ds.rolling() drops attributes and name

[ALDepp]

Hi all,

I just played around with some data and found that xarray's "rolling" drops the attributes and name (even if I say keep_attrs=True):

nt, nx = 100, 30
da = xr.DataArray(np.random.randn(nt, nx), dims=['time', 'x'],
                  name='foo') 
da.attrs['place'] = 'here'
da.attrs['long_name'] = 'test'

da
xarray.DataArray'foo'time: 100x: 30
array([[ 0.14720402, -0.29625209, -0.13164254, ...,  0.58363874,
         0.20588748,  1.21594309],
       [ 1.23770654, -0.18156258,  0.9182397 , ...,  0.16810624,
        -0.40726509,  0.2328856 ],
       [-0.10127142,  0.55696125,  0.7765333 , ..., -1.24054728,
        -0.3520287 ,  0.34090885],
       ...,
       [-0.62290589,  0.95234302,  1.33738597, ...,  1.25784705,
         0.32367764,  1.7907127 ],
       [ 0.2987966 , -0.9820949 , -1.33291223, ..., -0.43975905,
         2.28465498,  0.43231269],
       [ 0.66635482,  0.74084712, -2.02589549, ...,  1.64077719,
         2.84362149, -0.36572597]])
Coordinates: (0)
Attributes:
place : here
long_name : test

da.rolling(time=5).mean(dim='time')
xarray.DataArraytime: 100x: 30
array([[        nan,         nan,         nan, ...,         nan,
                nan,         nan],
       [        nan,         nan,         nan, ...,         nan,
                nan,         nan],
       [        nan,         nan,         nan, ...,         nan,
                nan,         nan],
       ...,
       [-0.56217953,  0.73100328,  0.03839124, ...,  0.60660493,
        -0.22207041,  0.72327949],
       [-0.31968275,  0.52925981,  0.00241698, ...,  0.70700785,
         0.34605282,  0.69566641],
       [-0.15442784,  0.78247162, -0.017953  , ...,  0.75334648,
         1.03670267,  0.89595308]])
Coordinates: (0)
Attributes: (0)

Again, this also happens when I include keep_attrs=True at both steps, rolling and mean. I think it should keep the name and attributes?

[mathause]

Yes, keep_attrs would only apply to global attrs of Datasets. See #4510 for a potential fix. I only just saw that you also mention the name this is not yet included.