Invalidate layer locks for launch module changes

Closes #89

Author: ncoghlan

docs/changelog.d/20250510_043944_ncoghlan_invalidate_app_layer_locks_on_launch_module_changes.rst

@@ -0,0 +1,5 @@
+Changed
+-------
+
+- Layer locks are now invalidated for launch module changes. This also means
+  that implicit versioning will update the layer version (resolves :issue:`89`).

src/venvstacks/stacks.py

@@ -2163,9 +2163,11 @@ def __post_init__(self) -> None:
         launch_module_path = self.env_spec.launch_module_path
         self.launch_module_name = launch_module_path.stem
         if launch_module_path.is_file():
-            self._launch_module_hash = _hash_file_name_and_contents(launch_module_path)
+            launch_module_hash = _hash_file_name_and_contents(launch_module_path)
         else:
-            self._launch_module_hash = _hash_directory(launch_module_path)
+            launch_module_hash = _hash_directory(launch_module_path)
+        self._launch_module_hash = launch_module_hash
+        self.env_lock.append_other_input(launch_module_hash)
 
     def _update_existing_environment(self, *, lock_only: bool = False) -> None:
         super()._update_existing_environment(lock_only=lock_only)

tests/expected-output-config.toml

@@ -33,4 +33,4 @@ UV_EXCLUDE_NEWER="2025-04-27 00:00:00+00:00"
 # Metadata updates can also be requested when the
 # launch module content in the sample project changes
 
-# Last requested update: include names in launch module hashes
+# Last requested update: include launch module hash in env lock checks

tests/test_env_locks.py

@@ -1,5 +1,6 @@
 """Test cases for environment lock management."""
 
+import shutil
 import tempfile
 import tomllib
 
@@ -137,8 +138,7 @@ def test_requirements_file_hashing(temp_dir_path: Path) -> None:
 ##################################
 
 EMPTY_SCRIPT_PATH = Path(__file__).parent / "minimal_project/empty.py"
-EMPTY_SCRIPT_PATH_POSIX = EMPTY_SCRIPT_PATH.as_posix()
-EXAMPLE_STACK_SPEC = f"""\
+EXAMPLE_STACK_SPEC = """\
 [[runtimes]]
 name = "cpython-to-be-modified"
 python_implementation = "[email protected]"
@@ -178,7 +178,7 @@ def test_requirements_file_hashing(temp_dir_path: Path) -> None:
 
 [[applications]]
 name = "to-be-modified"
-launch_module = "{EMPTY_SCRIPT_PATH_POSIX}"
+launch_module = "launch.py"
 # "dependent" must be first here to allow linearisation
 # when that layer also depends on "other-app-dependency"
 frameworks = ["dependent", "other-app-dependency"]
@@ -187,12 +187,12 @@ def test_requirements_file_hashing(temp_dir_path: Path) -> None:
 [[applications]]
 name = "runtime-only"
 runtime = "cpython-to-be-modified"
-launch_module = "{EMPTY_SCRIPT_PATH_POSIX}"
+launch_module = "launch2.py"
 requirements = []
 
 [[applications]]
 name = "unaffected"
-launch_module = "{EMPTY_SCRIPT_PATH_POSIX}"
+launch_module = "launch2.py"
 frameworks = ["unaffected"]
 requirements = []
 """
@@ -243,6 +243,9 @@ def _modified_file(file_path: Path, contents: str) -> Generator[Any, None, None]
 def test_build_env_layer_locks(temp_dir_path: Path, subtests: SubTests) -> None:
     # Built as a monolithic tests with subtests for performance reasons
     # (initial setup takes ~10 seconds, subsequent checks are fractions of a second)
+    launch_module_path = temp_dir_path / "launch.py"
+    shutil.copyfile(EMPTY_SCRIPT_PATH, launch_module_path)
+    shutil.copyfile(EMPTY_SCRIPT_PATH, temp_dir_path / "launch2.py")
     spec_path = temp_dir_path / "venvstacks.toml"
     updated_spec_path = temp_dir_path / "venvstacks_updated.toml"
     spec_data = tomllib.loads(EXAMPLE_STACK_SPEC)
@@ -505,6 +508,22 @@ def test_build_env_layer_locks(temp_dir_path: Path, subtests: SubTests) -> None:
         assert invalid_locks == expected_invalid_locks
         assert build_env._needs_lock()
         subtests_passed += 1
+    with subtests.test("Change launch module name at application layer"):
+        # Even if the locked requirements don't change, the layer needs updating
+        # This is due to the launch module needing to be invoked differently
+        subtests_started += 1
+        spec_data_to_check = tomllib.loads(EXAMPLE_STACK_SPEC)
+        env_spec_to_modify = spec_data_to_check["applications"][0]
+        assert env_spec_to_modify["name"] == "to-be-modified"
+        env_spec_to_modify["launch_module"] = "launch2.py"
+        build_env = _define_lock_testing_env(updated_spec_path, spec_data_to_check)
+        expected_invalid_locks = {"app-to-be-modified"}
+        expected_valid_locks = all_layer_names - expected_invalid_locks
+        valid_locks, invalid_locks = _partition_envs(build_env)
+        assert valid_locks == expected_valid_locks
+        assert invalid_locks == expected_invalid_locks
+        assert build_env._needs_lock()
+        subtests_passed += 1
     # Remaining subtests need to modify the actual envs, not just the layer specifications
     env_to_modify: LayerEnvBase
     with subtests.test("Change locked requirements at runtime layer"):
@@ -558,6 +577,22 @@ def test_build_env_layer_locks(temp_dir_path: Path, subtests: SubTests) -> None:
             assert invalid_locks == expected_invalid_locks
             assert build_env._needs_lock()
         subtests_passed += 1
+    with subtests.test("Change launch module content at application layer"):
+        # Even if the locked requirements don't change, the layer needs updating
+        # This is due to the launch module needing to be invoked differently
+        subtests_started += 1
+        spec_data_to_check = tomllib.loads(EXAMPLE_STACK_SPEC)
+        with _modified_file(
+            launch_module_path, "# Changed launch module contents"
+        ):
+            build_env = _define_lock_testing_env(updated_spec_path, spec_data_to_check)
+            expected_invalid_locks = {"app-to-be-modified"}
+            expected_valid_locks = all_layer_names - expected_invalid_locks
+            valid_locks, invalid_locks = _partition_envs(build_env)
+            assert valid_locks == expected_valid_locks
+            assert invalid_locks == expected_invalid_locks
+            assert build_env._needs_lock()
+        subtests_passed += 1
 
     # Work around pytest-subtests not failing the test case when subtests fail
     # https://github.com/pytest-dev/pytest-subtests/issues/76