issues

The documentation about iteration with rolling contains the following:

We can also manually iterate through Rolling objects:

for label, arr_window in r:
   # arr_window is a view of x

It would help to explain what sort of thing label is, and how we can use it. It seems to be an object of type Coordinate, but Coordinate does not appear in the API reference.

So I would suggest the two following improvements:

1. Add an example of using the label and arr_window, maybe plotting the arr_window with the label used to give title.
2. Tweak the document generator to include docstrings from Coordinate in the API reference section.
3. Crossreference (hyperlink) from this section of the doc text to the API reference.

Using to_xarray() on a DataFrame doesn't get me to the Dataset I want, because typically I want some of the DataFrame's columns to be coordinates, and some Data Variables.

This Stack Exchange question and answer give part of the solution: https://stackoverflow.com/questions/59672658/reassigning-xarray-data-variable-to-xarray-coordinate

This tells us how to turn some of the data variables in the translated data set into coordinates. But that's not the full solution, because we also need the coordinates to be applied to the data variables. Somehow we need to add the variables that are now coordinates as dimensions.

This is probably somewhere in the docs, but it isn't pulled together into a big picture answer. It would be great if there was an end-to-end example of taking a pandas DataFrame, partitioning variables into coordinates and data variables, and being able to use the resulting coordinates (with Dataset.sel()) to address the data variables.

If that is in the online docs, I don't know how to find it. Thanks!

When I have a bad dimension in an xarray I get this error:

ValueError: different number of dimensions on data and dims: 2 vs 3

Expected Output

what I would like to see is

ValueError: different number of dimensions on data and dims: 2 vs 3 for variable Foo

The improvement I am asking for is here: https://github.com/pydata/xarray/blob/c8dac5866d2c54ee6b262b5060a701e0be1e40cb/xarray/core/dataarray.py#L368-L371

What I claim would be an improvement would be something like this: data = _check_data_shape(data, coords, dims) data = as_compatible_data(data) try: coords, dims = _infer_coords_and_dims(data.shape, coords, dims) except Exception as e: if name: raise Exception("Error finding coordinates and dims for variable %s: %s"%(name, e) else: raise e variable = Variable(dims, data, attrs, encoding, fastpath=True)

I'm not an expert on python exception handling, so this is probably wrong.

Problem Description

The programmer cannot tell what variable causes the shape and dimension issue.

Output of xr.show_versions()

There's a great answer to a reshaping question -- how do I subdivide a dimension -- in this StackExchange Q&A: https://stackoverflow.com/questions/43015638/xarray-reshape-data-split-dimension

It would be great to get this moved into the "How do I?" section of the documentation.

ll_means.nc.gz

MCVE Code Sample

I'm attaching the relevant .nc file.

python foo = xr.open_dataset('ll_means.nc') print(foo) foo.sum(axis=1) # or foo.sum(axis=0)

Instead of getting a sum (either across the data variables or across the dimension shared between the variables), I get the error in the title: ValueError: cannot supply both 'axis' and 'dim' arguments even though I did not supply a dim argument. Furthermore, I get this error even when passing dim=None.

It's possible that I am doing something wrong, and it is just the error message that is bad, and not the behavior. I don't know enough to tell.

Expected Output

TBQH, I didn't know whether this would sum across the (single) dimension of this dataset, or if it would sum across the data variables along that dimension. I was experimenting to try to figure this out.

What I am trying to do is sum across the data variables, "perpendicular" to a dimension, instead of along it.

Problem Description

I assumed that summing across one of the axes would sum across the data variables, but perhaps that was a bad assumption. At any rate, the current behavior is undesirable.

Output of xr.show_versions()