help with class and constructors

i am a real new with c++, so bare with me here. i have this assignment that asks for a rational number class that overloads the +,-,/,*. I pretty much know how i am going to do it, but there is some specs i don't understand.

the problem says to use a constructor to make sure the denoninator is never zero. now i thought that a constructor was only called as the object was being created, not as it everytime it is used. i am not sure how do do this. i have only used constructors for initializations before now....

for example...

Code:

class RationalNumber {
  public:
      RationalNumber(int 0, int 1); // this would justt be a default initialization to 0/1
       RationalNumber();  // this one would need to be called every time i need to check to see if the denominator is zero?????
 };

    RatoinalNumber::RationalNumber( int num, int den) 
{
    num = 0;
    den = 1;
}

RationalNUmber::RationalNumber()
{
  //do some checking
  //do what ever needs to be done
}

i don't really get it. am i misuderstanding the constructor?

You seem to be on the right track here. And you seem to understand what you are doing. You don't need the void constructor. But your syntax is a little off for the default numbers for your parameters in your fist constructor. You should be using =1 not just 1. Unless of course your compiler is a genious and knows that you aren't trying to use 1 as the name of a parameter, in which case go ahead

as far as I know the constructor only gets called on initilization,
so if you were to check something with it, it's when you create
the instance. So maybe somethin like:

Code:

class RationalNumber {
  public:
  int denom,numer;  
    RationalNumber(int num, int den); // this would justt be a default initialization to 0/1
       RationalNumber();  // this one would need to be called every time i need to check to see if the denominator is zero?????
 };

    RatoinalNumber::RationalNumber( int num, int den) 
{
    numer=num;
    if(!den)  //checking if den==0, if so set it to 1..
        denom = 1;
    else
         denom=den;    

}

RationalNUmber::RationalNumber()
{
  //do some checking
  //do what ever needs to be done
}

now if you call it like: RationalNumber myNum(2,0)
you'll have a number 2/1...

/btq

You can set the den to 1 if they try use zero, but persoanlly i would prefer to throw an exception.....maybe a bit harsh, but I dont like classes and functions that change params without notification

Also, you can use a constructor with default values...this can act as a normal default constructor, and you can add param checking in the body of the function (in this case I throw an exception)

Code:

#include <iostream>

class RationalNumber {
	
public:
	int num,den;//these would normally be private!
    RationalNumber(int n = 0, int d = 1):num(n),den(d)
    	{if(!den)throw DenIsSetAsZeroException();}
    class DenIsSetAsZeroException{};//exception for use with class
    //...extra stuff...assignments operators, copy constructors....
};



int main(){	

	try{
		RationalNumber a;//acts like default constructor (1,0)
		std::cout << "num & den = " << a.num << " " << a.den << std::endl;
	}
	catch(RationalNumber::DenIsSetAsZeroException&){
		std::cout << "Error caught! den is zero!" << std::endl;
	}
	
	try{
		RationalNumber b(10,20);//fine
		std::cout << "num & den = " << b.num << " " << b.den << std::endl;
	}
	catch(RationalNumber::DenIsSetAsZeroException&){
		std::cout << "Error caught! den is zero!" << std::endl;
	}
	
	try{
		RationalNumber c(1,0);//oh dear!!!
		std::cout << "num & den = " << c.num << " " << c.den << std::endl;
	}
	catch(RationalNumber::DenIsSetAsZeroException&){
		std::cout << "Error caught! den is zero!" << std::endl;
	}

}

I think everything is OK but for this:
RationalNumber(int 0, int 1);
I think the syntax is wrong., it should be:
RationalNumber(int num, int den);
then is the definition you do the default arguments.

I don't get why Fordy used the member initializer?

when you wrote: RationalNUmber::RationalNumber()
you overloaded the constructor which is not what I think you need, then checking should be in the pervious function where you receive arguments because it you don't receice arguments how can you check them

Originally posted by ammar
I don't get why Fordy used the member initializer?

Why not?

If I pass no params, the default values {0,1} kick in, but if I pass an explicit value to the constructor, those defaults are not used, and the members are intialised to the proper values

RationalNumber(int n = 0, int d = 1):num(n),den(d)

I mean why didn't we use the normal assignment operator : num=n and den=d,num and den are not const.
I'm new to C++ so can you tell me if it's better to use the member initializer syntax, maybe it's better because if the member was const we don't have to re write the ctor.

fordy, thanks i like how you did that, and i want to use it. But i don't understand everthing you did.

In particular the line under public:
class DenIsSetAsZeroException{};//exception for use with class

i don't really get how that works...

Originally posted by stautze
fordy, thanks i like how you did that, and i want to use it. But i don't understand everthing you did.

In particular the line under public:
class DenIsSetAsZeroException{};//exception for use with class

i don't really get how that works...

Your createing another class within a class...so when I encounter an error, my class throws an exception based on the inner class (that inner class may have members to say what's wrong etc....)...

Then in my main code I just watch out for an exception of type RationalNumber:enIsSetAsZeroException (Outerclass::Innerclass)....

correct me if i am wrong, but the way you have it written the class DenominatorIsSetAsZeroException needs no members?

Would it look like this though if I added members:

Code:

class DenominatorIsSetAsZeroException{
  public:
    DenominatorIsSetAsZeroException()
       :message("attempted to set denominator to zero") {}
    const char *what() const {return message;}
  private:
     const char *message;
};

And this could be just placed inside the other class definition? Sort of like embedding a class in another class?
Sorry i am new to C++....

You can add members if you wish...Many people dont, but the option is there....I think this should work;

Code:

#include <iostream>
#include <string>


class RationalNumber {

public:
	int num,den;//these would normally be private!
	
	RationalNumber(int n = 0, int d = 1):num(n),den(d)
		{if(!den)throw DenIsSetAsZeroException("Divide by Zero Exception");}
    
	class DenIsSetAsZeroException{
	public:
		std::string m_str;
		DenIsSetAsZeroException(const char* str):m_str(str){};
	};

    
};



int main(){

	//..............

	try{
		RationalNumber c(1,0);//oh dear!!!
			std::cout << "num & den = " << c.num << " " << c.den << std::endl;
	}
	catch(RationalNumber::DenIsSetAsZeroException& e){
		std::cout << e.m_str << std::endl;
	}

}

Of course your version will likely be more robust than mine (access restictions...) but that's a basic idea

>Of course your version will likely be more robust than mine (access restictions...) but that's a basic idea

probably not , whats an access restriction?

"I am new to c++", and you're telling him to throw and catch exceptions already? And throwing classes at that?? Yeesh, might as well tell him to learn Winsock API! (I heard that's difficult to figure out )

P.S.

I think that what he means by "access restrictions" means something like making some of the variables private/protected... don't quote me or anything, tho

> "I am new to c++", and you're telling him to throw and catch exceptions already? And throwing classes at that?? Yeesh, might as well tell him to learn Winsock API! (I heard that's difficult to figure out )


I might be a bit misleading, because i have been programming in C for a few years....

Alrighty then, so you're not an easily confused one.