floating point rounding errors

[momesana]

I can't explain this:

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

int main()
{
    // n
    double n = log(125)/log(5);

    // n2
    int base = 5;
    int number = 30;
    double n2  = log10((number + 1) * (base - 1) + 1) / log10(base);

    cout << (int) n <<  " --> " << (int) n2 << endl;
}

log(125)/log(5) returns three as expected.
(number + 1) * (base - 1) + 1 is 125. But If I put that as an argument into the log function and cast it into an int it gives me 2. I am seriously confused.

Thanks in advance

[Bazzy]

Doubles aren't much precise, so the value of n2 may be 2.99999 or something like that.
When you cast a double into an integer the decimal digits are truncated.

[Disch]

Floating point math is an approximation. Read sectsion 29.18:

http://www.parashift.com/c++-faq-lite/newbie.html#faq-29.18


29.16 and 29.17 are also good reads but do not directly address this particular problem

EDIT: way too slow -- that's what I get for going to the bathroom before posting XD