Why does this C program display different output when compiled on Windows?

Question

Now, I'm doing some problems on ProjectEuler.net, and this is the code I write for Problem #4:

#include <assert.h>
#include <errno.h>
#include <math.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <time.h>

int isPalindrome(int num)
{
    int length = floor(log10(abs(num))) + 1;
    int index = 0;
    int firstChar, lastChar;

    while (index <= (length / 2)) {

        firstChar = (num % (int)pow(10, length - index)) / pow(10, length - 1 - index);
        lastChar = (num % (int)pow(10, (index + 1))) / (pow(10, index));

        if (firstChar != lastChar) {
            return 0;
        }
        index++;
    }
    return 1;
}

int main(int argc, char *argv[])
{
    clock_t begin, end;
    double time_spent;
    int result = 0;
    int x, y;

    printf("Is 998001 a palidrome? %d\n", isPalindrome(998001));
    printf("Is 987789 a palidrome? %d\n", isPalindrome(987789));
    printf("Is 884448 a palidrome? %d\n", isPalindrome(884448));

    /* clock start */
    begin = clock();

    for (x = 999; x > 99; x--) {
        for (y = 999; y > 99; y--) {
            if (isPalindrome(x * y) && x * y > result) {
                result = x * y;
                printf("Found palindrome: %d\tX: %d\tY: %d\n", result, x, y);
            }
        }
    }

    end = clock();
    /* clock end */
    time_spent = (double)(end - begin) / CLOCKS_PER_SEC;

    printf("ANSWER: %d\n", result);
    printf("ELAPSED TIME: %f\n", time_spent);

    return 0;
}

Not beautiful, but it works. When I compile this on GNU/Linux, it works fine. However on Windows 7 64-bit, I get this output:

Expected output:

Here's where it gets weird. If you swap lines 17 and 18 (the ones that start with firstChar and lastChar) it works fine on both Windows and GNU/Linux.

What is going on here? I compiled with mingw32 gcc like this:

gcc -v prob4.c -o prob4.exe -lm

Here's the compiler output: http://pastebin.com/rtarBtNY

Seriously guys what the hell is happening?

Answer 1

The problem with floating points. If you have the following code:

pow(10, 2)

you expect the value returned to be 100. It might be but there's no guarantee. You will get a value close to 100 within some margin of error.

If the value returned is 100+d (where d is within the precision) then when that value is converted to an int you get back 100 . If the value returned is 100-d then when that value is converted to an int you get back 99 .

Answer 2

That code shouldn't even compile. You're declaring a function inside of main.

If you move the isPalindrome() function definition outside of main I can verify that it compiles in linux with no errors and no warnings.