Conversion of vector<CustomType> fails with 'assign' error about mexplus HOT 4 CLOSED

@patrikhuber I am not familiar with Eigen library, but my bet is that Eigen::MatrixXd itself is a template and that might be treating Eigen::MatrixXd as a container just like a std::vector. The exact check is happening here

For a quick workaround for this issue, you can specialize std::vector<Eigen::MatrixXd>.

To make conversion more efficient, you can instantiate numeric matrix first and use std::copy or ANSI C memcpy to copy data, assuming everything is column-major order.
https://github.com/patrikhuber/eos/blob/7f0e46179112c44c26321491ea0246394a9a5a8a/matlab/include/mexplus_eigen.hpp#L48

from mexplus.

@patrikhuber For now, I recommend std::vector<Eigen::MatrixXd> specialization. The desired solution is a fix for a better template support (c.f., #25 ), but that requires major API updates.

from mexplus.

I am not familiar with Eigen library, but my bet is that Eigen::MatrixXd itself is a template and that might be treating Eigen::MatrixXd as a container just like a std::vector. The exact check is happening here

Yes, it is a template indeed. Aah! So the check, if I understand correctly, thinks that Eigen::MatrixXd is a std::vector, because it has a first template parameter (and it's a double), so the check here fires:

Do you think we could do a better solution than the "specialise std::vectorEigen::MatrixXd" workaround?

from mexplus.

@kyamagu Ok, I will! I'll close this then in favour of #25.
Thank you very much for the quick replies! :-)

from mexplus.