[BUG] Error when doing a collect aggregation on an empty Series

[shwina]

In [1]: import cudf

In [2]: df = cudf.DataFrame({'a': [], 'b': []})

In [3]: df.groupby('a').agg(list) # equivalently, agg("collect")
---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-3-0f69423f9563> in <module>
----> 1 df.groupby('a').agg(list)

~/NVTX/python/nvtx/nvtx.py in inner(*args, **kwargs)
     99         def inner(*args, **kwargs):
    100             libnvtx_push_range(self.attributes, self.domain.handle)
--> 101             result = func(*args, **kwargs)
    102             libnvtx_pop_range(self.domain.handle)
    103             return result

~/cudf/python/cudf/cudf/core/groupby/groupby.py in agg(self, func)
    169         # a Float64Index, while Pandas returns an Int64Index
    170         # (GH: 6945)
--> 171         result = self._groupby.aggregate(self.obj, normalized_aggs)
    172
    173         result = cudf.DataFrame._from_table(result)

~/cudf/python/cudf/cudf/_lib/groupby.pyx in cudf._lib.groupby.GroupBy.aggregate()
    153             c_result = move(
    154                 self.c_obj.get()[0].aggregate(
--> 155                     c_agg_requests
    156                 )
    157             )

RuntimeError: cuDF failure at: ../../../../src/column/column_factories.cpp:68: make_empty_column is invalid to call on nested types

[kkraus14]

@ttnghia since you implemented collect_set / collect, any chance you could take a look at this?

[ttnghia]

Hi! This is a current bug in collect aggregate. An issue for it has been filed here: #7611

[kkraus14]

Closing as duplicated of #7611