test: add a small set of ops.testing benchmark tests

.gitignore

@@ -33,3 +33,6 @@ test/charms/test_smoke/.charmcraft_output_packages.txt
 test/charms/test_smoke/requirements.txt
 test/charms/test_smoke/charmcraft.yaml
 juju-crashdump*.tar.xz
+
+# Benchmark test artifacts
+.benchmarks

test/benchmark/__init__.py

@@ -0,0 +1,24 @@
+# Copyright 2024 Canonical Ltd.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Benchmark tests for ops.
+
+Optimising performance is not a current goal with ops - any gains are
+unlikely to be significant compared with ones from Juju or the charm and
+its workload. However, we do want to ensure that we do not unknowingly
+regress in performance.
+
+This package is for tests that cover core functionality, to be used for
+performance benchmarking.
+"""

test/charms/test_benchmark/charmcraft.yaml

@@ -0,0 +1,40 @@
+name: benchmark
+type: charm
+title: ops-benchmark
+summary: A simple charm used for benchmark tests
+description: Read the summary.
+bases:
+  - build-on:
+    - name: ubuntu
+      channel: "22.04"
+    run-on:
+    - name: ubuntu
+      channel: "22.04"
+config:
+  options:
+    log-level:
+      description: Configures the log level.
+      default: "info"
+      type: string
+actions:
+  act:
+    description: Do something to the workload.
+containers:
+  foo:
+resources:
+  baz:
+    type: oci-image
+storage:
+  bar:
+    type: filesystem
+requires:
+  rel:
+    interface: qux
+peers:
+  peer:
+    interface: chat
+extra-bindings:
+  MySpace: null
+parts:
+  charm:
+    charm-entrypoint: src/bcharm.py

test/charms/test_benchmark/requirements.txt

@@ -0,0 +1 @@
+ops ~= 2.17

test/charms/test_benchmark/src/benchmark_charm.py

@@ -0,0 +1,43 @@
+#!/usr/bin/env python3
+# Copyright 2024 Canonical Ltd.
+# See LICENSE file for licensing details.
+
+"""Basic benchmarking charm.
+
+Note that this is named benchmark_charm rather than charm as is typical to
+avoid conflicts with ops.charm in the testing runs.
+"""
+
+import logging
+
+import ops
+
+logger = logging.getLogger('__name__')
+
+
+class BenchmarkCharm(ops.CharmBase):
+    """Charm the service."""
+
+    _stored = ops.StoredState()
+
+    def __init__(self, framework: ops.Framework):
+        super().__init__(framework)
+        framework.observe(self.on.update_status, self._on_update_status)
+        framework.observe(self.on.stop, self._on_stop)
+        framework.observe(self.on.config_changed, self._on_config_changed)
+
+    def _on_update_status(self, _: ops.UpdateStatusEvent):
+        # Say a bunch of things.
+        for level in ('debug', 'info', 'warning', 'error'):
+            for i in range(50):
+                getattr(logger, level)('This is message %s', i)
+
+    def _on_stop(self, _: ops.StopEvent):
+        """Do nothing - this exists to benchmark having an observer."""
+
+    def _on_config_changed(self, event: ops.ConfigChangedEvent):
+        event.defer()
+
+
+if __name__ == '__main__':  # pragma: nocover
+    ops.main(BenchmarkCharm)

testing/tests/benchmark/__init__.py

@@ -0,0 +1,22 @@
+# Copyright 2024 Canonical Ltd.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Benchmark tests for ops-scenario.
+
+Optimising performance is not a current goal with ops-scenario. However,
+we do want to ensure that we do not unknowingly regress in performance.
+
+This package contains a small set of tests that cover core functionality,
+to be used for performance benchmarking.
+"""

testing/tests/benchmark/test_testing.py

@@ -0,0 +1,154 @@
+# Copyright 2024 Canonical Ltd.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Benchmark tests for ops-scenario."""
+
+import dataclasses
+import pathlib
+import sys
+
+import ops
+from ops import testing
+
+sys.path.append(
+    str(
+        pathlib.Path(__file__).parent.parent.parent.parent
+        / "test"
+        / "charms"
+        / "test_benchmark"
+        / "src"
+    )
+)
+
+from benchmark_charm import BenchmarkCharm
+
+
+# Note: the 'benchmark' argument here is a fixture that pytest-benchmark
+# automatically makes available to all tests.
+def test_context_explicit_meta(benchmark):
+    ctx = benchmark(testing.Context, ops.CharmBase, meta={"name": "foo"})
+    assert isinstance(ctx, testing.Context)
+
+
+def test_run_no_observer(benchmark):
+    ctx = testing.Context(BenchmarkCharm)
+    benchmark(ctx.run, ctx.on.start(), testing.State())
+    assert len({e.handle.kind for e in ctx.emitted_events}) == 1
+
+
+def test_run_observed(benchmark):
+    ctx = testing.Context(BenchmarkCharm)
+    benchmark(ctx.run, ctx.on.stop(), testing.State())
+    assert len({e.handle.kind for e in ctx.emitted_events}) == 1
+
+
+def test_context_explicit_meta_config_actions(benchmark):
+    ctx = benchmark(
+        testing.Context,
+        ops.CharmBase,
+        meta={"name": "foo"},
+        actions={"act": {"description": "foo"}},
+        config={"options": {"conf": {"type": "int", "description": "bar"}}},
+    )
+    ctx.run(ctx.on.action("act"), testing.State(config={"conf": 10}))
+    assert len({e.handle.kind for e in ctx.emitted_events}) == 1
+
+
+def test_context_autoload_meta(benchmark):
+    ctx = benchmark(testing.Context, BenchmarkCharm)
+    assert isinstance(ctx, testing.Context)
+
+
+def test_many_tests_explicit_meta(benchmark):
+    def mock_pytest():
+        """Simulate running multiple tests against the same charm."""
+        for event in ("install", "start", "stop", "remove"):
+            for _ in range(5):
+                ctx = testing.Context(ops.CharmBase, meta={"name": "foo"})
+                ctx.run(getattr(ctx.on, event)(), testing.State())
+                assert len({e.handle.kind for e in ctx.emitted_events}) == 1
+
+    benchmark(mock_pytest)
+
+
+def test_many_tests_autoload_meta(benchmark):
+    def mock_pytest():
+        """Simulate running multiple tests against the same charm."""
+        for event in ("install", "start", "stop", "remove"):
+            for _ in range(5):
+                ctx = testing.Context(BenchmarkCharm)
+                ctx.run(getattr(ctx.on, event)(), testing.State())
+                assert len({e.handle.kind for e in ctx.emitted_events}) == 1
+
+    benchmark(mock_pytest)
+
+
+def test_lots_of_logs(benchmark):
+    ctx = testing.Context(BenchmarkCharm)
+    benchmark(ctx.run, ctx.on.update_status(), testing.State())
+    assert len(ctx.juju_log) > 200
+
+
+def test_full_state(benchmark):
+    def fill_state():
+        rel = testing.Relation("rel")
+        peer = testing.PeerRelation("peer")
+        network = testing.Network("MySpace")
+        container = testing.Container("foo")
+        storage = testing.Storage("bar")
+        tcp = testing.TCPPort(22)
+        icmp = testing.ICMPPort()
+        udp = testing.UDPPort(8000)
+        secret = testing.Secret({"password": "admin"})
+        resource = testing.Resource(name="baz", path=".")
+        stored_state = testing.StoredState(owner_path="BenchMarkCharm")
+        state = testing.State(
+            relations={rel, peer},
+            networks={network},
+            containers={container},
+            storages={storage},
+            opened_ports={tcp, icmp, udp},
+            secrets={secret},
+            resources={resource},
+            stored_states={stored_state},
+            app_status=testing.ActiveStatus(),
+            unit_status=testing.BlockedStatus("I'm stuck!"),
+        )
+        return state
+
+    ctx = testing.Context(BenchmarkCharm)
+    state_in = benchmark(fill_state)
+    state_out = ctx.run(ctx.on.start(), state_in)
+    # stored_states is complicated: it will contain a stored state the
+    # framework itself added (counting the number of events), so the
+    # input and output state doesn't naively match. We strip that out and
+    # compare it separately.
+    state_in_dict = dataclasses.asdict(state_in)
+    state_out_dict = dataclasses.asdict(state_out)
+    # An owner_path of None means that it's owned by the framework.
+    assert state_in_dict["stored_states"] == {
+        ss for ss in state_out_dict["stored_states"] if ss.owner_path is not None
+    }
+    del state_in_dict["stored_states"]
+    del state_out_dict["stored_states"]
+    assert state_in_dict == state_out_dict
+
+
+def test_deferred_events(benchmark):
+    ctx = testing.Context(BenchmarkCharm, capture_deferred_events=True)
+    deferred = ctx.on.stop().deferred(BenchmarkCharm._on_stop)
+    state_in = testing.State(deferred=[deferred])
+    state_out = benchmark(ctx.run, ctx.on.config_changed(), state_in)
+    assert len(state_out.deferred) == 1
+    assert len({e.handle.kind for e in ctx.emitted_events}) == 2

tox.ini

@@ -23,6 +23,8 @@ envlist = lint, static, unit
 src_path = ops/
 tst_path = test/
 all_path = {[vars]src_path} {[vars]tst_path}
+testing_src_path = testing/src/scenario/
+testing_tst_path = testing/tests/
 
 [testenv]
 basepython = python3
@@ -104,7 +106,9 @@ deps =
     -e .
     -e testing
 commands =
-    pytest -n auto --ignore={[vars]tst_path}smoke -v --tb native \
+    pytest -n auto --ignore={[vars]tst_path}smoke \
+        --ignore={[vars]tst_path}benchmark --ignore={[vars]testing_tst_path}benchmark \
+        -v --tb native \
         -W 'ignore:Harness is deprecated:PendingDeprecationWarning' {posargs}
 
 [testenv:coverage]
@@ -124,11 +128,30 @@ deps =
     -e testing
 commands =
     mkdir -p .report
-    coverage run --source={[vars]src_path},testing/src/scenario \
-             -m pytest --ignore={[vars]tst_path}smoke -v --tb native {posargs}
+    coverage run --source={[vars]src_path},{[vars]testing_src_path} \
+             -m pytest --ignore={[vars]tst_path}smoke \
+             --ignore={[vars]tst_path}benchmark --ignore={[vars]testing_tst_path}benchmark \
+             -v --tb native \
+             -W 'ignore:Harness is deprecated:PendingDeprecationWarning' {posargs}
     coverage xml -o .report/coverage.xml
     coverage report
 
+[testenv:benchmark]
+description = Run benchmark tests
+passenv =
+    RUN_REAL_PEBBLE_TESTS
+    PEBBLE
+deps =
+    PyYAML==6.*
+    websocket-client==1.*
+    pytest~=7.2
+    pytest-benchmark
+    typing_extensions~=4.2
+    -e .
+    -e testing
+commands =
+    pytest -v --tb native {[vars]tst_path}benchmark {[vars]testing_tst_path}benchmark {posargs}
+
 [testenv:pebble]
 description = Run real pebble tests
 allowlist_externals = pebble