Check for non-zero length frame on success cases and added a release note

Author: jameshiebert

doc/source/whatsnew/v0.17.0.txt

@@ -57,3 +57,5 @@ Performance Improvements
 
 Bug Fixes
 ~~~~~~~~~
+
+- Bug in ``DataFrame.to_hdf()`` where table format would raise a seemingly unrelated error for invalid (non-string) column names. This is now explicitly forbidden. (:issue:`9057`)

pandas/io/tests/test_pytables.py

@@ -4621,8 +4621,6 @@ def test_to_hdf_with_object_column_names(self):
         # GH9057
         # Writing HDF5 table format should only work for string-like
         # column types
-        def column_index_data_frame(index):
-            return DataFrame(np.random.randn(10,2), columns=index(2))
 
         types_should_fail = [ tm.makeIntIndex, tm.makeFloatIndex,
                                 tm.makeDateIndex, tm.makeTimedeltaIndex,
@@ -4635,7 +4633,7 @@ def column_index_data_frame(index):
             types_should_fail.append(tm.makeUnicodeIndex)
 
         for index in types_should_fail:
-            df = column_index_data_frame(index)
+            df = DataFrame(np.random.randn(10, 2), columns=index(2))
             with ensure_clean_path(self.path) as path:
                 with self.assertRaises(ValueError,
                         msg="cannot have non-object label DataIndexableCol"):
@@ -4645,6 +4643,8 @@ def column_index_data_frame(index):
             df = DataFrame(np.random.randn(10, 2), columns=index(2))
             with ensure_clean_path(self.path) as path:
                 df.to_hdf(path, 'df', format='table', data_columns=True)
+                result = pd.read_hdf(path, 'df', where="index = [{0}]".format(df.index[0]))
+                assert(len(result))
 
 
 def _test_sort(obj):