Figure out my C++ program's problem. YOU CAN FIND THE MISSING PENNY!

[aKobrakai]

I have a program in C++ that lets users enter in the amount of a purchase and the amount they paid. It will then calculate their change in dollars, quarters, dimes, nickels and pennies. Here it is:

PHP Code:

* 
Programmer: Joseph Malicke 
Date: 06-10-05 
Exercise: 2 
Version: 1 
Problem: Calculate change ($) as long as user wants 
*/ 

#include <iostream> 
#include <conio.h> // for holding DOS window 

using namespace std; 

// function calculates coins from the change 
void calculateCoins(double theChange); 
// function prints 
void printData(); 

// define variables to hold user input 
double costOfPurchase, amountPaid;     
// holds change 
double change; 
// hold loop control 
char loopControl = 'Y'; 
// hold change by coins 
int changeDollar, changeQuarter, changeDime, changeNickel, changePenny; 
     
int main() 
{ 
    // format cout to 2 places 
    cout.setf(ios::fixed); 
    cout.setf(ios::showpoint); 
    cout.precision(2); 
     
    do 
    { 
          // collect the data 
          cout << "Please enter the cost of purchase, a space, and the amount paid.\n"; 
          cin >> costOfPurchase >> amountPaid; 
          // calculate change 
          change = amountPaid - costOfPurchase; 
          // pass change on to calculate coins 
          calculateCoins(change); 
          // now print 
          printData(); 
          // ask whether user wants to continue 
          cout << "Would you like to enter more data?\n"; 
          cin >> loopControl; 
          cout << endl; 
    } while (loopControl == 'Y' || loopControl == 'y'); 
     
    // Hold DOS window 
    cout << "\nHit any key to exit:\n"; 
    getch(); 
     
    return 0; 
} 

void calculateCoins(double theChange) 
{ 
         // first calculate how many dollars can go into it 
         // we don't care about fractional parts 
         changeDollar = static_cast<int>(theChange / 1); 
         // now subtract those dollars 
         theChange = theChange - (1.00 * changeDollar); 
         // calculate for quarters 
         changeQuarter = static_cast<int>(theChange / .25); 
         // now subtract those quarters 
         theChange = theChange - (.25 * changeQuarter); 
         // calculate for dimes 
         changeDime = static_cast<int>(theChange / .10); 
         // now subtract those dimes 
         theChange = theChange - (.10 * changeDime); 
         // calculate for nickels 
         changeNickel = static_cast<int>(theChange / .05); 
         // now subtract those nickels 
         theChange = theChange - (.05 * changeNickel); 
         // calculate for pennies 
         changePenny = static_cast<int>(theChange / .01); 
         // now subtract those pennies 
         theChange = theChange - (.01 * changePenny); 
} 

void printData() 
{ 
     cout << "The cost of purchase was $" << costOfPurchase << " and the paid amount was $" << amountPaid << endl; 
     cout << "Your change is $" << change << endl; 
     cout << "You should recieve..." << endl; 
     cout << changeDollar << " dollars, " << changeQuarter << " quarters, " << changeDime << " dimes, " << changeNickel << " nickels, " << changePenny << " pennies.\n" << endl; 
      
} 


So, for example, let's say the change is 2.76.

2.75 / 1 = 2.75, casted to an int, drops the decimal
changeDollar = 2
2.75 - (1.00 * 2) = .76

So my algorithm broke down the dollars.

.76 / .25 = 3.04, casted to an int, drops the decimal
changeQuarter = 3
.76 - (.25 * 3) = .1

Now we're down to one penny.

It'll continue to do this until it gets to the pennies and eventually finishes. Now, for all my
test data, the output is logically correct. HOWEVER, when I enter 50.00 for the purchase
amount and 50.01 for the amount paid, the result says this:

Your change is $0.01
You should be paid in...
0 dollars, 0 quarters, 0 dimes, 0 nickels, 0 pennies

It's off by a penny. Why??? It works when I do total 100 and 100.01 paid. In fact, it
even works when I enter 20 and 20.01 paid. I'm almost tempted to write a loop to try
every value up to 500 to see how many are off by a penny.

What's going on? :O

[alfie2]

C++ floating point is incapable of exactly representing most decimal fractions, regardless of the choice of precision
Use a BCD library( Binary Coded Decimal)

we, engineers have a saying: "precision arithmetic calculations are best done by FORTRAN( computer language/compiler)"
and "dont subtract two number of similar magnitude."

[james8547]

Even the program we use at work (Portfolio Accounting System) is off a penny some of the time (sometimes even up to 15 cents). We use 6-8 decimal places. We deal with billions of dollars and we usually make an adjusting entry to reconcile it out.

I also had the same problem back then when I was taking an Assembly class.

[alfie2]

Originally Posted by james8547 Even the program we use at work (Portfolio Accounting System) is off a penny some of the time (sometimes even up to 15 cents). We use 6-8 decimal places. We deal with billions of dollars and we usually make an adjusting entry to reconcile it out. I also had the same problem back then when I was taking an Assembly class.

My firm uses a "string library" (two's complement) to handle precision, its accurate to 127 decimal places; I hate to think that financial institutions use any thing worse. These "library/objects" can be brought in the open market cheap.