I am trying to use to_numpy(allow_copy=False) on my polars dataframe, but I am getting:
RuntimeError: copy not allowed: only numeric data without nulls in Fortran-like order can be converted without copy
Since I checked that df.select(pl.all().is_null().sum()) gives 0 for every column (I have 3), I suppose that the problem is in the fortran-like indexing. How to enforce it?
I would also like to know what does the fortran-like indexing mean for polars (I failed to find any sources on that).
I was unfortunately not able to reproduce an example since the zero copy mode works just fine for basic examples and my data creation process is a too long to reproduce.
The exception also raises for order = "c" and writable = False options.
EDIT: After some thinking I decided to reformulate this question in more general form here.