can not create a function pointer of class

[cplusplusstudent]

Hi Galik,i have one problem.You may remember the problem we were discussing here. I am pasting the code again here as you have written it.I have changed it a bit in get_example() function.

#include <iostream>
#include <cstdlib>

template<typename Data_T>
class test
{
public:
	void test_func_1(const Data_T &var)
	{
		std::cout << "test_func_1: " << var << std::endl;
	}
};

template<typename Data_T>
struct example
{
	int a;
	float b;
	void (test<Data_T>::*fun_ptr)(const Data_T& a);
};

template<typename Data_T>
std::vector<example<Data_T> > get_example(const Data_T&, size_t i)
{
      std::vector<example<Data_T> > test_vect;
	static example<Data_T> example_array[] =
	{
		{1, 2, &test<Data_T>::test_func_1}
	};

for(int i=0;i<size;i++)
		test_vect.push_back(example_array[i]);
	return test_vect;

}

int main()
{
	int i = 3;
	double d = 23.78;

	test<int> a;
	test<double> b;

	(a.*get_example(int(), 0).fun_ptr)(i);
	(b.*get_example(double(), 0).fun_ptr)(d);
}

Now how can i access the fuc_ptr()?

i am doing (test_vect.*fun_ptr)(i);. But it gives error.

[Galik]

Oh my goodness. I thought you'd forgotten about this crazy scheme ;o)

Well you changed the get_example() function to return a vector of examples rather than just examples. So you need to dereference each element of that vector to get to your example:

#include <iostream>
#include <cstdlib>
#include <vector>

template<typename Data_T>
class test
{
public:
	void test_func_1(const Data_T &var)
	{
		std::cout << "test_func_1: " << var << std::endl;
	}
};

template<typename Data_T>
struct example
{
	int a;
	float b;
	void (test<Data_T>::*fun_ptr)(const Data_T& a);
};

template<typename Data_T>
std::vector<example<Data_T> > get_example(const Data_T&)
{
	std::vector<example<Data_T> > test_vect;
	static example<Data_T> example_array[] =
	{
		{1, 2.9, &test<Data_T>::test_func_1}
		, {3, 4.7, &test<Data_T>::test_func_1}
	};

	int size = sizeof(example_array)/sizeof(example_array[0]);

	for(int i = 0; i < size; i++)
	{
		test_vect.push_back(example_array[i]);
	}
	return test_vect;

}

int main()
{
	int i = 3;
	double d = 23.78;

	test<int> a;
	test<double> b;

	std::vector<example<int> > test_vec_i = get_example(int());

	for(size_t n = 0; n < test_vec_i.size(); ++n)
	{
		(a.*test_vec_i[n].fun_ptr)(i);
	}

	std::vector<example<double> > test_vec_d = get_example(double());

	for(size_t n = 0; n < test_vec_i.size(); ++n)
	{
		(b.*test_vec_d[n].fun_ptr)(i);
	}
}

But why you would ever use such an elaborate scheme is still kind of beyond me...

[cplusplusstudent]

No it stills there ;)
I am modifying it a bit further to replicate exactly what i am facing

template<typename Data_T>
class test
{
public:
	void test_func_1(const Data_T &var)
	{
		std::cout << "test_func_1: " << var << std::endl;
	}
    void do_stuff()
    {
    std::vector<example<int> > test_vec_i = get_example(int());
//now here i want to access the test_func_1 function,means i want to execute it here,not directly calling it but through the vector
   }
};
int main()
{
test<int> a;
a.do_stuff();
}

[Galik]

There has to be a better way...

#include <iostream>
#include <cstdlib>
#include <vector>
template<typename Data_T>
class test;

template<typename Data_T>
struct example
{
	int a;
	float b;
	void (test<Data_T>::*fun_ptr)(const Data_T& a);
};

template<typename Data_T>
std::vector<example<Data_T> > get_example(const Data_T&);

template<typename Data_T>
class test
{
public:
	void test_func_1(const Data_T &var)
	{
		std::cout << "test_func_1: " << var << std::endl;
	}

    void do_stuff()
	{
		std::vector<example<Data_T> > test_vec = get_example(Data_T());

		for(size_t i = 0; i < test_vec.size(); ++i)
		{
			(this->*test_vec[i].fun_ptr)(i);
		}
	}
};

template<typename Data_T>
std::vector<example<Data_T> > get_example(const Data_T&)
{
	std::vector<example<Data_T> > test_vect;
	static example<Data_T> example_array[] =
	{
		{1, 2.9, &test<Data_T>::test_func_1}
		, {3, 4.7, &test<Data_T>::test_func_1}
	};

	int size = sizeof(example_array)/sizeof(example_array[0]);

	for(int i = 0; i < size; i++)
	{
		test_vect.push_back(example_array[i]);
	}
	return test_vect;

}

int main()
{
	test<int> a;
	a.do_stuff();

	test<double> b;
	b.do_stuff();
}

[m4ster r0shi]

Would you mind saying again what exactly you want to do? I read the previous posts but didn't quite understand... :/ It would be great if you posted some pseudo-code showing how you would like things to work ideally.

[cplusplusstudent]

Galik,is there any way to declare the test_func_1 as static. Because if i do so compiler gives me an error in the following definition

static example<Data_T> example_array[] =
	{
		{1, 2.9, &test<Data_T>::test_func_1}
		, {3, 4.7, &test<Data_T>::test_func_1}
	};

because i wan to call the *test_vec[i].fun_ptr)(i) without this pointer. I don't want to make the "test_func_1" part of every object.

[Galik]

You are trying to call a member function. You can only do that from an object. So either it has to be from this-> or it has to be from a variable of the right class.

[cplusplusstudent]

Galik,i got the point . actually i wanted to declare these functions as static so that ot to make all these functions to make part of every object. But the isseu is that if i make these static i have to make lots of member variables static which doesn't make sense.

[Galik]

If the functions need to access the member variables then there is no benefit to making them static because you will still have to pass in an object as a parameter.