Segfault in .groupby() with empty groups

[twiecki]

I get a segfault when running:

pandas.DataFrame([1,2,3]).groupby([]).groups

As an aside, in my scenario (which I don't know how general it is) the user can supply the grouping. If no grouping should take place (i.e. groupby([])) I would like .groupby() to return the whole data, instead there is a:

/usr/local/lib/python2.7/dist-packages/pandas/core/groupby.pyc in _get_compressed_labels(self)
547 group_index = get_group_index(all_labels, self.shape)
548 else:
--> 549 group_index = all_labels[0]
550 comp_ids, obs_group_ids = _compress_group_index(group_index)
551 return comp_ids, obs_group_ids

IndexError: list index out of range

I'm not sure whether it would be sensible to change the behavior so that the whole DataFrame is returned, but to me it would be more useful than raising an error.

[cdeil]

I just updated to 9385b62 and now also see a segfault when running the tests:

test_aggregate_item_by_item (pandas.tests.test_groupby.TestGroupBy) ... Segmentation fault: 11

as well as when running this line of code

pandas.DataFrame([1,2,3]).groupby([]).groups

[wesm]

Now raise ValueError in this case