Results 506 comments of Tom White

I added some notes on how to write a new executor in #498.

I've added initial support for Ray in #687. I tested it on a local cluster using this `cubed.yaml` file: ``` spec: allowed_mem: "2GB" executor_name: "ray" executor_options: ray_init: runtime_env: pip: "/Users/tom/workspace/cubed/requirements.txt"...

I've now run on a small Ray cluster on AWS: see #769.

Thanks Alex!

Thank for opening this issue @songhan89. It's not immediately obvious why this is slower - it will need a bit of investigation.

Yes, although it will be quite a bit of work since we are implicitly using NumPy dtypes everywhere, which are not interchangeable with dtypes from the array API standard. E.g....

Tests pass on Python 3.13 (with one exception, which seems to be a flaky test). This can be left until Python 3.13 is released.

Thanks for opening this @TomNicholas! > Perhaps we should move the existing text to e.g. `design/motivation` Definitely. It's mainly of historical interest, so should be moved down to the bottom...