Why do floating-point numbers print without decimal unless std::fixed and std::setprecision are used?

Question

I'm experimenting with floating-point output in C++. I have a simple calculation which is 1359270 / 4, in a calculator this results in 339817.5, but when I print it in C++ using std::cout, I get 339818.

Here’s an example:

#include <iostream>
using namespace std;

int main() {
    int var1 = 1359270;
    int var2 = 4;

    cout << static_cast<double>(var1) / var2;
}

I tried to change the variable types to double, but same result.

However, when I used std::setprecision and std::fixed like the example below:

#include <iostream>
#include <iomanip>
using namespace std;

int main() {
    int var1 = 1359270;
    int var2 = 4;

    cout << fixed << setprecision(1) << static_cast<double>(var1) / var2;
}

This prints the expected result: 339817.5.

Therefore, is this a floating-point precision issue, or is it related to how std::cout formats numbers?

Note: I'm using MinGW-64 GCC C++20 with Code::Blocks IDE In Windows.

Answer 1

As you can see in std::ios_base::precision documentation:

The default precision, as established by std::basic_ios::init, is 6.

This is the reason you see 339818 as the output in the first snippet:
std::cout will use a default of 6 digits, and so 339817.5 is rounded to 339818 when printed.

If your var1 was smaller, e.g. 359270 the exact result value (89817.5) would fit into 6 digits and you would see the exact value with the fraction (see demo).

As you already noticed, you can change the default precision.