OT: More Elementary C++ Madness

[Karim]

Purpose: Determine least amount of coins and bills for change.
Specific problem: value returned.

~~~
double change, money, bill;
int hundred; //includes more variables but hundred will suffice for this example

change = money - bill; //user inputs both values in the form 123.45
change = change * 100; //to convert everything to pennies

value = int(change);
~~~

Using the example of a bill of $65.87 and money of $100.00, yields:

change = 34.13
change = 3413
value = 3412 !!!!!!!!!!!!!!!!!!

hundred = value/10000; //returns integer equating to #of bill required
value = value - (10000 * hundred); //calculates remaining change

Can anyone explain why the cast is causing this to happen, and more importantly, how I can prevent it? When you get down to the dimes, nickels and pennies, it throws everything off.

[sabotai]

That one's a head scratcher. When I do 65.80, it comes out right. I have no idea why it sends it down to 3412. Weird. I tried a couple of things, but nothing worked.

[Karim]

Yeah. It took me a long time to realize that the cast expression was the problem because my instructor specifically told me to use it to convert from float to int. I was using cout statements checking everything BUT value = int(change).

Somewhere along the way I'm supposed to use modulus division to carry the remainder but that's pretty pointless if the initial data type conversion doesn't work.

I don't think I'll complain about anyone's text sims anymore. :D

[Mr. Wednesday]

Your problem is the same one I noted in the other post about floating point math -- you have to be careful about rounding. 34.13 happens not to be exactly representable in your computer (when multiplied by 100, it's stored as 3412.9999999999995 [from my watch window]). And casting doesn't round, it truncates. If you want to round, you have to do so explicitly -- try adding 0.5 to change. You'll still run into corner cases, but not for the stuff that your interested in right now.


On a pedantic note, if you're doing C++, you should (strictly speaking) be using static_cast rather than (unsafe) C-style casts.

(OK, I'm being a little pedantic, I still use C-style casts for arithmatic stuff, but I never use them any other time.)

[Mr. Wednesday]

Here's an example program that exhibits the correct behavior:
#include <iostream>
#include <iomanip>

int main(void)
{
double change = 0.0, money = 100.00, bill = 65.87;
int hundred;

change = money - bill; //user inputs both values in the form 123.45
change = change * 100; //to convert everything to pennies

hundred = int(change + 0.5);

std::cout << "value = " << hundred << std::endl;

char c = 0;

std::cin >> c;
}(to be really pedantic, I think I also need to include istream and ostream, but I'm not going to worry about that)

Edit: It tried to parse the includes as HTML. Doh!

[Fido]

Or...

#include <iostream>
#include <iomanip>
using namespace std;
int main(void)
{
double money = 100.00;
double bill = 65.87;

double change = (money - bill) * 100.0;
int hundred = int (floor(change + 0.5));

change = double ((hundred % 100) / 100.0);
hundred = hundred/100;

cout << "Bill Total: $" << bill << "\n"
<< "Amount Tendered: $" << money << "\n"
<< "Change Due: $" << hundred << "." << change << endl;

char c = 0;
cout << "Hit a key to exit" << endl;
cin >> c;
}

[Karim]

Thanks, guys. It's starting to make sense. Representation at the bit level is something I have to keep in mind. From now on I think I'll do a bunch of cout statements early to see what values are actually being generated and then go from there.