I'm trying to implement templates to work with Eigen3 matrices and arrays. Generally, my implementation seems to be working just fine, but I fail to extend my implementation via template spezialization, to be able to either use Eigen3 types or standard numeric types (e.g. int, float, ...).
This is a shortened version of my current code:
#include <Eigen/Dense>
using namespace Eigen;
template<typename T>
void myFunc(Eigen::MatrixBase<T>& matrix)
{
cout << "Eigen type" << endl;
}
template<typename T>
void myFunc(T& matrix)
{
cout << "numeric type" << endl;
}
void main (void)
{
int var=9;
Eigen::Matrix<double,1,1> mat;
myFunc(mat); // This should uset the first template, but it doesn't !
myFunc(var);
}
This compiles fine but when I run this, both calls to myFunc will be directed to the second template (-> "numeric type"), which is of course not what I want to achieve.
Any hints on solving this issue would be greatly appreciated.
Sebastian
PS: Using MSVC 2012
The second function is chosen because it's a better match than the first when instantiated with Eigen::Matrix<double,1,1> as T. You need to constrain the second function so it's only valid with the types you intend. Have a look at std::enable_if, the examples on that page have pretty much exactly what you want.
If you want to learn more also look into SFINAE in general, that's what std::enable_if does.