Why are these 2 cout statements giving opposite result?

Question

I can't understand why the statements are giving different results as According to Me a==b is same as b==a

#include<iostream>
#include<cmath>
using namespace std;
int main()
{  
   cout<<(pow(10,2)==(pow(8,2)+pow(6,2)))<<endl;
   cout<<((pow(8,2)+pow(6,2))==pow(10,2))<<endl;
   return 0;
}

OUTPUT IS-

1
0

Cannot [reproduce](https://godbolt.org/z/5GKoTcPTz), BTW. What compiler/version/standard are you using? — Bob__, Nov 02 '21 at 10:31
anyway I wonder which compiler gives this result. Outcome should be same even for floating point lack of precision. Gcc gives `1` twice: https://godbolt.org/z/T6j3dz7aK — Marek R, Nov 02 '21 at 10:31
An implementation of `pow` that doesn't return the best possible `double` for integral arguments ought to be considered defective. Of course neither the C++ nor IEEE754 mandates that (cf. `sqrt` where IEEE754 does). A good toolchain ought to give you the result you expect. You might get better results with `std::pow` which has overloads for integral types. — Bathsheba, Nov 02 '21 at 10:40
I can recreate the behaviour with gcc-6.3.0 that ships with mingw32 — Peter, Nov 02 '21 at 10:51
Just out of curiosity, what's the output you get from this [snippet](https://godbolt.org/z/1o8c41YKh)? — Bob__, Nov 02 '21 at 10:59

score 0 · Accepted Answer · answered Nov 02 '21 at 10:33

This is double comparison issue. You can use something like:

   cout<<(fabs(pow(10,2) - (pow(8,2)+pow(6,2))) <  std::numeric_limits<double>::epsilon() ) <<endl;
   cout<<(fabs((pow(8,2)+pow(6,2))-pow(10,2)) <  std::numeric_limits<double>::epsilon() ) <<endl;

Ref: What is the most effective way for float and double comparison?

Why are these 2 cout statements giving opposite result?

1 Answers1