fix(log): keep log emitter running until the run managed (#290)

The log emitter paused too early that not logging the command it runs
in the managed mode.

craft_application/application.py

@@ -344,14 +344,19 @@ def run_managed(self, platform: str | None, build_for: str | None) -> None:
             with self.services.provider.instance(
                 build_info, work_dir=self._work_dir
             ) as instance:
+                cmd = [self.app.name, *sys.argv[1:]]
+                craft_cli.emit.debug(
+                    f"Executing {cmd} in instance location {instance_path} with {extra_args}."
+                )
                 try:
-                    # Pyright doesn't fully understand craft_providers's CompletedProcess.
-                    instance.execute_run(  # pyright: ignore[reportUnknownMemberType,reportUnknownVariableType]
-                        [self.app.name, *sys.argv[1:]],
-                        cwd=instance_path,
-                        check=True,
-                        **extra_args,
-                    )
+                    with craft_cli.emit.pause():
+                        # Pyright doesn't fully understand craft_providers's CompletedProcess.
+                        instance.execute_run(  # pyright: ignore[reportUnknownMemberType,reportUnknownVariableType]
+                            cmd,
+                            cwd=instance_path,
+                            check=True,
+                            **extra_args,
+                        )
                 except subprocess.CalledProcessError as exc:
                     raise craft_providers.ProviderError(
                         f"Failed to execute {self.app.name} in instance."

craft_application/services/provider.py

@@ -133,8 +133,7 @@ def instance(
             emit.debug("Instance launched and working directory mounted")
             self._setup_instance_bashrc(instance)
             try:
-                with emit.pause():
-                    yield instance
+                yield instance
             finally:
                 self._capture_logs_from_instance(instance)
 

tests/unit/services/test_provider.py

@@ -290,15 +290,13 @@ def test_instance(
     provider_service,
     base_name,
     allow_unstable,
-    mocker,
 ):
     mock_provider = mock.MagicMock(spec=craft_providers.Provider)
     monkeypatch.setattr(
         provider_service,
         "get_provider",
         lambda name: mock_provider,  # noqa: ARG005 (unused argument)
     )
-    spy_pause = mocker.spy(provider.emit, "pause")
     arch = util.get_host_architecture()
     build_info = models.BuildInfo("foo", arch, arch, base_name)
 
@@ -326,8 +324,6 @@ def test_instance(
         )
     with check:
         emitter.assert_progress("Launching managed .+ instance...", regex=True)
-    with check:
-        assert spy_pause.call_count == 1
 
 
 def test_load_bashrc(emitter):

tests/unit/test_application.py

@@ -237,11 +237,12 @@ def test_log_path(monkeypatch, app, provider_managed, expected):
     assert actual == expected
 
 
-def test_run_managed_success(app, fake_project, fake_build_plan):
+def test_run_managed_success(mocker, app, fake_project, fake_build_plan):
     mock_provider = mock.MagicMock(spec_set=services.ProviderService)
     app.services.provider = mock_provider
     app.project = fake_project
     app._build_plan = fake_build_plan
+    mock_pause = mocker.spy(craft_cli.emit, "pause")
     arch = get_host_architecture()
 
     app.run_managed(None, arch)
@@ -253,6 +254,7 @@ def test_run_managed_success(app, fake_project, fake_build_plan):
         )
         in mock_provider.instance.mock_calls
     )
+    mock_pause.assert_called_once()
 
 
 def test_run_managed_failure(app, fake_project, fake_build_plan):