Rémi Lafage

I understand that it has not a very high priority. Such problem may not arise that often.... It has not so far. At the moment, ML in Rust is not...

Regarding the first point, unless there is a compelling reason not to do so, it would be better to follow `scikit-learn` conventions to ease comparison and reach a wider adoption.

Excellent! Didn't see the commit! Not only Rust is fast! :rocket:

Blocked by https://github.com/rust-ndarray/ndarray-linalg/issues/274, waiting for fix in https://github.com/rust-ndarray/ndarray-linalg/pull/275 to be released.

Thanks for the swift reply. Actually I went down that way, but in that case I got: ```rust error[E0308]: mismatched types --> src\main.rs:14:18 | 14 | let b = &a...

Yep did that too! :eyes: Then I opened the issue thinking I was missing something.

I confirm it compiles with 0.9.3 adding only the `Default` trait as you mentioned. As far as I am concerned, I want to use ndarray 0.14, so...

Yes, I think it would be great to have that backup solution in the meantime, not to fall too much behind as `ndarray 0.15.1` is already out.

@mulimoen FWIW, your `0.9.4` branch works for `linfa` with `ndarray 0.14` (https://github.com/rust-ml/linfa/pull/110).

Thanks @vbarrielle for your hard work!