Casting or pow()?

[Volatile Pulse]

I was messing around with a small program and I happened to notice something that didn't seem right to me.

Here was the code:

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

#include <iostream> #include <cmath> #include <ctime> #include <cstdlib> int main() { srand(time(0)); // Get a randomNumber greater than 10,000 long randomNumber = rand() + 10000; int numberOfDigits = 1; for(int i = 1; (randomNumber / static_cast<int>(pow(10, i)) != 0); i ++){ std::cout << "10^" << i << " = " << static_cast<int>(pow(10, i)) << "\n"; std::cout << static_cast<int>(pow(10, i)) << "\n"; numberOfDigits ++; } return 0; }

10^1 = 10 10 10^2 = 100 99 10^3 = 1000 1000 10^4 = 10000 9999

Why do I get a value one less than 10^n ever n/2th time?

I'm using g++ by the way.

[Cubbi]

Doesn't happen on my g++, and doesn't happen on ideone.com's http://ideone.com/JnJAt
could you give your cout a std::setprecision(100) and print the pow() without a cast?

[Volatile Pulse]

I did, and I had to manually step the program with pauses. I noticed it was perfect powers of 10 until it reached 10^23...However, my code never got to that point originally, what does this mean? I have a fresh install of MingW as of last night as well.

Edit:

Console Output:http://prntscr.com/aibxf

Code:http://prntscr.com/aibyw

[Cubbi]

That looks like Windows. Their C library isn't that good on math -- in particular, pow() for integers is nowhere as precise as it could be.
Workaround: add 0.5 before casting to int.

[Volatile Pulse]

You're right, it's WinXP. And I found a different work around, but I figured this would be more of a precise answer than just multiplying by 10 over and over again.

And I figured that the compiler made the difference with the libraries, not the OS.

[ne555]

Workaround

while(n){
  n/=10;
  ++digits;
}

Learn about errors.

Edit:

and I had to manually step the program with pauses.

You could have used a debugger.
Still ~40 lines is not something that you can't scroll.