c pow function printing wrong value but function returns correct results

Question

I have done a little function that converts hex colour values to rgb values in c.

The function is working correctly, if I pass it in "#008080" I get:

HEX: #008080 RGB: 0, 128, 128

The issue I am having is when I am trying to print the value returned by pow() (for debugging purposes) this is what I get:

EX::: 0
POWER::: 0.000000
EX::: 1
POWER::: 0.000000
EX::: 0
POWER::: 0.000000
EX::: 1
POWER::: 0.000000
EX::: 0
POWER::: 0.000000
EX::: 1
POWER::: 0.00000

This is the code that outputs this:

int num = 0;
char c;
double power;

for (int i = strlen(hexString), ex = 0; i > 0; i--, ex++) {
    c = hexString[i - 1];
    power = pow(16.0, ex);
    printf("EX::: %d\n", ex);
    printf("POWER::: %lf\n", power);       
    num += convertHexCharToDec(&c) * power;
}

If I run this with GDB I can see that the power variable has the values of 1 and then 16.

What is the problem here? The function outputs the correct value, I can see the correct value in the debugger so why can't I see the correct values??

Thanks

Edit: Added power variable type

Edit2: After messing a bit more with printfs specifiers I discovered that I should have used %f specifier.

Thanks for the help.

Answer 1

You should first check the type of power , which you haven't shown.

Disagreements between format specifiers and the variables you pass to printf are undefined behaviour, which is why:

#include <stdio.h>

int main (void) {
    int i = 42;
    double d = 3.14159;
    printf("POWER::: %lf\n", i);
    printf("POWER::: %lf\n", d);
    return 0;
}

can give you:

POWER::: 0.000000
POWER::: 3.141590

with impunity.

For a more detailed reason why these incorrect results can happen, see this answer . It doesn't specifically cover floating point values but the theory is the same. If you tell printf to expect one thing then give it a totally different thing, you can hardly complain if it becomes confused :-)

If power is an integral type, you need to either change it to a non-integral type, or change the format specifier. One way or another, they should match.

---

In any case, you don't need to resort to floating point at all for this task provided you have a 32-bit integer (and the color codes are all 24-bit). You can just use code like:

for (int i = strlen(hexString), mult = 1; i > 0; i--, mult *= 16) {
    c = hexString[i - 1];
    printf("MULT::: %d\n", mult);       
    num += convertHexCharToDec(&c) * mult;
}