It took 10 days of on and off programming with #rust and #ndarray crate to solve #AdventOfCode day 4. It was bizzarely tough.
NDarray is getting there (close to #numpy) but some key features like indexing using arrays and boolean masks was missing. Which made the implementation painful.
https://codeberg.org/ashwinvis/advent-of-code/src/branch/main/2024/examples/day04.rs
Could it be better?
ndarray 0.16 is released: https://github.com/rust-ndarray/ndarray/releases/tag/0.16.0
A release has been needed for a long time to say the least, and it's a very big one too. I'm happy it's out.
@smranaldi Coming from #Python and #JuliaLang, I was annoyed at first by things like
1 + 1.0
which does not compile. You have to cast often, but you understand after a while why this is important to prevent any surprises!
I missed Arrays in Julia (multidimensional). I needed a matrix but felt like it would be an overkill to use #ndarray for it. So I wrote my small matrix for fun. It was not hard! Next time I will try #ndarray if I need multidimensional arrays more often.
(2)
hidefix (https://github.com/gauteh/hidefix) can now be used as an #xarray backend: more than 6x faster on a 8 CPU machine!
This reads a full variable of about 370mb.
ndarray 0.15 is finally released. ๐
The following are some highlights, see the release note for the whole story.
https://github.com/rust-ndarray/ndarray/blob/master/RELEASES.md#version-0150-2021-03-25
- Inserting new axes while slicing
- two-sided broadcasting for arithmetic operations
- `.cell_view()
- Improved `Array::uninit`
Thanks to **jturner314**, **SparrowLii**, **xd009642**, **dam5h**, **insideoutclub** and **bluss** who contributed the changes for this release.
ashwinvis
bluss
Mo Bitar 
gauteh
ใฆใใขใน๐
bluss