Add inconsistencies keyword to derived_from

[avriiil]

This is a work in progress.
This PR does not yet build correctly.

As part of the discussion in #9177

• Closes #xxxx
• Tests added / passed
• Passes pre-commit run --all-files

[GPUtester]

Can one of the admins verify this patch?

[quasiben]

add to allowlist

[jrbourbeau]

Thanks @rrpelgrim! I see this is marked as a draft PR. Often this means someone is working on something but it may not currently be in a state were other people should take time to review. Is this PR where you would you like a review, or is that premature?

[avriiil]

Let me give this one more go and then I’ll ping you for a review. Thanks for taking the time to teach me the ropes here, really appreciate it!

[avriiil]

@jrbourbeau - this is ready for review

[pavithraes]

Thanks for this PR, @rrpelgrim! I looked at this with @ncclementi, and this is close!

test_ufuc_meta is failing because it checks for docstring similarity with numpy, after removing the Dask specific disclaimers. It'll be tricky to include the inconsistencies in the disclaimer test because it's a parameter we pass to derived_from.

Taking a step back though, we probably don't want to always display "Known inconsistencies: None" because it might suggest that there are no inconsistencies. It'll be nicer to display it only if there are any inconsistencies included. The code we've suggested is implementing this, and they seem to work locally. :)

Also, the linting tests are failing, you can run the pre-commit hooks to fix that locally and then commit: https://docs.dask.org/en/stable/develop.html#code-formatting

[avriiil]

@pavithraes - I accepted your suggestions and ran the pre-commit hooks. LMK if this looks OK now?

[ncclementi]

@rrpelgrim it looks like when running the pre-commit hooks there was a file that wasn't added to the commit: accessor.py. See https://github.com/dask/dask/runs/6985825841?check_suite_focus=true for reference, can you run the pre-commit again and include the changes to that file?

Excluding linting, the resto LGTM.
@jrbourbeau I think this is ready for final review.

[ncclementi]

There is a failure on a test, but it seems unrelated https://github.com/dask/dask/runs/7006605767?check_suite_focus=true#step:6:21930

@jsignell since James is out for a couple of weeks, would you be able to give this PR a final review?

[jsignell]

This looks great! For posterity: here is what the docstring looks like rendered: