Hello!

I am trying to take the lpNorm<1> of a matrix in a function that takes a MatrixBase argument.

For some reason, it gives me a syntax error, but works fine if I make the argument a concrete matrix type instead.

Any idea why that may be?

Smallest compilable code (this is OK) -

Code: Select all

#include <Eigen/Dense>

typedef Eigen::Matrix<float, Eigen::Dynamic, Eigen::Dynamic> MyMatrix;

void f(const MyMatrix &x)
{
        x.lpNorm<1>();
}

int main()
{
        f(MyMatrix::Zero(3, 3));
}


This gives me a syntax error:

Code: Select all

#include <Eigen/Dense>

typedef Eigen::Matrix<float, Eigen::Dynamic, Eigen::Dynamic> MyMatrix;

template <typename Derived>
void f(const Eigen::MatrixBase<Derived> &x)
{
        x.lpNorm<1>();
}

int main()
{
        f(MyMatrix::Zero(3, 3));
}


Code: Select all

test.cpp:8:14: error: expected expression
        x.lpNorm<1>();
                    ^
1 error generated.

Many thanks!

EDIT:

AH I figured it out. Need to use "template" keyword for disambiguation. Leaving this here for Googlers' benefit.

Code: Select all

#include <Eigen/Dense>

typedef Eigen::Matrix<float, Eigen::Dynamic, Eigen::Dynamic> MyMatrix;

template <typename Derived>
void f(const Eigen::MatrixBase<Derived> &x)
{
        x.template lpNorm<1>();
}

int main()
{
        f(MyMatrix::Zero(3, 3));
}