Opinions on next release

[tpeulen]

Since the last release in 2018 the the calculation of distance distributions among other features has been added to LabelLib.

If there are no new features planned in the near future, it may be time for a new release.

[mdimura]

We still probably should change all the std::array<> and std::vector<> in the interfaces to the Eigen:: equivivalents to achieve better consistency. It will change the API. I will probably not be able to do it in the next 2 weeks, but later I could implement that. Should we postpone the release until then or just make another release?

[tpeulen]

I think consistency is a good idea. I don't know about Pybind11 but in SWIG Eigen vectors and arrays can be directly mapped to numpy arrays. If possible, mapping to python lists could be avoided, as a direct map from Eigen -> Numpy array could be more efficient as opposed to Eigen -> List -> Numpy.

[mdimura]

Yes, Pybind11 maps Eigen -> Numpy by default.

[tpeulen]

Great, I think the support of Python lists can be dropped. Mapping Eigen3 arrays to numpy should be anyway more performant.

[tpeulen]

For the next release we could/should link labellib to Zenodo (see: link). This way, LabelLib gets a DOI and could be cited.