Why is the compiler calling the default constructor?

Question

Why do I receive the error below? (Why is the compiler trying to call the default constructor?)

#include <cmath>

template<typename F> struct Foo { Foo(F) { } };

int main()
{
    Foo<double(double)>(sin);   // no appropriate default constructor available
}

Answer 1

It is because there is no difference between

 Foo<double(double)>(sin);   

and

 Foo<double(double)> sin;   

Both declare a variable of name sin.

The parens are superfluous. You can put as many parens as you want.

int x;             //declares a variable of name x
int (x);           //declares a variable of name x
int ((x));         //declares a variable of name x
int (((x)));       //declares a variable of name x
int (((((x)))));   //declares a variable of name x

All are same!

If you want to create temporary instance of the class, passing sin as argument to the constructor, then do this:

#include<iostream>
#include <cmath>

template<typename F> 
struct Foo { Foo(F) { std::cout << "called" << std::endl; } };

int main()
{
    (void)Foo<double(double)>(sin); //expression, not declaration
    (Foo<double(double)>(sin));     //expression, not declaration
    (Foo<double(double)>)(sin);     //expression, not declaration
}

Output:

called
called
called

Demo : http://ideone.com/IjFUe

They work, because all three syntaxes force them to be expressions, rather than variable declarations.

However, if you try this (as @fefe sugguested in the comment):

 Foo<double(double)>(&sin);  //declaration, expression

It is not going to work, for it declares a reference variable, and since it is not initialized, you will get compilation error. See : http://ideone.com/HNt2Z