From 911156021b75f5852ebd934c0ee817bab5b362e1 Mon Sep 17 00:00:00 2001
From: Matthew Roeschke <emailformattr@gmail.com>
Date: Sat, 29 Jan 2022 14:44:30 -0800
Subject: [PATCH 1/3] TST: Disable functionality that hangs Azure CI

---
 ci/run_tests.sh                       |  2 +-
 pandas/tests/io/excel/conftest.py     |  4 +++-
 pandas/tests/io/excel/test_readers.py |  6 ++----
 pandas/tests/io/parser/conftest.py    | 18 ++++++++++++++++--
 pandas/tests/io/test_sql.py           |  8 ++++++++
 pandas/tests/window/test_numba.py     | 15 +++++++++++++++
 pandas/tests/window/test_online.py    | 15 +++++++++++++++
 pandas/util/_test_decorators.py       |  3 ++-
 8 files changed, 62 insertions(+), 9 deletions(-)

diff --git a/ci/run_tests.sh b/ci/run_tests.sh
index 020a3ed7f265c..344982920273e 100755
--- a/ci/run_tests.sh
+++ b/ci/run_tests.sh
@@ -24,7 +24,7 @@ if [[ $(uname) == "Linux" && -z $DISPLAY ]]; then
     XVFB="xvfb-run "
 fi
 
-PYTEST_CMD="${XVFB}pytest -r fEs -m \"$PATTERN\" -n $PYTEST_WORKERS --dist=loadfile $TEST_ARGS $COVERAGE $PYTEST_TARGET"
+PYTEST_CMD="${XVFB}pytest -v -r fEs -m \"$PATTERN\" -n $PYTEST_WORKERS --dist=loadfile $TEST_ARGS $COVERAGE $PYTEST_TARGET"
 
 if [[ $(uname) != "Linux"  && $(uname) != "Darwin" ]]; then
     PYTEST_CMD="$PYTEST_CMD --ignore=pandas/tests/plotting/"
diff --git a/pandas/tests/io/excel/conftest.py b/pandas/tests/io/excel/conftest.py
index 0455e0d61ad97..4ce06c01892d9 100644
--- a/pandas/tests/io/excel/conftest.py
+++ b/pandas/tests/io/excel/conftest.py
@@ -1,5 +1,6 @@
 import pytest
 
+from pandas.compat import is_platform_windows
 import pandas.util._test_decorators as td
 
 import pandas._testing as tm
@@ -43,7 +44,8 @@ def read_ext(request):
     return request.param
 
 
-@pytest.fixture(autouse=True)
+# Checking for file leaks can hang on Windows CI
+@pytest.fixture(autouse=not is_platform_windows())
 def check_for_file_leaks():
     """
     Fixture to run around every test to ensure that we are not leaking files.
diff --git a/pandas/tests/io/excel/test_readers.py b/pandas/tests/io/excel/test_readers.py
index 589c98721f139..119457a3d4025 100644
--- a/pandas/tests/io/excel/test_readers.py
+++ b/pandas/tests/io/excel/test_readers.py
@@ -150,9 +150,7 @@ def parser(self, *args, **kwargs):
             expected = expected_defaults[read_ext[1:]]
         assert result == expected
 
-    def test_usecols_int(self, read_ext, df_ref):
-        df_ref = df_ref.reindex(columns=["A", "B", "C"])
-
+    def test_usecols_int(self, read_ext):
         # usecols as int
         msg = "Passing an integer for `usecols`"
         with pytest.raises(ValueError, match=msg):
@@ -1311,7 +1309,7 @@ def cd_and_set_engine(self, engine, datapath, monkeypatch):
         monkeypatch.chdir(datapath("io", "data", "excel"))
         monkeypatch.setattr(pd, "ExcelFile", func)
 
-    def test_engine_used(self, read_ext, engine, monkeypatch):
+    def test_engine_used(self, read_ext, engine):
         expected_defaults = {
             "xlsx": "openpyxl",
             "xlsm": "openpyxl",
diff --git a/pandas/tests/io/parser/conftest.py b/pandas/tests/io/parser/conftest.py
index 2070057aff10b..e98d23d12bbd0 100644
--- a/pandas/tests/io/parser/conftest.py
+++ b/pandas/tests/io/parser/conftest.py
@@ -12,6 +12,9 @@
 )
 import pandas._testing as tm
 
+from pandas.io.parsers import readers
+from pandas.io.parsers.arrow_parser_wrapper import ArrowParserWrapper
+
 
 class BaseParser:
     engine: str | None = None
@@ -101,7 +104,7 @@ def csv1(datapath):
 
 
 @pytest.fixture(params=_all_parsers, ids=_all_parser_ids)
-def all_parsers(request):
+def all_parsers(request, monkeypatch):
     """
     Fixture all of the CSV parsers.
     """
@@ -109,10 +112,21 @@ def all_parsers(request):
     if parser.engine == "pyarrow":
         pytest.importorskip("pyarrow", VERSIONS["pyarrow"])
         # Try setting num cpus to 1 to avoid hangs?
+        # Disable threads in CI environment to avoid timeouts
         import pyarrow
 
         pyarrow.set_cpu_count(1)
-    return parser
+
+        class NoThreadArrowParserWrapper(ArrowParserWrapper):
+            def _get_pyarrow_options(self):
+                super()._get_pyarrow_options()
+                self.read_options["use_threads"] = False
+
+        with monkeypatch.context() as m:
+            m.setattr(readers, "ArrowParserWrapper", NoThreadArrowParserWrapper)
+            yield parser
+    else:
+        yield parser
 
 
 @pytest.fixture(params=_c_parsers_only, ids=_c_parser_ids)
diff --git a/pandas/tests/io/test_sql.py b/pandas/tests/io/test_sql.py
index 741af4324c1a6..584308db3bae8 100644
--- a/pandas/tests/io/test_sql.py
+++ b/pandas/tests/io/test_sql.py
@@ -498,6 +498,7 @@ def sqlite_buildin_iris(sqlite_buildin, iris_path):
 all_connectable_iris = sqlalchemy_connectable_iris + ["sqlite_buildin_iris"]
 
 
+@pytest.mark.db
 @pytest.mark.parametrize("conn", all_connectable)
 @pytest.mark.parametrize("method", [None, "multi"])
 def test_to_sql(conn, method, test_frame1, request):
@@ -508,6 +509,7 @@ def test_to_sql(conn, method, test_frame1, request):
     assert count_rows(conn, "test_frame") == len(test_frame1)
 
 
+@pytest.mark.db
 @pytest.mark.parametrize("conn", all_connectable)
 @pytest.mark.parametrize("mode, num_row_coef", [("replace", 1), ("append", 2)])
 def test_to_sql_exist(conn, mode, num_row_coef, test_frame1, request):
@@ -519,6 +521,7 @@ def test_to_sql_exist(conn, mode, num_row_coef, test_frame1, request):
     assert count_rows(conn, "test_frame") == num_row_coef * len(test_frame1)
 
 
+@pytest.mark.db
 @pytest.mark.parametrize("conn", all_connectable)
 def test_to_sql_exist_fail(conn, test_frame1, request):
     conn = request.getfixturevalue(conn)
@@ -531,6 +534,7 @@ def test_to_sql_exist_fail(conn, test_frame1, request):
         pandasSQL.to_sql(test_frame1, "test_frame", if_exists="fail")
 
 
+@pytest.mark.db
 @pytest.mark.parametrize("conn", all_connectable_iris)
 def test_read_iris(conn, request):
     conn = request.getfixturevalue(conn)
@@ -539,6 +543,7 @@ def test_read_iris(conn, request):
     check_iris_frame(iris_frame)
 
 
+@pytest.mark.db
 @pytest.mark.parametrize("conn", sqlalchemy_connectable)
 def test_to_sql_callable(conn, test_frame1, request):
     conn = request.getfixturevalue(conn)
@@ -557,6 +562,7 @@ def sample(pd_table, conn, keys, data_iter):
     assert count_rows(conn, "test_frame") == len(test_frame1)
 
 
+@pytest.mark.db
 @pytest.mark.parametrize("conn", mysql_connectable)
 def test_default_type_conversion(conn, request):
     conn = request.getfixturevalue(conn)
@@ -575,6 +581,7 @@ def test_default_type_conversion(conn, request):
     assert issubclass(df.BoolColWithNull.dtype.type, np.floating)
 
 
+@pytest.mark.db
 @pytest.mark.parametrize("conn", mysql_connectable)
 def test_read_procedure(conn, request):
     conn = request.getfixturevalue(conn)
@@ -611,6 +618,7 @@ def test_read_procedure(conn, request):
     tm.assert_frame_equal(df, res2)
 
 
+@pytest.mark.db
 @pytest.mark.parametrize("conn", postgresql_connectable)
 def test_copy_from_callable_insertion_method(conn, request):
     # GH 8953
diff --git a/pandas/tests/window/test_numba.py b/pandas/tests/window/test_numba.py
index a9e38751c9a2f..36625d86873f6 100644
--- a/pandas/tests/window/test_numba.py
+++ b/pandas/tests/window/test_numba.py
@@ -1,6 +1,12 @@
+import os
+
 import numpy as np
 import pytest
 
+from pandas.compat import (
+    is_platform_mac,
+    is_platform_windows,
+)
 from pandas.errors import NumbaUtilError
 import pandas.util._test_decorators as td
 
@@ -13,6 +19,15 @@
 import pandas._testing as tm
 from pandas.core.util.numba_ import NUMBA_FUNC_CACHE
 
+# TODO: Mark these as pytest.mark.single GH 44584
+pytestmark = pytest.mark.skipif(
+    os.environ.get("PANDAS_CI", "0") == "1"
+    and (is_platform_windows() or is_platform_mac()),
+    reason="On Azure CI, Windows can fail with "
+    "'Windows fatal exception: stack overflow' "
+    "and MacOS can timeout",
+)
+
 
 @pytest.fixture(params=["single", "table"])
 def method(request):
diff --git a/pandas/tests/window/test_online.py b/pandas/tests/window/test_online.py
index 80cf1c55958ee..debf87bf6ade1 100644
--- a/pandas/tests/window/test_online.py
+++ b/pandas/tests/window/test_online.py
@@ -1,6 +1,12 @@
+import os
+
 import numpy as np
 import pytest
 
+from pandas.compat import (
+    is_platform_mac,
+    is_platform_windows,
+)
 import pandas.util._test_decorators as td
 
 from pandas import (
@@ -9,6 +15,15 @@
 )
 import pandas._testing as tm
 
+# TODO: Mark these as pytest.mark.single GH 44584
+pytestmark = pytest.mark.skipif(
+    os.environ.get("PANDAS_CI", "0") == "1"
+    and (is_platform_windows() or is_platform_mac()),
+    reason="On Azure CI, Windows can fail with "
+    "'Windows fatal exception: stack overflow' "
+    "and MacOS can timeout",
+)
+
 
 @td.skip_if_no("numba")
 @pytest.mark.filterwarnings("ignore:\n")
diff --git a/pandas/util/_test_decorators.py b/pandas/util/_test_decorators.py
index 78ef335adf948..10322a25ffd18 100644
--- a/pandas/util/_test_decorators.py
+++ b/pandas/util/_test_decorators.py
@@ -266,7 +266,8 @@ def file_leak_context():
     ContextManager analogue to check_file_leaks.
     """
     psutil = safe_import("psutil")
-    if not psutil:
+    if not psutil or is_platform_windows():
+        # Checking for file leaks can hang on Windows CI
         yield
     else:
         proc = psutil.Process()

From b904b8818e77a62beec2f4ffeb6d95b1295c94da Mon Sep 17 00:00:00 2001
From: Matthew Roeschke <emailformattr@gmail.com>
Date: Sat, 29 Jan 2022 18:29:55 -0800
Subject: [PATCH 2/3] Revert pyarrow change and change todo format

---
 pandas/tests/io/parser/conftest.py | 22 +++++-----------------
 pandas/tests/window/test_numba.py  |  2 +-
 pandas/tests/window/test_online.py |  2 +-
 3 files changed, 7 insertions(+), 19 deletions(-)

diff --git a/pandas/tests/io/parser/conftest.py b/pandas/tests/io/parser/conftest.py
index e98d23d12bbd0..b2d2be362d0d3 100644
--- a/pandas/tests/io/parser/conftest.py
+++ b/pandas/tests/io/parser/conftest.py
@@ -12,9 +12,6 @@
 )
 import pandas._testing as tm
 
-from pandas.io.parsers import readers
-from pandas.io.parsers.arrow_parser_wrapper import ArrowParserWrapper
-
 
 class BaseParser:
     engine: str | None = None
@@ -104,29 +101,20 @@ def csv1(datapath):
 
 
 @pytest.fixture(params=_all_parsers, ids=_all_parser_ids)
-def all_parsers(request, monkeypatch):
+def all_parsers(request):
     """
     Fixture all of the CSV parsers.
     """
     parser = request.param()
     if parser.engine == "pyarrow":
         pytest.importorskip("pyarrow", VERSIONS["pyarrow"])
-        # Try setting num cpus to 1 to avoid hangs?
-        # Disable threads in CI environment to avoid timeouts
+        # Try setting num cpus to 1 to avoid hangs on Azure MacOS/Windows builds
+        # or better yet find a way to disable threads
+        # TODO(GH#44584) pytest.mark.single these tests
         import pyarrow
 
         pyarrow.set_cpu_count(1)
-
-        class NoThreadArrowParserWrapper(ArrowParserWrapper):
-            def _get_pyarrow_options(self):
-                super()._get_pyarrow_options()
-                self.read_options["use_threads"] = False
-
-        with monkeypatch.context() as m:
-            m.setattr(readers, "ArrowParserWrapper", NoThreadArrowParserWrapper)
-            yield parser
-    else:
-        yield parser
+    return parser
 
 
 @pytest.fixture(params=_c_parsers_only, ids=_c_parser_ids)
diff --git a/pandas/tests/window/test_numba.py b/pandas/tests/window/test_numba.py
index 36625d86873f6..2c9ae3d70f218 100644
--- a/pandas/tests/window/test_numba.py
+++ b/pandas/tests/window/test_numba.py
@@ -19,7 +19,7 @@
 import pandas._testing as tm
 from pandas.core.util.numba_ import NUMBA_FUNC_CACHE
 
-# TODO: Mark these as pytest.mark.single GH 44584
+# TODO(GH#44584): Mark these as pytest.mark.single
 pytestmark = pytest.mark.skipif(
     os.environ.get("PANDAS_CI", "0") == "1"
     and (is_platform_windows() or is_platform_mac()),
diff --git a/pandas/tests/window/test_online.py b/pandas/tests/window/test_online.py
index debf87bf6ade1..543e5e4e545e0 100644
--- a/pandas/tests/window/test_online.py
+++ b/pandas/tests/window/test_online.py
@@ -15,7 +15,7 @@
 )
 import pandas._testing as tm
 
-# TODO: Mark these as pytest.mark.single GH 44584
+# TODO(GH#44584): Mark these as pytest.mark.single
 pytestmark = pytest.mark.skipif(
     os.environ.get("PANDAS_CI", "0") == "1"
     and (is_platform_windows() or is_platform_mac()),

From ca0b64211b37e5dd4bbaa4e0562c3b24fa29a9cc Mon Sep 17 00:00:00 2001
From: Matthew Roeschke <emailformattr@gmail.com>
Date: Sat, 29 Jan 2022 23:49:07 -0800
Subject: [PATCH 3/3] Remove -v

---
 ci/run_tests.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ci/run_tests.sh b/ci/run_tests.sh
index 344982920273e..020a3ed7f265c 100755
--- a/ci/run_tests.sh
+++ b/ci/run_tests.sh
@@ -24,7 +24,7 @@ if [[ $(uname) == "Linux" && -z $DISPLAY ]]; then
     XVFB="xvfb-run "
 fi
 
-PYTEST_CMD="${XVFB}pytest -v -r fEs -m \"$PATTERN\" -n $PYTEST_WORKERS --dist=loadfile $TEST_ARGS $COVERAGE $PYTEST_TARGET"
+PYTEST_CMD="${XVFB}pytest -r fEs -m \"$PATTERN\" -n $PYTEST_WORKERS --dist=loadfile $TEST_ARGS $COVERAGE $PYTEST_TARGET"
 
 if [[ $(uname) != "Linux"  && $(uname) != "Darwin" ]]; then
     PYTEST_CMD="$PYTEST_CMD --ignore=pandas/tests/plotting/"