How to use the UserDefinedEvaluator

[steffen-limmer]

With an older release I was able to use the UserDefinedEvaluator by initiating it with a custom Python function my_fit as follows:

evaluator = op.UserDefinedEvaluator(problem, my_fit)

Now I get the following error when I do this:

TypeError: __init__(): incompatible function arguments. The following argument types are supported:
    1. __init__(self, arg0: pyoperon.pyoperon.Problem, arg1: std::function<std::vector<float, std::allocator<float> > (Operon::Random::RomuTrio*, Operon::Individual const&)>, /) -> None

Invoked with types: pyoperon.pyoperon.UserDefinedEvaluator, pyoperon.pyoperon.Problem, function

[foolnotion]

Hi, this is a particularly tricky topic that unfortunately does not work well at the moment.
This type of code used to work in older releases that are based on pybind11, but currently does not work with the new nanobind-based release.

The TypeError you are seeing is indeed a bug in the bindings, but beyond that there are still some issues causing cyclic references when Python calls C++ which calls a Python function (wrapped as std::function). We are investigating the additional issues and hope to find a working solution.