Changed behavior for replacing coordinates on dataset.

MCVE Code Sample

We noticed a change in behavior in xarray that broke a test in xgcm.

Consider this code example:

import xarray as xr
import numpy as np
x = np.arange(5)
data = xr.DataArray(np.random.rand(5), coords={'x':x
}, dims=['x'])
x_c = np.arange(5) + 0.5
data_c = xr.DataArray(np.random.rand(5), coords={'x_c':x_c
}, dims=['x_c'])
ds = xr.Dataset({'data':data, 'data_c':data_c})
del ds['data_c']

ds['x_c'] = ds['x_c'][:3]
ds

Expected Output

In previous versions of xarray this resulted in

<xarray.Dataset>
Dimensions:  (x: 5, x_c: 3)
Coordinates:
  * x_c      (x_c) float64 0.5 1.5 2.5
  * x        (x) int64 0 1 2 3 4
Data variables:
    data     (x) float64 0.3828 0.6016 0.3603 0.414 0.35

(I tested right now with an older v0.11.3 but this works with v0.13 as well.

In the current master branch instead the coordinate gets padded with nans:

<xarray.Dataset>
Dimensions:  (x: 5, x_c: 5)
Coordinates:
  * x_c      (x_c) float64 0.5 1.5 2.5 nan nan
  * x        (x) int64 0 1 2 3 4
Data variables:
    data     (x) float64 0.8369 0.4197 0.3058 0.7419 0.8126

Problem Description

We fixed the test in a new PR, but @dcherian encouraged me to submit this.

Output of xr.show_versions()

INSTALLED VERSIONS ------------------ commit: None python: 3.7.4 (default, Aug 13 2019, 15:17:50) [Clang 4.0.1 (tags/RELEASE_401/final)] python-bits: 64 OS: Darwin OS-release: 18.6.0 machine: x86_64 processor: i386 byteorder: little LC_ALL: None LANG: en_US.UTF-8 LOCALE: en_US.UTF-8 libhdf5: None libnetcdf: None

xarray: 0.13.0+24.g4254b4af pandas: 0.25.1 numpy: 1.17.2 scipy: None netCDF4: None pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: None nc_time_axis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: None dask: 2.5.0 distributed: 2.5.1 matplotlib: None cartopy: None seaborn: None numbagg: None setuptools: 41.2.0 pip: 19.2.3 conda: None pytest: 5.2.0 IPython: 7.8.0 sphinx: None

[shoyer]

@jbusecke kudos to you for catching this so early, before we even made a release!

https://github.com/pydata/xarray/pull/3393 should fix the immediate issue, though it’s a pretty awkward fix.

[jbusecke]

Glad that this orphaned test (we ended up removing it, because the function was not called anymore) was still useful!

And many thanks to @dcherian for suggesting to test xgxm with the upstream master!

_{Sent with GitHawk}