Thread: Problem with simple IF

for some reason, this doesn't work, and I can't figure out why.

I'm probaby missing something easy, but here it is.. this is only part of the code I am testing, because it passes every time no matter what I put it..

Code:

#include <cstdlib>
#include <iostream>
#include <iomanip>

using namespace std;

int main(int argc, char *argv[])
{
 		double minutes, total, rate, num, num2;
		cout << "Enter time call was started in HH.MM format\n\n\n";
		cout << "Example: 07.51 is 7:51 \n\n";
		cout << "Time: ";
		
		cin >> fixed >> setprecision(2) >> showpoint;
		cin >> num;
		cout << fixed << setprecision(2) << showpoint;
		num2 = num - static_cast<int>(num); //should give the fractional part
		cout << endl;
		cout << num2;
		cout << endl;
		if (num2 < 0.6)
	  	 cout << "\nok, this worked.\n";
  	else
	  		cout << "ok, this didn't work.\n";

thanks for any advice.

What input are you giving? What output are you getting? What output are you expecting?

BTW, this appeared to work for me.

It worked for me as well. Of course, your output is wonky if you enter a value that doesn't have a fractional part to it...

what's your input? what's the output? what's the expected output?

Originally Posted by matsp

I'm also a bit confused as to what you are wanting to happen vs. what actually happens.

I ran it with two different inputs, and it gives the expected result from what I read from the code itself - but I don't know what YOU EXPECT it to do.

--
Mats

This is just part of a small program in my college book.

The goal is to make a program, in which you enter a time and an amount of time, and calculate the results based on what rate the time you entered is at. So, if you start your call at 7:00 at night, the rate is .55 cents, if you start after 12:00am the rate is .22 cents (hypothetical).

So, you have to make sure that noone enters a number > 23.59 (23:59), or < 0. Also, none of the decimal parts can be over .59. So, you can't enter 12.6 because thats not a valid time, but 12.59 is perfectly valid.

That little part is just testing that you can find the last two digits of a Float by subtracting the number from itself with a static_cast<int> in the mix.

When I run this, it always comes out true no matter what I enter. If I enter 12.59 it doesn't work either.

12.59 is supposed to come out ok because 0.59 is less than 0.6.

Can you be more specific?

Originally Posted by Daved

12.59 is supposed to come out ok because 0.59 is less than 0.6.

Can you be more specific?

Well, I think I got it figured out, but still not sure why it's having trouble knowing that .59 is < .6

Specifically, you have it right.

.59 is ok, .6 is not
1.59 is ok, 1.6 is not,
2.59 is ok, 2.6 is not

.59 - .99 are not ok
1.59 - 1.99 are not ok
2.59 - 2.99 are not ok

Do you see where I'm going. Just like you can't say "Hey the time is twelve sixty five!" because there is no such thing as twelve sixty five when you are dealing with time, therefore a person cannot enter 12.65 into this program.

So, sometimes I enter 1.6 and it says it's ok, sometimes it says it's bad. (or 12.6 or 20.6 or 2.6, doesn't matter..) I just can't have it say that anything above .59 is ok.

I ended up putting if (num < .6 && num != .6) and it seemed to work, even though it's redundant.

Maybe it's because I'm using a free compiler.. (Dev c++)

I hope that helps.

So, sometimes I enter 1.6 and it says it's ok, sometimes it says it's bad.

Without reading your code, I am guessing it's because floating point numbers don't have infinite precision. Not all floating point numbers can be represented with infinite precision, so when you type in 2.6, it's possible that 2.60000001 or 2.599999999 is stored instead.

You will have to do something like this:

Code:

const int EPSILON = 0.000001; /* a small number. error threshold */
...
if (num < (0.6-EPSILON)) {
...
}

I ended up putting if (num < .6 && num != .6) and it seemed to work, even though it's redundant.

It is indeed redundant, and I doubt it is actually what fixes your problem.

Nothing wrong with the compiler. Free software does not have to be inferior to commercial software. The compiler that Dev-C++ uses, GCC, is a state of art compiler, not any worse than commercial compilers.

you do perform an operation where you are mixing variable types. could that be the cause of precision loss?

>> I hope that helps. <<
It shouldn't be too hard to give the exact input that isn't working for you, the exact output you get, and the exact output you expect.

>> So, sometimes I enter 1.6 and it says it's ok, sometimes it says it's bad. (or 12.6 or 20.6 or 2.6, doesn't matter..) <<
If you're testing exactly against something with a .6, then your problem is likely what cyberfish stated. See http://www.parashift.com/c++-faq-lit...html#faq-29.16 and 29.17 for more information.

Originally Posted by cyberfish

Without reading your code, I am guessing it's because floating point numbers don't have infinite precision. Not all floating point numbers can be represented with infinite precision, so when you type in 2.6, it's possible that 2.60000001 or 2.599999999 is stored instead.

You will have to do something like this:

Code:

const int EPSILON = 0.000001; /* a small number. error threshold */
...
if (num < (0.6-EPSILON)) {
...
}

It is indeed redundant, and I doubt it is actually what fixes your problem.

Nothing wrong with the compiler. Free software does not have to be inferior to commercial software. The compiler that Dev-C++ uses, GCC, is a state of art compiler, not any worse than commercial compilers.

Thats right, but I thought the "cin >> setprecision(2)" would take care of that? I need to look at my first message and see if I forgot to include that for the 'cin' statement.

I'm going to dink around with your suggestion and see what I can make happen.

thank you!


Thanks!

Originally Posted by Daved

>> I hope that helps. <<
It shouldn't be too hard to give the exact input that isn't working for you, the exact output you get, and the exact output you expect.

>> So, sometimes I enter 1.6 and it says it's ok, sometimes it says it's bad. (or 12.6 or 20.6 or 2.6, doesn't matter..) <<
If you're testing exactly against something with a .6, then your problem is likely what cyberfish stated. See http://www.parashift.com/c++-faq-lit...html#faq-29.16 and 29.17 for more information.

Well, I'm sure when I get past chapter 4 of my book, it'll talk more about floats. It has mentioned that you cannot use == to compare floating point numbers or doubles, and a few other things, but nothing real complicated yet.

Anyways, thats a good article, and definatly something to keep in memory.

thanks.

Originally Posted by Zzaacchh

for some reason, this doesn't work, and I can't figure out why.

I'm probaby missing something easy, but here it is.. this is only part of the code I am testing, because it passes every time no matter what I put it..

Code:

#include <cstdlib>
#include <iostream>
#include <iomanip>

using namespace std;

int main(int argc, char *argv[])
{
 		double minutes, total, rate, num, num2;
		cout << "Enter time call was started in HH.MM format\n\n\n";
		cout << "Example: 07.51 is 7:51 \n\n";
		cout << "Time: ";
		
		cin >> fixed >> setprecision(2) >> showpoint;
		cin >> num;
		cout << fixed << setprecision(2) << showpoint;
		num2 = num - static_cast<int>(num); //should give the fractional part
		cout << endl;
		cout << num2;
		cout << endl;
		if (num2 < 0.6)
	  	 cout << "\nok, this worked.\n";
  	else
	  		cout << "ok, this didn't work.\n";

thanks for any advice.

im no pro but you missed the { } on your if and else try this

Code:

if (num2 < 0.6){
	  	 cout << "\nok, this worked.\n";
}
  	else {
	  		cout << "ok, this didn't work.\n";
}

hope it helps