Encoding not preserved when using where function

[jlandmann]

My first issue in xarray, so I'd first of all thank you so much for this great package!

Code Sample, a copy-pastable example if possible

da = xr.DataArray(np.random.randn(2, 3), encoding={'_FillValue': np.nan})
da.encoding 
# {'_FillValue': nan}
da_where = da.where(1)
da_where.encoding
# {}

Problem description

This is a problem I stumbled upon when using the roi method of salem (see here), but @fmaussion brought up the very good idea to maybe fix this upstream: when using the .where method, the encoding of the dataset/the dataarray is not preserved. Could this be changed or is there any specific reason (conflict?) why this is dropped?

Expected Output

da_where.encoding
# {'_FillValue': nan}

Output of xr.show_versions()

INSTALLED VERSIONS ------------------ commit: None python: 3.5.2.final.0 python-bits: 64 OS: Windows OS-release: 10 machine: AMD64 processor: Intel64 Family 6 Model 78 Stepping 3, GenuineIntel byteorder: little LC_ALL: None LANG: None LOCALE: None.None xarray: 0.10.4 pandas: 0.22.0 numpy: 1.11.3 scipy: 1.0.1 netCDF4: 1.2.7 h5netcdf: 0.3.1 h5py: 2.6.0 Nio: None zarr: None bottleneck: 1.2.0 cyordereddict: 1.0.0 dask: 0.15.1 distributed: 1.17.0 matplotlib: 2.2.0 cartopy: None seaborn: None setuptools: 40.0.0 pip: 10.0.1 conda: 4.5.8 pytest: 2.9.2 IPython: 5.1.0 sphinx: 1.7.5

[shoyer]

Thanks for raising this. Please take a look at #1614. I'd love to figure out a more comprehensive set of rules for what xarray should do for propagating encoding/attributes in every operation. Proposals and opinions would be very welcome!

[jlandmann]

Thanks for your comment @shoyer and sorry for not having found the issue you refer to!
So in favor of a low number of open issues, what would you suggest? Should I just add the where case to #1614 and we close this?

[dcherian]

I've added where to the table in #1614 with keep as default for both encoding and attrs.