I was making a little program to test floats in the C: The program itself is very simple, I just want to based on the User's input, return how much Dollar(s), Quarter(s)... etc, his number has.

//------------------------------> First Part: All the necessary Variables <-----------------------------
 
int main (void)
{
    //Getting the user Input
    float number = get_float("Number: ");
   
    //Checking if is a positive number
    if (number < 0)
    {
        printf("A positive number, Please: ");
    }

    //Declaring my Constant/Temporary variables.
    float coinValues[] = {1.00, 0.25, 0.10, 0.5, 0.01};
    char  *coinNames[] = {"Dollar(s): ", "Quarter(s): ", "Dime(s): ", "Nickel(s): ", "Penny(ies): "};
    int   i            = 0;
    int   tmp          = 0;
    
//-----------------------------------> Second Part: The code Itself <-----------------------------------

    //Checking/Printing the necessary coins.
    while (number > 0)
    {
        //Until the loop stops, check if the number can be divided by the CoinValue.
        if (number >= coinValues[i])
        {
            //Print the current Coin Name from the divided value.
            printf("%s", coinNames[i]);
            //Check if the Current Number still contains Coin Values inside of it, if True counts one in your "Coin Score".
            while (number >= coinValues[i])
            {
                number -= coinValues[i];
                tmp  ;
            }
            //Print the Current "Coin Score", then resets TMP.
            printf("%i\n", tmp);
            tmp = 0;

        }
        else
        {   
            //Updating the Coin value
            i  ;
        }

    }
    
}

My program was running very well as long I use Integers, but when I converted this code for Floats the values (Dime(s), Nickel(s), and Penny(ies)) start to return non-expected results in the Int variable tmp.

An expected result for a number like 2.6, will be 2 Dollars, 2 Quarters, and 1 Dime, but sometimes, instead of use the Dime(s), the program skips them entire and make the operation with the Nickel(s), however, what is bugging me is that the program is always returning AWL= without any value and then the program stay froze forever.

Considering that my only thought is that I'm "suffering" from Float Imprecision, and I don't know how to solve it, so can anyone Help me?

Ps. The program needs to always return the maximum value from each coin before pass forward.

CodePudding user response：

Floating-point arithmetic is designed to approximate real-number arithmetic. IEEE 754-2008 says “Floating-point arithmetic is a systematic approximation of real arithmetic…” So there is no way to “lock” decimal digits in binary floating-point. It is intended to be used where you want an approximation, such as modeling physics. That even includes some financial applications, such as options evaluation. Although floating-point arithmetic can be used for exact calculations in some situations, these require particular care and are generally only pursued in special circumstances.

So the answer to your question on how to lock decimal places is there is no good way to do that with binary floating-point. Attempting to do so generally just yields the same effects as integer arithmetic but less efficiently and with more difficult code. So use integer arithmetic and scale the amounts according to the smallest unit of currency desired, such as a penny.