what are the cases when a floating point exception happens other than a divide by zero?

Question

i was writing some code to test the checksum for a credit card via an algorithm known as Luhn’s algorithm and verify if it's a valid card number or not the code i wrote is the program compiled fine but when i gave an input it said floating point exception i did some reading into when a floating point exception gets invoked but there was only one case which said that it occurs when you divide by zero but my program does not do that so i want to know what other cases can occur which lead to floating point exception.

#include <stdio.h>
int main(void)
{
     int a = get_int("enter your card number:");
    int x = 1  ;
    int z ;
    int y = 0 ;
    int s = 0  ;
    int d,k,m,g,h;
     int c ;
    int f = 1;
    int p = 0;
    bool checksum ;
    int l ;
    while ((a / (10^x) )> 1)
    {
        c = (a/10^x) ;
        z= c % 10 ;
        k = 2*z ;
        if  (k<10)
        {
            d = k;
        }
        else if (k>10||k==10)
        {
          d = (k%10) ;
        }
        y +=d;
        x+=2;
    }
    x=p;
    while ( (a / (10 ^ x) )>1 ) 
    {
        l= a/(10^x);
        m = l % 10 ;
        s += m;
        x+=2;
    }
     g= s+y;
     if(g%10==0)
     {
         checksum = 1 ;
        printf("valid");
     }
     else
     {
         checksum = 0 ; 
         printf("invalid");
     }
}```

`that it occurs when you divide by zero but my program does not do that` how do you know? `a / (10 ^ x)` for sure looks like divide by 0 when `x = 10`. — KamilCuk, Dec 05 '21 at 16:27
'my program does not do that' well, you should back up that claim by loading every divisor into a temp var, before using it, and then either printffing the temp or use a debugger to inspect the value. — Martin James, Dec 05 '21 at 16:30
@MartinJames ^ worked in other languages for me it even worked in previous programs..whats the symbol in c ? — dxuian, Dec 05 '21 at 16:31
(a) In C, `^` is a bitwise XOR operator. (b) When `x` is 10, `x^10` is zero, so `a / (10 ^ x)` divides by zero. (c) For historic reasons, integer divide-by-zero is reported under the floating-point exception signal. (d) For other historic reasons, floating-point exceptions are largely suppressed and alternate results (like NaN) are delivered. Exceptions include invalid operation (there is no usefully definable result), division by zero, overflow (finite range exceeded), underflow (normal range exceeded), and inexact (limited precision of format cannot represent the ideal mathematical result). — Eric Postpischil, Dec 05 '21 at 16:32
@dxuian: Other languages have other rules, and `^` did not perform exponentiation in any of your previous C programs. — Eric Postpischil, Dec 05 '21 at 16:33
@EricPostpischil thanks ..maybe i was mistaken ..does ++thanks work in stack? please forgive my etiquette im new to stack — dxuian, Dec 05 '21 at 16:35
To use Luhn’s algorithm with credit card numbers, read the number as a string and convert it to an array of 16 digits. (To convert a character `x` to the value of the digit it represents, use `x - '0'`.) In common C implementations, `int` and `long` are not wide enough to hold 16-digit decimal numbers. — Eric Postpischil, Dec 05 '21 at 16:40
@dxuian, "what are the cases when a floating point exception happens other than a divide by zero?" --> Perhaps `Inf/Inf`? — chux - Reinstate Monica, Dec 05 '21 at 17:28

Luis Colorado · Answer 1 · 2021-12-05T18:23:21.787

In my opinon, you think that 10 to the power of x is written in C as 10^x, but nothing farthest from that. 10^x is indeed 10 xor x (bitwise exclusive or) and not a to the power of operator. So if x happens to be 10, you get 0 as the denominator of the expresion and you will be, indeed, dividing by zero.

Anyway, It's my idea that you can have easier to handle your input if you accept indeed a string of digits, instead of an integer. As integer have limited precission, using an int will end you limiting your range of numbers from 0 to 4,204,967,295, while a credit card has between 15 and 20 digits, this can make inssuficient the range of numbers, to process a single card number.

On other side, if you analize your code, you will see that no floating point (neither division nor any other operation) is used in your program. This is because integer division produces the same exception as floating point division by zero. Both are signalled with the same exception, despite of the cause of the exception.

what are the cases when a floating point exception happens other than a divide by zero?

1 Answers1