C++ Primer Plus

To terminate this loop early, you enter a negative number. This restricts input to non-negative values. This restriction fits the needs of some programs, but more typically you would want a means of terminating a loop that doesn’t exclude certain numeric values. Using cin >> as the test condition eliminates such restrictions because it accepts all valid numeric input. You should keep this trick in mind when you need an input loop for numbers. Also, you should keep in mind that non-numeric input sets an error condition that prevents the reading of any more input. If a program needs input subsequent to the input loop, you must use cin.clear() to reset input, and you might then need to get rid of the offending input by reading it. Listing 7.7 illustrates those techniques.

Passing Structure Addresses

Suppose you want to save time and space by passing the address of a structure instead of passing the entire structure. This requires rewriting the functions so that they use pointers to structures. First, let’s look at how you rewrite the show_polar() function. You need to make three changes:

• When calling the function, pass it the address of the structure (&pplace) rather than the structure itself (pplace).

• Declare the formal parameter to be a pointer-to-polar—that is, type polar *. Because the function shouldn’t modify the structure, use the const modifier.

• Because the formal parameter is a pointer instead of a structure, use the indirect membership operator (->) rather than the membership operator (dot).

After these changes are made, the function looks like this:

// show polar coordinates, converting angle to degrees

void show_polar (const polar * pda)

{

    using namespace std;

    const double Rad_to_deg = 57.29577951;

    cout << "distance = " << pda->distance;

    cout << ", angle = " << pda->angle * Rad_to_deg;

    cout << " degrees\n";

}

Next, let’s alter rect_to_polar. This is more involved because the original rect_to_polar function returns a structure. To take full advantage of pointer efficiency, you should use a pointer instead of a return value. The way to do this is to pass two pointers to the function. The first points to the structure to be converted, and the second points to the structure that’s to hold the conversion. Instead of returning a new structure, the function modifies an existing structure in the calling function. Hence, although the first argument is const pointer, the second is not const. Otherwise, you apply the same principles used to convert show_polar() to pointer arguments. Listing 7.13 shows the reworked program.

Listing 7.13. strctptr.cpp

// strctptr.cpp -- functions with pointer to structure arguments

#include

#include

// structure templates

struct polar

{

    double distance;      // distance from origin

    double angle;         // direction from origin

};

struct rect

{

    double x;             // horizontal distance from origin

    double y;             // vertical distance from origin

};

// prototypes

void rect_to_polar(const rect * pxy, polar * pda);

void show_polar (const polar * pda);

int main()

{

    using namespace std;

    rect rplace;

    polar pplace;

    cout << "Enter the x and y values: ";

    while (cin >> rplace.x >> rplace.y)

    {

        rect_to_polar(&rplace, &pplace);    // pass addresses

        show_polar(&pplace);        // pass address

        cout << "Next two numbers (q to quit): ";

    }

    cout << "Done.\n";

    return 0;

}

// show polar coordinates, converting angle to degrees

void show_polar (const polar * pda)

{

    using namespace std;

    const double Rad_to_deg = 57.29577951;

    cout << "distance = " << pda->distance;

    cout << ", angle = " << pda->angle * Rad_to_deg;

    cout << " degrees\n";

}

// convert rectangular to polar coordinates

void rect_to_polar(const rect * pxy, polar * pda)

{

    using namespace std;

    pda->distance =

        sqrt(pxy->x * pxy->x + pxy->y * pxy->y);

    pda->angle = atan2(pxy->y, pxy->x);

}

Note

Some compilers require explicit instructions to search the math library. For example, older versions of g++ use this command line:

g++ structfun.C -lm

From the user’s standpoint, the program in Listing 7.13 behaves like that in Listing 7.12. The hidden difference is that Listing 7.12 works with copies of structures, whereas Listing 7.13 uses pointers, allowing the functions to operate on the original structures.

Functions and string Class Objects

Although C-style strings and string class objects serve much the same purpose, a string class object is more closely related to a structure than to an array. For example, you can assign a structure to another structure and an object to another object. You can pass a structure as a complete entity to a function, and you can pass an object as a complete entity. If you need several strings, you can declare a one-dimensional array of string objects instead of a two-dimensional array of char.

Listing 7.14 provides a short example that declares an array of string objects and passes the array to a function that displays the contents.