implementation of isnan() function

Question

I am a beginner to c++ programming and I am given a task of implementation of fixed point math arithmetic in c++. here I am trying to implementation a function isnan() which returns true if the number is not-a-number else will return false.

Test file

#include "fixed_point_header.h"
int main()
{
fp::fixed_point<long long int, 63> a=fp::fixed_point<long long int, 63>::positive_infinity(); // will assign positive infinity value to a from an function from header 
fp::fixed_point<long long int, 63> b=fp::fixed_point<long long int, 63>::negative_infinity(); // will assign positive infinity value to b from an function from header 
float nan=fp::fixed_point<long long int, 63>::isnan(a,b);
printf( "fixed point nan value  == %f\n", float (nan));
}

In the header I want to do somewhat like the code shown below if positive and negative infinity values are added, the isnan function should return 1 else 0.

Header file

#include fixed_point_header
static fp::fixed_point<FP, I, F> isnan (fp::fixed_point<FP, I, F> x,fp::fixed_point<FP, I, F> y){
/*if ( x + y ) happens, ie. x and y are infinities
      {
 should return 1; }
       else {
 should return 0; }
      } */

can anyone please tell how to proceed with it? or how to solve this paradigm

@Mike Seymouroh yes, `isnan` function will take an argument and return 1 or 0 based on whether it is nan or not. so for that first i need to the define the nan's? — Dr7, May 13 '13 at 13:43
Yes, you'll need to define a value to represent `nan`. By the way, I removed my comment when I expanded it to an answer. — Mike Seymour, May 13 '13 at 14:07

score 0 · Accepted Answer · answered May 13 '13 at 13:39

I am trying to implementation a function isnan() which returns true if the number is not-a-number else will return false.

That's simple enough; define a reserved value to represent nan (as you have for the infinities), and compare with that:

bool isnan(fixed_point x) {
    return x == fixed_point::nan();
}

I want to do somewhat like the code shown below if positive and negative infinity values are added, the isnan function should return 1 else 0

It would be the responsibility of the addition operator to check the inputs and return nan if appropriate:

fixed_point operator+(fixed_point x, fixed_point y) {
    if (x == fixed_point::nan() || y == fixed_point::nan()) {
        return nan;
    }
    if (x == fixed_point::positive_infinity()) {
        return y == fixed_point::negative_infinity() ? fixed_point::nan() : x;
    }
    // and so on
}

then the test in main becomes:

bool nan = fixed_point::isnan(a+b);

Thank you for reply, i have a doubt, calling `bool nan = fixed_point::isnan(a+b);` will call which function? can you please explain bit more how to code flows when `bool nan = fixed_point::isnan(a+b);` is called? — Dr7, May 13 '13 at 14:15
@Dr7: First it will call `operator+` to perform the addition; then it will call `isnan` to check whether the result is a number. — Mike Seymour, May 13 '13 at 16:11

implementation of isnan() function

1 Answers1