Series cast operations result in "ValueError: Length of values does not match index length"

Note: df['b'].astype('int') works when using cudf directly

Repro:

df = cudf.DataFrame()
df['a'] = [0, 1, 2, 3, 4]

dgd = dask_cudf.from_cudf(df, npartitions=1)
dgd['b'] = dgd['b'].astype('int')

Result:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-131-c6dfd269c825> in <module>
      4 dgd = dask_cudf.from_cudf(df, npartitions=1)
      5 #dgd['b'] = dgd['a']+1
----> 6 dgd['b'] = dgd['a'].astype(np.int32)

/conda/envs/cudf/lib/python3.7/site-packages/dask/dataframe/core.py in __setitem__(self, key, value)
   2521             df = self.assign(**{k: value for k in key})
   2522         else:
-> 2523             df = self.assign(**{key: value})
   2524 
   2525         self.dask = df.dask

/conda/envs/cudf/lib/python3.7/site-packages/dask/dataframe/core.py in assign(self, **kwargs)
   2718 
   2719         # Figure out columns of the output
-> 2720         df2 = self._meta.assign(**_extract_meta(kwargs))
   2721         return elemwise(methods.assign, self, *pairs, meta=df2)
   2722 

/conda/envs/cudf/lib/python3.7/site-packages/cudf-0.6.0.dev0+708.gc5b9d3e.dirty-py3.7-linux-x86_64.egg/cudf/dataframe/dataframe.py in assign(self, **kwargs)
    320         new = self.copy()
    321         for k, v in kwargs.items():
--> 322             new[k] = v
    323         return new
    324 

/conda/envs/cudf/lib/python3.7/site-packages/cudf-0.6.0.dev0+708.gc5b9d3e.dirty-py3.7-linux-x86_64.egg/cudf/dataframe/dataframe.py in __setitem__(self, name, col)
    272             self._cols[name] = self._prepare_series_for_add(col)
    273         else:
--> 274             self.add_column(name, col)
    275 
    276     def __delitem__(self, name):

/conda/envs/cudf/lib/python3.7/site-packages/cudf-0.6.0.dev0+708.gc5b9d3e.dirty-py3.7-linux-x86_64.egg/cudf/dataframe/dataframe.py in add_column(self, name, data, forceindex)
    835         if isinstance(data, GeneratorType):
    836             data = Series(data)
--> 837         series = self._prepare_series_for_add(data, forceindex=forceindex)
    838         series.name = name
    839         self._cols[name] = series

/conda/envs/cudf/lib/python3.7/site-packages/cudf-0.6.0.dev0+708.gc5b9d3e.dirty-py3.7-linux-x86_64.egg/cudf/dataframe/dataframe.py in _prepare_series_for_add(self, col, forceindex)
    807         """
    808         self._sanitize_columns(col)
--> 809         col = self._sanitize_values(col)
    810 
    811         empty_index = len(self._index) == 0

/conda/envs/cudf/lib/python3.7/site-packages/cudf-0.6.0.dev0+708.gc5b9d3e.dirty-py3.7-linux-x86_64.egg/cudf/dataframe/dataframe.py in _sanitize_values(self, col)
    791             return Series(arr)
    792         elif len(self) > 0 and len(sind) != len(index):
--> 793             raise ValueError('Length of values does not match index length')
    794         return col
    795 

ValueError: Length of values does not match index length

rapidsai / dask-cudf

Series cast operations result in "ValueError: Length of values does not match index length" #104