Whats function of this class?

Question

2.00/5 (2 votes)

See more:

While reading some tutorial on template, i stumbled upon a class which was used as a float later while declaring template class object, but couldn't understand why its needed, if any one could explain

C++

class Number {
  float f;
public:
  Number(float ff = 0.0f) : f(ff) {}
  Number& operator=(const Number& n) {
    f = n.f;
    return *this;
  }
  operator float() const { return f; }
  friend ostream&
    operator<<(ostream& os, const Number& x) {
      return os << x.f;
  }
};

Posted 21-Oct-15 0:52am

the_beginner

Add a Solution

Comments

Jochen Arndt 21-Oct-15 7:01am

Why it is needed can be only answered when seeing where and how it is used.

The only reason for its existance seems the output operator. When that is not used, there is no reason to use such a class.

[EDIT]
Because you found that in a tutorial, it might be just an example for a template class whithout practical usage.

3 solutions

Solution 3

This snippet comes from here (the last example): http://www.fi.muni.cz/usr/jkucera/tic/tic0169.html[^] and the question demonstrates the problem with taking code out of context.

Eckel is talking about using built in types as template arguments. In my opinion the example is not explained fully.

He could have written:

C++

Holder<float,> h;

and got the same output. However the use of the wrapper class for float, Number allows initialization of the value in the constructor.

Doing this:

C++

int main() {
  Holder<number,> h;
  for(int j = 0; j < 20; j++)
    cout << h[j] << endl;
} ///:~

we see that the the values are set to 0 in the constructor of Number.

If we do this:

C++

int main() {
  Holder<float,> h;
  for(int j = 0; j < 20; j++)
    cout << h[j] << endl;
} ///:~

then we have uninitialized values.

Posted 22-Oct-15 0:39am

User 59241

Updated 22-Oct-15 0:45am

v2

Comments

the_beginner 22-Oct-15 7:10am

So except for initializing values to zero does it serve any other purpose

Andreas Gieriet 22-Oct-15 8:58am

It seems to serve the purpose to show you how to pass *any* type to a template argument. In the examples before (Array, Stack), there are always built-in types used as template parameters. In the last example, the Holder class gets another class as template parameter. The Number class is just a simple class to show this. Don't take any intrinsic meaning out of the fact that this Number class is not useful for anything.
Do not project too much (meaning) into the example - it only shows various usages of template parameters.
Regards
Andi

Solution 2

A float is a so called value type. They are just primary values and don't have any instance or even class methods or properties. Value types don't have a class backing them up. When a template<class T> is required, you wouldn't be able to use float or int directly. But when encapsulated in a class like number you can. You now have a reference type instead of just a value type. This Number class is like boxing/unboxing in Java.

Good luck!

Posted 21-Oct-15 1:05am

E.F. Nijboer

Updated 21-Oct-15 12:36pm

v2

Comments

Andreas Gieriet 21-Oct-15 10:42am

This is C++, not C# or alike. So, no such thing like value versus reference type. Or am I missing something?
Regards
Andi

E.F. Nijboer 21-Oct-15 16:16pm

Fundamental type would be a better name in the context of C++.

Andreas Gieriet 21-Oct-15 17:27pm

Your solution is incorrect also with respect of if float is usable in templates. C++ allows to use any type as template parameter as long as the applied operations (e.g. like comparison or arithmetic, etc.) is supported by that given type.
Regards
Andi
PS: I was tempted to down-vote your solution, but it was not me. I opted for discussion the issue ;-)

E.F. Nijboer 21-Oct-15 18:41pm

Now that you said I was wrong because templates can use fundamental types I saw that in my answer the "class T" was filtered out. I updated my answer and used html encoding so it is now visible. So what I meant that encapsulating the float in a Number class makes it possible to use them when a template class has this restriction.

Andreas Gieriet 22-Oct-15 3:21am

Still incorrect information. You seem to mix up C# generics with C++ templates. The class in template<class T> is not restricting the usage of types here - you may use any type, no matter if a class or a struct or an enum or a composed type or a built-in type, etc. Instead of class T you can likewise write typename T without changing the meaning. I wonder if I misread completely your text...? What I read simply does not make sense for C++.
Regards
Andi

E.F. Nijboer 22-Oct-15 6:16am

I think you're right that I might indeed have mixed up when it comes to usage of restrictions. So, that wasn't a valid argument for wrapping a fundamental type into a reference type. It is only a valid point that wrapping it into a reference type might give some benefit in using it. A number can be retrieved from a vector and you would be able to change the value. The Number object in the vector (referencing the same object) doesn't need to be written back into the vector explicitly. When using int or float directly it would be copied into a variable and you would need to write it back to the vector. Otherwise the vector would still contain the same value. Sorry for the misunderstanding.

Andreas Gieriet 22-Oct-15 8:21am

You still talk about value- versus reference type. There is no such thing in C++! Everything is by value. If you want to pass something "by reference" you pass the address of the object, either by pointer or by reference operator. This can be done for any instance of any type. If you have a vector of *pointers* to some type, than it can be a pointer to a class as well as to a float, etc. The object that is stored in the vector is then the *address* to some entity (note that you cannot store a *reference* to any type in a vector).
Regards
Andi

E.F. Nijboer 22-Oct-15 13:01pm

hmmm, it has been quite a while back using c++... and now you mention it, you're completely right. It seemed very logical to wrap a value type as a reference for reasons already mentioned. It would be just like boxing/unboxing in Java. But I have to admit I was, well... just wrong. ;-)
I appreciate you didn't give up on me so I learned something today :-D Thanks!

Andreas Gieriet 22-Oct-15 15:25pm

No problem.
Cheers
Andi

Add a Solution

This content, along with any associated source code and files, is licensed under The Code Project Open License (CPOL)

CPallini · Accepted Answer · 2015-10-21T01:03:00

Solution 1

The class actually wraps a float in a object. I don't see any reason why it is needed, however (I don't see any code using it as well, however).

Posted 21-Oct-15 1:03am

CPallini

Whats function of this class?

3 solutions

Solution 3

Solution 1

Solution 2

Add your solution here

Preview 0

Whats function of this class?

3 solutions

Solution 3

Solution 1

Solution 2

Add your solution here

Preview 0

Existing Members

...or Join us