Chris Havlin

oh, just read throuhg the longer discussion in the PR that was closed so my comment above isn't super relevant. Ignore :)

Yup, a rebase and some rewriting!

latest commits include a rebase, some formatting fixes, a bug fix (slicing was returning a plain dask array rather than a unyt-dask array).

> Hi @chrishavlin -- with your latest changes, have you evaluated any impact on performance? Not with the latest changes, but I'll redo the analysis I did for scipy and...

No docs yet, but I think this is ready for an initial review. I'll add the performance evaluation results here soon. Let me know if there's anything else that would...

Some performance testing results! The following 2 plots are the median and minimum computation time as a function of array size for standard numpy arrays, standard unyt arrays, standard dask...

thanks for reviewing, @jzuhone ! I know I still need to write some documentation for this -- I'm thinking it'll probably be more straightforward as its own PR. If that...

Re: documentation, I can certainly draft some docs. Should I just add a dask section to the `docs/usage.rst` page?

@neutrinoceros at some point ya, but I can also split https://github.com/yt-project/yt/pull/3511 . The dask-unyt array functionality there is only for returning the dask-unyt arrays to the user, the actual read...

Just added some docs -- will look into the test failures, don't think they're related to my changes, might be a tox issue (update -- there's already a PR to...