CODEWITHSUNDEEP
c++operator-overloadingsubclasseigen
nkint
nkint

Reputation: 11733

C++ Eigen: subclassed vector get error in conversion with simple operation

I've subclassed Eigen::Vector2d for have some convenience methods that i don't will write here (like MyVec.randomize(), MyVec.distanceWithThreshold(), etc..).

But I'm facing with a error in conversion when I try to assign to a new vector some simple operation. Let's see my code:

#include <iostream>
#include "Eigen/Core"

class Vec2D : public Eigen::Vector2d {
    public:
        Vec2D() : Eigen::Vector2d() {};
        Vec2D(double x, double y) : Eigen::Vector2d(x,y) {}
};

std::ostream& operator<< (std::ostream &out, Vec2D &cPoint) {
    out << "(" << cPoint.x() << ", " << cPoint.y() << ")";
    return out;
}

int main() {
    // test with base class
    Eigen::Vector2d A = Eigen::Vector2d(0,0);
    Eigen::Vector2d B = Eigen::Vector2d(5,5);
    Eigen::Vector2d C = A - B;
    std::cout << C.x() << " " << C.y() << std::endl;

    // test with my subclassed Vec2D
    Vec2D _A,_B;
    _A = Vec2D(0, 0);
    _B = Vec2D(5, 5);
    std::cout << _A-_B << std::endl; // my stream overload is not called
    std::cout << _A << std::endl; // my stream overload is called

    // now the problem
    Vec2D dudeee = _A - _B;            // if I comment this line, no error
    std::cout << dudee << std::endl;  // if I comment this line, no error

    return 0;
}

and the error is:

test.cpp: In function 'int main()':
test.cpp:34: error: conversion from 'const Eigen::CwiseBinaryOp<Eigen::internal::scalar_difference_op<double>, const Eigen::Matrix<double, 2, 1, 0, 2, 1>, const Eigen::Matrix<double, 2, 1, 0, 2, 1> >' to non-scalar type 'Vec2D' requested
test.cpp:35: error: 'dudee' was not declared in this scope

I think (because of call on my stream operator overload) that in some way I have to overwrite the normal operators (+, - *, /) in Vec2D too saying in some way to allocate a new Vec2D object, but I don't know how to do. Some advice?

Upvotes: 0

Views: 1831

Answers (1)

ggael
ggael

Reputation: 29205

Regarding the compilation error, in C++ constructors as well as assignment operators are not automatically inherited. See this doc. In particular you have to re-implement all constructors of the Matrix<> class, and add:

using Vector2d::operator=;

to forward assignment operators.

Regarding A-B, its type is a CwiseBinaryOp<...> expression. If you want it to be a Vec2D, then you have to cast it:

cout << Vec2D(A-B);

Finally, if your only purpose is to extend Eigen's API, then the best is to use Eigen's plugin plugin mechanism. The following plugins exist:

  • EIGEN_DENSEBASE_PLUGIN
    • EIGEN_MATRIXBASE_PLUGIN
      • EIGEN_MATRIX_PLUGIN
    • EIGEN_ARRAYBASE_PLUGIN
      • EIGEN_ARRAY_PLUGIN

Upvotes: 2

Related Questions