pip install fails on ppc64le

[thilo-maurer]

Installing pandas using pip install pandas fails on architecture ppc64le.

$> pip install pandas --user
Collecting pandas
  Using cached https://files.pythonhosted.org/packages/45/12/1e1ba99fb65df9f7f3724d3232feef35cc044d18604d57492d561e90219f/pandas-0.23.0.tar.gz
Collecting python-dateutil>=2.5.0 (from pandas)
  Using cached https://files.pythonhosted.org/packages/cf/f5/af2b09c957ace60dcfac112b669c45c8c97e32f94aa8b56da4c6d1682825/python_dateutil-2.7.3-py2.py3-none-any.whl
Requirement already satisfied (use --upgrade to upgrade): pytz>=2011k in /usr/lib/python2.7/site-packages (from pandas)
Requirement already satisfied (use --upgrade to upgrade): numpy>=1.9.0 in ./.local/lib/python2.7/site-packages (from pandas)
Requirement already satisfied (use --upgrade to upgrade): six>=1.5 in /usr/lib/python2.7/site-packages (from python-dateutil>=2.5.0->pandas)
Building wheels for collected packages: pandas
  Running setup.py bdist_wheel for pandas ... error
  Complete output from command /usr/bin/python2 -u -c "import setuptools, tokenize;__file__='/tmp/pip-build-WyedBH/pandas/setup.py';exec(compile(getattr(tokenize, 'open', open)(__file__).read().replace('\r\n', '\n'), __file__, 'exec'))" bdist_wheel -d /tmp/tmp6F7saUpip-wheel- --python-tag cp27:
  running bdist_wheel
  [...]
  copying pandas/tests/io/data/test2.xlsm -> build/lib.linux-ppc64le-2.7/pandas/tests/io/data
  copying pandas/tests/io/data/test_types.xls -> build/lib.linux-ppc64le-2.7/pandas/tests/io/data
  error: can't copy 'pandas/tests/io/data/legacy_hdf': doesn't exist or not a regular file
  
  ----------------------------------------
  Failed building wheel for pandas

Issue may be resolved by the

[tmaurer@ibm pandas]$ git diff
diff --git a/setup.py b/setup.py
index 90ec8e91a..b38deedce 100755
--- a/setup.py
+++ b/setup.py
@@ -733,7 +733,7 @@ setup(name=DISTNAME,
       maintainer=AUTHOR,
       version=versioneer.get_version(),
       packages=find_packages(include=['pandas', 'pandas.*']),
-      package_data={'': ['data/*', 'templates/*', '_libs/*.dll'],
+      package_data={'': ['data/*.*', 'templates/*', '_libs/*.dll'],
                     'pandas.tests.io': ['data/legacy_hdf/*.h5',
                                         'data/legacy_pickle/*/*.pickle',
                                         'data/legacy_msgpack/*/*.msgpack',

Is seems that on ppc64le data/* selects the directory data/legacy_hdf, while on x86, it does not.
In turn, on ppc64le, the directory cannot be copied because is it not a file.

The fix is IMHO a workaround, since by accident, using the glob data/*.* instead of data/* selects only files, since all the directories have no file-name extension.

To my understanding, on x86, data/* should select directories as well, and the same error should have been raised.

OS used: Red Hat Enterprise Linux Server release 7.3 (Maipo)

[TomAugspurger]

I'll try to finish up #19535 now, so that this becomes irrelevant.