Feature request: example converting to ndarray

[drewm1980]

We're experiencing a lot of friction just figuring out how to get our data all of the way out of a pyarray.

rust-ndarray/ndarray#493

I suspect the ergonomics could be improved (or better documented) in the ndarray crate, but since this might be a more common pain point for new users of rust-numpy (compared to people just learning ndarray in isolation), I thought I'd mention it here as well.

New users of the ndarray crate are probably going to start out with owned arrays of static number of dimensions, whereas a numpy users starting from the rust-numpy example get handed an array of dynamic dimensions and non-owned data with lifetime bound to the python interpreter. I imagine that would be nice for performance if we ever manage to get our code to compile, but for a numpy user trying to get up to speed in rust, it really throws you into the deep end of the pool.

Thanks!

[kngwyu]

We're experiencing a lot of friction just figuring out how to get our data all of the way out of a pyarray.

Thanks for reporting!
I'm going to write some examples soon.

Also, I want to enhance converting APIs' ergonomics entirely, though I don't have a concrete roadmap now.

[kngwyu]

I made 2 examples(as_array, to_owned_array).

@drewm1980
Please let me know if you want additional examples 🙂

[rth]

Looks like this issue was resolved and can be closed?