Update min pandas to 0.23.0

This is the first version that passing a dtype of
pandas.api.types.DatetimeTZDtype(tz="UTC") to the Series constructor
actually works. I get "TypeError: data type not understood" in lower
versions.

.circleci/config.yml

@@ -45,12 +45,12 @@ jobs:
       - run: nox -s lint
 
   # Conda
-  "conda-3.6-0.21.1":
+  "conda-3.6-0.23.0":
     docker:
       - image: continuumio/miniconda3
     environment:
       PYTHON: "3.6"
-      PANDAS: "0.21.1"
+      PANDAS: "0.23.0"
     steps:
       - checkout
       - run: ci/config_auth.sh
@@ -65,4 +65,4 @@ workflows:
       - "pip-3.6"
       - "pip-3.7"
       - lint
-      - "conda-3.6-0.21.1"
+      - "conda-3.6-0.23.0"

ci/requirements-2.7.pip

@@ -1,5 +1,5 @@
 mock
-pandas==0.20.0
+pandas==0.23.0
 google-auth==1.4.1
 google-auth-oauthlib==0.0.1
 google-cloud-bigquery==1.9.0

ci/requirements-3.5.pip

@@ -1,4 +1,4 @@
-pandas==0.20.0
+pandas==0.23.0
 google-auth==1.4.1
 google-auth-oauthlib==0.0.1
 google-cloud-bigquery==1.9.0

docs/source/changelog.rst

@@ -11,7 +11,7 @@ Dependency updates
 
 - Update the minimum version of ``google-cloud-bigquery`` to 1.9.0.
   (:issue:`247`)
-- Update the minimum version of ``pandas`` to 0.20.0. (:issue:`263`)
+- Update the minimum version of ``pandas`` to 0.23.0. (:issue:`263`)
 
 Internal changes
 ~~~~~~~~~~~~~~~~

setup.py

@@ -18,7 +18,7 @@ def readme():
 
 INSTALL_REQUIRES = [
     "setuptools",
-    "pandas>=0.20.0",
+    "pandas>=0.23.0",
     "pydata-google-auth",
     "google-auth",
     "google-auth-oauthlib",

tests/system/test_gbq.py

@@ -314,7 +314,7 @@ def test_should_properly_handle_timestamp_unix_epoch(self, project_id):
             df,
             DataFrame(
                 {"unix_epoch": ["1970-01-01T00:00:00.000000Z"]},
-                dtype="datetime64[ns]",
+                dtype=pandas.api.types.DatetimeTZDtype(tz="UTC"),
             ),
         )
 
@@ -330,7 +330,7 @@ def test_should_properly_handle_arbitrary_timestamp(self, project_id):
             df,
             DataFrame(
                 {"valid_timestamp": ["2004-09-15T05:00:00.000000Z"]},
-                dtype="datetime64[ns]",
+                dtype=pandas.api.types.DatetimeTZDtype(tz="UTC"),
             ),
         )
 
@@ -403,7 +403,11 @@ def test_should_properly_handle_null_timestamp(self, project_id):
             dialect="legacy",
         )
         tm.assert_frame_equal(
-            df, DataFrame({"null_timestamp": [NaT]}, dtype="datetime64[ns]")
+            df,
+            DataFrame(
+                {"null_timestamp": [NaT]},
+                dtype=pandas.api.types.DatetimeTZDtype(tz="UTC"),
+            ),
         )
 
     def test_should_properly_handle_null_datetime(self, project_id):
@@ -589,7 +593,9 @@ def test_zero_rows(self, project_id):
             "title": pandas.Series([], dtype=object),
             "id": pandas.Series([], dtype=np.dtype(int)),
             "is_bot": pandas.Series([], dtype=np.dtype(bool)),
-            "ts": pandas.Series([], dtype="datetime64[ns]"),
+            "ts": pandas.Series(
+                [], dtype=pandas.api.types.DatetimeTZDtype(tz="UTC")
+            ),
         }
         expected_result = DataFrame(
             empty_columns, columns=["title", "id", "is_bot", "ts"]