Self-evolving Context Database for AI Agents. Unify Agent Memory, Knowledge RAG and Skills.
Python agent-contextagent-memoryagent-pluginsagentic-ragcontext-databasedsh-pluginknowledge-basememorymemory-managementself-evolvingvector-database
fix(storage): 缩小写锁范围并跳过父目录刷新冲突 (#5014)
Qin Haojiea9d908e1313d
diff --git a/docs/en/api/12-content.md b/docs/en/api/12-content.mdindex 5d4bc1643..f1bf45ccc 100644--- a/docs/en/api/12-content.md+++ b/docs/en/api/12-content.md@@ -231,7 +231,8 @@ Write a file and automatically refresh related semantics and vectors.- Explicit `create` only accepts text-writable extensions: `.md`, `.txt`, `.json`, `.yaml`, `.yml`, `.toml`, `.py`, `.js`, `.ts`. Parent directories are created automatically for every write mode.- Existing `.abstract.md` and `.overview.md` bodies may be updated, but public APIs cannot create them. A body-only request preserves stored OKF metadata; a full-OKF request must match the stored metadata. Unknown metadata fields are silently dropped. A sidecar body write rebuilds only the directory's existing L0/L1 vectors and does not regenerate semantics.- File content is updated before the API returns. `wait` only controls whether the call waits for semantic/vector refresh to finish.-- The public API no longer accepts `regenerate_semantics` or `revectorize`; write always refreshes related semantics and vectors.+- The public API no longer accepts `regenerate_semantics` or `revectorize`; write automatically schedules related semantic and vector processing.+- Parent L0/L1 refreshes for resource writes are best-effort: a parent lock conflict skips that directory refresh while preserving the file write and its own summary/vector work. Skipping L0/L1 persistence also skips directory vector updates; a later refresh is not guaranteed. Locks on the written file itself still raise conflicts. Contention detected before enqueueing returns `semantic_status: "skipped"`; skips during background execution are logged, and `wait=true` does not guarantee updated parent summaries.- When `tags` is supplied, tags are included in the file's first vector upsert rather than updated after processing. Omitting `tags` preserves existing tags; explicit `tags: []` with `tag_mode: "replace"` clears them.@@ -365,10 +366,11 @@ Each operation contains:- All targets must be files below `root_uri`, use the same context type, and have unique canonical URIs.- Resource targets may use any safe file extension; Memory targets retain the text extension allowlist and do not accept binary content.- `replace`, `append`, and `create` match `write()` semantics. `upsert` replaces an existing file or creates a missing file.-- The batch holds one target tree lock while writing. Semantic processing starts only after every file is written and the lock is released, so `.overview.md` and `.abstract.md` are refreshed once for the batch.+- The batch acquires exact locks for all target files before validating file state and writing. Writes to disjoint files in the same directory can proceed concurrently; overlapping writes and parent-directory deletion or moves still conflict. Semantic processing starts after all writes finish and the locks are released, refreshing the affected `.overview.md` and `.abstract.md` files together.+- Resource parent refreshes use the same best-effort behavior as `write()`: L0/L1 lock conflicts skip the directory refresh and its vector updates while preserving file writes and file processing. A later refresh is not guaranteed.- An underlying I/O failure can still leave writes completed earlier in the batch visible.- Existing `.abstract.md` and `.overview.md` bodies may be replaced or appended. OpenViking preserves and validates protected OKF metadata and rebuilds only the directory's existing L0/L1 vectors for these operations.-- In the response body, `semantic_status` (`queued`, `complete`, or `deferred`) reports the directory aggregation status, while `vector_status` reports vector maintenance for changed files.+- In the response body, `semantic_status` (`queued`, `complete`, `deferred`, or `skipped`) reports the directory aggregation status; it is `skipped` if any directory encounters contention before enqueueing. Meanwhile, `vector_status` reports vector maintenance for changed files.**Python SDK**diff --git a/docs/zh/api/12-content.md b/docs/zh/api/12-content.mdindex c2b765010..9d9556841 100644--- a/docs/zh/api/12-content.md+++ b/docs/zh/api/12-content.md@@ -231,7 +231,8 @@ openviking read viking://resources/docs/api.md- 显式 `create` 只允许以下文本类扩展名:`.md`、`.txt`、`.json`、`.yaml`、`.yml`、`.toml`、`.py`、`.js`、`.ts`。所有写入模式都会自动创建父目录。- 已存在的 `.abstract.md` / `.overview.md` 可以修改正文,但不能通过公共 API 创建;只提交正文时会保留现有 OKF metadata,提交完整 OKF 时 metadata 必须与存量值一致。未知 metadata 字段会静默丢弃。sidecar 正文写入只重建该目录实际存在的 L0/L1 向量,不触发语义重新生成。- 文件内容会在 API 返回前完成更新;`wait` 只控制是否等待语义/向量刷新完成。-- 公共 API 已不再接受 `regenerate_semantics` 或 `revectorize`;写入后一定会自动刷新相关语义与向量。+- 公共 API 已不再接受 `regenerate_semantics` 或 `revectorize`;写入后会自动调度相关语义与向量处理。+- 资源写入附带的父目录 L0/L1 刷新采用尽力更新:父目录锁冲突时跳过本次目录刷新,保留原文写入及文件自身的摘要、向量处理。跳过 L0/L1 写回时也跳过目录向量更新,不保证自动补刷;原文文件本身的锁冲突仍报错。提交任务前发现冲突时返回 `semantic_status: "skipped"`;后台运行期间的跳过记录在日志中,`wait=true` 也不保证父目录摘要更新。- 提供 `tags` 时,标签会在该文件首次向量 upsert 时写入,而非在处理完成后再单独更新。省略 `tags` 不改变已有标签;显式 `tags: []` 配合 `tag_mode: "replace"` 可清空标签。@@ -365,10 +366,11 @@ openviking write viking://resources/docs/api.md \- 所有目标必须是 `root_uri` 下的文件、属于同一 context type,且 canonical URI 不能重复。- Resource 目标允许任意安全文件扩展名;Memory 目标仍使用文本扩展名白名单,且不接受二进制内容。- `replace`、`append`、`create` 与 `write()` 语义一致;`upsert` 会覆盖已有文件或创建缺失文件。-- 写入期间整批共用一个目标 tree lock。所有文件写完并释放锁后才启动语义处理,因此 `.overview.md` / `.abstract.md` 每批只统一刷新一次。+- 整批先获取所有目标文件的精确锁,再校验文件状态并写入;同一目录下不涉及相同文件的写入可以并行,重叠文件的写入或父目录删除、移动仍会冲突。所有文件写完并释放锁后才启动语义处理,统一刷新受影响的 `.overview.md` / `.abstract.md`。+- 资源父目录刷新与 `write()` 一样采用尽力更新:L0/L1 锁冲突时跳过目录刷新及对应目录向量更新,保留原文和文件自身的处理,不保证自动补刷。- 底层 I/O 中途失败时,本批次较早完成的写入仍可能已经可见。- 已存在的 `.abstract.md` / `.overview.md` 可以 replace 或 append;系统会保留并校验受保护的 OKF metadata,并只重建对应目录实际存在的 L0/L1 向量。-- 响应体中,通过 `semantic_status`(`queued`、`complete` 或 `deferred`)表达目录聚合状态,通过 `vector_status` 表达变化文件的向量维护状态。+- 响应体中,通过 `semantic_status`(`queued`、`complete`、`deferred` 或 `skipped`)表达目录聚合状态;提交任务前任一目录因锁冲突跳过时为 `skipped`,通过 `vector_status` 表达变化文件的向量维护状态。**Python SDK**diff --git a/openviking/storage/content_write.py b/openviking/storage/content_write.pyindex 90b0e2cbf..a236d20ff 100644--- a/openviking/storage/content_write.py+++ b/openviking/storage/content_write.py@@ -95,7 +95,9 @@ class _BatchRefreshOutcome:embedding_requested: bool = Falsedef statuses(self, *, wait: bool) -> tuple[str, str]:- if self.semantic_actions:+ if FreshnessAction.NOOP in self.semantic_actions:+ semantic_status = "skipped"+ elif self.semantic_actions:semantic_status = ("deferred"if all(action is FreshnessAction.MARK_PENDING for action in self.semantic_actions)@@ -215,8 +217,8 @@ class ContentWriteCoordinator:Each operation follows the same create/replace/append semantics as ``write``;``upsert`` is available for callers that already hold the desired final tree.- Refresh runs only after every write and after releasing the tree lock, so derived- summaries are generated once per batch.+ All target files stay locked from state validation through the last write,+ while unrelated files remain writable. Refresh starts after the locks are released."""normalized_root = self._validate_uri_path(root_uri, field_name="root_uri")await self._validate_batch_root(normalized_root, ctx=ctx)@@ -224,9 +226,12 @@ class ContentWriteCoordinator:normalized_root, operations, ctx=ctx)- root_path = self._viking_fs._uri_to_path(normalized_root, ctx=ctx)+ target_paths = [+ self._viking_fs._uri_to_path(operation["uri"], ctx=ctx)+ for operation in normalized_operations+ ]try:- lease = await self._viking_fs._async_agfs.pathlock_acquire_tree(root_path)+ lease = await self._viking_fs._async_agfs.pathlock_acquire_exact_batch(target_paths)except LockAcquisitionError as exc:raise ResourceBusyError(f"resource is busy and cannot be written now: {normalized_root}",@@ -578,19 +583,25 @@ class ContentWriteCoordinator:) -> FreshnessAction:changed_entries = len({uri for values in changes.values() for uri in values})semantic_config = get_openviking_config().semantic- decision = await plan_abstract_overview_refresh(- viking_fs=self._viking_fs,- dir_uri=root_uri,- changed_entries=changed_entries,- ctx=ctx,- overview_sample_limit=getattr(semantic_config, "overview_sample_limit", 32),- refresh_ratio=getattr(semantic_config, "freshness_refresh_ratio", 0.10),- force_refresh=force_refresh,- )- aggregate_directory = decision.action is FreshnessAction.REFRESH_NOW+ try:+ decision = await plan_abstract_overview_refresh(+ viking_fs=self._viking_fs,+ dir_uri=root_uri,+ changed_entries=changed_entries,+ ctx=ctx,+ overview_sample_limit=getattr(semantic_config, "overview_sample_limit", 32),+ refresh_ratio=getattr(semantic_config, "freshness_refresh_ratio", 0.10),+ force_refresh=force_refresh,+ )+ action = decision.action+ except LockAcquisitionError:+ # Parent aggregation is best-effort; changed-file work must still run.+ logger.info("Skipping busy parent semantic refresh: %s", root_uri)+ action = FreshnessAction.NOOP+ aggregate_directory = action is FreshnessAction.REFRESH_NOWhas_live_files = any(changes.get(kind) for kind in ("added", "modified"))if not aggregate_directory and not has_live_files:- return decision.action+ return actionqueue_manager = get_queue_manager()semantic_queue = queue_manager.get_queue(queue_manager.SEMANTIC, allow_create=True)telemetry = get_current_telemetry()@@ -632,7 +643,7 @@ class ContentWriteCoordinator:if msg.telemetry_id:get_request_wait_tracker().mark_semantic_failed(msg.telemetry_id, msg.id, str(exc))raise- return decision.action+ return action@staticmethoddef _raise_refresh_errors(queue_status: Dict[str, Any]) -> None:@@ -865,12 +876,14 @@ class ContentWriteCoordinator:"semantic_status": "skipped","vector_status": vector_status,}- elif refresh_action is FreshnessAction.MARK_PENDING:+ elif refresh_action in {FreshnessAction.MARK_PENDING, FreshnessAction.NOOP}:# Changed-file semantic/vector work may still be queued, while- # the directory aggregation itself is intentionally deferred.+ # directory aggregation is deferred or skipped on contention._, vector_status = self._refresh_statuses(wait=wait, queue_status=queue_status)result_kwargs = {- "semantic_status": "deferred",+ "semantic_status": (+ "skipped" if refresh_action is FreshnessAction.NOOP else "deferred"+ ),"vector_status": vector_status,}return self._build_write_result(diff --git a/openviking/storage/queuefs/semantic_dag.py b/openviking/storage/queuefs/semantic_dag.pyindex 0ddccfbd4..95ca59a1b 100644--- a/openviking/storage/queuefs/semantic_dag.py+++ b/openviking/storage/queuefs/semantic_dag.py@@ -23,6 +23,7 @@ from openviking.storage.abstract_overview import (write_abstract_overview,)from openviking.storage.acl import CreatorAclGrant+from openviking.storage.errors import LockAcquisitionErrorfrom openviking.storage.viking_fs import LS_ALL_NODES, get_viking_fsfrom openviking.telemetry import bind_telemetry, get_current_telemetryfrom openviking.utils.ingest_options import IngestOptions@@ -403,16 +404,21 @@ class SemanticDagExecutor:sampled_entries = deterministic_sample(direct_entries, sample_limit)sampled_children_dirs = {uri for kind, uri in sampled_entries if kind == "directory"}sampled_file_paths = {uri for kind, uri in sampled_entries if kind == "file"}- pending_snapshot = (- await read_abstract_overview_pending_snapshot(- viking_fs=self._viking_fs,- dir_uri=dir_uri,- ctx=self._ctx,- lock=self._lock,- )- if self._aggregate_directory- else 0- )+ pending_snapshot = 0+ if self._aggregate_directory:+ try:+ pending_snapshot = await read_abstract_overview_pending_snapshot(+ viking_fs=self._viking_fs,+ dir_uri=dir_uri,+ ctx=self._ctx,+ lock=self._lock,+ )+ except LockAcquisitionError:+ if self._generation_trigger != "content_write":+ raise+ # Content writes run one directory; retain their file work.+ self._aggregate_directory = False+ logger.info("Skipping busy parent semantic refresh: %s", dir_uri)file_index = {path: idx for idx, path in enumerate(file_paths)}child_index = {path: idx for idx, path in enumerate(children_dirs)}# Recursive/initial work still maintains every file. Incremental@@ -1056,7 +1062,7 @@ class SemanticDagExecutor:if self._closed:return- # Write directly, protected by the outer semantic lock.+ # Persist sidecars before publishing their directory vectors.if should_write:assert overview is not None and abstract is not Nonetry:@@ -1076,6 +1082,7 @@ class SemanticDagExecutor:except AbstractOverviewFormatError:raiseexcept Exception:+ need_vectorize = Falselogger.info(f"[SemanticDag] {dir_uri} write failed, skipping")except AbstractOverviewFormatError:diff --git a/tests/server/test_content_batch_write.py b/tests/server/test_content_batch_write.pyindex 085f39214..0d2384927 100644--- a/tests/server/test_content_batch_write.py+++ b/tests/server/test_content_batch_write.py@@ -23,8 +23,8 @@ class _PathLockClient:self.held = Falseself.releases = 0- async def pathlock_acquire_tree(self, path):- del path+ async def pathlock_acquire_exact_batch(self, paths):+ del pathsself.held = Truereturn {"lease_ref": "lock-1"}@@ -156,7 +156,7 @@ async def test_batch_validates_all_modes_before_any_write(monkeypatch):@pytest.mark.asyncio-async def test_batch_releases_tree_lock_before_one_aggregated_refresh(monkeypatch):+async def test_batch_releases_file_locks_before_one_aggregated_refresh(monkeypatch):root = "viking://resources/wiki"a = f"{root}/a.md"b = f"{root}/b.md"@@ -230,20 +230,15 @@ async def test_batch_replace_memory_preserves_metadata(monkeypatch):@pytest.mark.asyncio-async def test_batch_reports_deferred_directory_and_queued_file_vectors(monkeypatch):+async def test_batch_reports_skipped_directory_and_queued_file_vectors(monkeypatch):root = "viking://resources/wide"page = f"{root}/page.md"coordinator = ContentWriteCoordinator(_VFS(root))- async def resolve_root(uri, **kwargs):- del uri, kwargs- return root-async def enqueue(**kwargs):del kwargs- return FreshnessAction.MARK_PENDING+ return FreshnessAction.NOOP- monkeypatch.setattr(coordinator, "_resolve_root_uri", resolve_root)monkeypatch.setattr(coordinator, "_enqueue_semantic_refresh_changes", enqueue)ctx = RequestContext(user=UserIdentifier.the_default_user(), role=Role.USER)@@ -260,7 +255,7 @@ async def test_batch_reports_deferred_directory_and_queued_file_vectors(monkeypawait=False,)- assert result["semantic_status"] == "deferred"+ assert result["semantic_status"] == "skipped"assert result["vector_status"] == "queued"assert result["queue_status"] is Nonediff --git a/tests/server/test_content_write_service.py b/tests/server/test_content_write_service.pyindex 136570bc2..a2c093d31 100644--- a/tests/server/test_content_write_service.py+++ b/tests/server/test_content_write_service.py@@ -20,7 +20,6 @@ from openviking.session.memory.utils.content_visibility import visible_contentfrom openviking.storage.acl import AclModefrom openviking.storage.content_write import ContentWriteCoordinatorfrom openviking.storage.errors import LockAcquisitionError, ResourceBusyError-from openviking.storage.queuefs.semantic_ops.freshness_policy import FreshnessActionfrom openviking_cli.exceptions import (AlreadyExistsError,DeadlineExceededError,@@ -640,28 +639,6 @@ async def test_resource_write_semantic_refresh_uses_coalesce_key(monkeypatch):assert queue.messages[0].lock_handoff is None-async def test_resource_write_wait_forces_directory_refresh(monkeypatch):- file_uri = "viking://resources/demo/doc.md"- root_uri = "viking://resources/demo"- ctx = RequestContext(user=UserIdentifier.the_default_user(), role=Role.USER)- coordinator = ContentWriteCoordinator(- viking_fs=_FakeVikingFS(file_uri=file_uri, root_uri=root_uri)- )- enqueue = AsyncMock(return_value=FreshnessAction.REFRESH_NOW)- monkeypatch.setattr(coordinator, "_enqueue_semantic_refresh", enqueue)- monkeypatch.setattr(coordinator, "_wait_for_request", AsyncMock(return_value=None))-- await coordinator.write(- uri=file_uri,- content="updated",- ctx=ctx,- wait=True,- )-- assert enqueue.await_args.kwargs["force_refresh"] is True--@pytest.mark.asyncioasync def test_write_timeout_after_enqueue_releases_resource_lock(monkeypatch):file_uri = "viking://resources/demo/doc.md"@@ -793,36 +770,35 @@ async def test_write_direct_reuses_outer_lease_for_viking_fs(monkeypatch):@pytest.mark.asyncio[email protected]("path", ["demo/doc.md", "demo#1/a#one.md"])-async def test_resource_write_updates_target_and_queues_refresh_before_return(monkeypatch, path):- file_uri = f"viking://resources/{path}"[email protected]("wait", [False, True])+async def test_resource_write_skips_busy_parent_and_keeps_file_work(monkeypatch, wait):+ file_uri = "viking://resources/demo/doc.md"root_uri = file_uri.rsplit("/", 1)[0]ctx = RequestContext(user=UserIdentifier.the_default_user(), role=Role.USER)viking_fs = _FakeVikingFS(file_uri=file_uri, root_uri=root_uri)coordinator = ContentWriteCoordinator(viking_fs=viking_fs)- captured_enqueue = {}-- async def _fake_enqueue_semantic_refresh(**kwargs):- captured_enqueue.update(kwargs)-- monkeypatch.setattr(coordinator, "_enqueue_semantic_refresh", _fake_enqueue_semantic_refresh)-- result = await coordinator.write(- uri=file_uri,- content="updated",- ctx=ctx,- mode="replace",- wait=False,+ queue = _FakeSemanticQueue()+ monkeypatch.setattr(+ "openviking.storage.content_write.get_queue_manager", lambda: _FakeQueueManager(queue))+ plan = AsyncMock(side_effect=LockAcquisitionError("parent sidecars are busy"))+ monkeypatch.setattr("openviking.storage.content_write.plan_abstract_overview_refresh", plan)+ monkeypatch.setattr(+ "openviking.storage.content_write.get_openviking_config",+ lambda: SimpleNamespace(semantic=SimpleNamespace()),+ )+ monkeypatch.setattr(coordinator, "_wait_for_request", AsyncMock(return_value=None))+ result = await coordinator.write(uri=file_uri, content="updated", ctx=ctx, wait=wait)++ assert plan.await_args.kwargs["force_refresh"] is waitassert viking_fs.content[file_uri] == "updated"assert result["content_updated"] is True- assert result["semantic_status"] == "queued"- assert result["vector_status"] == "queued"- assert captured_enqueue["root_uri"] == root_uri- assert captured_enqueue["changed_uri"] == file_uri- assert captured_enqueue["change_type"] == "modified"- assert viking_fs.delete_temp_calls == []+ assert result["semantic_status"] == "skipped"+ assert result["vector_status"] == ("complete" if wait else "queued")+ assert len(queue.messages) == 1+ assert queue.messages[0].changes == {"modified": [file_uri]}+ assert queue.messages[0].aggregate_directory is Falseassert viking_fs._async_agfs.release_calls == ["lock-1"]diff --git a/tests/storage/test_semantic_dag_incremental.py b/tests/storage/test_semantic_dag_incremental.pyindex 898873b14..308f6c741 100644--- a/tests/storage/test_semantic_dag_incremental.py+++ b/tests/storage/test_semantic_dag_incremental.py@@ -15,6 +15,7 @@ from openviking.storage.abstract_overview import (parse_abstract_overview,render_abstract_overview,)+from openviking.storage.errors import LockAcquisitionErrorfrom openviking.storage.queuefs.semantic_dag import SemanticDagExecutorfrom openviking.utils.ingest_options import IngestOptionsfrom openviking_cli.session.user_id import UserIdentifier@@ -340,15 +341,7 @@ async def test_directory_vectorization_retries_after_matching_sidecar_write(monkfile_path = f"{root_uri}/a.txt"fake_fs = _FakeVikingFS(tree={root_uri: [{"name": "a.txt", "isDir": False}]},- file_contents={- file_path: "new content",- f"{root_uri}/.overview.md": render_abstract_overview(- ContextLevel.OVERVIEW, root_uri, "FILES:\n- a.txt: old-summary"- ),- f"{root_uri}/.abstract.md": render_abstract_overview(- ContextLevel.ABSTRACT, root_uri, "old abstract"- ),- },+ file_contents={file_path: "new content"},)monkeypatch.setattr("openviking.storage.queuefs.semantic_dag.get_viking_fs", lambda: fake_fs)monkeypatch.setattr(@@ -373,6 +366,16 @@ async def test_directory_vectorization_retries_after_matching_sidecar_write(monkchanges={"modified": [file_path]},)+ acquire = fake_fs.pathlock_acquire_exact_batch+ fake_fs.pathlock_acquire_exact_batch = AsyncMock(+ side_effect=[{"paths": []}, LockAcquisitionError("parent sidecars are busy")]+ )+ await make_executor().run(root_uri)+ assert fake_fs.writes == []+ assert processor.vectorized_files == [file_path]+ vectorize_directory.assert_not_awaited()+ fake_fs.pathlock_acquire_exact_batch = acquire+with pytest.raises(RuntimeError, match="temporary failure"):await make_executor().run(root_uri)await make_executor().run(root_uri)diff --git a/tests/storage/test_semantic_dag_stats.py b/tests/storage/test_semantic_dag_stats.pyindex f4cfc9471..145597e43 100644--- a/tests/storage/test_semantic_dag_stats.py+++ b/tests/storage/test_semantic_dag_stats.py@@ -3,6 +3,7 @@import asynciofrom types import SimpleNamespace+from unittest.mock import AsyncMockimport pytest@@ -14,6 +15,7 @@ from openviking.service.task_work_index import (get_task_context,)from openviking.storage.abstract_overview import parse_abstract_overview+from openviking.storage.errors import LockAcquisitionErrorfrom openviking.storage.queuefs.named_queue import NamedQueuefrom openviking.storage.queuefs.semantic_dag import (DagStats,@@ -323,7 +325,7 @@ async def test_non_recursive_memory_samples_files_and_reads_child_abstracts(monk@pytest.mark.asyncio-async def test_deferred_aggregation_processes_only_changed_files(monkeypatch):+async def test_busy_parent_snapshot_preserves_changed_file_work(monkeypatch):root_uri = "viking://resources/wide"changed = f"{root_uri}/file-020.txt"tree = {@@ -332,6 +334,9 @@ async def test_deferred_aggregation_processes_only_changed_files(monkeypatch):fake_fs = _FakeVikingFS(tree)monkeypatch.setattr("openviking.storage.queuefs.semantic_dag.get_viking_fs", lambda: fake_fs)_patch_semantic_config(monkeypatch, overview_sample_limit=4)+ monkeypatch.setattr(+ fake_fs, "pathlock_acquire_exact_batch", AsyncMock(side_effect=LockAcquisitionError("busy"))+ )processor = _FakeProcessor()ctx = RequestContext(user=UserIdentifier("acc1", "user1"), role=Role.USER)@@ -344,7 +349,7 @@ async def test_deferred_aggregation_processes_only_changed_files(monkeypatch):target_uri=root_uri,recursive=False,changes={"modified": [changed]},- aggregate_directory=False,+ generation_trigger="content_write",)await executor.run(root_uri)