category dtype coerced to object after pd.concat(axis = 1)

[Kodiologist]

import pandas as pd

x = pd.DataFrame(pd.Series(["a", "b", "c"], dtype = "category"))
print x.dtypes
df = pd.concat(
    [x, pd.Series([0, 1, 2], index = [0, 1, 3])],
    axis = 1)
print df.dtypes

This prints:

0    category
dtype: object
0     object
0    float64
dtype: object

Notice that category has unexpectedly changed to object, even though Categoricals can hold missing values (as code -1).

pd.show_versions() yields:

INSTALLED VERSIONS
------------------
commit: None
python: 2.7.10.final.0
python-bits: 64
OS: Linux
OS-release: 4.4.0-994-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8

pandas: 0.17.1
nose: None
pip: 1.5.6
setuptools: 18.4
Cython: None
numpy: 1.10.2
scipy: 0.16.1
statsmodels: None
IPython: None
sphinx: None
patsy: None
dateutil: 2.4.2
pytz: 2015.7
blosc: None
bottleneck: None
tables: None
numexpr: None
matplotlib: 1.5.0
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: None
html5lib: 0.999
httplib2: None
apiclient: None
sqlalchemy: None
pymysql: None
psycopg2: None
Jinja2: None

[jreback]

xref #10409

yeh I think the logic in the reindex part is not exactly right.

if you'd like to have a look would be great!

[mroeschke]

Looks fixed on master. Could use a test.

In [112]: df = pd.concat(
     ...:     [x, pd.Series([0, 1, 2], index = [0, 1, 3])],
     ...:     axis = 1)

In [113]: df.dtypes
Out[113]:
0    category
0     float64
dtype: object

In [115]: pd.__version__
Out[115]: '0.26.0.dev0+555.gf7d162b18'