C++Talk.NET

Ravi · Guest

One of the most important lessons I learnt from "Modern C++ design" is
that there is always a way (sometimes requiring considerable
ingenuity) to make the compiler deduce types at compile time if I can
deduce it at compile time. However, the following simple problem has
vexed me for several days.

I am trying to create a functor that applies a function to each
element of a vector in the following manner:

std::vector<float> myvec(10);
// Fill up vector here.
....
// Want the following to replace each element with its sine.
onEachElement( myvec, std::sin );

My first naive attempt at defining such a functor (which works for the
most part):
template <typename T> void onEachElement( std::vector<T> &vec,
T(&func)(T) )
{
for ( int indx = 0; indx < vec.size(); ++indx )
vec[indx] = func(vec[indx]);
}

However, the template above fails when "func" takes a template
parameter such as in the case below when called with a vector of
floats:
double mu( double x ) { return x+2.; }
T mu( T x ) { return x; }

Is there a way of doing this that would work with templates as well?
The compiler must somehow be forced to instantiate the template for
the type deduced from the first argument, and then take the address to
pass as the reference. Any help is greatly appreciated. If this is a
FAQ, I must have been searching for the wrong terms.

Regards,
Ravi

[ See http://www.gotw.ca/resources/clcm.htm for info about ]
[ comp.lang.c++.moderated. First time posters: Do this! ]

Gianluca Silvestri · Guest

"Ravi" <ravi (AT) gmx (DOT) de> ha scritto nel messaggio
news:c01b487b.0411221612.8c6b901 (AT) posting (DOT) google.com...

jjr2004a · Guest

The real question is why are you trying to reinvent the wheel? Chapter 18
of Stroustrup's book "Algorithms and Function Objects" explains the host of
algorithms available in the standard library to do almost anything to a
sequence of objects. It also contains tools to help adapt items like functions
for use with these algorithms.

I used the transform algorithm and the ptr_fun adapter to do your task.
You could use for_each but this solution seemed more straight forward.
Here's the complete example:

#include <iostream>
#include <vector>
#include <algorithm> // for transform
#include <functional> // for ptr_fun
#include <math.h> // for sin

using namespace std;

//
// sample transform program
//
int main(int argc, char** argv)
{
vector<double> myvec(10);

for (int i=0; i < 10; i++)
myvec[i] = i * 10.0 * 3.14159265 / 180.0;

transform(myvec.begin(), myvec.end(), myvec.begin(), ptr_fun(sin));

for (int i=0; i < 10; i++)
cout << "sine of " << i * 10 << " degrees = " << myvec[i] << endl;

return 0;
}

[ See http://www.gotw.ca/resources/clcm.htm for info about ]
[ comp.lang.c++.moderated. First time posters: Do this! ]