Remove df_value parameter, and unnecessary astype call from test

Author: kinow

pandas/tests/io/formats/test_to_csv.py

@@ -538,12 +538,10 @@ def test_to_csv_compression(self, compression_only,
                                  compression=read_compression)
             tm.assert_frame_equal(result, df)
 
-    @pytest.mark.parametrize("df_value", [float('nan')])
     @pytest.mark.parametrize("df_new_type", ["Int64"])
-    def test_to_csv_na_rep_long_string(self, df_value, df_new_type):
+    def test_to_csv_na_rep_long_string(self, df_new_type):
         # see gh-25099
-        df = pd.DataFrame({"c": [df_value] * 3})
-        df = df.astype(df_new_type)
+        df = pd.DataFrame([np.nan] * 3, columns=['c'], dtype=dtype)
         expected_rows = ['c',
                          'mynull',
                          'mynull',