How to create an initialise pointer to pointer to function variable;

[iceColdFire]

HI,

I have a function as

void f(int p)
{
return p++;
}

now I have created a function pointer as
void(**pf)(int);

and initialization as
*pf=f;

but compiler gives error...

Why ?

a.a.cpp

[Alf P. Steinbach]

* iceColdFire:

I have a function as

void f(int p)
{
return p++;
}
First, you cannot return non-void when the function is declared
as returning void.

Second, with 'void' changed to 'int', that has the same final effect as

int f( int p )
{
return p;
}

You might want to take a look at

<url: http://home.no.net/dubjai/win32cpptut/html/w32cpptut_01_02_11.html>

for other reasons why you should avoid postfix increment and decrement.

now I have created a function pointer as
void(**pf)(int);

and initialization as
*pf=f;

but compiler gives error...

The compiler shouldn't flag _that_ as an error, but a good compiler will
warn you that you're derefencing an uninitialized pointer.

Possibly what you wanted was

int (*pf)(int) = f;

However, as a beginner try to avoid using pointers directly. In this case
you would probably (here I'm guessing, but probably) be much better served
by using a virtual member function instead. Try to read up on that.

--
A: Because it messes up the order in which people normally read text.
Q: Why is it such a bad thing?
A: Top-posting.
Q: What is the most annoying thing on usenet and in e-mail?

[iceColdFire]

Hi Alf::

Well...that was great explanaiton indeed...however I need to use
pointer to pointer to function as

int(**pf)(int);

Now thing is how do I initialise the variable pf with f, and is my
variable declaration correct for (pointer to (pointer to function))

Thanks,
a.a.cpp

[Alf P. Steinbach]

* iceColdFire:

Well...that was great explanaiton indeed...however I need to use
pointer to pointer to function as

int(**pf)(int);

Now thing is how do I initialise the variable pf with f
int (*pf)(int) = f;
int (**ppf)(int) = &pf;
and is my
variable declaration correct for (pointer to (pointer to function))

I think so (modulo bad eyesight & auto filtering of on-screen text).

But if you'd describe what you're trying to achieve instead of how you're
intending to implement part of the solution, I'm reasonably sure that I or
someone else can tell you how to do that much more safely and easily via
virtual member functions -- or perhaps templates, or whatever.

--
A: Because it messes up the order in which people normally read text.
Q: Why is it such a bad thing?
A: Top-posting.
Q: What is the most annoying thing on usenet and in e-mail?

[John Carson]

"iceColdFire" <ic*******@yahoo.com> wrote in message
news:11**********************@g49g2000cwa.googlegr oups.com

Hi Alf::

Well...that was great explanaiton indeed...however I need to use
pointer to pointer to function as

int(**pf)(int);

Now thing is how do I initialise the variable pf with f, and is my
variable declaration correct for (pointer to (pointer to function))

Thanks,
a.a.cpp

An alternative to Alf's scheme that is closer to your original is as follow:

// pointer to function
int(*pf)(int);

// pointer to pointer to function
int(**ppf)(int);

int main()
{
//initialise double pointer to point to single pointer
ppf=&pf;
// use double pointer to initialise single pointer
*ppf = &f;
// call function from double pointer
int x = (**ppf)(5);
return 0;
}

Note that the & in

*ppf = &f;

is optional as is one of the asterisks in

int x = (**ppf)(5);

but it is more logical to include them (if working with pointers to member
functions, then you must include them --- you don't get a choice).
--
John Carson

[iceColdFire]

::Alf

Allright,
That was nice and neat...can we initialize ppf with f instead of
initialising with pf...

And here is the explanation, I intend to build an array of pointer to
pointer to functions which shall be used in a module for runtime
function reference...I am avoiding using templates and virtual member
functions, as what I have as resource is just the function addresses
and not the code...
so I can only use pointers to build up the entire system....

Thanks,
a.a.cpp

[iceColdFire]

::John,,,thanks for the nice explanation...
Check my last posting for what I actually intend to build..

a.a.cpp

[Donovan Rebbechi]

On 2005-05-23, iceColdFire <ic*******@yahoo.com> wrote:

::Alf

Allright,
That was nice and neat...can we initialize ppf with f instead of
initialising with pf...

And here is the explanation, I intend to build an array of pointer to
pointer to functions which shall be used in a module for runtime
function reference...
Yes, fine -- but why do you need pointer to (pointer to functions) ? Why is
a pointer to function not good enough ?

Or do you mean that you want to create a dynamic array of pointer-to-functions
and the type of that dynamic array is pointer to (pointer to function) ?
I am avoiding using templates and virtual member
functions, as what I have as resource is just the function addresses
and not the code...

You can wrap a function pointer in a function object (for example). Most of the
solutions that would be suggested do not require you to have access to the source
code for your functions.

Cheers,
--
Donovan Rebbechi
http://pegasus.rutgers.edu/~elflord/

[iceColdFire]

::Donovan,
Yup !!!.

I intend to build a dynamic array of [......]
Also I liked your wrapping idea ...means wrapping a function pointer
into a function object...
But how do you do that....do you have any related docs for the trick..

Thanks,...
a.a.cpp

[John Carson]

"iceColdFire" <ic*******@yahoo.com> wrote in message
news:11**********************@g43g2000cwa.googlegr oups.com

John,,,thanks for the nice explanation...

Check my last posting for what I actually intend to build..

If you want a dynamic array of function pointers, then you can build it with
new or with vector or...

With vector, you can just go:

// creates a 10 element vector of function pointers
std::vector<int(*)(int)> vec(10);

//Then you can set

vec[0] = &f;

// and so on. You would call the function from
// the first vector element as follows:

int y = (*vec[0])(7);

Naturally, you can expand the size of the vector using push_back or various
other member functions.

--
John Carson