C Programming/Procedures and functions

In C programming, all executable code resides within a function. Note that other programming languages may distinguish between a "function", "subroutine", "subprogram", "procedure", or "method" -- in C, these are all functions. Functions are a fundamental feature of any high level programming language and make it possible to tackle large, complicated tasks by breaking tasks into smaller, more manageable pieces of code.

At a lower level, a function is nothing more than a memory address where the instructions associated with a function reside in your computer's memory. In the source code, this memory address is usually given a descriptive name which programmers can use to call the function and execute the instructions that begin at the function's starting address. The instructions associated with a function are frequently referred to as a block of code. After the function's instructions finish executing, the function can return a value and code execution will resume with the instruction that immediately follows the initial call to the function. If this doesn't make immediate sense to you, don't worry. Understanding what is happening inside your computer at the lowest levels can be confusing at first, but will eventually become very intuitive as you develop your C programming skills.

For now, it's enough to know that a function and its associated block of code is often executed (called) several times, from several different places, during a single execution of a program.

As a basic example, suppose you are writing a program that calculates the distance of a given (x,y) point to the x-axis and to the y-axis. You will need to compute the absolute value of the whole numbers x and y. We could write it like this (assuming we don't have a predefined function for absolute value in any library):

#include <stdio.h>

/*this function computes the absolute value of a whole number.*/
int abs(int x)
{
        if (x>=0) return x;
        else return -x;
}

/*this program calls the abs() function defined above twice.*/
int main()
{
        int x, y;
        
        printf("Type the coordinates of a point in 2-plane, say P = (x,y). First x=");
        scanf("%d", &x);
        printf("Second y=");
        scanf("%d", &y);
        
        printf("The distance of the P point to the x-axis is %d. \n Its distance to the y-axis is %d. \n", abs(y), abs(x));

        return 0;
}

The next example illustrates the usage of a function as a procedure. It's a simplistic program that asks students for their grade for three different courses and tells them if they passed a course. Here, we created a function, called check() that can be called as many times as we need to. The function saves us from having to write the same set of instructions for each class the student has taken.

#include<stdio.h>

/*the 'check' function is defined here.*/
void check(int x)
{
        if (x<60)
                printf("Sorry! You will need to try this course again.\n");
        else
                printf("Enjoy your vacation! You have passed.\n");
}

/*the program starts here at the main() function, which calls the check() function three times.*/
int main()
{
        int a, b, c;
        
        printf("Type your grade in Mathematics (whole number). \n");
        scanf("%d", &a);
        check(a);
        
        printf("Type your grade in Science (whole number). \n");
        scanf("%d", &b);
        check(b);
        
        printf("Type your grade in Programming (whole number). \n");
        scanf("%d", &c);
        check(c);

        /* this program should be replaced by something more meaningful.*/
        return 0;
}

Notice that in the program above, there is no outcome value for the 'check' function. It only executes a procedure.

This is precisely what functions are for.

More on functions

It's useful to conceptualize a function like a machine in a factory. On the input side of the machine, you dump in the "raw materials," or the input data, that you want the machine to process. Then the machine goes to work and and spits out a finished product, the "return value," to the output side of the machine which you can collect and use for other purposes.

In C, you must tell the machine exactly what raw materials it is expected to process and what kind of finished product you want the machine to return to you. If you supply the machine with different raw materials than it expects, or if you try to return a product that's different than what you told the machine to produce, the C compiler will throw an error.

Note that a function isn't required to take any inputs. It doesn't have to return anything back to us, either. If we modify the example above to ask the user for their grade inside the check function, there would be no need to pass the grade value into the function. And notice that the check doesn't pass a value back. The function just prints out a message to the screen.

You should be familiar with some basic terminology related to functions:

• A function, call it f, that uses another function g, is said to call g. For example, f calls g to print the squares of ten numbers. f is referred to as the caller function and g is the callee.
• The inputs we send to a function are called its arguments. When we declare our function, we describe the parameters that determine what type of arguments are acceptable to pass into the function. We describe these parameters to the compiler inside a set of parentheses next to the function's name.
• A function g that gives some kind of answer back to f is said to return that answer or value. For example, g returns the sum of its arguments.

Writing functions in C

It's always good to learn by example. Let's write a function that will return the square of a number.

int square(int x)
{
   int square_of_x;
   square_of_x = x * x;
   return square_of_x;
}

To understand how to write such a function like this, it may help to look at what this function does as a whole. It takes in an int, x, and squares it, storing it in the variable square_of_x. Now this value is returned.

The first int at the beginning of the function declaration is the type of data that the function returns. In this case when we square an integer we get an integer, and we are returning this integer, and so we write int as the return type.

Next is the name of the function. It is good practice to use meaningful and descriptive names for functions you may write. It may help to name the function after what it is written to do. In this case we name the function "square", because that's what it does - it squares a number.

Next is the function's first and only argument, an int, which will be referred to in the function as x. This is the function's input.

In between the braces is the actual guts of the function. It declares an integer variable called square_of_x that will be used to hold the value of the square of x. Note that the variable square_of_x can only be used within this function, and not outside. We'll learn more about this sort of thing later, and we will see that this property is very useful.

We then assign x multiplied by x, or x squared, to the variable square_of_x, which is what this function is all about. Following this is a return statement. We want to return the value of the square of x, so we must say that this function returns the contents of the variable square_of_x.

Our brace to close, and we have finished the declaration.

Written in a more concise manner, this code performs exactly the same function as the above:

int square(int x)
{
   return x * x;
}

Note this should look familiar - you have been writing functions already, in fact - main is a function that is always written.

In general

In general, if we want to declare a function, we write

 type name(type1 arg1, type2 arg2, ...)
 {
   /* code */
 } 

We've previously said that a function can take no arguments, or can return nothing, or both. What do we write if we want the function to return nothing? We use C's void keyword. void basically means "nothing" - so if we want to write a function that returns nothing, for example, we write

void sayhello(int number_of_times)
{
  int i;
  for(i=1; i <= number_of_times; i++) {
     printf("Hello!\n");
  }
}

Notice that there is no return statement in the function above. Since there's none, we write void as the return type. (Actually, one can use the return keyword in a procedure to return to the caller before the end of the procedure, but one cannot return a value as if it were a function.)

What about a function that takes no arguments? If we want to do this, we can write for example

float calculate_number(void)
{
  float to_return=1;
  int i;
  for(i=0; i < 100; i++) {
     to_return += 1;
     to_return = 1/to_return;
  }
  return to_return;
}

Notice this function doesn't take any inputs, but merely returns a number calculated by this function.

Naturally, you can combine both void return and void in arguments together to get a valid function, also.

Recursion

Here's a simple function that does an infinite loop. It prints a line and calls itself, which again prints a line and calls itself again, and this continues until the stack overflows and the program crashes. A function calling itself is called recursion, and normally you will have a conditional that would stop the recursion after a small, finite number of steps.

      // don't run this!
void infinite_recursion()
{
    printf("Infinite loop!\n");
    infinite_recursion();
}

A simple check can be done like this. Note that ++depth is used so the increment will take place before the value is passed into the function. Alternatively you can increment on a separate line before the recursion call. If you say print_me(3,0); the function will print the line Recursion 3 times.

void print_me(int j, int depth)
{
   if(depth < j) {
       printf("Recursion! depth = %d j = %d\n",depth,j); //j keeps its value
       print_me(j, ++depth);
   }
}

Recursion is most often used for jobs such as directory tree scans, seeking for the end of a linked list, parsing a tree structure in a database and factorising numbers (and finding primes) among other things.

Static functions

If a function is to be called only from within the file in which it is declared, it is appropriate to declare it as a static function. When a function is declared static, the compiler will know to compile an object file in a way that prevents the function from being called from code in other files. Example:

static int compare( int a, int b )
{
    return (a+4 < b)? a : b;
}

Using C functions

We can now write functions, but how do we use them? When we write main, we place the function outside the braces that encompass main.

When we want to use that function, say, using our calculate_number function above, we can write something like

 float f;
 f = calculate_number();

If a function takes in arguments, we can write something like

 int square_of_10;
 square_of_10 = square(10);

If a function doesn't return anything, we can just say

 say_hello();

since we don't need a variable to catch its return value.

Functions from the C Standard Library

While the C language doesn't itself contain functions, it is usually linked with the C Standard Library. To use this library, you need to add an #include directive at the top of the C file, which may be one of the following from C89/C90:

<assert.h> <ctype.h> <errno.h> <float.h>

<limits.h> <locale.h> <math.h> <setjmp.h>

<signal.h> <stdarg.h> <stddef.h> <stdio.h>

<stdlib.h> <string.h> <time.h>

The functions available are:

<assert.h>	<limits.h>	<signal.h>	<stdlib.h>
assert(int)	(constants only)	int raise(int sig). This void* signal(int sig, void (*func)(int))	atof(char*), atoi(char*), atol(char*) strtod(char * str, char ** endptr ), strtol(char *str, char **endptr), strtoul(char *str, char **endptr) rand(), srand() malloc(size_t), calloc (size_t elements, size_t elementSize), realloc(void*, int) free (void*) exit(int), abort() atexit(void (*func)()) getenv system qsort(void *, size_t number, size_t size, int (*sortfunc)(void*, void*)) abs, labs div, ldiv
<ctype.h>	<locale.h>	<stdarg.h>	<string.h>
isalnum, isalpha, isblank iscntrl, isdigit, isgraph islower, isprint, ispunct isspace, isupper, isxdigit tolower, toupper	struct lconv* localeconv(void); char* setlocale(int, const char*);	va_start (va_list, ap) va_arg (ap, (type)) va_end (ap) va_copy (va_list, va_list)	memcpy, memmove memchr, memcmp, memset strcat, strncat, strchr, strrchr strcmp, strncmp, strccoll strcpy, strncpy strerror strlen strspn, strcspn strpbrk strstr strtok strxfrm
errno.h	math.h	stddef.h	time.h
(errno)	sin, cos, tan asin, acos, atan, atan2 sinh, cosh, tanh ceil exp fabs floor fmod frexp ldexp log, log10 modf pow sqrt	offsetof macro	asctime (struct tm* tmptr) clock_t clock() char* ctime(const time_t* timer) double difftime(time_t timer2, time_t timer1) struct tm* gmtime(const time_t* timer) struct tm* gmtime_r(const time_t* timer, struct tm* result) struct tm* localtime(const time_t* timer) time_t mktime(struct tm* ptm) time_t time(time_t* timer) char * strptime(const char* buf, const char* format, struct tm* tptr) time_t timegm(struct tm *brokentime)
float.h	setjmp.h	stdio.h
(constants)	int setjmp(jmp_buf env) void longjmp(jmp_buf env, int value)	fclose fopen, freopen remove rename rewind tmpfile clearerr feof, ferror fflush fgetpos, fsetpos fgetc, fputc fgets, fputs ftell, fseek	fread, fwrite getc, putc getchar, putchar, fputchar gets, puts printf, vprintf fprintf, vfprintf sprintf, snprintf, vsprintf, vsnprintf perror scanf, vscanf fscanf, vfscanf sscanf, vsscanf setbuf, setvbuf tmpnam ungetc

Variable-length argument lists

Functions with variable-length argument lists are functions that can take a varying number of arguments. An example in the C standard library is the printf function, which can take any number of arguments depending on how the programmer wants to use it.

C programmers rarely find the need to write new functions with variable-length arguments. If they want to pass a bunch of things to a function, they typically define a structure to hold all those things -- perhaps a linked list, or an array -- and call that function with the data in the arguments.

However, you may occasionally find the need to write a new function that supports a variable-length argument list. To create a function that can accept a variable-length argument list, you must first include the standard library header stdarg.h. Next, declare the function as you would normally. Next, add as the last argument an ellipsis ("..."). This indicates to the compiler that a variable list of arguments is to follow. For example, the following function declaration is for a function that returns the average of a list of numbers:

  float average (int n_args, ...);

Note that because of the way variable-length arguments work, we must somehow, in the arguments, specify the number of elements in the variable-length part of the arguments. In the average function here, it's done through an argument called n_args. In the printf function, it's done with the format codes that you specify in that first string in the arguments you provide.

Now that the function has been declared as using variable-length arguments, we must next write the code that does the actual work in the function. To access the numbers stored in the variable-length argument list for our average function, we must first declare a variable for the list itself:

  va_list myList;

The va_list type is a type declared in the stdarg.h header that basically allows you to keep track of your list. To start actually using myList, however, we must first assign it a value. After all, simply declaring it by itself wouldn't do anything. To do this, we must call va_start, which is actually a macro defined in stdarg.h. In the arguments to va_start, you must provide the va_list variable you plan on using, as well as the name of the last variable appearing before the ellipsis in your function declaration:

#include <stdarg.h>
float average (int n_args, ...)
{
    va_list myList;
    va_start (myList, n_args);
    va_end (myList);
}

Now that myList has been prepped for usage, we can finally start accessing the variables stored in it. To do so, use the va_arg macro, which pops off the next argument on the list. In the arguments to va_arg, provide the va_list variable you're using, as well as the primitive data type (e.g. int, char) that the variable you're accessing should be:

#include <stdarg.h>
float average (int n_args, ...)
{
    va_list myList;
    va_start (myList, n_args);
    
    int myNumber = va_arg (myList, int);
    va_end (myList);
}

By popping n_args integers off of the variable-length argument list, we can manage to find the average of the numbers:

#include <stdarg.h>
float average (int n_args, ...)
{
    va_list myList;
    va_start (myList, n_args);
    
    int numbersAdded = 0;
    int sum = 0;
     
    while (numbersAdded < n_args) {
        int number = va_arg (myList, int); // Get next number from list
        sum += number;
        numbersAdded += 1;
    }
    va_end (myList);
     
    float avg = (float)(sum) / (float)(numbersAdded); // Find the average
    return avg;
}

By calling average (2, 10, 20), we get the average of 10 and 20, which is 15.