feature: Detect ecosystem mismatches (#45)

Ecosystem language and version is added to both lock file manifests and
to environment state files. This allows detecting when configuration is
changed and there are mismatches between either state and manifest, or
between config and state.

When a mismatch is discovered, the existing environment is deleted and
rebuilt. When a mismatch between manifest and config is discovered, this
requires running `goose upgrade`.

This breaks existing configurations, and requires rebuilding
configuration as well as manually deleting existing environments.

.goose/node/manifest.json

@@ -1 +1 @@
-{"source_dependencies":["prettier"],"lock_files":[{"path":"package-lock.json","checksum":"sha256:c72f3206d345d7285c175a9363dc4d7efbb56729dd02e4828c9356ecfaaad0ca"},{"path":"package.json","checksum":"sha256:c5008cb00d6009f6d901cfe35523169a3deb91979b71fce65d6ac79fdace548b"}],"checksum":"sha256:e130609d6f3cb3681a612f936c5fac5c409019253d18071772d86949766d0bbb"}
+{"source_ecosystem":{"language":"node","version":"23.0.0"},"source_dependencies":["prettier"],"lock_files":[{"path":"package-lock.json","checksum":"sha256:c72f3206d345d7285c175a9363dc4d7efbb56729dd02e4828c9356ecfaaad0ca"},{"path":"package.json","checksum":"sha256:c5008cb00d6009f6d901cfe35523169a3deb91979b71fce65d6ac79fdace548b"}],"checksum":"sha256:e130609d6f3cb3681a612f936c5fac5c409019253d18071772d86949766d0bbb"}

.goose/python/manifest.json

@@ -1 +1 @@
-{"source_dependencies":["pre-commit-hooks","ruff"],"lock_files":[{"path":"requirements.txt","checksum":"sha256:f649a1f943c1cfbd71571de02aedcf7d88e7427c0e985ec26f399d8c40970c8e"}],"checksum":"sha256:2087aaf2b95093aae8f2cfef2b6a55aaddd9e94d1814b268ea2fd699f0a10f11"}
+{"source_ecosystem":{"language":"python","version":"3.13"},"source_dependencies":["pre-commit-hooks","ruff"],"lock_files":[{"path":"requirements.txt","checksum":"sha256:f649a1f943c1cfbd71571de02aedcf7d88e7427c0e985ec26f399d8c40970c8e"}],"checksum":"sha256:2087aaf2b95093aae8f2cfef2b6a55aaddd9e94d1814b268ea2fd699f0a10f11"}

.goose/type-check/manifest.json

@@ -1 +1 @@
-{"source_dependencies":["identify==2.6.0","mypy","pydantic==2.8.2","rich==13.7.1","typer==0.12.3","types-pyyaml==6.0.12.20240724"],"lock_files":[{"path":"requirements.txt","checksum":"sha256:30a32dcd426ab99471b1d43f5a575db3d8df2b5b64ce3dc756ce730f7f0b388c"}],"checksum":"sha256:40b81307cd69b710976f92453cdf92ef743fdd9882eabea9c6b6b6688779a516"}
+{"source_ecosystem":{"language":"python","version":"3.13"},"source_dependencies":["identify==2.6.0","mypy","pydantic==2.8.2","rich==13.7.1","typer==0.12.3","types-pyyaml==6.0.12.20240724"],"lock_files":[{"path":"requirements.txt","checksum":"sha256:30a32dcd426ab99471b1d43f5a575db3d8df2b5b64ce3dc756ce730f7f0b388c"}],"checksum":"sha256:40b81307cd69b710976f92453cdf92ef743fdd9882eabea9c6b6b6688779a516"}

src/goose/backend/node.py

@@ -116,6 +116,7 @@ async def freeze(
 
     package_lock_json_path = lock_files_path / "package-lock.json"
     manifest = build_manifest(
+        source_ecosystem=config.ecosystem,
         source_dependencies=config.dependencies,
         lock_files=(
             package_json_path,

src/goose/backend/python.py

@@ -138,6 +138,7 @@ async def freeze(
         raise RuntimeError("Failed freezing dependencies {process.returncode=}")
 
     manifest = build_manifest(
+        source_ecosystem=config.ecosystem,
         source_dependencies=config.dependencies,
         lock_files=(requirements_txt,),
         lock_files_path=lock_files_path,

src/goose/environment.py

@@ -1,5 +1,7 @@
+import asyncio
 import enum
 import os
+import shutil
 import sys
 from collections.abc import Mapping
 from pathlib import Path
@@ -15,6 +17,7 @@
 from .backend.base import RunResult
 from .backend.index import load_backend
 from .config import Config
+from .config import EcosystemConfig
 from .config import EnvironmentConfig
 from .config import EnvironmentId
 from .executable_unit import ExecutableUnit
@@ -27,7 +30,6 @@ class NeedsFreeze(Exception): ...
 
 
 class InitialStage(enum.Enum):
-    new = "new"
     bootstrapped = "bootstrapped"
     frozen = "frozen"
 
@@ -39,23 +41,30 @@ class SyncedStage(enum.Enum):
 class SyncedState(BaseModel):
     stage: SyncedStage
     checksum: str
+    ecosystem: EcosystemConfig
 
 
 class InitialState(BaseModel):
     stage: InitialStage
+    ecosystem: EcosystemConfig
 
 
-_PersistedState = RootModel[InitialState | SyncedState]
+class UninitializedState: ...
 
 
-def read_state(env_dir: Path) -> InitialState | SyncedState:
+type State = SyncedState | InitialState | UninitializedState
+
+_PersistedState = RootModel[SyncedState | InitialState]
+
+
+def read_state(env_dir: Path) -> State:
     state_file = env_dir / "goose-state.json"
     if not state_file.exists():
-        return InitialState(stage=InitialStage.new)
+        return UninitializedState()
     return _PersistedState.model_validate_json(state_file.read_bytes()).root
 
 
-def write_state(env_dir: Path, state: InitialState | SyncedState) -> None:
+def write_state(env_dir: Path, state: SyncedState | InitialState) -> None:
     state_file = env_dir / "goose-state.json"
     state_file.write_text(state.model_dump_json())
 
@@ -67,7 +76,7 @@ def __init__(
         config: EnvironmentConfig,
         path: Path,
         lock_files_path: Path,
-        discovered_state: InitialState | SyncedState,
+        discovered_state: State,
     ) -> None:
         self.config: Final = config
         self._backend: Final = load_backend(config.ecosystem)
@@ -80,8 +89,13 @@ def __init__(
     def __repr__(self) -> str:
         return f"Environment(id={self.config.id}, ecosystem={self._backend.ecosystem})"
 
+    def check_should_teardown(self) -> bool:
+        if isinstance(self.state, UninitializedState):
+            return False
+        return self.config.ecosystem != self.state.ecosystem
+
     def check_should_bootstrap(self) -> bool:
-        if self.state.stage is InitialStage.new:
+        if isinstance(self.state, UninitializedState):
             return True
         if not self._path.exists():
             print("State mismatch: environment does not exist")
@@ -153,12 +167,19 @@ def check_should_sync(self) -> bool:
         else:
             assert_never(state)
 
+    async def teardown(self) -> None:
+        await asyncio.to_thread(shutil.rmtree, self._path)
+        self.state = UninitializedState()
+
     async def bootstrap(self) -> None:
         await self._backend.bootstrap(
             env_path=self._path,
             config=self.config,
         )
-        self.state = InitialState(stage=InitialStage.bootstrapped)
+        self.state = InitialState(
+            stage=InitialStage.bootstrapped,
+            ecosystem=self.config.ecosystem,
+        )
         write_state(self._path, self.state)
 
     async def freeze(self) -> None:
@@ -167,7 +188,10 @@ async def freeze(self) -> None:
             config=self.config,
             lock_files_path=self.lock_files_path,
         )
-        self.state = InitialState(stage=InitialStage.frozen)
+        self.state = InitialState(
+            stage=InitialStage.frozen,
+            ecosystem=self.config.ecosystem,
+        )
         write_state(self._path, self.state)
         os.sync()
 
@@ -181,6 +205,7 @@ async def sync(self) -> None:
         self.state = SyncedState(
             stage=SyncedStage.synced,
             checksum=manifest.checksum,
+            ecosystem=self.config.ecosystem,
         )
         write_state(self._path, self.state)
 
@@ -228,6 +253,17 @@ async def prepare_environment(
 ) -> None:
     log_prefix = f"[{environment.config.id}] "
 
+    if environment.check_should_teardown():
+        print(
+            f"{log_prefix}Environment needs rebuilding, tearing down ...",
+            file=sys.stderr,
+        )
+        await environment.teardown()
+        print(
+            f"{log_prefix}Environment deleted.",
+            file=sys.stderr,
+        )
+
     if environment.check_should_bootstrap():
         print(f"{log_prefix}Bootstrapping environment ...", file=sys.stderr)
         await environment.bootstrap()

src/goose/manifest.py

@@ -13,6 +13,7 @@
 from pydantic import field_validator
 
 from ._utils.pydantic import BaseModel
+from .config import EcosystemConfig
 from .config import EnvironmentConfig
 
 
@@ -35,6 +36,7 @@ def __lt__(self, other: Self) -> bool:
 
 
 class LockManifest(BaseModel):
+    source_ecosystem: EcosystemConfig
     source_dependencies: tuple[str, ...]
     lock_files: tuple[LockFile, ...]
     checksum: str
@@ -94,6 +96,7 @@ def read_lock_file(lock_files_path: Path, path: Path) -> LockFile:
 
 
 def build_manifest(
+    source_ecosystem: EcosystemConfig,
     source_dependencies: Iterable[str],
     lock_files: Iterable[Path],
     lock_files_path: Path,
@@ -102,6 +105,7 @@ def build_manifest(
         sorted(read_lock_file(lock_files_path, path) for path in lock_files)
     )
     return LockManifest(
+        source_ecosystem=source_ecosystem,
         source_dependencies=tuple(sorted(source_dependencies)),
         lock_files=lock_file_instances,
         checksum=_get_accumulated_checksum(lock_file_instances),
@@ -140,6 +144,9 @@ def check_lock_files(
     except FileNotFoundError:
         return LockFileState.config_manifest_mismatch
 
+    if config.ecosystem != manifest.source_ecosystem:
+        return LockFileState.config_manifest_mismatch
+
     if set(config.dependencies) ^ set(manifest.source_dependencies):
         return LockFileState.config_manifest_mismatch
 

src/goose/scheduler.py

@@ -7,7 +7,6 @@
 from itertools import chain
 from typing import Final
 from typing import NamedTuple
-from typing import TypeAlias
 
 from .backend.base import RunResult
 from .config import HookConfig
@@ -26,11 +25,10 @@ class UnitFinished(NamedTuple):
     result: RunResult
 
 
-SchedulerEvent: TypeAlias = UnitScheduled | UnitFinished
-
-
-SchedulerState: TypeAlias = Mapping[
-    HookConfig, Mapping[ExecutableUnit, RunResult | asyncio.Task[RunResult] | None]
+type SchedulerEvent = UnitScheduled | UnitFinished
+type SchedulerState = Mapping[
+    HookConfig,
+    Mapping[ExecutableUnit, RunResult | asyncio.Task[RunResult] | None],
 ]