tests for datasets with units

[keewis]

As a follow-up to #3238, this adds tests for datasets. Replacing assert_equal_with_units with assert_identical and adding tests for the toplevel functions are not included, these will be new PRs.

• Tests added
• Passes black . && mypy . && flake8

As a reference for myself, this is the list of methods from the documentation of Dataset:

• creation: Dataset()
• contents: copy, assign, assign_coords, assign_attrs, pipe, merge, rename, rename_vars, rename_dims, swap_dims, expand_dims, drop, drop_dims, set_coords, reset_coords
• comparisons: equals, broadcast_equals, identical
• indexing: loc, isel, sel, head, tail, thin, squeeze, interp, interp_like, reindex, reindex_like, set_index, reset_index, reorder_levels
• missing value handling: isnull, notnull, combine_first, count, dropna, fillna, ffill, bfill, interpolate_na, where, isin
• computation: apply, reduce, groupby, groupby_bins, rolling, rolling_exp, coarsen, resample, diff, quantile, differentiate, integrate
• aggregation: all, any, argmax, argmin, max, min, mean, median, prod, sum, std, var
• ndarray methods: astype, argsort, clip, conj, conjugate, imag, round, real, cumsum, cumprod, rank
• grouped operations: assign, assign_coords, first, last, fillna, where, quantile
• reshaping and reorganizing: transpose, stack, unstack, to_stacked_array, shift, roll, sortby

these methods are not covered:

• creation: decode_cf
• dictionary interface: __getitem__, __setitem__, __delitem__, update, items, values
• broadcast_like (since I also forgot this for DataArray, I'll create a new PR for it)

[max-sixty]

👏 on the thoroughness of these!

[keewis]

I think this is ready for review and merge. I am especially interested in comments on Dataset.__init__ and Dataset.test_merge, since at the moment they seem to test the same code.

There is still a lot to do, and I will also have to go over all tests (DataArray and Dataset) to remove any obvious mistakes (I found a lot from the code in #3238), but if there are any that are not so obvious, they probably won't be found until someone works on making the tests pass.

[max-sixty]

Looks really good. I spent 20 mins looking through and looks very thorough. I didn't read through every function. Let me know if there are any other ones you want someone to look through

I am especially interested in comments on Dataset.__init__ and Dataset.test_merge, since at the moment they seem to test the same code.

I had a good look through these. (though I didn't fully understand re 'test the same code', they seem to test the relevant init / merge functions?)

[keewis]

I think they at least partially test the same code: if two DataArray objects that get passed to the Dataset constructor as data variables have the same dimensions / coordinates, they get merged. I think that happens even if the values of the dims / coords are equal.

[max-sixty]

I think they at least partially test the same code: if two DataArray objects that get passed to the Dataset constructor as data variables have the same dimensions / coordinates, they get merged. I think that happens even if the values of the dims / coords are equal.

Ah great, because merge happens in a dataset's init, yes. Let's keep both tests though

[keewis]

Is there anything left to do before this is ready to merge? I'd like to use some of the changes to the helper functions made here in #3493.

[max-sixty]

Great @keewis . As above, I haven't looked through all (1750!!) LoC, but looks good on the code I have. Let me / us know if there's anything you want us to look into more.

Thank you v much!

[keewis]

thanks, @max-sixty

[keewis]

I think we need to review most if not all tests once we try to make them pass, right now there are too many errors (mostly from pint) to know if they actually do what they should.