CSC 161 Grinnell College Spring, 2013 Imperative Problem Solving and Data Structures

# Function Pointers

As with Scheme, C allows variables to refer to functions, and functions can be passed as parameters. However, since C requires the type of each variable or parameter be declared, variables that may refer to functions must indicate the function's signature (number and type of parameters, return type).

The remainder of this reading presents an extended example that illustrates:

## Initial Example

Suppose you are asked to write a program to compute a circle's circumference and area for radii between 0 and 9. The desired output might be:

```      radius   circumference   area
0.0000      0.0000      0.0000
1.0000      6.2832      3.1416
2.0000     12.5664     12.5664
3.0000     18.8496     28.2743
4.0000     25.1327     50.2655
5.0000     31.4159     78.5398
6.0000     37.6991    113.0973
7.0000     43.9823    153.9380
8.0000     50.2655    201.0619
9.0000     56.5487    254.4690
```

One simple approach for writing this type of program might utilize these features:

• Separate functions are defined for the circumference and the area of a circle.
• The first line of the main program prints a header.
• A main loop iterates through the desired radius values.
• The loop itself contains three printf states, one for each value to be printed on a given line.

The resulting C program follows:

```/* program to compute a circle's circumference and area
*/

# include <stdio.h>

const double pi = 3.1415926535;

/* circumference function */
{
return 2 * pi * radius;
}

/* area function */
{
}

int main ()
{
{
printf ("\n");
}
return 0;
}
```

## Function Parameters

The above C program works fine and likely is completely satisfactory for the simple problem given. However, several elements in the program contain some common elements. For example,

• The first three printf statements all use the same format string, and they each print just one value.
• The second and third printf statements call a function, the function requires one parameter, and radius is used in the function call.
• Arguably, the first printf statement also has this format, where the function called is the identify function x = iden(x)

Although these common elements are hardly earth shaking, it can be helpful to take advantage of such common elements in more complicated programs.

### Printing Function

In exploiting the similarities in the printf statements, we might write a function that takes a radius and a function as parameters and then performs the required printing. The following code is an example:

```/* printing function */
double myPrint (double x, double f (double))
{
printf ("%12.4lf", f(x));
}
```

This function has two formal parameters, the number x and a function f. Further, f is identified as a function that will utilize one double value as an input parameter, and f will return a double when it completes execution.

The function myPrint utilizes the function f when it performs its printing.

### Full Program

The following program replaces the printf statements in the main loop by calls to myPrint.

Note that myPrint is also used for printing the value of radius by using an identify function. The code observes, however, that a simple printf for radius could be used as a reasonable alternative.

```/* program to compute a circle's circumference and area

example using function with a function parameter
*/

# include <stdio.h>

const double pi = 3.1415926535;

/* identity function */
{
}

/* circumference function */
{
return 2 * pi * radius;
}

/* area function */
{
}

/* printing function */
void myPrint (double x, double f (double))
{
printf ("%12.4lf", f(x));
}

int main ()
{
{
/* could also be
*/
printf ("\n");
}
return 0;
}
```

## Arrays of Functions

As noted earlier in this lab, functions as parameters provide one mechanism to take advantage of common elements within an algorithm. A second approach involves utilizing an array of functions.

### Declaration

Already, we have observed that the main loop in our example utilizes the same printf statement — with different functions being called. This suggests that we might declare an array of the relevant functions:

```  double (*funcarr[3]) (double)= {iden, circum, area};
```

Let's unpack this syntax:

• We can declare a simple variable as a pointer to a function with the statement:

```   double (*f) (double)
```

This states that f will refer to a function that has one parameter (a double) and it will return a double. (The * means that f will be a function pointer — a reference to a function.)

• With this declaration, f can be assigned a function:

```f = area;
```

so the statement

```
printf ("%5.2", f(3.5));
```

will cause the area function to be called, executed with the value 3.5, and used as the value to be printed.

• As with many C declarations, a variable can be defined and initialized in a single statement. The following declaration initializes f as a reference to the area function.

```   double (*f) (double) = area;
```
• Arrays are declared by placing [] within the declaration:

```  double (*funcarr[3]) (double);
```

Here funcarr is declared as an array of 3 function pointers.

• Arrays can be initialized by placing desired values within braces { }:
```  double (*funcarr[3]) (double)= {iden, circum, area};
```
• With this declaration, we can refer to the circum function as

```funcarr[1]
```

and we can call this function with the parameter 3.5 by writing

```funcarr[1] (3.5)
```

### Full Program

The following program uses this declaration of funcarr in the main loop by iterating through the three functions of interest for printing.

```/* program to compute a circle's circumference and area

example using an array of functions
*/

# include <stdio.h>

const double pi = 3.1415926535;

/* identity function */
{
}

/* circumference function */
{
return 2 * pi * radius;
}

/* area function */
{
}

int main ()
{
double (*funcarr[3]) (double) = {iden, circum, area};
{
int i;
for (i = 0; i < 3; i++)
{