Constructor question

[spencer.d]

I was writing a brief program testing scope with a class and discovered a line that I do not understand. I fixed the problem but my question remains.

Here is my class.

class aClass
{
public:
  aClass()
  { 
	  std::cout << "i'm alive\n";
  }
  ~aClass()
  { 
	  std::cout << "i'm dead\n";
  }
};

Here is the fixed code. It ouputs "i'm alive" and "i'm dead" then finishes.

int main(int argc, char * argv[])
{
  aClass A;
}

Here is my questionable code.

int main(int argc, char * argv[])
{
  aClass A();
}

I thought that by using A() I would be calling it's constructor. The code compiles but outputs nothing. I tried overloading the () operator, with a similar short message but nothing was outputted. What is the code aClass A() doing? Newbie question, thanks.

[Joeman]

Take this code as an example

#include <iostream>

using namespace std;

class Obj
{
};

int main()
{
    Obj Test();
    Test();
    return 0;
}

Obj Test()
{
    cout << "Ran" << endl;
    return Obj();
}

What is highlighted will tell you what you did . You did a forward declare on a function you didn't define

[Ledidas]

Take this code as an example

#include <iostream>

using namespace std;

class Obj
{
};

int main()
{
    Obj Test();
    Test();
    return 0;
}

Obj Test()
{
    cout << "Ran" << endl;
    return Obj();
}

What is highlighted will tell you what you did . You did a forward declare on a function you didn't define

Your code doesn't at all imply any educational purpose, only for those who want to expose their 'obfuscatabity' of code at spare time. Calling ctor 2 times exhibits a combination of low and fat code leading to bloats.

[Joeman]

The code wasn't to show how to keep code optimized. You're pointing out code I placed there as something to fill the return with. I could have easily wrote non compiling code or undefined code, but I thought it would be better as a working example. The return had nothing to do with the answer or the question.

I thought that by using A() I would be calling it's constructor.

I showed him why this wasn't so. It was an example and not for production purposes

Also

Calling ctor 2 times exhibits a combination of low and fat code leading to bloats.

Why do you think 2 ctors are getting called for?

#include <iostream>
#include <vector>

using namespace std;

class Object
{
    public:

    Object()
    {
        cout << "constructed" << endl;
    }

    Object( const Object& Obj )
    {
        cout << "copy constructed" << endl;
    }

    ~Object()
    {
        cout << "destructed" << endl;
    }

    Object& operator = ( const Object& Obj )
    {
        cout << "assignment" << endl;
        return *this;
    }
};

Object Construct()
{
    return Object();
}

int main()
{
    Object Test = Construct();
    return 0;
}

This outputs one constructor and one destructor. not 2

[treuss]

Your code doesn't at all imply any educational purpose, only for those who want to expose their 'obfuscatabity' of code at spare time. Calling ctor 2 times exhibits a combination of low and fat code leading to bloats.

What exactly is it, you trying to state?

[Paul McKenzie]

Your code doesn't at all imply any educational purpose,

Sure it does. It shows where that line of code will actually mean something, and that is to forward declare a function.

As a matter of fact, it teaches a lot -- that functions can be declared (and the proof is in the example), and that creating no parameter, non-dynamic objects is not done using the () syntax.

Calling ctor 2 times exhibits a combination of low and fat code leading to bloats.

So were you fooled by this?

Object Construct()
{
    return Object();
}
//...
Object Test = Construct();

A great trick interview question -- how many times is the constructor for Object called? If you answer this by saying a number, you're already wrong.

Regards,

Paul McKenzie

[treuss]

On related note, does anybody know of a good reason, why declaring a function within a function block is legal? I mean, if functions could only be declared outside of function blocks, this typical beginner's mistake would not be possible and probably the C++ parsers would be a little easier to write. So there must be some counterargument.

[Paul McKenzie]

On related note, does anybody know of a good reason, why declaring a function within a function block is legal? I mean, if functions could only be declared outside of function blocks, this typical beginner's mistake would not be possible and probably the C++ parsers would be a little easier to write. So there must be some counterargument.

It's a holdover from 'C'.

Regards,

Paul McKenzie

[superbonzo]

On related note, does anybody know of a good reason, why declaring a function within a function block is legal? I mean, if functions could only be declared outside of function blocks, this typical beginner's mistake would not be possible and probably the C++ parsers would be a little easier to write. So there must be some counterargument.

It's a holdover from 'C'.

Regards,

Paul McKenzie

I read somewhere that C coders used local function declarations as a primitive form of encapsulation; like, say, you have a complex data structure, a set of functions operating on its "internals" and a set of "public" functions; the latter declares locally the former passing them around between each other as function pointers... I know, this is programming paleontology ...