Adam Ehlers Nyholm Thomsen

https://github.com/rpy2/rpy2/blob/bc91910994af7028d777c5854eed9918c53db2e0/rpy2/rinterface_lib/R_API.h#L378-L380 on the version of R 4.0 I am using this should be ```c extern char *getDLLVersion(void); extern char *getRUser(void); extern char *get_R_HOME(void); ``` note the last 2 lines are...

I am not really sure what is wrong with the lint or how to fix it, so any help would be greatly appreciated.

@xhochy If I want to add tests for #316 as requested by @fjetter testing the module on Windows and Python >= 3.8, should I just merge this pull request into...

@kwgoodman I am not sure about the smaller example presented in your next to last post. If I understand the code correctly you run a 3 element window over the...

I think this is a small illustrative example (though not minimal): ```python a = np.empty(10) a[:5] = 1.0e9 a[5::2] = 1.0 a[6::2] = 0.0 bn.move_std(a, 3) bn.move_std(a.astype(np.float32), 3) ``` In...

@kwgoodman I am sorry for the late reply, you can fix the test case such that it works for the 64-bit case by making it a bit less extreme: ```python...

Ah thank you that makes sense, it seems that python/mypy#6178 is exactly what is happening here, unfortunately it does not seem like it has any satisfying resolution.

Thank you for the details. Thinking about it, I think mypy is right that model_construct is not valid for covariant types, since it uses cls which will have T_co bound...

Thinking a bit more, the same issue crops up for `__init__` and type[..] types. This again allows us to put a base value inside a container for derived instances. ```python...

I still expect it to be valid to define immutable containers with covariant generic types. However I do see that the presence of alternative constructors on subtypes makes this unsound....