issues

What happened?

We've been getting this AttributeError on PyGMT's dev CI which tests nightly packages of numpy/xarray/etc. See https://github.com/GenericMappingTools/pygmt/actions/runs/6043991082/job/16401918043#step:16:40

What did you expect to happen?

Running import xarray as xr should work with numpy 2.0.

Minimal Complete Verifiable Example

```Python mamba create --name xr_with_np2 python=3.11 mamba activate xr_with_np2 python -m pip install --pre --extra-index https://pypi.anaconda.org/scientific-python-nightly-wheels/simple numpy pandas xarray

python -c "import xarray as xr" ```

MVCE confirmation

• [X] Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray.
• [X] Complete example — the example is self-contained, including all data and the text of any traceback.
• [X] Verifiable example — the example copy & pastes into an IPython prompt or Binder notebook, returning the result.
• [X] New issue — a search of GitHub Issues suggests this is not a duplicate.

Relevant log output

Python Traceback (most recent call last): File "<string>", line 1, in <module> File "/home/user/mambaforge/envs/xr_with_np2/lib/python3.11/site-packages/xarray/__init__.py", line 1, in <module> from xarray import testing, tutorial File "/home/user/mambaforge/envs/xr_with_np2/lib/python3.11/site-packages/xarray/testing.py", line 10, in <module> from xarray.core import duck_array_ops, formatting, utils File "/home/user/mambaforge/envs/xr_with_np2/lib/python3.11/site-packages/xarray/core/duck_array_ops.py", line 36, in <module> from xarray.core import dask_array_ops, dtypes, nputils File "/home/user/mambaforge/envs/xr_with_np2/lib/python3.11/site-packages/xarray/core/dask_array_ops.py", line 3, in <module> from xarray.core import dtypes, nputils File "/home/user/mambaforge/envs/xr_with_np2/lib/python3.11/site-packages/xarray/core/dtypes.py", line 43, in <module> (np.bytes_, np.unicode_), # numpy promotes to unicode ^^^^^^^^^^^ File "/home/user/mambaforge/envs/xr_with_np2/lib/python3.11/site-packages/numpy/__init__.py", line 372, in __getattr__ raise AttributeError( AttributeError: `np.unicode_` was removed in the NumPy 2.0 release. Use `np.str_` instead.

Anything else we need to know?

See https://numpy.org/devdocs/release/2.0.0-notes.html#numpy-2-0-python-api-removals. Relevant code at xarray is here https://github.com/pydata/xarray/blob/1043a9e13574e859ec08d19425341b2e359d2802/xarray/core/dtypes.py#L36-L44

Environment

Instead of raising a ValueError about the file pointer not being at the start of the file, reset the file pointer automatically to zero, and warn that the pointer has been reset.

• [ ] Closes #6813
• [ ] Tests added
• [x] User visible changes (including notable bug fixes) are documented in whats-new.rst
• [ ] New functions/methods are listed in api.rst

MCVE Code Sample

Adapted from the test_serialization unit test at here.

```python import xarray as xr from xarray.tests.test_backends import assert_identical, create_tmp_geotiff, create_tmp_file

with create_tmp_geotiff(additional_attrs={}) as (tmp_file, expected): # write it to a netcdf and read again (roundtrip) using open_rasterio with xr.open_rasterio(tmp_file) as rioda: with create_tmp_file(suffix='.nc') as tmp_nc_file: # Write geotiff to netcdf file rioda.to_netcdf(tmp_nc_file)

        # Read using open_dataarray works
        with xr.open_dataarray(tmp_nc_file) as ncds:
            assert_identical(rioda, ncds)

        # Read using open_rasterio doesn't work!!
        with xr.open_rasterio(tmp_nc_file) as ncds:
            assert_identical(rioda, ncds)

```

Actual Output (using netCDF4>=1.4.2)

```python-traceback AssertionError: Left and right DataArray objects are not identical

Differing coordinates: L * x (x) float64 5.5e+03 6.5e+03 7.5e+03 8.5e+03 R * x (x) float64 0.5 1.5 2.5 3.5 L * y (y) float64 7.9e+04 7.7e+04 7.5e+04 R * y (y) float64 0.5 1.5 2.5 Differing attributes: L transform: (1000.0, 0.0, 5000.0, 0.0, -2000.0, 80000.0) R transform: (1.0, 0.0, 0.0, 0.0, 1.0, 0.0) L res: (1000.0, 2000.0) R res: (1.0, -1.0) Attributes only on the left object: crs: +init=epsg:32618 ```

Expected Output (using netCDF4==1.4.1)

```python-traceback AssertionError: Left and right DataArray objects are not identical

Differing attributes: L nodatavals: (nan, nan, nan) R nodatavals: (nan, nan, nan) Attributes only on the left object: crs: +init=epsg:32618 ```

Problem Description

I have a script which takes in either NetCDF or GeoTIFF files as an input and I've been using xr.open_rasterio to read them quite successfully before, as I'm basically just interested in parsing the correct XY coordinates out. However, upgrading from NetCDF 1.4.1 to 1.4.2 breaks this behaviour (xr.open_rasterio no longer parses the coordinates properly from a .nc file).

My hunch is that it has something to do with different NetCDF4 formats (e.g. ‘NETCDF4’, 'NETCDF4_CLASSIC’, ‘NETCDF3_64BIT’) but looking at the xr.open_rasterio code, I'm not too sure what's going on... Also not very sure if this is an upstream netcdf4-python or rasterio issue but I thought I'd report it here first.

Output of xr.show_versions()

Work on enabling xr.open_dataset(..., engine="zarr"), to ~~replace~~ complement xr.open_zarr. This also allows `xr.open_mfdataset(..., engine="zarr") to be used.

Note: Credit should be given to @Mikejmnez, I'm just continuing this on from #4003.

• [x] Closes #3668, closes #4003
• [x] Tests added
• [x] Passes isort -rc . && black . && mypy . && flake8
• [x] User visible changes (including notable bug fixes) are documented in whats-new.rst