default: cbrt, \
                              float: cbrtf)(X)

Answer 6

This may not help at all, but if you're using clang you can use the overloadable attribute - This works even when compiling as C

http://clang.llvm.org/docs/AttributeReference.html#overloadable

Header

extern void DecodeImageNow(CGImageRef image, CGContextRef usingContext) __attribute__((overloadable));
extern void DecodeImageNow(CGImageRef image) __attribute__((overloadable));

Implementation

void __attribute__((overloadable)) DecodeImageNow(CGImageRef image, CGContextRef usingContext { ... }
void __attribute__((overloadable)) DecodeImageNow(CGImageRef image) { ... }

Answer 7

Yes, sort of.

Here you go by example :

void printA(int a){
printf("Hello world from printA : %d\n",a);
}

void printB(const char *buff){
printf("Hello world from printB : %s\n",buff);
}

#define Max_ITEMS() 6, 5, 4, 3, 2, 1, 0 
#define __VA_ARG_N(_1, _2, _3, _4, _5, _6, N, ...) N
#define _Num_ARGS_(...) __VA_ARG_N(__VA_ARGS__) 
#define NUM_ARGS(...) (_Num_ARGS_(_0, ## __VA_ARGS__, Max_ITEMS()) - 1) 
#define CHECK_ARGS_MAX_LIMIT(t) if(NUM_ARGS(args)>t)
#define CHECK_ARGS_MIN_LIMIT(t) if(NUM_ARGS(args) 
#define print(x , args ...) \
CHECK_ARGS_MIN_LIMIT(1) printf("error");fflush(stdout); \
CHECK_ARGS_MAX_LIMIT(4) printf("error");fflush(stdout); \
({ \
if (__builtin_types_compatible_p (typeof (x), int)) \
printA(x, ##args); \
else \
printB (x,##args); \
})

int main(int argc, char** argv) {
    int a=0;
    print(a);
    print("hello");
    return (EXIT_SUCCESS);
}

It will output 0 and hello .. from printA and printB.

Answer 8

The following approach is similar to a2800276's, but with some C99 macro magic added:

// we need `size_t`
#include <stddef.h>

// argument types to accept
enum sum_arg_types { SUM_LONG, SUM_ULONG, SUM_DOUBLE };

// a structure to hold an argument
struct sum_arg
{
    enum sum_arg_types type;
    union
    {
        long as_long;
        unsigned long as_ulong;
        double as_double;
    } value;
};

// determine an array's size
#define count(ARRAY) ((sizeof (ARRAY))/(sizeof *(ARRAY)))

// this is how our function will be called
#define sum(...) _sum(count(sum_args(__VA_ARGS__)), sum_args(__VA_ARGS__))

// create an array of `struct sum_arg`
#define sum_args(...) ((struct sum_arg []){ __VA_ARGS__ })

// create initializers for the arguments
#define sum_long(VALUE) { SUM_LONG, { .as_long = (VALUE) } }
#define sum_ulong(VALUE) { SUM_ULONG, { .as_ulong = (VALUE) } }
#define sum_double(VALUE) { SUM_DOUBLE, { .as_double = (VALUE) } }

// our polymorphic function
long double _sum(size_t count, struct sum_arg * args)
{
    long double value = 0;

    for(size_t i = 0; i < count; ++i)
    {
        switch(args[i].type)
        {
            case SUM_LONG:
            value += args[i].value.as_long;
            break;

            case SUM_ULONG:
            value += args[i].value.as_ulong;
            break;

            case SUM_DOUBLE:
            value += args[i].value.as_double;
            break;
        }
    }

    return value;
}

// let's see if it works

#include <stdio.h>

int main()
{
    unsigned long foo = -1;
    long double value = sum(sum_long(42), sum_ulong(foo), sum_double(1e10));
    printf("%Le\n", value);
    return 0;
}

Answer 9

In the sense you mean — no, you cannot.

You can declare a va_arg function like

void my_func(char* format, ...);

, but you'll need to pass some kind of information about number of variables and their types in the first argument — like printf() does.

Answer 10

Leushenko's answer is really cool - solely: the foo example does not compile with GCC, which fails at foo(7), stumbling over the FIRST macro and the actual function call ((_1, __VA_ARGS__), remaining with a surplus comma. Additionally, we are in trouble if we want to provide additional overloads, such as foo(double).

So I decided to elaborate the answer a little further, including to allow a void overload (foo(void) – which caused quite some trouble...).

Idea now is: Define more than one generic in different macros and let select the correct one according to the number of arguments!

Number of arguments is quite easy, based on this answer:

#define foo(...) SELECT(__VA_ARGS__)(__VA_ARGS__)

#define SELECT(...) CONCAT(SELECT_, NARG(__VA_ARGS__))(__VA_ARGS__)
#define CONCAT(X, Y) CONCAT_(X, Y)
#define CONCAT_(X, Y) X ## Y

That's nice, we resolve to either SELECT_1 or SELECT_2 (or more arguments, if you want/need them), so we simply need appropriate defines:

#define SELECT_0() foo_void
#define SELECT_1(_1) _Generic ((_1),    \
        int: foo_int,                   \
        char: foo_char,                 \
        double: foo_double              \
)
#define SELECT_2(_1, _2) _Generic((_1), \
        double: _Generic((_2),          \
                int: foo_double_int     \
        )                               \
)

OK, I added the void overload already – however, this one actually is not covered by the C standard, which does not allow empty variadic arguments, i. e. we then rely on compiler extensions!

At very first, an empty macro call (foo()) still produces a token, but an empty one. So the counting macro actually returns 1 instead of 0 even on empty macro call. We can "easily" eliminate this problem, if we place the comma after __VA_ARGS__ conditionally, depending on the list being empty or not:

#define NARG(...) ARG4_(__VA_ARGS__ COMMA(__VA_ARGS__) 4, 3, 2, 1, 0)

That looked easy, but the COMMA macro is quite a heavy one; fortunately, the topic is already covered in a blog of Jens Gustedt (thanks, Jens). Basic trick is that function macros are not expanded if not followed by parentheses, for further explanations, have a look at Jens' blog... We just have to modify the macros a little to our needs (I'm going to use shorter names and less arguments for brevity).

#define ARGN(...) ARGN_(__VA_ARGS__)
#define ARGN_(_0, _1, _2, _3, N, ...) N
#define HAS_COMMA(...) ARGN(__VA_ARGS__, 1, 1, 1, 0)

#define SET_COMMA(...) ,

#define COMMA(...) SELECT_COMMA             \
(                                           \
        HAS_COMMA(__VA_ARGS__),             \
        HAS_COMMA(__VA_ARGS__ ()),          \
        HAS_COMMA(SET_COMMA __VA_ARGS__),   \
        HAS_COMMA(SET_COMMA __VA_ARGS__ ()) \
)

#define SELECT_COMMA(_0, _1, _2, _3) SELECT_COMMA_(_0, _1, _2, _3)
#define SELECT_COMMA_(_0, _1, _2, _3) COMMA_ ## _0 ## _1 ## _2 ## _3

#define COMMA_0000 ,
#define COMMA_0001
#define COMMA_0010 ,
// ... (all others with comma)
#define COMMA_1111 ,

And now we are fine...

The complete code in one block:

/*
 * demo.c
 *
 *  Created on: 2017-09-14
 *      Author: sboehler
 */

#include <stdio.h>

void foo_void(void)
{
    puts("void");
}
void foo_int(int c)
{
    printf("int: %d\n", c);
}
void foo_char(char c)
{
    printf("char: %c\n", c);
}
void foo_double(double c)
{
    printf("double: %.2f\n", c);
}
void foo_double_int(double c, int d)
{
    printf("double: %.2f, int: %d\n", c, d);
}

#define foo(...) SELECT(__VA_ARGS__)(__VA_ARGS__)

#define SELECT(...) CONCAT(SELECT_, NARG(__VA_ARGS__))(__VA_ARGS__)
#define CONCAT(X, Y) CONCAT_(X, Y)
#define CONCAT_(X, Y) X ## Y

#define SELECT_0() foo_void
#define SELECT_1(_1) _Generic ((_1), \
        int: foo_int,                \
        char: foo_char,              \
        double: foo_double           \
)
#define SELECT_2(_1, _2) _Generic((_1), \
        double: _Generic((_2),          \
                int: foo_double_int     \
        )                               \
)

#define ARGN(...) ARGN_(__VA_ARGS__)
#define ARGN_(_0, _1, _2, N, ...) N

#define NARG(...) ARGN(__VA_ARGS__ COMMA(__VA_ARGS__) 3, 2, 1, 0)
#define HAS_COMMA(...) ARGN(__VA_ARGS__, 1, 1, 0)

#define SET_COMMA(...) ,

#define COMMA(...) SELECT_COMMA             \
(                                           \
        HAS_COMMA(__VA_ARGS__),             \
        HAS_COMMA(__VA_ARGS__ ()),          \
        HAS_COMMA(SET_COMMA __VA_ARGS__),   \
        HAS_COMMA(SET_COMMA __VA_ARGS__ ()) \
)

#define SELECT_COMMA(_0, _1, _2, _3) SELECT_COMMA_(_0, _1, _2, _3)
#define SELECT_COMMA_(_0, _1, _2, _3) COMMA_ ## _0 ## _1 ## _2 ## _3

#define COMMA_0000 ,
#define COMMA_0001
#define COMMA_0010 ,
#define COMMA_0011 ,
#define COMMA_0100 ,
#define COMMA_0101 ,
#define COMMA_0110 ,
#define COMMA_0111 ,
#define COMMA_1000 ,
#define COMMA_1001 ,
#define COMMA_1010 ,
#define COMMA_1011 ,
#define COMMA_1100 ,
#define COMMA_1101 ,
#define COMMA_1110 ,
#define COMMA_1111 ,

int main(int argc, char** argv)
{
    foo();
    foo(7);
    foo(10.12);
    foo(12.10, 7);
    foo((char)'s');

    return 0;
}

Answer 11

Normally a wart to indicate the type is appended or prepended to the name. You can get away with macros is some instances, but it rather depends what you're trying to do. There's no polymorphism in C, only coercion.

Simple generic operations can be done with macros:

#define max(x,y) ((x)>(y)?(x):(y))

If your compiler supports typeof, more complicated operations can be put in the macro. You can then have the symbol foo(x) to support the same operation different types, but you can't vary the behaviour between different overloads. If you want actual functions rather than macros, you might be able to paste the type to the name and use a second pasting to access it (I haven't tried).

Answer 12

I think, I've just found a very elegant solution.

You could (ab)use function pointers to combine types.

_Generic(((void (*)(typeof(1.4) , typeof(9.9) , typeof(4) ))0),
  void (*)(int, double) : printf("int double\n"),
  void (*)(double, double) : printf("double double\n"),
  void (*)(double, double, int) : printf("double double int\n")
  );

With a little bit of macro magic ...

#define PARENS ()

#define EXPAND(...) EXPAND4(EXPAND4(EXPAND4(EXPAND4(__VA_ARGS__))))
#define EXPAND4(...) EXPAND3(EXPAND3(EXPAND3(EXPAND3(__VA_ARGS__))))
#define EXPAND3(...) EXPAND2(EXPAND2(EXPAND2(EXPAND2(__VA_ARGS__))))
#define EXPAND2(...) EXPAND1(EXPAND1(EXPAND1(EXPAND1(__VA_ARGS__))))
#define EXPAND1(...) __VA_ARGS__

#define FOR_EACH(macro, ...)                                    \
  __VA_OPT__(EXPAND(FOR_EACH_HELPER(macro, __VA_ARGS__)))

#define FOR_EACH_HELPER(macro, a1, ...)                         \
  macro(a1)                                                     \
  __VA_OPT__(, FOR_EACH_AGAIN PARENS (macro, __VA_ARGS__))

#define FOR_EACH_AGAIN() FOR_EACH_HELPER




#define mGROUP_TYPE(...)      void (*)(__VA_ARGS__)
#define mGROUP_TYPE_INST(...) ((void (*)(FOR_EACH(typeof, __VA_ARGS__)))0)

... you can do very beautiful things:

#define myfoo(...) \
  _Generic(mGROUP_TYPE_INST(__VA_ARGS__), \
      mGROUP_TYPE(int, double) : printf("int double\n"), \
      mGROUP_TYPE(double, double) : printf("double double\n"), \
      mGROUP_TYPE(double, double, int) : printf("double double int\n"), \
      mGROUP_TYPE(int, double, int) : printf("int double int\n") \
      )

myfoo(1, 0.4);
myfoo(1, 0.4, 4);

In case the return value is always the same you could even go a step further and use actual function pointers:

#define mFUNC_CASE(fn) typeof(&(fn)) : fn

#define mFUNC_GENERIC_INST(ret, ...) (ret (*) (FOR_EACH(typeof, __VA_ARGS__)))0


#define to_string(...) \
  _Generic(mFUNC_GENERIC_INST(char *, __VA_ARGS__), \
      mFUNC_CASE(int_to_string), \
      mFUNC_CASE(float_to_string), \
      mFUNC_CASE(file_to_string), \
      mFUNC_CASE(vec3_to_string) \
      ) (__VA_ARGS__)

---

The FOR_EACH() macro originated from here (with a small modifications). There are also other variants of this macro around, that do not need the __VA_OPT__. In case your compiler doesn't support that.

Answer 13

Can't you just use C++ and not use all other C++ features except this one?

If still no just strict C then I would recommend variadic functions instead.

Answer 14

An object-oriented way to emulate function overloading in C that doesn't rely on using preprocessor macros consists of creating a struct for each parameter list that you want your function to accept, and making them all "inherit" from the same parent struct. This "parent" struct will have as a member a function pointer representing the function to emulate. Upon the creation of a child struct this member will be initialized to a corresponding function implementation.

For instance, say that you want to emulate the following behaviour:

int add(int n, int m);
int add(double x, int k, double z);

First, you'll start by creating a struct abstracting whatever is that the different parameter lists represent:

struct summable_t {
    int (*add)(struct summable_t *summable);
};

Then, you create a different struct for each parameter list:

struct summable1_t {
   struct summable_t parent;
   int n;
   int m;
}

struct summable2_t {
   struct summable_t parent;
   double x;
   int k;
   double z;
}

After this, you write a concrete implementation of the function for each child struct (I'll only write the first one to showcase the need for casting):

int add1(struct summable_t *summable) {
    int n = ((struct summable1_t *) summable)->n;
    int m = ((struct summable1_t *) summable)->m;
    return n + m;
};

int add2(struct summable_t *summable) {// some instructions};

Then you glue everything together by defining a "constructor" for each struct:

struct summable_t* new_summable1(int n, int m) {
    struct summable1_t* summable1 = malloc(sizeof(struct summable1_t));
    summable1->n = n;
    summable1->m = m;
    summable1->parent.add = add1;
    return (struct summable_t*) summable1;
};

struct summable_t* new_summable2(double x, int k, double z) {
    struct summable2_t* summable1 = malloc(sizeof(struct summable2_t));
    summable2->x = x;
    summable2->k = k;
    summable2->z = z;

    summable2->parent.add = add2;

    return (struct summable_t*) summable2;
};

Finally, you write a function accepting the parent struct and calling the associated callback:

int add(struct summable_t* summable) {
    return summable->add(summable);
}

In userland, you could consume the above as follows:

int main(int argc, char** argv) {
    struct summable_t* summable1 = new_summable1(1, 2);
    struct summable_t* summable2 = new_summable2(0.5, 2, 0.9);

    int res1 = add(summable1);
    int res2 = add(summable2);

    free(summable1);
    free(summable2);

    return 0;
}

Answer 15

Try to declare these functions as extern "C++" if your compiler supports this, http://msdn.microsoft.com/en-us/library/s6y4zxec(VS.80).aspx

Answer 16

I hope the below code will help you to understand function overloading

#include <stdio.h>
#include<stdarg.h>

int fun(int a, ...);
int main(int argc, char *argv[]){
   fun(1,10);
   fun(2,"cquestionbank");
   return 0;
}
int fun(int a, ...){
  va_list vl;
  va_start(vl,a);

  if(a==1)
      printf("%d",va_arg(vl,int));
   else
      printf("\n%s",va_arg(vl,char *));
}