Add a tests for our tests helpers

[TomAugspurger]

Dask uses a couple assert_eq helpers for easily testing the equality of the result of a computation (a dask.array or dask.dataframe say) to the expected value (np.array, pd.DataFrame). It'd be nice to have some tests for these helpers to ensure they're functioning properly

• array:

  dask/dask/array/utils.py

  Line 62 in a1a75a7

  def assert_eq(a, b, check_shape=True, **kwargs):

• dataframe:

  dask/dask/dataframe/utils.py

  Line 553 in a1a75a7

  def assert_eq(a, b, check_names=True, check_dtypes=True,

See https://github.com/dask/dask/pull/2681/files#diff-20d741e3b309789af8da0fbc2663f964 for an example of the kinds of tests we would be good to have.

[ryankarlos]

@TomAugspurger Does this issue still require a PR, if so Im happy to work on it ?

[TomAugspurger]

I suspect it's still open. At a glance, I didn't see any tests named "test_assert_eq".

…

On Mon, Sep 30, 2019 at 4:33 PM Ryan Nazareth ***@***.***> wrote: @TomAugspurger <https://github.com/TomAugspurger> Does this issue still require a PR, if so Im happy to work on it ? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#2682?email_source=notifications&email_token=AAKAOIQIRBOLLLOC46CFNQTQMJWD7A5CNFSM4D3BBQOKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD77FRSQ#issuecomment-536762570>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAKAOISFJTKH6427BU6U6HDQMJWD7ANCNFSM4D3BBQOA> .

[ryankarlos]

Ok will have a look and submit a PR

[exitflynn]

hey there, i'd like to help with this but im unsure how much work is already done since the time the issue was opened. I could use some help with what stuff I could contribute, I think the dask.arrays.test.test_testing module could still be expanded as mentioned in #2681?

[Quantum0uasar]

I'd like to work on adding tests for the assert_eq and related test helpers!

My plan:

1. Write unit tests for dask.array.utils.assert_eq covering edge cases: shape mismatch, dtype mismatch, NaN equality, and scalar vs array comparisons
2. Write unit tests for dask.dataframe.utils.assert_eq covering index equality, column ordering, and dtype handling
3. Place tests in dask/tests/test_testing_utils.py or similar

This would help ensure the test helpers themselves are reliable. Would these tests live in dask/array/tests/ and dask/dataframe/tests/ or in a shared dask/tests/ folder?