pandas concat working in 0.12 but not 0.13

[MichaelWS]

I have a long list of dataframes that concat fine in 0.12 but not in 0.13. Has anyone seen this error?
I am concatenating using the following

pd.concat(all_gics, ignore_index=True) 

My traceback:

/usr/anaconda/lib/python2.7/site-packages/pandas/tools/merge.pyc in concat(objs, axis, join, join_axes, ignore_index, keys, levels, names, verify_integrity)
925 keys=keys, levels=levels, names=names,
926 verify_integrity=verify_integrity)
--> 927 return op.get_result()
928
929

/usr/anaconda/lib/python2.7/site-packages/pandas/tools/merge.pyc in get_result(self)
1016 return tmpdf
1017 else:
-> 1018 new_data = self._get_concatenated_data()
1019 new_data = self._post_merge(new_data)
1020 return self.objs[0]._from_axes(new_data, self.new_axes)

/usr/anaconda/lib/python2.7/site-packages/pandas/tools/merge.pyc in _get_concatenated_data(self)
1075 new_data = {}
1076 for item in self.new_axes[0]:
-> 1077 new_data[item] = self._concat_single_item(rdata, item)
1078
1079 return new_data

/usr/anaconda/lib/python2.7/site-packages/pandas/tools/merge.pyc in _concat_single_item(self, objs, item)
1184 elif isnull(v).all():
1185
-> 1186 alls.remove('other')
1187 alls.remove('object')
1188

KeyError: 'other'

[jreback]

pls put up a reproducible example

[MichaelWS]

I will try to create an example that mimics it. (Its a private data source or else I would immediately post it immediately.)

[jreback]

@MichaelWS I have reproduced the problem; was really tested code, try with PR #5757

I believe you were appending row-wise a datetime64[ns] dtype and something else (with a non-empty type, say a nan / string)

[MichaelWS]

This now works and you are right. Thanks

[jreback]

gr8

in the future if cannot submit a link to your actual data just make a copy of what reproduces it and replace with transformed/modified data

and as small a set as u can

it helps in finding the errors

thanks for the reports

[MichaelWS]

Thanks, I will post a sample going forward.

I would have mocked up a sample but I wasn't sure what the issue was.