Added new How To (simple io with NumPy)

[melissawm]

Hi all,

This is the draft for the How to on storing and reading arrays from disk with NumPy. A couple of comments:

1. This does not cover all topics raised in this NumPy issue. That was too much for one How To. The current plan is to do another How To specifically mentioning cloud storage and more complicated issues, and an "Explanation" or "Discussions" page on the NumPy storage model and advanced performance comments.
2. There is definitely a bit more discussion than what would be recommended by the How To guide we're using. I'm aware of that, and I'm happy to take feedback on this, but I just couldn't see how to explain all this without some minor explanations.

Suggestions and feedback are much appreciated!

[review-notebook-app]

Check out this pull request on 

Review Jupyter notebook visual diffs & provide feedback on notebooks.

---

Powered by ReviewNB

[eric-wieser]

@rossbar, was the intent to use the mystnb format here for a better plaintext view when reviewing?

[eric-wieser]

It might be good to show where 181 and 212 came from, or a way of finding them programmatically.

[rossbar]

was the intent to use the mystnb format here for a better plaintext view when reviewing?

The primary goal of using something like myst-nb for the numpy-tutorials repo is to allow users to submit content in formats other than rST. The text-based notebook format is a nice feature, but I don't think the numpy-tutorials repo will have a policy recommending one format over any other. Hopefully the text-based format catches on with Jupyter users and we start seeing submissions with that format.

[rgommers]

My main comment is that I think it would be best to put the things most users are looking for upfront. Those are:

• loading and saving arrays (binary); simple (.npy) and large (Zarr)
• loading and saving multiple arrays in a single file (.npz, Zarr)
• loadings and saving arrays in text format (savetxt & friends)

NetCDF4 and structured arrays are a bit niche in comparison. I'd probably leave them out completely. Maybe structured arrays can come at the end. NetCDF4 seems like a separate topic, one of many other formats users can encounter. Maybe worth as an overview table of important formats and how they're supported, but not as the very first example on the page I'd think.

[melissawm]

Thanks for the comments. Since some of the discussion has moved to #15 , it might be better to continue there and closing (or at least pausing) this issue. If there is still value in this document after that, I'll retake it.