A simple bug that defies logic(solved)

[huike]

I've been working on implementing a string library just for the fun of it, and I've come across the most insane bug... I have it narrowed down to this piece of simple code:

                int rh = 444;
		double numd = 0;
		
		while(rh>0)
		{
			numd = rh/10.0;
			rh = rh/10;
			
			double TheValue = (numd-rh)*10+48;
			
			std::cout << "The value as an int: " <<  TheValue << std::endl;
			
			char num = TheValue;
			std::cout << "The value as a char: " << num << std::endl;
			
			std::cout << "numd: " << numd << " rh: " << rh << std::endl;
		}	

You can see it clearly prints out that the value of the integer is 52 (which is ASCII character 4) however the char prints out a value of 3, but only on the first loop. The worst part is it will randomly work for some numbers and not others.

If someone could explain this and save me my sanity I would appreciate it.

SOLUTION: apparently subtracting an integer from a double can lead to unexpected results (rounding errors), I'm sure that's what the problem is here

[coder777]

SOLUTION: apparently subtracting an integer from a double can lead to unexpected results (rounding errors), I'm sure that's what the problem is here

No, it's a rounding problem.

You can see it clearly prints out that the value of the integer is 52

Yes, because it rounds up the value of 51.999999999999... to 52 while the cast to num just removes the digits -> 51

the reason for 51.999999999999... is that not any floating point number can be expressed in binary

[MiiNiPaa]

First rule of floating point numbers: do not use them unless really needed.

You do not need any doubles here. As I can see you are just iteration over rh digits and adding 48 to them. All you need is modulo operator and integer division. That is all.