GitLab

Eigen::Ref is a common way to pass eigen dense types without needing a
template, e.g. the single definition `void
func(Eigen::Ref<Eigen::MatrixXd> x)` can be called with any double
matrix-like object.

The current pybind11 eigen support fails with internal errors if
attempting to bind a function with an Eigen::Ref<...> argument because
Eigen::Ref<...> satisfies the "is_eigen_dense" requirement, but can't
compile if actually used: Eigen::Ref<...> itself is not default
constructible, and so the argument std::tuple containing an
Eigen::Ref<...> isn't constructible, which results in compilation
failure.

This commit adds support for Eigen::Ref<...> by giving it its own
type_caster implementation which consists of an internal type_caster of
the referenced type, load/cast methods that dispatch to the internal
type_caster, and a unique_ptr to an Eigen::Ref<> instance that gets
set during load().

There is, of course, no performance advantage for pybind11-using code of
using Eigen::Ref<...>--we are allocating a matrix of the derived type
when loading it--but this has the advantage of allowing pybind11 to bind
transparently to C++ methods taking Eigen::Refs.

scipy is imported in pybind11/eigen.h when it encounters a sparse
matrix, which gets tested in the eigen test.

This allows (and changes the current examples) to exit with status 99 to
skip a test instead of outputting a special string ("NumPy missing").

This also fixes the eigen test, which currently fails when eigen
headers are available but NumPy is not, to skip instead of failing when
NumPy isn't available.

Previous version would give false 'OK' if, for example, we were supposed
to get [1, 2, 3] but instead got [2, 1, 3].

Fails --- next commit will tighten test.

Passing a non-contiguous one-dimensional numpy array gives incorrect
results, so three of these tests fail.  The only one passing is the
simple case where the numpy array is contiguous and we are building a
column-major vector.  Subsequent commit will fix the three failing
cases.