libcmaes interface to R

[berndbischl]

Dear @beniz ,

thx for this quite nice library. I am currently trying to create an R wrapper for this -- and have partially succeeded.

1. Are you fine with this? Anything I should consider when publishing this?

2. I would really like an interface where I can create an objective function where I evaluate all lambda-descendents in one go --
   to minimize call-overhead into R / overhead in general.

(That's seems pretty much the same issue as this very old one here: #159 )

I might have figured out a way to do that, but I am feeling a bit "uncomfortable" about doing this, as I think I am calling into only "semi-exposed" API / might overlook a problem?

Would you be willing to take a glance at this at some point / answer some brief questions?

Best, Bernd

[nikohansen]

Just to be sure, the C++ code has the ask-and-tell interface, so are you trying to use the Python interface?

[beniz]

@berndbischl I have seen https://github.com/mlr-org/libcmaesr but am not familiar with R and thus can't comment. From your existing code it seems you are linking against the C++ code, but let me know if there's a Python issue.

[nikohansen]

Are you fine with this? Anything I should consider when publishing this?

Of course, go for it. I guess mention this repository?

I would really like an interface where I can create an objective function where I evaluate all lambda-descendents in one go --
to minimize call-overhead into R / overhead in general.

I don't exactly understand the issue, because the ask-and-tell interface should make this straight forward?