Merge branch 'main' into delta-channel-writes-based

- Add test_add_messages_to_delta_migration_preserves_message_history (sync
  + async) covering the primary real-world BinaryOperatorAggregate →
  DeltaChannel migration path with real Message objects and IDs
- Hoist all in-function imports to module level in test_channels.py and
  fix _delta_channel_with_type helper accordingly
- Add section headers in test_channels.py for better navigation

libs/checkpoint-postgres/langgraph/checkpoint/postgres/__init__.py

@@ -19,7 +19,6 @@ from langgraph.checkpoint.base import (
     get_serializable_checkpoint_metadata,
 )
 from langgraph.checkpoint.serde.base import SerializerProtocol
-from langgraph.checkpoint.serde.types import _DeltaSnapshot
 from psycopg import Capabilities, Connection, Cursor, Pipeline
 from psycopg.rows import DictRow, dict_row
 from psycopg.types.json import Jsonb
@@ -27,11 +26,9 @@ from psycopg_pool import ConnectionPool
 
 from langgraph.checkpoint.postgres import _internal
 from langgraph.checkpoint.postgres.base import (
-    SELECT_DELTA_STAGE1_SQL,
-    SELECT_DELTA_STAGE2_SQL,
+    SELECT_DELTA_COMBINED_SQL,
     BasePostgresSaver,
-    _DeltaStage1Row,
-    _DeltaStage2Row,
+    _DeltaCombinedRow,
 )
 from langgraph.checkpoint.postgres.shallow import ShallowPostgresSaver
 
@@ -311,12 +308,7 @@ class PostgresSaver(BasePostgresSaver):
         # others are stored in blobs table
         blob_values = {}
         for k, v in checkpoint["channel_values"].items():
-            if v is DELTA_SENTINEL:
-                copy["channel_values"].pop(k)
-            elif isinstance(v, _DeltaSnapshot):
-                blob_values[k] = copy["channel_values"].pop(k)
-                copy["channel_values"][k] = True
-            elif v is None or isinstance(v, (str, int, float, bool)):
+            if v is None or isinstance(v, (str, int, float, bool)):
                 pass
             else:
                 blob_values[k] = copy["channel_values"].pop(k)
@@ -449,49 +441,41 @@ class PostgresSaver(BasePostgresSaver):
     ) -> _ChannelWritesHistory:
         """Fast-path override of `BaseCheckpointSaver._get_channel_writes_history`.
 
-        Two-stage query: stage 1 scans checkpoint metadata to walk the parent
-        chain and locate the nearest snapshot; stage 2 fetches only the
-        chain-limited writes and single seed blob.
+        One combined UNION ALL query (`SELECT_DELTA_COMBINED_SQL`) fetches rows
+        from `checkpoints`, `checkpoint_writes`, and `checkpoint_blobs` in a
+        single roundtrip; the ancestor walk runs in Python.
         """
         thread_id = config["configurable"]["thread_id"]
         checkpoint_ns = config["configurable"].get("checkpoint_ns", "")
         checkpoint_id = get_checkpoint_id(config)
         if checkpoint_id is None:
+            # Caller didn't specify a target — resolve to the latest
+            # checkpoint on the thread. `get_tuple` without `checkpoint_id`
+            # returns the newest; its config carries the resolved id.
             target = self.get_tuple(config)
             if target is None:
                 return _ChannelWritesHistory(seed=DELTA_SENTINEL, writes=[])
             checkpoint_id = target.config["configurable"]["checkpoint_id"]
-
         with self._cursor() as cur:
             cur.execute(
-                SELECT_DELTA_STAGE1_SQL,
-                (channel, channel, thread_id, checkpoint_ns),
-            )
-            stage1_rows = cur.fetchall()
-        chain_cids, seed_version = self._walk_stage1(
-            cast("list[_DeltaStage1Row]", stage1_rows), checkpoint_id
-        )
-        seed_versions = [seed_version] if seed_version else []
-        with self._cursor() as cur:
-            cur.execute(
-                SELECT_DELTA_STAGE2_SQL,
+                SELECT_DELTA_COMBINED_SQL,
                 (
+                    channel,
+                    thread_id,
+                    checkpoint_ns,
                     thread_id,
                     checkpoint_ns,
                     channel,
-                    chain_cids,
                     thread_id,
                     checkpoint_ns,
                     channel,
-                    seed_versions,
                 ),
             )
-            stage2_rows = cur.fetchall()
+            rows = cur.fetchall()
         return self._build_delta_channel_writes_history(
             channel=channel,
-            chain_cids=chain_cids,
-            seed_version=seed_version,
-            stage2_rows=cast("list[_DeltaStage2Row]", stage2_rows),
+            target_id=checkpoint_id,
+            rows=cast("list[_DeltaCombinedRow]", rows),
         )
 
     def _load_checkpoint_tuple(self, value: DictRow) -> CheckpointTuple:

libs/checkpoint-postgres/langgraph/checkpoint/postgres/aio.py

@@ -19,7 +19,6 @@ from langgraph.checkpoint.base import (
     get_serializable_checkpoint_metadata,
 )
 from langgraph.checkpoint.serde.base import SerializerProtocol
-from langgraph.checkpoint.serde.types import _DeltaSnapshot
 from psycopg import AsyncConnection, AsyncCursor, AsyncPipeline, Capabilities
 from psycopg.rows import DictRow, dict_row
 from psycopg.types.json import Jsonb
@@ -27,11 +26,9 @@ from psycopg_pool import AsyncConnectionPool
 
 from langgraph.checkpoint.postgres import _ainternal
 from langgraph.checkpoint.postgres.base import (
-    SELECT_DELTA_STAGE1_SQL,
-    SELECT_DELTA_STAGE2_SQL,
+    SELECT_DELTA_COMBINED_SQL,
     BasePostgresSaver,
-    _DeltaStage1Row,
-    _DeltaStage2Row,
+    _DeltaCombinedRow,
 )
 from langgraph.checkpoint.postgres.shallow import AsyncShallowPostgresSaver
 
@@ -270,12 +267,7 @@ class AsyncPostgresSaver(BasePostgresSaver):
         # others are stored in blobs table
         blob_values = {}
         for k, v in checkpoint["channel_values"].items():
-            if v is DELTA_SENTINEL:
-                copy["channel_values"].pop(k)
-            elif isinstance(v, _DeltaSnapshot):
-                blob_values[k] = copy["channel_values"].pop(k)
-                copy["channel_values"][k] = True
-            elif v is None or isinstance(v, (str, int, float, bool)):
+            if v is None or isinstance(v, (str, int, float, bool)):
                 pass
             else:
                 blob_values[k] = copy["channel_values"].pop(k)
@@ -410,9 +402,10 @@ class AsyncPostgresSaver(BasePostgresSaver):
     ) -> _ChannelWritesHistory:
         """Fast-path override of `BaseCheckpointSaver._aget_channel_writes_history`.
 
-        Two-stage query: stage 1 scans checkpoint metadata to walk the parent
-        chain and locate the nearest snapshot; stage 2 fetches only the
-        chain-limited writes and single seed blob.
+        One combined UNION ALL query (`SELECT_DELTA_COMBINED_SQL`) fetches rows
+        from `checkpoints`, `checkpoint_writes`, and `checkpoint_blobs` in a
+        single roundtrip; rows are assembled by the shared pure helper on
+        `BasePostgresSaver`.
         """
         thread_id = config["configurable"]["thread_id"]
         checkpoint_ns = config["configurable"].get("checkpoint_ns", "")
@@ -422,37 +415,26 @@ class AsyncPostgresSaver(BasePostgresSaver):
             if target is None:
                 return _ChannelWritesHistory(seed=DELTA_SENTINEL, writes=[])
             checkpoint_id = target.config["configurable"]["checkpoint_id"]
-
         async with self._cursor() as cur:
             await cur.execute(
-                SELECT_DELTA_STAGE1_SQL,
-                (channel, channel, thread_id, checkpoint_ns),
-            )
-            stage1_rows = await cur.fetchall()
-        chain_cids, seed_version = self._walk_stage1(
-            cast("list[_DeltaStage1Row]", stage1_rows), checkpoint_id
-        )
-        seed_versions = [seed_version] if seed_version else []
-        async with self._cursor() as cur:
-            await cur.execute(
-                SELECT_DELTA_STAGE2_SQL,
+                SELECT_DELTA_COMBINED_SQL,
                 (
+                    channel,
+                    thread_id,
+                    checkpoint_ns,
                     thread_id,
                     checkpoint_ns,
                     channel,
-                    chain_cids,
                     thread_id,
                     checkpoint_ns,
                     channel,
-                    seed_versions,
                 ),
             )
-            stage2_rows = await cur.fetchall()
+            rows = await cur.fetchall()
         return self._build_delta_channel_writes_history(
             channel=channel,
-            chain_cids=chain_cids,
-            seed_version=seed_version,
-            stage2_rows=cast("list[_DeltaStage2Row]", stage2_rows),
+            target_id=checkpoint_id,
+            rows=cast("list[_DeltaCombinedRow]", rows),
         )
 
     async def _load_checkpoint_tuple(self, value: DictRow) -> CheckpointTuple:

libs/checkpoint-postgres/langgraph/checkpoint/postgres/base.py

@@ -156,62 +156,62 @@ INSERT_CHECKPOINT_WRITES_SQL = """
 """
 
 
-class _DeltaStage2Row(TypedDict, total=False):
-    """One row from `SELECT_DELTA_STAGE2_SQL` (a UNION ALL of writes and blobs)."""
+class _DeltaCombinedRow(TypedDict, total=False):
+    """One row from `SELECT_DELTA_COMBINED_SQL` (a UNION ALL of three tables).
 
-    _kind: str  # "w" or "b"
-    checkpoint_id: str | None  # "w" rows only
+    Every row carries `_kind` ("p" / "w" / "b") plus whichever columns are
+    relevant for that kind; irrelevant columns are NULL and typed as `None`.
+    """
+
+    _kind: str  # always present: "p", "w", or "b"
+    # checkpoint row ("p")
+    checkpoint_id: str | None
+    parent_checkpoint_id: str | None
+    ver: str | None
+    # write / blob rows ("w", "b")
     type: str | None
     blob: bytes | None
-    task_id: str | None  # "w" rows only
-    idx: int | None  # "w" rows only
-    version: str | None  # "b" rows only
+    # write row only ("w")
+    task_id: str | None
+    idx: int | None
+    # blob row only ("b")
+    version: str | None
 
 
-# Two-stage DeltaChannel reconstruction.  Stage 1 scans checkpoint
-# metadata (no blob bytes) to walk the parent chain and locate the
-# nearest snapshot marker.  Stage 2 fetches only the chain-limited
-# writes and the single seed snapshot blob.
+# DeltaChannel reconstruction: one UNION ALL query fetches checkpoints,
+# writes, and blobs for `channel` in one roundtrip; the ancestor walk runs
+# in Python in `_build_delta_channel_writes_history`.
 #
-# Parameter order:
-#   stage1: (channel, channel, thread_id, checkpoint_ns)
-#   stage2: (thread_id, checkpoint_ns, channel, chain_cids[],
-#            thread_id, checkpoint_ns, channel, seed_versions[])
-
-SELECT_DELTA_STAGE1_SQL = """
-    SELECT checkpoint_id,
+# Parameter order: (channel, thread_id, checkpoint_ns,
+#                   thread_id, checkpoint_ns, channel,
+#                   thread_id, checkpoint_ns, channel)
+SELECT_DELTA_COMBINED_SQL = """
+    SELECT 'p'::text       AS _kind,
+           checkpoint_id,
            parent_checkpoint_id,
            checkpoint -> 'channel_versions' ->> %s AS ver,
-           (checkpoint -> 'channel_values' -> %s) IS NOT NULL AS has_snapshot
+           NULL::text       AS type,
+           NULL::bytea      AS blob,
+           NULL::text       AS task_id,
+           NULL::int        AS idx,
+           NULL::text       AS version
     FROM checkpoints
     WHERE thread_id = %s AND checkpoint_ns = %s
-"""
-
-SELECT_DELTA_STAGE2_SQL = """
-    SELECT 'w'::text AS _kind,
-           checkpoint_id,
-           type, blob, task_id, idx, NULL::text AS version
+    UNION ALL
+    SELECT 'w',
+           checkpoint_id, NULL, NULL,
+           type, blob, task_id, idx, NULL
     FROM checkpoint_writes
     WHERE thread_id = %s AND checkpoint_ns = %s AND channel = %s
-      AND checkpoint_id = ANY(%s)
     UNION ALL
-    SELECT 'b', NULL,
+    SELECT 'b',
+           NULL, NULL, NULL,
            type, blob, NULL, NULL, version
     FROM checkpoint_blobs
     WHERE thread_id = %s AND checkpoint_ns = %s AND channel = %s
-      AND version = ANY(%s)
 """
 
 
-class _DeltaStage1Row(TypedDict):
-    """One row from `SELECT_DELTA_STAGE1_SQL`."""
-
-    checkpoint_id: str
-    parent_checkpoint_id: str | None
-    ver: str | None
-    has_snapshot: bool
-
-
 class BasePostgresSaver(BaseCheckpointSaver[str]):
     SELECT_SQL = SELECT_SQL
     SELECT_PENDING_SENDS_SQL = SELECT_PENDING_SENDS_SQL
@@ -254,59 +254,38 @@ class BasePostgresSaver(BaseCheckpointSaver[str]):
             if t.decode() != "empty"
         }
 
-    @staticmethod
-    def _walk_stage1(
-        stage1_rows: Sequence[_DeltaStage1Row],
-        target_id: str,
-    ) -> tuple[list[str], str | None]:
-        """Walk the parent chain from stage 1 metadata rows.
-
-        Returns (chain_cids, seed_version):
-          chain_cids: ancestor checkpoint IDs from target's parent down to
-                      the seed (or root), in newest-first order.
-          seed_version: the channel blob version at the nearest ancestor
-                        with has_snapshot=True, or None if pure delta.
-        """
-        parent_of: dict[str, str | None] = {}
-        ver_of: dict[str, str | None] = {}
-        snapshot_of: dict[str, bool] = {}
-        for r in stage1_rows:
-            cid = r["checkpoint_id"]
-            parent_of[cid] = r["parent_checkpoint_id"]
-            ver_of[cid] = r["ver"]
-            snapshot_of[cid] = r["has_snapshot"]
-
-        chain_cids: list[str] = []
-        seed_version: str | None = None
-        cur_cid: str | None = parent_of.get(target_id)
-        while cur_cid is not None:
-            chain_cids.append(cur_cid)
-            if snapshot_of.get(cur_cid, False):
-                seed_version = ver_of.get(cur_cid)
-                break
-            cur_cid = parent_of.get(cur_cid)
-        return chain_cids, seed_version
-
     def _build_delta_channel_writes_history(
         self,
         *,
         channel: str,
-        chain_cids: list[str],
-        seed_version: str | None,
-        stage2_rows: Sequence[_DeltaStage2Row],
+        target_id: str,
+        rows: Sequence[_DeltaCombinedRow],
     ) -> _ChannelWritesHistory:
-        """Reconstruct delta channel history from two-stage query results.
+        """Reconstruct one delta channel's history from the combined UNION ALL rows.
 
-        chain_cids are in newest-first order (target's parent first).
-        stage2_rows contain only writes for chain_cids and the single
-        seed blob at seed_version.
+        Pure data transform shared by sync (`PostgresSaver`) and async
+        (`AsyncPostgresSaver`); both paths run `SELECT_DELTA_COMBINED_SQL`
+        and feed the tagged rows here.
+
+        Walk is newest → oldest from the target's parent. A non-sentinel
+        blob in `checkpoint_blobs` (a pre-delta snapshot) terminates the
+        walk and is returned as the seed so replay starts from it.
+
+        Writes stored at `target_id` itself are pending writes for the next
+        step and are excluded — the walk begins at the target's parent.
         """
+        parent_of: dict[str, str | None] = {}
+        ver_of: dict[str, str | None] = {}
         writes_by_cid: dict[str, list[tuple[str, bytes, str, int]]] = {}
-        seed_blob: tuple[str, bytes] | None = None
+        blob_by_ver: dict[str, tuple[str, bytes]] = {}
 
-        for r in stage2_rows:
+        for r in rows:
             kind = r["_kind"]
-            if kind == "w":
+            if kind == "p":
+                cid = cast(str, r["checkpoint_id"])
+                parent_of[cid] = r["parent_checkpoint_id"]
+                ver_of[cid] = r["ver"]
+            elif kind == "w":
                 cid = cast(str, r["checkpoint_id"])
                 writes_by_cid.setdefault(cid, []).append(
                     cast(
@@ -315,26 +294,42 @@ class BasePostgresSaver(BaseCheckpointSaver[str]):
                     )
                 )
             else:  # kind == "b"
-                seed_blob = cast("tuple[str, bytes]", (r["type"], r["blob"]))
+                blob_by_ver[cast(str, r["version"])] = cast(
+                    "tuple[str, bytes]", (r["type"], r["blob"])
+                )
 
+        # newest write first per ancestor (task_id DESC, idx DESC)
         for ws in writes_by_cid.values():
             ws.sort(key=lambda w: (w[2], w[3]), reverse=True)
 
-        if not chain_cids:
+        ancestors: list[str] = []
+        cur_cid: str | None = parent_of.get(target_id)
+        while cur_cid is not None:
+            ancestors.append(cur_cid)
+            cur_cid = parent_of.get(cur_cid)
+        if not ancestors:
             return _ChannelWritesHistory(seed=DELTA_SENTINEL, writes=[])
 
-        collected: list[PendingWrite] = []
-        for cid in chain_cids:
+        collected: list[PendingWrite] = []  # newest first; reversed at the end
+        for cid in ancestors:
+            # Collect writes first — they encode the transition FROM this
+            # ancestor's state to its child's and must be included even if
+            # this ancestor is also the seed checkpoint.
             for type_tag, write_blob, task_id, _idx in writes_by_cid.get(cid, []):
                 val = self.serde.loads_typed((type_tag, write_blob))
                 collected.append((task_id, channel, val))
+            # Then check seed terminator.
+            ver = ver_of.get(cid)
+            if ver is not None:
+                seed_blob = blob_by_ver.get(ver)
+                if seed_blob is not None and seed_blob[0] != "empty":
+                    blob_value = self.serde.loads_typed(seed_blob)
+                    if blob_value is not DELTA_SENTINEL:
+                        collected.reverse()
+                        return _ChannelWritesHistory(seed=blob_value, writes=collected)
 
-        seed: Any = DELTA_SENTINEL
-        if seed_blob is not None and seed_blob[0] != "empty":
-            seed = self.serde.loads_typed(seed_blob)
-
-        collected.reverse()
-        return _ChannelWritesHistory(seed=seed, writes=collected)
+        collected.reverse()  # oldest → newest
+        return _ChannelWritesHistory(seed=DELTA_SENTINEL, writes=collected)
 
     def _dump_blobs(
         self,

libs/checkpoint-postgres/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "langgraph-checkpoint-postgres"
-version = "3.1.0a3"
+version = "3.0.5"
 description = "Library with a Postgres implementation of LangGraph checkpoint saver."
 authors = []
 requires-python = ">=3.10"
@@ -12,7 +12,7 @@ readme = "README.md"
 license = "MIT"
 license-files = ['LICENSE']
 dependencies = [
-  "langgraph-checkpoint>=4.1.0a3,<5.0.0",
+  "langgraph-checkpoint>=4.0.3,<5.0.0",
   "orjson>=3.11.5",
   "psycopg>=3.2.0",
   "psycopg-pool>=3.2.0",

libs/checkpoint-postgres/uv.lock

@@ -259,7 +259,7 @@ wheels = [
 
 [[package]]
 name = "langgraph-checkpoint"
-version = "4.1.0a3"
+version = "4.0.3"
 source = { editable = "../checkpoint" }
 dependencies = [
     { name = "langchain-core" },
@@ -307,7 +307,7 @@ test = [
 
 [[package]]
 name = "langgraph-checkpoint-postgres"
-version = "3.1.0a3"
+version = "3.0.5"
 source = { editable = "." }
 dependencies = [
     { name = "langgraph-checkpoint" },

libs/checkpoint-sqlite/uv.lock

@@ -268,7 +268,7 @@ wheels = [
 
 [[package]]
 name = "langgraph-checkpoint"
-version = "4.1.0a3"
+version = "4.0.3"
 source = { editable = "../checkpoint" }
 dependencies = [
     { name = "langchain-core" },

libs/checkpoint/langgraph/checkpoint/memory/__init__.py

@@ -452,9 +452,7 @@ class InMemorySaver(
         values: dict[str, Any] = c.pop("channel_values")  # type: ignore[misc]
         for k, v in new_versions.items():
             self.blobs[(thread_id, checkpoint_ns, k, v)] = (
-                self.serde.dumps_typed(values[k])
-                if k in values and values[k] is not DELTA_SENTINEL
-                else ("empty", b"")
+                self.serde.dumps_typed(values[k]) if k in values else ("empty", b"")
             )
         self.storage[thread_id][checkpoint_ns].update(
             {

libs/checkpoint/langgraph/checkpoint/serde/jsonplus.py

@@ -34,7 +34,9 @@ from langgraph.checkpoint.serde import _msgpack as _lg_msgpack
 from langgraph.checkpoint.serde.base import SerializerProtocol
 from langgraph.checkpoint.serde.event_hooks import emit_serde_event
 from langgraph.checkpoint.serde.types import (
+    DELTA_SENTINEL,
     SendProtocol,
+    _DeltaSentinel,
     _DeltaSnapshot,
 )
 from langgraph.store.base import Item
@@ -320,11 +322,14 @@ EXT_PYDANTIC_V1 = 4
 EXT_PYDANTIC_V2 = 5
 EXT_NUMPY_ARRAY = 6
 EXT_DELTA_SNAPSHOT = 7
+EXT_DELTA_SENTINEL = 8
 
 
 def _msgpack_default(obj: Any) -> str | ormsgpack.Ext:
     if isinstance(obj, _DeltaSnapshot):
         return ormsgpack.Ext(EXT_DELTA_SNAPSHOT, _msgpack_enc(obj.value))
+    elif isinstance(obj, _DeltaSentinel):
+        return ormsgpack.Ext(EXT_DELTA_SENTINEL, b"")
     elif hasattr(obj, "model_dump") and callable(obj.model_dump):  # pydantic v2
         return ormsgpack.Ext(
             EXT_PYDANTIC_V2,
@@ -651,7 +656,9 @@ def _create_msgpack_ext_hook(
         return False
 
     def ext_hook(code: int, data: bytes) -> Any:
-        if code == EXT_DELTA_SNAPSHOT:
+        if code == EXT_DELTA_SENTINEL:
+            return DELTA_SENTINEL
+        elif code == EXT_DELTA_SNAPSHOT:
             return _DeltaSnapshot(
                 ormsgpack.unpackb(
                     data, ext_hook=ext_hook, option=ormsgpack.OPT_NON_STR_KEYS

libs/checkpoint/langgraph/checkpoint/serde/types.py

@@ -17,10 +17,12 @@ TASKS = "__pregel_tasks"
 
 
 class _DeltaSentinel:
-    """In-memory marker for a DeltaChannel field with no snapshot.
+    """Singleton marker stored (as zero bytes) in checkpoint_blobs for a
+    DeltaChannel field. The actual per-step writes live in checkpoint_writes
+    and are replayed through the reducer at load time.
 
-    Never serialized to storage — checkpointers strip it before writing.
-    Compare with `is DELTA_SENTINEL`; always the same module-level instance.
+    Compare with `is DELTA_SENTINEL` — `loads_typed` always returns the same
+    module-level instance.
     """
 
     __slots__ = ()

libs/checkpoint/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "langgraph-checkpoint"
-version = "4.1.0a3"
+version = "4.0.3"
 description = "Library with base interfaces for LangGraph checkpoint savers."
 authors = []
 requires-python = ">=3.10"

libs/checkpoint/tests/test_jsonplus.py

@@ -1048,3 +1048,15 @@ def test_msgpack_nested_pydantic_serializes_as_dict(
     # No blocking should occur - inner is serialized as dict, not ext
     assert "blocked" not in caplog.text.lower()
     assert result == obj
+
+
+def test_delta_sentinel_serde_round_trip() -> None:
+    from langgraph.checkpoint.base import DELTA_SENTINEL
+    from langgraph.checkpoint.serde.jsonplus import JsonPlusSerializer
+
+    serde = JsonPlusSerializer()
+    type_tag, blob = serde.dumps_typed(DELTA_SENTINEL)
+    assert type_tag == "msgpack"
+    assert blob  # non-empty ext envelope
+    loaded = serde.loads_typed((type_tag, blob))
+    assert loaded is DELTA_SENTINEL

libs/checkpoint/tests/test_memory.py

@@ -322,17 +322,26 @@ def test_memory_saver_with_allowlist_proxy_isolated() -> None:
 
 
 class TestInMemorySaverDeltaChannel:
-    def test_load_blobs_omits_delta_channel(self) -> None:
-        """_load_blobs omits delta channels (stored as 'empty'); reconstruction deferred."""
+    def test_load_blobs_returns_sentinel_for_delta_channel(self) -> None:
+        """_load_blobs returns DELTA_SENTINEL for delta channels (reconstruction deferred)."""
         saver = InMemorySaver()
+        serde = JsonPlusSerializer()
 
         thread_id, ns, channel = "t1", "", "messages"
         v1 = "00000000000000000000000000000001.0000000000000000"
 
-        saver.blobs[(thread_id, ns, channel, v1)] = ("empty", b"")
+        saver.blobs[(thread_id, ns, channel, v1)] = serde.dumps_typed(DELTA_SENTINEL)
+
+        cp1 = empty_checkpoint()
+        cp1["id"] = "cp1"
+        cp1["channel_versions"][channel] = v1
+        saver.storage[thread_id][ns] = {
+            "cp1": (serde.dumps_typed(cp1), serde.dumps_typed({}), None),
+        }
 
         result = saver._load_blobs(thread_id, ns, {channel: v1})
-        assert channel not in result
+        assert channel in result
+        assert result[channel] is DELTA_SENTINEL
 
     def test_get_channel_writes_collects_ancestor_writes_only(self) -> None:
         """_get_channel_writes_history collects ancestor writes oldest→newest,
@@ -573,9 +582,9 @@ class TestPreDeltaBlobTerminator:
 
         # Pre-delta: cp1 stored a real blob for the channel.
         saver.blobs[(thread_id, ns, channel, v1)] = serde.dumps_typed(["A"])
-        # Delta-era: cp2 and cp3 store "empty"; real writes in checkpoint_writes.
-        saver.blobs[(thread_id, ns, channel, v2)] = ("empty", b"")
-        saver.blobs[(thread_id, ns, channel, v3)] = ("empty", b"")
+        # Delta-era: cp2 and cp3 store sentinels; real writes in checkpoint_writes.
+        saver.blobs[(thread_id, ns, channel, v2)] = serde.dumps_typed(DELTA_SENTINEL)
+        saver.blobs[(thread_id, ns, channel, v3)] = serde.dumps_typed(DELTA_SENTINEL)
 
         cp1 = empty_checkpoint()
         cp1["id"] = "cp1"

libs/checkpoint/uv.lock

@@ -286,7 +286,7 @@ wheels = [
 
 [[package]]
 name = "langgraph-checkpoint"
-version = "4.1.0a3"
+version = "4.0.3"
 source = { editable = "." }
 dependencies = [
     { name = "langchain-core" },

libs/langgraph/langgraph/_internal/_constants.py

@@ -12,9 +12,6 @@ RESUME = sys.intern("__resume__")
 # for values passed to resume a node after an interrupt
 ERROR = sys.intern("__error__")
 # for errors raised by nodes
-ERROR_SOURCE_NODE = sys.intern("__error_source_node__")
-# failed source node name for node-level error handlers
-# value format in pending writes: `(task_id, ERROR_SOURCE_NODE, node_name: str)`
 NO_WRITES = sys.intern("__no_writes__")
 # marker to signal node didn't write anything
 TASKS = sys.intern("__pregel_tasks")
@@ -74,10 +71,6 @@ CONFIG_KEY_RESUME_MAP = sys.intern("__pregel_resume_map")
 CONFIG_KEY_STREAM_MESSAGES_V2 = sys.intern("__pregel_stream_messages_v2")
 # when True, attach StreamMessagesHandlerV2 so content-block (v2) events
 # flow through stream_mode="messages"; set by StreamingHandler only.
-CONFIG_KEY_NODE_ERROR = sys.intern("__pregel_node_error")
-# holds a `NodeError` (failed source node + exception) for the current
-# node-level error handler invocation, injected when handler signature
-# requests `error: NodeError`
 
 # --- Other constants ---
 PUSH = sys.intern("__pregel_push")
@@ -105,7 +98,6 @@ RESERVED = {
     INTERRUPT,
     RESUME,
     ERROR,
-    ERROR_SOURCE_NODE,
     NO_WRITES,
     # reserved config.configurable keys
     CONFIG_KEY_SEND,

libs/langgraph/langgraph/_internal/_runnable.py

@@ -51,11 +51,9 @@ from langgraph._internal._config import (
 )
 from langgraph._internal._constants import (
     CONF,
-    CONFIG_KEY_NODE_ERROR,
     CONFIG_KEY_RUNTIME,
 )
 from langgraph._internal._typing import MISSING
-from langgraph.errors import NodeError
 from langgraph.types import StreamWriter
 
 try:
@@ -196,15 +194,6 @@ KWARGS_CONFIG_KEYS: tuple[tuple[str, tuple[Any, ...], str, Any], ...] = (
         "N/A",
         inspect.Parameter.empty,
     ),
-    (
-        "error",
-        (NodeError, "NodeError"),
-        # we never hit this block, we read directly from configurable
-        "N/A",
-        # default to None so non-handler nodes that happen to type a parameter
-        # `error: NodeError` don't blow up; handlers always receive a NodeError.
-        None,
-    ),
 )
 """List of kwargs that can be passed to functions, and their corresponding
 config keys, default values and type annotations.
@@ -378,8 +367,6 @@ class RunnableCallable(Runnable):
             kw_value: Any = MISSING
             if kw == "config":
                 kw_value = config
-            elif kw == "error":
-                kw_value = config.get(CONF, {}).get(CONFIG_KEY_NODE_ERROR, MISSING)
             elif runtime:
                 if kw == "runtime":
                     kw_value = runtime
@@ -452,8 +439,6 @@ class RunnableCallable(Runnable):
             kw_value: Any = MISSING
             if kw == "config":
                 kw_value = config
-            elif kw == "error":
-                kw_value = config.get(CONF, {}).get(CONFIG_KEY_NODE_ERROR, MISSING)
             elif runtime:
                 if kw == "runtime":
                     kw_value = runtime

libs/langgraph/langgraph/errors.py

@@ -1,7 +1,6 @@
 from __future__ import annotations
 
 from collections.abc import Sequence
-from dataclasses import dataclass
 from enum import Enum
 from typing import Any, Literal
 from warnings import warn
@@ -16,12 +15,10 @@ from langgraph.warnings import LangGraphDeprecatedSinceV10
 __all__ = (
     "EmptyChannelError",
     "ErrorCode",
-    "GraphDrained",
     "GraphRecursionError",
     "InvalidUpdateError",
     "GraphBubbleUp",
     "GraphInterrupt",
-    "NodeError",
     "NodeInterrupt",
     "NodeTimeoutError",
     "ParentCommand",
@@ -46,23 +43,6 @@ def create_error_message(*, message: str, error_code: ErrorCode) -> str:
     )
 
 
-class GraphBubbleUp(Exception):
-    pass
-
-
-class GraphDrained(GraphBubbleUp):
-    """Raised when a graph run exits early due to a drain request.
-
-    This indicates the graph stopped cooperatively at a superstep boundary
-    because `RunControl.request_drain()` was called (e.g., in response to
-    SIGTERM). The checkpoint is saved and the run can be resumed later.
-    """
-
-    def __init__(self, reason: str = "shutdown") -> None:
-        self.reason = reason
-        super().__init__(f"Graph drained: {reason}")
-
-
 class GraphRecursionError(RecursionError):
     """Raised when the graph has exhausted the maximum number of steps.
 
@@ -98,6 +78,10 @@ class InvalidUpdateError(Exception):
     pass
 
 
+class GraphBubbleUp(Exception):
+    pass
+
+
 class GraphInterrupt(GraphBubbleUp):
     """Raised when a subgraph is interrupted, suppressed by the root graph.
     Never raised directly, or surfaced to the user."""
@@ -144,31 +128,12 @@ class TaskNotFound(Exception):
     pass
 
 
-@dataclass(frozen=True, slots=True)
-class NodeError:
-    """Failure context passed to a node-level error handler.
-
-    Inject by adding a parameter typed `NodeError` to a handler registered via
-    `StateGraph.add_node(..., error_handler=...)`:
-
-    ```python
-    def handler(state: State, error: NodeError) -> Command:
-        return Command(update={"status": f"recovered from {error.node}: {error.error}"})
-    ```
-    """
-
-    node: str
-    """Name of the node whose execution failed."""
-
-    error: BaseException
-    """Exception raised by the failed node."""
-
-
-class NodeTimeoutError(Exception):
+class NodeTimeoutError(TimeoutError):
     """Raised when a node invocation exceeds one of its configured timeouts.
 
-    Does **not** inherit from the built-in `TimeoutError` (a subclass of
-    `OSError`) so that the default `RetryPolicy` treats it as retryable.
+    Subclasses the built-in `TimeoutError`, so existing `except TimeoutError`
+    handlers keep working. If the node has a `retry_policy` whose `retry_on`
+    permits `TimeoutError`, the attempt will be retried.
 
     Both `idle_timeout` and `run_timeout` reflect the configured policy at the
     time of the failure (each is `None` if not configured). `kind` and

libs/langgraph/langgraph/graph/_node.py

@@ -88,8 +88,6 @@ class StateNodeSpec(Generic[NodeInputT, ContextT]):
     input_schema: type[NodeInputT]
     retry_policy: RetryPolicy | Sequence[RetryPolicy] | None
     cache_policy: CachePolicy | None
-    is_error_handler: bool = False
-    error_handler_node: str | None = None
     ends: tuple[str, ...] | dict[str, str] | None = EMPTY_SEQ
     defer: bool = False
     timeout: TimeoutPolicy | None = None

libs/langgraph/langgraph/graph/state.py

@@ -303,7 +303,6 @@ class StateGraph(Generic[StateT, ContextT, InputT, OutputT]):
         input_schema: None = None,
         retry_policy: RetryPolicy | Sequence[RetryPolicy] | None = None,
         cache_policy: CachePolicy | None = None,
-        error_handler: StateNode[Any, ContextT] | None = None,
         destinations: dict[str, str] | tuple[str, ...] | None = None,
         timeout: float | timedelta | TimeoutPolicy | None = None,
         **kwargs: Unpack[DeprecatedKwargs],
@@ -372,7 +371,6 @@ class StateGraph(Generic[StateT, ContextT, InputT, OutputT]):
         input_schema: type[NodeInputT],
         retry_policy: RetryPolicy | Sequence[RetryPolicy] | None = None,
         cache_policy: CachePolicy | None = None,
-        error_handler: StateNode[Any, ContextT] | None = None,
         destinations: dict[str, str] | tuple[str, ...] | None = None,
         timeout: float | timedelta | TimeoutPolicy | None = None,
         **kwargs: Unpack[DeprecatedKwargs],
@@ -446,7 +444,6 @@ class StateGraph(Generic[StateT, ContextT, InputT, OutputT]):
         input_schema: None = None,
         retry_policy: RetryPolicy | Sequence[RetryPolicy] | None = None,
         cache_policy: CachePolicy | None = None,
-        error_handler: StateNode[Any, ContextT] | None = None,
         destinations: dict[str, str] | tuple[str, ...] | None = None,
         timeout: float | timedelta | TimeoutPolicy | None = None,
         **kwargs: Unpack[DeprecatedKwargs],
@@ -515,7 +512,6 @@ class StateGraph(Generic[StateT, ContextT, InputT, OutputT]):
         input_schema: type[NodeInputT],
         retry_policy: RetryPolicy | Sequence[RetryPolicy] | None = None,
         cache_policy: CachePolicy | None = None,
-        error_handler: StateNode[Any, ContextT] | None = None,
         destinations: dict[str, str] | tuple[str, ...] | None = None,
         timeout: float | timedelta | TimeoutPolicy | None = None,
         **kwargs: Unpack[DeprecatedKwargs],
@@ -591,7 +587,6 @@ class StateGraph(Generic[StateT, ContextT, InputT, OutputT]):
         input_schema: type[NodeInputT] | None = None,
         retry_policy: RetryPolicy | Sequence[RetryPolicy] | None = None,
         cache_policy: CachePolicy | None = None,
-        error_handler: StateNode[Any, ContextT] | None = None,
         destinations: dict[str, str] | tuple[str, ...] | None = None,
         timeout: float | timedelta | TimeoutPolicy | None = None,
         **kwargs: Unpack[DeprecatedKwargs],
@@ -612,7 +607,6 @@ class StateGraph(Generic[StateT, ContextT, InputT, OutputT]):
 
                 If a sequence is provided, the first matching policy will be applied.
             cache_policy: The cache policy for the node.
-            error_handler: Optional node-level error handler callable for this node.
             destinations: Destinations that indicate where a node can route to.
 
                 Useful for edgeless graphs with nodes that return `Command` objects.
@@ -772,25 +766,6 @@ class StateGraph(Generic[StateT, ContextT, InputT, OutputT]):
         if destinations is not None:
             ends = destinations
 
-        resolved_input_schema: type[Any] = (
-            input_schema or inferred_input_schema or self.state_schema
-        )
-        handler_node_name: str | None = None
-        if error_handler is not None:
-            handler_node_name = f"__error_handler__{node}"
-            if handler_node_name in self.nodes:
-                raise ValueError(
-                    f"Auto-generated error handler node `{handler_node_name}` already exists."
-                )
-            self.nodes[handler_node_name] = StateNodeSpec[Any, ContextT](
-                coerce_to_runnable(error_handler, name=handler_node_name, trace=False),  # type: ignore[arg-type]
-                metadata=None,
-                input_schema=resolved_input_schema,
-                retry_policy=None,
-                cache_policy=None,
-                is_error_handler=True,
-            )
-
         if input_schema is not None:
             self.nodes[node] = StateNodeSpec[NodeInputT, ContextT](
                 coerce_to_runnable(action, name=node, trace=False),  # type: ignore[arg-type]
@@ -798,7 +773,6 @@ class StateGraph(Generic[StateT, ContextT, InputT, OutputT]):
                 input_schema=input_schema,
                 retry_policy=retry_policy,
                 cache_policy=cache_policy,
-                error_handler_node=handler_node_name,
                 ends=ends,
                 defer=defer,
                 timeout=timeout,
@@ -810,7 +784,6 @@ class StateGraph(Generic[StateT, ContextT, InputT, OutputT]):
                 input_schema=inferred_input_schema,
                 retry_policy=retry_policy,
                 cache_policy=cache_policy,
-                error_handler_node=handler_node_name,
                 ends=ends,
                 defer=defer,
                 timeout=timeout,
@@ -822,7 +795,6 @@ class StateGraph(Generic[StateT, ContextT, InputT, OutputT]):
                 input_schema=self.state_schema,
                 retry_policy=retry_policy,
                 cache_policy=cache_policy,
-                error_handler_node=handler_node_name,
                 ends=ends,
                 defer=defer,
                 timeout=timeout,
@@ -1080,6 +1052,7 @@ class StateGraph(Generic[StateT, ContextT, InputT, OutputT]):
             for node in interrupt:
                 if node not in self.nodes:
                     raise ValueError(f"Interrupt node `{node}` not found")
+
         self.compiled = True
         return self
 
@@ -1128,7 +1101,7 @@ class StateGraph(Generic[StateT, ContextT, InputT, OutputT]):
             name: The name to use for the compiled graph.
             transformers: Optional sequence of `StreamTransformer` classes or
                 configured factories. Classes and factories are instantiated
-                per run whenever `stream_events(version="v3")` / `astream_events(version="v3")` is called and are
+                per run whenever `stream_v2` / `astream_v2` is called and are
                 propagated to subgraph scopes. Custom factories should follow
                 the standard `StreamTransformer` constructor shape by
                 accepting `scope` as their first argument. Appended after the
@@ -1193,11 +1166,6 @@ class StateGraph(Generic[StateT, ContextT, InputT, OutputT]):
                 key for key, val in self.channels.items() if not is_managed_value(val)
             ]
         )
-        node_error_handler_map = {
-            node_name: spec.error_handler_node
-            for node_name, spec in self.nodes.items()
-            if spec.error_handler_node is not None
-        }
 
         compiled = CompiledStateGraph[StateT, ContextT, InputT, OutputT](
             builder=self,
@@ -1220,7 +1188,6 @@ class StateGraph(Generic[StateT, ContextT, InputT, OutputT]):
             debug=debug,
             store=store,
             cache=cache,
-            node_error_handler_map=node_error_handler_map,
             name=name or "LangGraph",
             stream_transformers=transformers,
         )
@@ -1395,8 +1362,6 @@ class CompiledStateGraph(
                 metadata=node.metadata,
                 retry_policy=node.retry_policy,
                 cache_policy=node.cache_policy,
-                is_error_handler=node.is_error_handler,
-                error_handler_node=node.error_handler_node,
                 bound=node.runnable,  # type: ignore[arg-type]
                 timeout=node.timeout,
             )

libs/langgraph/langgraph/pregel/_algo.py

@@ -39,7 +39,6 @@ from langgraph._internal._constants import (
     CONFIG_KEY_CHECKPOINT_MAP,
     CONFIG_KEY_CHECKPOINT_NS,
     CONFIG_KEY_CHECKPOINTER,
-    CONFIG_KEY_NODE_ERROR,
     CONFIG_KEY_READ,
     CONFIG_KEY_RESUME_MAP,
     CONFIG_KEY_RUNTIME,
@@ -48,7 +47,6 @@ from langgraph._internal._constants import (
     CONFIG_KEY_TASK_ID,
     CONFIG_KEY_THREAD_ID,
     ERROR,
-    ERROR_SOURCE_NODE,
     INTERRUPT,
     NO_WRITES,
     NS_END,
@@ -68,7 +66,6 @@ from langgraph.channels.base import BaseChannel
 from langgraph.channels.topic import Topic
 from langgraph.channels.untracked_value import UntrackedValue
 from langgraph.constants import TAG_HIDDEN
-from langgraph.errors import NodeError
 from langgraph.managed.base import ManagedValueMapping
 from langgraph.pregel._call import get_runnable_for_task, identifier
 from langgraph.pregel._io import read_channels
@@ -295,15 +292,7 @@ def apply_writes(
     pending_writes_by_channel: dict[str, list[Any]] = defaultdict(list)
     for task in tasks:
         for chan, val in task.writes:
-            if chan in (
-                NO_WRITES,
-                PUSH,
-                RESUME,
-                INTERRUPT,
-                RETURN,
-                ERROR,
-                ERROR_SOURCE_NODE,
-            ):
+            if chan in (NO_WRITES, PUSH, RESUME, INTERRUPT, RETURN, ERROR):
                 pass
             elif chan in channels:
                 pending_writes_by_channel[chan].append(val)
@@ -761,42 +750,6 @@ def prepare_single_task(
                 return PregelTask(task_id, name, task_path[:3])
 
 
-def _coerce_pending_error(value: Any) -> BaseException:
-    if isinstance(value, BaseException):
-        return value
-    return Exception(str(value))
-
-
-def _read_errors_from_pending_writes(
-    pending_writes: list[PendingWrite],
-) -> list[BaseException]:
-    errors: list[BaseException] = []
-    for _, channel, value in pending_writes:
-        if channel == ERROR:
-            errors.append(_coerce_pending_error(value))
-    return errors
-
-
-def _read_error_for_task_id_from_pending_writes(
-    pending_writes: list[PendingWrite], task_id: str
-) -> BaseException | None:
-    for pending_task_id, channel, value in reversed(pending_writes):
-        if pending_task_id == task_id and channel == ERROR:
-            return _coerce_pending_error(value)
-    return None
-
-
-def _read_error_source_node_from_pending_writes(
-    pending_writes: list[PendingWrite], task_id: str
-) -> str | None:
-    for pending_task_id, channel, value in reversed(pending_writes):
-        if pending_task_id == task_id and channel == ERROR_SOURCE_NODE:
-            if isinstance(value, str):
-                return value
-            return str(value)
-    return None
-
-
 def prepare_push_task_functional(
     task_path: tuple[str, tuple, int, str, Call],
     # (PUSH, parent task path, idx of PUSH write, id of parent task, Call)
@@ -1107,147 +1060,6 @@ def prepare_push_task_send(
         return PregelTask(task_id, packet.node, translated_task_path)
 
 
-def prepare_node_error_handler_task(
-    failed_task: PregelExecutableTask,
-    *,
-    handler_node_name: str,
-    failed_error: BaseException,
-    checkpoint: Checkpoint,
-    pending_writes: list[PendingWrite],
-    processes: Mapping[str, PregelNode],
-    channels: Mapping[str, BaseChannel],
-    managed: ManagedValueMapping,
-    config: RunnableConfig,
-    step: int,
-    stop: int,
-    store: BaseStore | None = None,
-    checkpointer: BaseCheckpointSaver | None = None,
-    manager: None | ParentRunManager | AsyncParentRunManager = None,
-    cache_policy: CachePolicy | None = None,
-    retry_policy: Sequence[RetryPolicy] = (),
-) -> PregelExecutableTask | None:
-    """Prepare an immediate node-level error handler task for a failed task."""
-    if handler_node_name not in processes:
-        return None
-    proc = processes[handler_node_name]
-    proc_node = proc.node
-    if proc_node is None:
-        return None
-
-    checkpoint_id_bytes = binascii.unhexlify(checkpoint["id"].replace("-", ""))
-    task_id_func = _xxhash_str if checkpoint["v"] > 1 else _uuid5_str
-    configurable = config.get(CONF, {})
-    parent_ns = configurable.get(CONFIG_KEY_CHECKPOINT_NS, "")
-    checkpoint_ns = (
-        f"{parent_ns}{NS_SEP}{handler_node_name}" if parent_ns else handler_node_name
-    )
-    task_id = task_id_func(
-        checkpoint_id_bytes,
-        checkpoint_ns,
-        str(step),
-        handler_node_name,
-        PUSH,
-        "node_error_handler",
-        failed_task.id,
-    )
-    task_checkpoint_ns = f"{checkpoint_ns}:{task_id}"
-    translated_task_path = (*failed_task.path[:3], "node_error_handler", False)
-    metadata = {
-        "langgraph_step": step,
-        "langgraph_node": handler_node_name,
-        "langgraph_triggers": PUSH_TRIGGER,
-        "langgraph_path": translated_task_path,
-        "langgraph_checkpoint_ns": task_checkpoint_ns,
-    }
-    if proc.metadata:
-        metadata.update(proc.metadata)
-    writes: deque[tuple[str, Any]] = deque()
-
-    effective_retry_policy = proc.retry_policy or retry_policy
-    effective_cache_policy = proc.cache_policy or cache_policy
-    if effective_cache_policy:
-        args_key = effective_cache_policy.key_func(failed_task.input)
-        cache_key = CacheKey(
-            (
-                CACHE_NS_WRITES,
-                (identifier(proc) or "__dynamic__"),
-                handler_node_name,
-            ),
-            xxh3_128_hexdigest(
-                args_key.encode() if isinstance(args_key, str) else args_key
-            ),
-            effective_cache_policy.ttl,
-        )
-    else:
-        cache_key = None
-
-    scratchpad = _scratchpad(
-        config[CONF].get(CONFIG_KEY_SCRATCHPAD),
-        pending_writes,
-        task_id,
-        xxh3_128_hexdigest(task_checkpoint_ns.encode()),
-        config[CONF].get(CONFIG_KEY_RESUME_MAP),
-        step,
-        stop,
-    )
-    runtime = cast(Runtime, configurable.get(CONFIG_KEY_RUNTIME, DEFAULT_RUNTIME))
-    runtime = runtime.override(
-        store=store, previous=checkpoint["channel_values"].get(PREVIOUS, None)
-    )
-    additional_config: RunnableConfig = {
-        "metadata": metadata,
-        "tags": proc.tags,
-    }
-    return PregelExecutableTask(
-        handler_node_name,
-        failed_task.input,
-        proc_node,
-        writes,
-        patch_config(
-            merge_configs(config, additional_config),
-            run_name=handler_node_name,
-            callbacks=manager.get_child(f"graph:step:{step}") if manager else None,
-            configurable={
-                CONFIG_KEY_TASK_ID: task_id,
-                CONFIG_KEY_SEND: writes.extend,
-                CONFIG_KEY_READ: partial(
-                    local_read,
-                    scratchpad,
-                    channels,
-                    managed,
-                    PregelTaskWrites(
-                        translated_task_path,
-                        handler_node_name,
-                        writes,
-                        PUSH_TRIGGER,
-                    ),
-                ),
-                CONFIG_KEY_CHECKPOINTER: (
-                    checkpointer or configurable.get(CONFIG_KEY_CHECKPOINTER)
-                ),
-                CONFIG_KEY_CHECKPOINT_MAP: {
-                    **configurable.get(CONFIG_KEY_CHECKPOINT_MAP, {}),
-                    parent_ns: checkpoint["id"],
-                },
-                CONFIG_KEY_CHECKPOINT_ID: None,
-                CONFIG_KEY_CHECKPOINT_NS: task_checkpoint_ns,
-                CONFIG_KEY_SCRATCHPAD: scratchpad,
-                CONFIG_KEY_RUNTIME: runtime,
-                CONFIG_KEY_NODE_ERROR: NodeError(
-                    node=failed_task.name, error=failed_error
-                ),
-            },
-        ),
-        PUSH_TRIGGER,
-        effective_retry_policy,
-        cache_key,
-        task_id,
-        translated_task_path,
-        writers=proc.flat_writers,
-        subgraphs=proc.subgraphs,
-    )
-
-
 def checkpoint_null_version(
     checkpoint: Checkpoint,
 ) -> V | None:

libs/langgraph/langgraph/pregel/_loop.py

@@ -45,13 +45,11 @@ from langgraph._internal._constants import (
     CONFIG_KEY_REPLAY_STATE,
     CONFIG_KEY_RESUME_MAP,
     CONFIG_KEY_RESUMING,
-    CONFIG_KEY_RUNTIME,
     CONFIG_KEY_SCRATCHPAD,
     CONFIG_KEY_STREAM,
     CONFIG_KEY_TASK_ID,
     CONFIG_KEY_THREAD_ID,
     ERROR,
-    ERROR_SOURCE_NODE,
     INPUT,
     INTERRUPT,
     NS_END,
@@ -89,7 +87,6 @@ from langgraph.pregel._algo import (
     checkpoint_null_version,
     increment,
     prepare_next_tasks,
-    prepare_node_error_handler_task,
     prepare_single_task,
     sanitize_untracked_values_in_send,
     should_interrupt,
@@ -122,7 +119,6 @@ from langgraph.pregel.debug import (
     map_debug_tasks,
 )
 from langgraph.pregel.protocol import StreamChunk, StreamProtocol
-from langgraph.runtime import RunControl, Runtime
 from langgraph.types import (
     All,
     CachePolicy,
@@ -210,12 +206,10 @@ class PregelLoop:
         "input",
         "pending",
         "done",
-        "draining",
         "interrupt_before",
         "interrupt_after",
         "out_of_steps",
     ]
-    control: RunControl | None
     tasks: dict[str, PregelExecutableTask]
     output: None | dict[str, Any] | Any = None
     updated_channels: set[str] | None = None
@@ -323,8 +317,6 @@ class PregelLoop:
             else ()
         )
         self.prev_checkpoint_config = None
-        runtime = self.config[CONF].get(CONFIG_KEY_RUNTIME)
-        self.control = runtime.control if isinstance(runtime, Runtime) else None
 
     def _push_graph_lifecycle_event(
         self,
@@ -332,16 +324,11 @@ class PregelLoop:
         *,
         interrupts: tuple[Interrupt, ...] = (),
     ) -> None:
-        # drain status never reaches lifecycle events: tick() returns False
-        # before pushing, and interrupts are raised through GraphInterrupt
-        if self.status == "draining":
-            raise RuntimeError("Draining status cannot emit lifecycle events")
-        status = self.status
         if kind == "resume":
             self._graph_lifecycle_events.append(
                 GraphResumeEvent(
                     run_id=None,
-                    status=status,
+                    status=self.status,
                     checkpoint_id=self.checkpoint["id"],
                     checkpoint_ns=self.checkpoint_ns,
                 )
@@ -350,7 +337,7 @@ class PregelLoop:
             self._graph_lifecycle_events.append(
                 GraphInterruptEvent(
                     run_id=None,
-                    status=status,
+                    status=self.status,
                     checkpoint_id=self.checkpoint["id"],
                     checkpoint_ns=self.checkpoint_ns,
                     interrupts=interrupts,
@@ -524,16 +511,6 @@ class PregelLoop:
             # return the new task, to be started if not run before
             return pushed
 
-    def schedule_error_handler(
-        self, failed_task: PregelExecutableTask, error: BaseException
-    ) -> PregelExecutableTask | None:
-        raise NotImplementedError
-
-    async def aschedule_error_handler(
-        self, failed_task: PregelExecutableTask, error: BaseException
-    ) -> PregelExecutableTask | None:
-        raise NotImplementedError
-
     def tick(self) -> bool:
         """Execute a single iteration of the Pregel loop.
 
@@ -592,10 +569,6 @@ class PregelLoop:
             self.status = "done"
             return False
 
-        if self.control is not None and self.control.drain_requested:
-            self.status = "draining"
-            return False
-
         # if there are pending writes from a previous loop, apply them
         if not self.is_replaying and self.checkpoint_pending_writes:
             self._match_writes(self.tasks)
@@ -662,7 +635,7 @@ class PregelLoop:
 
     def _match_writes(self, tasks: Mapping[str, PregelExecutableTask]) -> None:
         for tid, k, v in self.checkpoint_pending_writes:
-            if k in (ERROR, ERROR_SOURCE_NODE, INTERRUPT, RESUME):
+            if k in (ERROR, INTERRUPT, RESUME):
                 continue
             if task := tasks.get(tid):
                 task.writes.append((k, v))
@@ -1239,45 +1212,6 @@ class SyncPregelLoop(PregelLoop, AbstractContextManager):
                 self.output_writes(task.id, task.writes, cached=True)
         return pushed
 
-    def schedule_error_handler(
-        self, failed_task: PregelExecutableTask, error: BaseException
-    ) -> PregelExecutableTask | None:
-        handler_node = self.nodes[failed_task.name].error_handler_node
-        if not handler_node:
-            return None
-        writes = list(failed_task.writes)
-        writes.append((ERROR_SOURCE_NODE, failed_task.name))
-        self.put_writes(
-            failed_task.id,
-            writes,
-        )
-        handler_task = prepare_node_error_handler_task(
-            failed_task,
-            handler_node_name=handler_node,
-            failed_error=error,
-            checkpoint=self.checkpoint,
-            pending_writes=self.checkpoint_pending_writes,
-            processes=self.nodes,
-            channels=self.channels,
-            managed=self.managed,
-            config=failed_task.config,
-            step=self.step,
-            stop=self.stop,
-            store=self.store,
-            checkpointer=self.checkpointer,
-            manager=self.manager,
-            retry_policy=self.retry_policy,
-            cache_policy=self.cache_policy,
-        )
-        if handler_task is None:
-            return None
-        self.tasks[handler_task.id] = handler_task
-        if not self.is_replaying:
-            self._match_writes({handler_task.id: handler_task})
-        for task in self.match_cached_writes():
-            self.output_writes(task.id, task.writes, cached=True)
-        return handler_task
-
     def put_writes(self, task_id: str, writes: WritesT) -> None:
         """Put writes for a task, to be read by the next tick."""
         super().put_writes(task_id, writes)
@@ -1485,45 +1419,6 @@ class AsyncPregelLoop(PregelLoop, AbstractAsyncContextManager):
                 self.output_writes(task.id, task.writes, cached=True)
         return pushed
 
-    async def aschedule_error_handler(
-        self, failed_task: PregelExecutableTask, error: BaseException
-    ) -> PregelExecutableTask | None:
-        handler_node = self.nodes[failed_task.name].error_handler_node
-        if not handler_node:
-            return None
-        writes = list(failed_task.writes)
-        writes.append((ERROR_SOURCE_NODE, failed_task.name))
-        self.put_writes(
-            failed_task.id,
-            writes,
-        )
-        handler_task = prepare_node_error_handler_task(
-            failed_task,
-            handler_node_name=handler_node,
-            failed_error=error,
-            checkpoint=self.checkpoint,
-            pending_writes=self.checkpoint_pending_writes,
-            processes=self.nodes,
-            channels=self.channels,
-            managed=self.managed,
-            config=failed_task.config,
-            step=self.step,
-            stop=self.stop,
-            store=self.store,
-            checkpointer=self.checkpointer,
-            manager=self.manager,
-            retry_policy=self.retry_policy,
-            cache_policy=self.cache_policy,
-        )
-        if handler_task is None:
-            return None
-        self.tasks[handler_task.id] = handler_task
-        if not self.is_replaying:
-            self._match_writes({handler_task.id: handler_task})
-        for task in await self.amatch_cached_writes():
-            self.output_writes(task.id, task.writes, cached=True)
-        return handler_task
-
     def put_writes(self, task_id: str, writes: WritesT) -> None:
         """Put writes for a task, to be read by the next tick."""
         super().put_writes(task_id, writes)

libs/langgraph/langgraph/pregel/_messages.py

@@ -349,7 +349,7 @@ class StreamMessagesHandlerV2(StreamMessagesHandler, _V2StreamingCallbackHandler
         tags: list[str] | None = None,
         **kwargs: Any,
     ) -> Any:
-        """Forward a protocol event from `stream_events(version="v3")` as a messages stream part.
+        """Forward a protocol event from `stream_v2` as a messages stream part.
 
         Fires once per `MessagesData` event (`message-start`, per-block
         `content-block-*`, `message-finish`). The transformer layer

libs/langgraph/langgraph/pregel/_read.py

@@ -138,12 +138,6 @@ class PregelNode:
     metadata: Mapping[str, Any] | None
     """Metadata to attach to the node for tracing."""
 
-    is_error_handler: bool
-    """Whether this node is registered as an error handler node."""
-
-    error_handler_node: str | None
-    """Optional handler node name for failures from this node."""
-
     subgraphs: Sequence[PregelProtocol]
     """Subgraphs used by the node."""
 
@@ -159,8 +153,6 @@ class PregelNode:
         bound: Runnable[Any, Any] | None = None,
         retry_policy: RetryPolicy | Sequence[RetryPolicy] | None = None,
         cache_policy: CachePolicy | None = None,
-        is_error_handler: bool = False,
-        error_handler_node: str | None = None,
         subgraphs: Sequence[PregelProtocol] | None = None,
         timeout: float | timedelta | TimeoutPolicy | None = None,
     ) -> None:
@@ -177,8 +169,6 @@ class PregelNode:
         self.timeout = coerce_timeout_policy(timeout)
         self.tags = tags
         self.metadata = metadata
-        self.is_error_handler = is_error_handler
-        self.error_handler_node = error_handler_node
         if subgraphs is not None:
             self.subgraphs = subgraphs
         elif self.bound is not DEFAULT_BOUND:

libs/langgraph/langgraph/pregel/_runner.py

@@ -10,10 +10,8 @@ from collections.abc import (
     AsyncIterator,
     Awaitable,
     Callable,
-    Collection,
     Iterable,
     Iterator,
-    Mapping,
     Sequence,
 )
 from functools import partial
@@ -74,10 +72,6 @@ SKIP_RERAISE_SET: weakref.WeakSet[concurrent.futures.Future | asyncio.Future] =
 class FuturesDict(Generic[F, E], dict[F, PregelExecutableTask | None]):
     event: E
     callback: weakref.ref[Callable[[PregelExecutableTask, BaseException | None], None]]
-    # Stop condition is injected by PregelRunner instead of hard-coded here.
-    # This lets the runner treat graph-error-handled exceptions as non-fatal
-    # so `on_done` does not trigger an early stop for those futures.
-    should_stop: Callable[[set[F]], bool]
     counter: int
     done: set[F]
     lock: threading.Lock
@@ -88,7 +82,6 @@ class FuturesDict(Generic[F, E], dict[F, PregelExecutableTask | None]):
         callback: weakref.ref[
             Callable[[PregelExecutableTask, BaseException | None], None]
         ],
-        should_stop: Callable[[set[F]], bool],
         future_type: type[F],
         # used for generic typing, newer py supports FutureDict[...](...)
     ) -> None:
@@ -96,7 +89,6 @@ class FuturesDict(Generic[F, E], dict[F, PregelExecutableTask | None]):
         self.lock = threading.Lock()
         self.event = event
         self.callback = callback
-        self.should_stop = should_stop
         self.counter = 0
         self.done: set[F] = set()
 
@@ -117,7 +109,6 @@ class FuturesDict(Generic[F, E], dict[F, PregelExecutableTask | None]):
         task: PregelExecutableTask,
         fut: F,
     ) -> None:
-        # Called automatically by future.add_done_callback registered in __setitem__.
         try:
             if cb := self.callback():
                 cb(task, _exception(fut))
@@ -125,9 +116,7 @@ class FuturesDict(Generic[F, E], dict[F, PregelExecutableTask | None]):
             with self.lock:
                 self.done.add(fut)
                 self.counter -= 1
-                # Wake waiter when all tracked futures are done, or when runner-level
-                # stop condition is met (for example, a non-handled fatal exception).
-                if self.counter == 0 or self.should_stop(self.done):
+                if self.counter == 0 or _should_stop_others(self.done):
                     self.event.set()
 
 
@@ -143,34 +132,11 @@ class PregelRunner:
         put_writes: weakref.ref[Callable[[str, Sequence[tuple[str, Any]]], None]],
         use_astream: bool = False,
         node_finished: Callable[[str], None] | None = None,
-        node_error_handler_map: Mapping[str, str] | None = None,
-        schedule_error_handler: Callable[
-            [PregelExecutableTask, BaseException], PregelExecutableTask | None
-        ]
-        | None = None,
-        aschedule_error_handler: Callable[
-            [PregelExecutableTask, BaseException],
-            Awaitable[PregelExecutableTask | None],
-        ]
-        | None = None,
     ) -> None:
         self.submit = submit
         self.put_writes = put_writes
         self.use_astream = use_astream
         self.node_finished = node_finished
-        self.node_error_handler_map = dict(node_error_handler_map or {})
-        self.error_handler_nodes = set(self.node_error_handler_map.values())
-        self.schedule_error_handler = schedule_error_handler
-        self.aschedule_error_handler = aschedule_error_handler
-        # Exception object ids that are already routed to graph-level error handler.
-        # These ids are consulted by stop/panic checks to avoid re-raising handled
-        # exceptions via the normal fatal path in the same run.
-        self._handled_exception_ids: set[int] = set()
-
-    def _should_route_to_error_handler(self, task: PregelExecutableTask) -> bool:
-        if task.name in self.error_handler_nodes:
-            return False
-        return task.name in self.node_error_handler_map
 
     def tick(
         self,
@@ -189,9 +155,6 @@ class PregelRunner:
         futures = FuturesDict(
             callback=weakref.WeakMethod(self.commit),
             event=threading.Event(),
-            should_stop=partial(
-                _should_stop_others, handled_exception_ids=self._handled_exception_ids
-            ),
             future_type=concurrent.futures.Future,
         )
         # give control back to the caller
@@ -201,7 +164,6 @@ class PregelRunner:
             return
         elif len(tasks) == 1 and timeout is None and get_waiter is None:
             t = tasks[0]
-            scheduled_error_handler = False
             try:
                 run_with_retry(
                     t,
@@ -220,23 +182,12 @@ class PregelRunner:
                 self.commit(t, None)
             except Exception as exc:
                 self.commit(t, exc)
-                if (
-                    not isinstance(exc, GraphBubbleUp)
-                    and self._should_route_to_error_handler(t)
-                    and self.schedule_error_handler is not None
-                ):
-                    self._handled_exception_ids.add(id(exc))
-                    if handler_task := self.schedule_error_handler(t, exc):
-                        tasks = (handler_task,)
-                        scheduled_error_handler = True
-                        # Continue to the regular scheduling path for handler execution.
                 if reraise and futures:
-                    if id(exc) not in self._handled_exception_ids:
-                        # will be re-raised after futures are done
-                        fut: concurrent.futures.Future = concurrent.futures.Future()
-                        fut.set_exception(exc)
-                        futures.done.add(fut)
-                elif reraise and id(exc) not in self._handled_exception_ids:
+                    # will be re-raised after futures are done
+                    fut: concurrent.futures.Future = concurrent.futures.Future()
+                    fut.set_exception(exc)
+                    futures.done.add(fut)
+                elif reraise:
                     if tb := exc.__traceback__:
                         while tb.tb_next is not None and any(
                             tb.tb_frame.f_code.co_filename.endswith(name)
@@ -245,12 +196,10 @@ class PregelRunner:
                             tb = tb.tb_next
                         exc.__traceback__ = tb
                     raise
-            if not futures and not scheduled_error_handler:
-                # maybe `t` scheduled another task
+            if not futures:  # maybe `t` scheduled another task
                 return
             else:
-                if not scheduled_error_handler:
-                    tasks = ()  # don't reschedule this task
+                tasks = ()  # don't reschedule this task
         # add waiter task if requested
         if get_waiter is not None:
             futures[get_waiter()] = None
@@ -277,7 +226,6 @@ class PregelRunner:
         # each task is independent from all other concurrent tasks
         # yield updates/debug output as each task finishes
         end_time = timeout + time.monotonic() if timeout else None
-        handled_futures: set[concurrent.futures.Future[Any]] = set()
         while len(futures) > (1 if get_waiter is not None else 0):
             done, inflight = concurrent.futures.wait(
                 futures,
@@ -286,49 +234,17 @@ class PregelRunner:
             )
             if not done:
                 break  # timed out
-            done_for_stop: set[concurrent.futures.Future[Any]] = set()
             for fut in done:
                 task = futures.pop(fut)
                 if task is None:
                     # waiter task finished, schedule another
                     if inflight and get_waiter is not None:
                         futures[get_waiter()] = None
-                elif (
-                    (task_exc := _exception(fut))
-                    and self._should_route_to_error_handler(task)
-                    and not isinstance(task_exc, GraphBubbleUp)
-                ):
-                    self._handled_exception_ids.add(id(task_exc))
-                    SKIP_RERAISE_SET.add(fut)
-                    handled_futures.add(fut)
-                    if self.schedule_error_handler is not None:
-                        if handler_task := self.schedule_error_handler(task, task_exc):
-                            handler_fut = self.submit()(  # type: ignore[misc]
-                                run_with_retry,
-                                handler_task,
-                                retry_policy,
-                                configurable={
-                                    CONFIG_KEY_CALL: partial(
-                                        _call,
-                                        weakref.ref(handler_task),
-                                        retry_policy=retry_policy,
-                                        futures=weakref.ref(futures),
-                                        schedule_task=schedule_task,
-                                        submit=self.submit,
-                                    ),
-                                },
-                                __reraise_on_exit__=reraise,
-                            )
-                            futures[handler_fut] = handler_task
-                else:
-                    done_for_stop.add(fut)
             else:
                 # remove references to loop vars
                 del fut, task
             # maybe stop other tasks
-            if _should_stop_others(
-                done_for_stop, handled_exception_ids=self._handled_exception_ids
-            ):
+            if _should_stop_others(done):
                 break
             # give control back to the caller
             yield
@@ -343,8 +259,6 @@ class PregelRunner:
             _panic_or_proceed(
                 futures.done.union(f for f, t in futures.items() if t is not None),
                 panic=reraise,
-                handled_exception_ids=self._handled_exception_ids,
-                handled_futures=handled_futures,
             )
         except Exception as exc:
             if tb := exc.__traceback__:
@@ -378,9 +292,6 @@ class PregelRunner:
         futures = FuturesDict(
             callback=weakref.WeakMethod(self.commit),
             event=asyncio.Event(),
-            should_stop=partial(
-                _should_stop_others, handled_exception_ids=self._handled_exception_ids
-            ),
             future_type=asyncio.Future,
         )
         # give control back to the caller
@@ -390,7 +301,6 @@ class PregelRunner:
             return
         elif len(tasks) == 1 and get_waiter is None and timeout is None:
             t = tasks[0]
-            scheduled_error_handler = False
             try:
                 await arun_with_retry(
                     t,
@@ -412,22 +322,12 @@ class PregelRunner:
                 self.commit(t, None)
             except Exception as exc:
                 self.commit(t, exc)
-                if (
-                    not isinstance(exc, GraphBubbleUp)
-                    and self._should_route_to_error_handler(t)
-                    and self.aschedule_error_handler is not None
-                ):
-                    self._handled_exception_ids.add(id(exc))
-                    if handler_task := await self.aschedule_error_handler(t, exc):
-                        tasks = (handler_task,)
-                        scheduled_error_handler = True
                 if reraise and futures:
-                    if id(exc) not in self._handled_exception_ids:
-                        # will be re-raised after futures are done
-                        fut: asyncio.Future = loop.create_future()
-                        fut.set_exception(exc)
-                        futures.done.add(fut)
-                elif reraise and id(exc) not in self._handled_exception_ids:
+                    # will be re-raised after futures are done
+                    fut: asyncio.Future = loop.create_future()
+                    fut.set_exception(exc)
+                    futures.done.add(fut)
+                elif reraise:
                     if tb := exc.__traceback__:
                         while tb.tb_next is not None and any(
                             tb.tb_frame.f_code.co_filename.endswith(name)
@@ -436,12 +336,10 @@ class PregelRunner:
                             tb = tb.tb_next
                         exc.__traceback__ = tb
                     raise
-            if not futures and not scheduled_error_handler:
-                # maybe `t` scheduled another task
+            if not futures:  # maybe `t` scheduled another task
                 return
             else:
-                if not scheduled_error_handler:
-                    tasks = ()  # don't reschedule this task
+                tasks = ()  # don't reschedule this task
         # add waiter task if requested
         if get_waiter is not None:
             futures[get_waiter()] = None
@@ -476,7 +374,6 @@ class PregelRunner:
         # each task is independent from all other concurrent tasks
         # yield updates/debug output as each task finishes
         end_time = timeout + loop.time() if timeout else None
-        handled_futures: set[asyncio.Future[Any]] = set()
         while len(futures) > (1 if get_waiter is not None else 0):
             done, inflight = await asyncio.wait(
                 futures,
@@ -485,59 +382,17 @@ class PregelRunner:
             )
             if not done:
                 break  # timed out
-            done_for_stop: set[asyncio.Future[Any]] = set()
             for fut in done:
                 task = futures.pop(fut)
                 if task is None:
                     # waiter task finished, schedule another
                     if inflight and get_waiter is not None:
                         futures[get_waiter()] = None
-                elif (
-                    (task_exc := _exception(fut))
-                    and self._should_route_to_error_handler(task)
-                    and not isinstance(task_exc, GraphBubbleUp)
-                ):
-                    self._handled_exception_ids.add(id(task_exc))
-                    SKIP_RERAISE_SET.add(fut)
-                    handled_futures.add(fut)
-                    if self.aschedule_error_handler is not None:
-                        if handler_task := await self.aschedule_error_handler(
-                            task, task_exc
-                        ):
-                            handler_fut = cast(
-                                asyncio.Future,
-                                self.submit()(  # type: ignore[misc]
-                                    arun_with_retry,
-                                    handler_task,
-                                    retry_policy,
-                                    stream=self.use_astream,
-                                    configurable={
-                                        CONFIG_KEY_CALL: partial(
-                                            _acall,
-                                            weakref.ref(handler_task),
-                                            retry_policy=retry_policy,
-                                            stream=self.use_astream,
-                                            futures=weakref.ref(futures),
-                                            schedule_task=schedule_task,
-                                            submit=self.submit,
-                                            loop=loop,
-                                        ),
-                                    },
-                                    __name__=handler_task.name,
-                                    __cancel_on_exit__=True,
-                                    __reraise_on_exit__=reraise,
-                                ),
-                            )
-                            futures[handler_fut] = handler_task
-                else:
-                    done_for_stop.add(fut)
             else:
                 # remove references to loop vars
                 del fut, task
             # maybe stop other tasks
-            if _should_stop_others(
-                done_for_stop, handled_exception_ids=self._handled_exception_ids
-            ):
+            if _should_stop_others(done):
                 break
             # give control back to the caller
             yield
@@ -557,8 +412,6 @@ class PregelRunner:
                 futures.done.union(f for f, t in futures.items() if t is not None),
                 timeout_exc_cls=asyncio.TimeoutError,
                 panic=reraise,
-                handled_exception_ids=self._handled_exception_ids,
-                handled_futures=handled_futures,
             )
         except Exception as exc:
             if tb := exc.__traceback__:
@@ -594,11 +447,6 @@ class PregelRunner:
             else:
                 # save error to checkpointer
                 task.writes.append((ERROR, exception))
-                if self._should_route_to_error_handler(task) and not isinstance(
-                    exception, GraphBubbleUp
-                ):
-                    # Mark early in commit path; loop-side routing may happen later.
-                    self._handled_exception_ids.add(id(exception))
                 self.put_writes()(task.id, task.writes)  # type: ignore[misc]
         else:
             if self.node_finished and (
@@ -614,8 +462,6 @@ class PregelRunner:
 
 def _should_stop_others(
     done: set[F],
-    *,
-    handled_exception_ids: set[int] | None = None,
 ) -> bool:
     """Check if any task failed, if so, cancel all other tasks.
     GraphInterrupts are not considered failures."""
@@ -623,11 +469,7 @@ def _should_stop_others(
         if fut.cancelled():
             continue
         elif exc := fut.exception():
-            if (
-                id(exc) not in (handled_exception_ids or set())
-                and not isinstance(exc, GraphBubbleUp)
-                and fut not in SKIP_RERAISE_SET
-            ):
+            if not isinstance(exc, GraphBubbleUp) and fut not in SKIP_RERAISE_SET:
                 return True
 
     return False
@@ -651,9 +493,6 @@ def _panic_or_proceed(
     *,
     timeout_exc_cls: type[Exception] = TimeoutError,
     panic: bool = True,
-    handled_exception_ids: set[int] | None = None,
-    handled_futures: Collection[concurrent.futures.Future[Any] | asyncio.Future[Any]]
-    | None = None,
 ) -> None:
     """Cancel remaining tasks if any failed, re-raise exception if panic is True."""
     done: set[concurrent.futures.Future[Any] | asyncio.Future[Any]] = set()
@@ -670,10 +509,6 @@ def _panic_or_proceed(
         # if any task failed
         fut = done.pop()
         if exc := _exception(fut):
-            if fut in (handled_futures or set()):
-                continue
-            if id(exc) in (handled_exception_ids or set()):
-                continue
             # cancel all pending tasks
             while inflight:
                 inflight.pop().cancel()

libs/langgraph/langgraph/pregel/main.py

@@ -30,7 +30,6 @@ from typing import (
 )
 from uuid import UUID, uuid5
 
-from langchain_core._api import beta
 from langchain_core.globals import get_debug
 from langchain_core.runnables import (
     RunnableSequence,
@@ -42,7 +41,6 @@ from langchain_core.runnables.config import (
     get_callback_manager_for_config,
 )
 from langchain_core.runnables.graph import Graph
-from langchain_core.runnables.schema import StreamEvent
 from langgraph.cache.base import BaseCache
 from langgraph.checkpoint.base import (
     BaseCheckpointSaver,
@@ -113,7 +111,6 @@ from langgraph.config import get_config
 from langgraph.constants import END
 from langgraph.errors import (
     ErrorCode,
-    GraphDrained,
     GraphRecursionError,
     InvalidUpdateError,
     create_error_message,
@@ -159,7 +156,6 @@ from langgraph.pregel.protocol import PregelProtocol, StreamChunk, StreamProtoco
 from langgraph.runtime import (
     DEFAULT_RUNTIME,
     BaseUser,
-    RunControl,
     Runtime,
     ServerInfo,
 )
@@ -378,7 +374,7 @@ def _collect_stream_modes(mux: Any) -> list[StreamMode]:
     """Return the union of `required_stream_modes` across registered transformers.
 
     Transformers declare the stream modes they need to function, and
-    `stream_events(version="v3")` asks the graph for exactly that union — no hardcoded
+    `stream_v2` asks the graph for exactly that union — no hardcoded
     default set. If zero transformers declare a given mode, the graph
     does not stream events for it.
     """
@@ -408,14 +404,14 @@ def _normalize_stream_transformer_factories(
     for spec in specs or ():
         if isinstance(spec, StreamTransformer):
             raise TypeError(
-                "stream_events(version='v3') transformers must be scope-aware callables, "
+                "stream_v2 transformers must be scope-aware callables, "
                 f"got pre-built instance {type(spec).__name__}. Pass the "
                 "transformer class or a factory like "
                 "`lambda scope: MyTransformer(scope, ...)`."
             )
         if not callable(spec):
             raise TypeError(
-                "stream_events(version='v3') transformers must be scope-aware callables, "
+                "stream_v2 transformers must be scope-aware callables, "
                 f"got {type(spec).__name__}."
             )
 
@@ -732,7 +728,6 @@ class Pregel(
     name: str = "LangGraph"
 
     trigger_to_nodes: Mapping[str, Sequence[str]]
-    node_error_handler_map: Mapping[str, str]
 
     def __init__(
         self,
@@ -757,7 +752,6 @@ class Pregel(
         context_schema: type[ContextT] | None = None,
         config: RunnableConfig | None = None,
         trigger_to_nodes: Mapping[str, Sequence[str]] | None = None,
-        node_error_handler_map: Mapping[str, str] | None = None,
         name: str = "LangGraph",
         stream_transformers: Sequence[Callable[[tuple[str, ...]], Any]] | None = None,
         **deprecated_kwargs: Unpack[DeprecatedKwargs],
@@ -805,7 +799,6 @@ class Pregel(
         self.context_schema = context_schema
         self.config = config
         self.trigger_to_nodes = trigger_to_nodes or {}
-        self.node_error_handler_map = node_error_handler_map or {}
         self.name = name
         self.stream_transformers: tuple[Callable[[tuple[str, ...]], Any], ...] = tuple(
             stream_transformers or ()
@@ -2577,7 +2570,6 @@ class Pregel(
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
         durability: Durability | None = None,
-        control: RunControl | None = None,
         subgraphs: bool = False,
         debug: bool | None = None,
         version: Literal["v2"],
@@ -2597,7 +2589,6 @@ class Pregel(
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
         durability: Durability | None = None,
-        control: RunControl | None = None,
         subgraphs: bool = False,
         debug: bool | None = None,
         version: Literal["v1"] = ...,
@@ -2616,7 +2607,6 @@ class Pregel(
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
         durability: Durability | None = None,
-        control: RunControl | None = None,
         subgraphs: bool = False,
         debug: bool | None = None,
         version: Literal["v1", "v2"] = "v1",
@@ -2661,7 +2651,6 @@ class Pregel(
                 - `"sync"`: Changes are persisted synchronously before the next step starts.
                 - `"async"`: Changes are persisted asynchronously while the next step executes.
                 - `"exit"`: Changes are persisted only when the graph exits.
-            control: Optional run control used to request cooperative drain.
             subgraphs: Whether to stream events from inside subgraphs, defaults to `False`.
 
                 If `True`, the events will be emitted as tuples `(namespace, data)`,
@@ -2826,7 +2815,6 @@ class Pregel(
                 previous=None,
                 execution_info=None,
                 server_info=server_info,
-                control=control or parent_runtime.control or RunControl(),
             )
             runtime = parent_runtime.merge(runtime)
             config[CONF][CONFIG_KEY_RUNTIME] = runtime
@@ -2876,8 +2864,6 @@ class Pregel(
                     ),
                     put_writes=weakref.WeakMethod(loop.put_writes),
                     node_finished=config[CONF].get(CONFIG_KEY_NODE_FINISHED),
-                    node_error_handler_map=self.node_error_handler_map,
-                    schedule_error_handler=loop.schedule_error_handler,
                 )
                 # enable subgraph streaming
                 if subgraphs:
@@ -2959,10 +2945,6 @@ class Pregel(
                     error_code=ErrorCode.GRAPH_RECURSION_LIMIT,
                 )
                 raise GraphRecursionError(msg)
-            elif loop.status == "draining":
-                if loop.control is None:
-                    raise RuntimeError("Draining status requires run control")
-                raise GraphDrained(loop.control.drain_reason or "shutdown")
             # set final channel values as run output
             run_manager.on_chain_end(loop.output)
         except BaseException as e:
@@ -2983,7 +2965,6 @@ class Pregel(
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
         durability: Durability | None = None,
-        control: RunControl | None = None,
         subgraphs: bool = False,
         debug: bool | None = None,
         version: Literal["v2"],
@@ -3003,7 +2984,6 @@ class Pregel(
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
         durability: Durability | None = None,
-        control: RunControl | None = None,
         subgraphs: bool = False,
         debug: bool | None = None,
         version: Literal["v1"] = ...,
@@ -3022,7 +3002,6 @@ class Pregel(
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
         durability: Durability | None = None,
-        control: RunControl | None = None,
         subgraphs: bool = False,
         debug: bool | None = None,
         version: Literal["v1", "v2"] = "v1",
@@ -3067,7 +3046,6 @@ class Pregel(
                 - `"sync"`: Changes are persisted synchronously before the next step starts.
                 - `"async"`: Changes are persisted asynchronously while the next step executes.
                 - `"exit"`: Changes are persisted only when the graph exits.
-            control: Optional run control used to request cooperative drain.
             subgraphs: Whether to stream events from inside subgraphs, defaults to `False`.
 
                 If `True`, the events will be emitted as tuples `(namespace, data)`,
@@ -3267,7 +3245,6 @@ class Pregel(
                 previous=None,
                 execution_info=None,
                 server_info=server_info,
-                control=control or parent_runtime.control or RunControl(),
             )
             runtime = parent_runtime.merge(runtime)
             config[CONF][CONFIG_KEY_RUNTIME] = runtime
@@ -3329,8 +3306,6 @@ class Pregel(
                     put_writes=weakref.WeakMethod(loop.put_writes),
                     use_astream=do_stream,
                     node_finished=config[CONF].get(CONFIG_KEY_NODE_FINISHED),
-                    node_error_handler_map=self.node_error_handler_map,
-                    aschedule_error_handler=loop.aschedule_error_handler,
                 )
                 # enable subgraph streaming
                 if subgraphs:
@@ -3438,10 +3413,6 @@ class Pregel(
                     error_code=ErrorCode.GRAPH_RECURSION_LIMIT,
                 )
                 raise GraphRecursionError(msg)
-            elif loop.status == "draining":
-                if loop.control is None:
-                    raise RuntimeError("Draining status requires run control")
-                raise GraphDrained(loop.control.drain_reason or "shutdown")
             # set final channel values as run output
             await run_manager.on_chain_end(loop.output)
         except BaseException as e:
@@ -3449,22 +3420,49 @@ class Pregel(
                 await asyncio.shield(run_manager.on_chain_error(e))
             raise
 
-    @beta(message="The v3 streaming protocol on Pregel is experimental.")
-    def _pregel_stream_v3(
+    def stream_v2(
         self,
         input: InputT | Command | None,
         config: RunnableConfig | None = None,
         *,
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
-        control: RunControl | None = None,
         transformers: Sequence[Callable[[tuple[str, ...]], Any]] | None = None,
     ) -> Any:
-        """Internal v3 sync streaming implementation. Public entry: stream_events(version='v3').
+        """Start a sync v2 streaming run driven by transformer projections.
 
-        !!! warning
+        Builds a `StreamMux` from the built-in transformers, this
+        graph's compile-time `stream_transformers`, and any additional
+        `transformers=` supplied at the call site. Returns a
+        `GraphRunStream` that the caller drives by iterating any
+        projection — no background thread.
 
-            The v3 streaming protocol is experimental and may change.
+        `run.output`, `run.interrupted` and `run.interrupts` work
+        regardless of which transformers are registered.
+
+        Note:
+            Nesting v1 `stream(stream_mode="messages")` inside a node
+            of a `stream_v2` run is not fully supported. The outer v2
+            messages handler reroutes `BaseChatModel.invoke` through
+            the v2 event protocol, so the inner v1 handler does not see
+            `on_llm_new_token` chunks. The inner stream still yields a
+            finalized message via `on_llm_end`. Use `stream_v2` for
+            the inner graph as well, or call
+            `chat_model.stream(...)` explicitly, to get token-level
+            streaming.
+
+        Args:
+            input: Graph input.
+            config: Optional runnable config forwarded to the graph.
+            interrupt_before: Nodes to interrupt before, if any.
+            interrupt_after: Nodes to interrupt after, if any.
+            transformers: Extra transformer classes or configured factories
+                appended after compile-time `stream_transformers`. Factories
+                are called as `factory(scope)` so they can propagate to
+                subgraph scopes.
+
+        Returns:
+            A `GraphRunStream` the caller iterates to drive the run.
         """
         parent_ns = _resolve_parent_ns(self.config, config)
         compiled_factories = _normalize_stream_transformer_factories(
@@ -3492,27 +3490,43 @@ class Pregel(
                 version="v2",
                 interrupt_before=interrupt_before,
                 interrupt_after=interrupt_after,
-                control=control,
             )
         )
         return GraphRunStream(graph_iter, mux)
 
-    @beta(message="The v3 streaming protocol on Pregel is experimental.")
-    async def _apregel_stream_v3(
+    async def astream_v2(
         self,
         input: InputT | Command | None,
         config: RunnableConfig | None = None,
         *,
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
-        control: RunControl | None = None,
         transformers: Sequence[Callable[[tuple[str, ...]], Any]] | None = None,
     ) -> Any:
-        """Internal v3 async streaming implementation. Public entry: astream_events(version='v3').
+        """Async counterpart to `stream_v2`.
 
-        !!! warning
+        Returns an `AsyncGraphRunStream` whose projections can be awaited
+        concurrently; each subscribed cursor drives the pump when its
+        buffer is empty.
 
-            The v3 streaming protocol is experimental and may change.
+        Note:
+            Same nesting limitation as `stream_v2`: nesting v1
+            `astream(stream_mode="messages")` inside a node of an
+            `astream_v2` run drops `on_llm_new_token` chunks because
+            the outer v2 handler reroutes `BaseChatModel.invoke`
+            through the v2 event protocol. Use `astream_v2` for the
+            inner graph as well, or call `chat_model.astream(...)`
+            explicitly, to get token-level streaming.
+
+        Args:
+            input: Graph input.
+            config: Optional runnable config forwarded to the graph.
+            interrupt_before: Nodes to interrupt before, if any.
+            interrupt_after: Nodes to interrupt after, if any.
+            transformers: Extra transformer classes or configured factories
+                appended after compile-time `stream_transformers`. Factories
+                are called as `factory(scope)` so they can propagate to
+                subgraph scopes.
         """
         parent_ns = _resolve_parent_ns(self.config, config)
         compiled_factories = _normalize_stream_transformer_factories(
@@ -3539,166 +3553,9 @@ class Pregel(
             version="v2",
             interrupt_before=interrupt_before,
             interrupt_after=interrupt_after,
-            control=control,
         ).__aiter__()
         return AsyncGraphRunStream(graph_aiter, mux)
 
-    @overload
-    def stream_events(
-        self,
-        input: InputT | Command | None,
-        config: RunnableConfig | None = None,
-        *,
-        version: Literal["v1", "v2"] = "v2",
-        **kwargs: Any,
-    ) -> Iterator[StreamEvent]: ...
-
-    @overload
-    def stream_events(
-        self,
-        input: InputT | Command | None,
-        config: RunnableConfig | None = None,
-        *,
-        version: Literal["v3"],
-        interrupt_before: All | Sequence[str] | None = None,
-        interrupt_after: All | Sequence[str] | None = None,
-        control: RunControl | None = None,
-        transformers: Sequence[Callable[[tuple[str, ...]], Any]] | None = None,
-    ) -> Any: ...
-
-    def stream_events(
-        self,
-        input: InputT | Command | None,
-        config: RunnableConfig | None = None,
-        *,
-        version: Literal["v1", "v2", "v3"] = "v2",
-        interrupt_before: All | Sequence[str] | None = None,
-        interrupt_after: All | Sequence[str] | None = None,
-        control: RunControl | None = None,
-        transformers: Sequence[Callable[[tuple[str, ...]], Any]] | None = None,
-        **kwargs: Any,
-    ) -> Any:
-        """Stream events from this graph.
-
-        For `version="v1"` / `"v2"`, yields `StreamEvent` dicts (see
-        `Runnable.stream_events`). For `version="v3"`, returns a
-        `GraphRunStream` whose typed projections the caller drives by
-        iterating — no background thread.
-
-        !!! warning
-
-            The `version="v3"` API is experimental and may change.
-
-        Builds a `StreamMux` from the built-in transformers, this
-        graph's compile-time `stream_transformers`, and any additional
-        `transformers=` supplied at the call site. `run.output`,
-        `run.interrupted`, and `run.interrupts` work regardless of
-        which transformers are registered.
-
-        Note:
-            Nesting v1 `stream(stream_mode="messages")` inside a node
-            of a `stream_events(version="v3")` run is not fully
-            supported. The outer v3 messages handler reroutes
-            `BaseChatModel.invoke` through the v2 event protocol, so
-            the inner v1 handler does not see `on_llm_new_token`
-            chunks. The inner stream still yields a finalized message
-            via `on_llm_end`. Use `stream_events(version="v3")` for the
-            inner graph as well, or call `chat_model.stream(...)`
-            explicitly, to get token-level streaming.
-
-        Args:
-            input: Graph input.
-            config: Optional runnable config.
-            version: Streaming-event schema version. `"v3"` selects the
-                content-block-centric streaming protocol.
-            interrupt_before: Nodes to interrupt before, if any. Only
-                used for `version="v3"`.
-            interrupt_after: Nodes to interrupt after, if any. Only
-                used for `version="v3"`.
-            control: Optional run control used to request cooperative
-                drain. Only used for `version="v3"`.
-            transformers: Extra transformer classes or configured
-                factories appended after compile-time
-                `stream_transformers`. Factories are called as
-                `factory(scope)` so they can propagate to subgraph
-                scopes. Only used for `version="v3"`.
-            **kwargs: Forwarded to the v1/v2 path.
-
-        Returns:
-            For `version="v3"`, a `GraphRunStream` the caller iterates
-            to drive the run. Otherwise an `Iterator[StreamEvent]`.
-        """
-        if version == "v3":
-            return self._pregel_stream_v3(
-                input,
-                config,
-                interrupt_before=interrupt_before,
-                interrupt_after=interrupt_after,
-                control=control,
-                transformers=transformers,
-            )
-        return super().stream_events(input, config, version=version, **kwargs)
-
-    @overload
-    def astream_events(
-        self,
-        input: InputT | Command | None,
-        config: RunnableConfig | None = None,
-        *,
-        version: Literal["v1", "v2"] = "v2",
-        **kwargs: Any,
-    ) -> AsyncIterator[StreamEvent]: ...
-
-    @overload
-    def astream_events(
-        self,
-        input: InputT | Command | None,
-        config: RunnableConfig | None = None,
-        *,
-        version: Literal["v3"],
-        interrupt_before: All | Sequence[str] | None = None,
-        interrupt_after: All | Sequence[str] | None = None,
-        control: RunControl | None = None,
-        transformers: Sequence[Callable[[tuple[str, ...]], Any]] | None = None,
-    ) -> Awaitable[Any]: ...
-
-    def astream_events(
-        self,
-        input: InputT | Command | None,
-        config: RunnableConfig | None = None,
-        *,
-        version: Literal["v1", "v2", "v3"] = "v2",
-        interrupt_before: All | Sequence[str] | None = None,
-        interrupt_after: All | Sequence[str] | None = None,
-        control: RunControl | None = None,
-        transformers: Sequence[Callable[[tuple[str, ...]], Any]] | None = None,
-        **kwargs: Any,
-    ) -> AsyncIterator[StreamEvent] | Awaitable[Any]:
-        """Async variant of `stream_events`.
-
-        For `version="v3"`, returns an `AsyncGraphRunStream` whose
-        projections can be awaited concurrently; each subscribed cursor
-        drives the pump when its buffer is empty. The same nesting
-        limitation as the sync path applies — see `stream_events` for
-        details.
-
-        !!! warning
-
-            The `version="v3"` API is experimental and may change.
-
-        See `stream_events` for full argument and return documentation.
-        """
-        if version == "v3":
-            return self._apregel_stream_v3(
-                input,
-                config,
-                interrupt_before=interrupt_before,
-                interrupt_after=interrupt_after,
-                control=control,
-                transformers=transformers,
-            )
-        return super().astream_events(input, config, version=version, **kwargs)
-
     @overload
     def invoke(
         self,
@@ -3712,7 +3569,6 @@ class Pregel(
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
         durability: Durability | None = None,
-        control: RunControl | None = None,
         version: Literal["v2"],
         **kwargs: Any,
     ) -> GraphOutput[OutputT]: ...
@@ -3730,7 +3586,6 @@ class Pregel(
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
         durability: Durability | None = None,
-        control: RunControl | None = None,
         version: Literal["v2"],
         **kwargs: Any,
     ) -> list[StreamPart[StateT, OutputT]]: ...
@@ -3748,7 +3603,6 @@ class Pregel(
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
         durability: Durability | None = None,
-        control: RunControl | None = None,
         version: Literal["v1"] = ...,
         **kwargs: Any,
     ) -> dict[str, Any] | Any: ...
@@ -3765,7 +3619,6 @@ class Pregel(
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
         durability: Durability | None = None,
-        control: RunControl | None = None,
         version: Literal["v1", "v2"] = "v1",
         **kwargs: Any,
     ) -> dict[str, Any] | Any:
@@ -3790,7 +3643,6 @@ class Pregel(
                 - `"sync"`: Changes are persisted synchronously before the next step starts.
                 - `"async"`: Changes are persisted asynchronously while the next step executes.
                 - `"exit"`: Changes are persisted only when the graph exits.
-            control: Optional run control used to request cooperative drain.
             version: The streaming format version. `"v1"` (default) returns the
                 traditional format, `"v2"` returns `StreamPart` typed dicts when
                 `stream_mode` is not `"values"`.
@@ -3818,7 +3670,6 @@ class Pregel(
                 interrupt_before=interrupt_before,
                 interrupt_after=interrupt_after,
                 durability=durability,
-                control=control,
                 version=version,
                 **kwargs,
             ):
@@ -3842,7 +3693,6 @@ class Pregel(
                 interrupt_before=interrupt_before,
                 interrupt_after=interrupt_after,
                 durability=durability,
-                control=control,
                 **kwargs,
             ):
                 if stream_mode == "values":
@@ -3889,7 +3739,6 @@ class Pregel(
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
         durability: Durability | None = None,
-        control: RunControl | None = None,
         version: Literal["v2"],
         **kwargs: Any,
     ) -> GraphOutput[OutputT]: ...
@@ -3907,7 +3756,6 @@ class Pregel(
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
         durability: Durability | None = None,
-        control: RunControl | None = None,
         version: Literal["v2"],
         **kwargs: Any,
     ) -> list[StreamPart[StateT, OutputT]]: ...
@@ -3925,7 +3773,6 @@ class Pregel(
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
         durability: Durability | None = None,
-        control: RunControl | None = None,
         version: Literal["v1"] = ...,
         **kwargs: Any,
     ) -> dict[str, Any] | Any: ...
@@ -3942,7 +3789,6 @@ class Pregel(
         interrupt_before: All | Sequence[str] | None = None,
         interrupt_after: All | Sequence[str] | None = None,
         durability: Durability | None = None,
-        control: RunControl | None = None,
         version: Literal["v1", "v2"] = "v1",
         **kwargs: Any,
     ) -> dict[str, Any] | Any:
@@ -3967,7 +3813,6 @@ class Pregel(
                 - `"sync"`: Changes are persisted synchronously before the next step starts.
                 - `"async"`: Changes are persisted asynchronously while the next step executes.
                 - `"exit"`: Changes are persisted only when the graph exits.
-            control: Optional run control used to request cooperative drain.
             version: The streaming format version. `"v1"` (default) returns the
                 traditional format, `"v2"` returns `StreamPart` typed dicts when
                 `stream_mode` is not `"values"`.
@@ -3995,7 +3840,6 @@ class Pregel(
                 interrupt_before=interrupt_before,
                 interrupt_after=interrupt_after,
                 durability=durability,
-                control=control,
                 version=version,
                 **kwargs,
             ):
@@ -4019,7 +3863,6 @@ class Pregel(
                 interrupt_before=interrupt_before,
                 interrupt_after=interrupt_after,
                 durability=durability,
-                control=control,
                 **kwargs,
             ):
                 if stream_mode == "values":
@@ -4193,7 +4036,7 @@ def _resolve_parent_ns(
 ) -> tuple[str, ...]:
     """Return the checkpoint namespace the caller is running under.
 
-    `stream_events(version="v3")` uses this to scope its native projections
+    `stream_v2` uses this to scope its native projections
     (`ValuesTransformer`, `MessagesTransformer`) to events emitted at
     the run's own level. A root call resolves to `()`; a call made
     from inside a node carries the outer graph's task namespace so the

libs/langgraph/langgraph/runtime.py

@@ -16,7 +16,6 @@ from langgraph.typing import ContextT
 __all__ = (
     "BaseUser",
     "ExecutionInfo",
-    "RunControl",
     "Runtime",
     "ServerInfo",
     "get_runtime",
@@ -76,34 +75,6 @@ class ServerInfo:
     """
 
 
-class RunControl:
-    """Run-scoped control surface for cooperative draining.
-
-    Intended for a single graph run. Create a fresh `RunControl` per run;
-    reusing a control after `request_drain()` leaves it drained.
-
-    Safe to call from any thread: the drain request is represented by a
-    single attribute write, so no lock is needed for this signal.
-    If more mutable state is added here, add synchronization.
-    """
-
-    __slots__ = ("_drain_reason",)
-
-    def __init__(self) -> None:
-        self._drain_reason: str | None = None
-
-    def request_drain(self, reason: str = "shutdown") -> None:
-        self._drain_reason = reason
-
-    @property
-    def drain_requested(self) -> bool:
-        return self._drain_reason is not None
-
-    @property
-    def drain_reason(self) -> str | None:
-        return self._drain_reason
-
-
 def _no_op_stream_writer(_: Any) -> None: ...
 
 
@@ -118,7 +89,6 @@ class _RuntimeOverrides(TypedDict, Generic[ContextT], total=False):
     previous: Any
     execution_info: ExecutionInfo
     server_info: ServerInfo | None
-    control: RunControl | None
 
 
 @dataclass(**_DC_KWARGS)
@@ -197,7 +167,7 @@ class Runtime(Generic[ContextT]):
 
     context: ContextT = field(default=None)  # type: ignore[assignment]
     """Static context for the graph run, like `user_id`, `db_conn`, etc.
-
+    
     Can also be thought of as 'run dependencies'."""
 
     store: BaseStore | None = field(default=None)
@@ -218,7 +188,7 @@ class Runtime(Generic[ContextT]):
 
     previous: Any = field(default=None)
     """The previous return value for the given thread.
-
+    
     Only available with the functional API when a checkpointer is provided.
     """
 
@@ -230,13 +200,6 @@ class Runtime(Generic[ContextT]):
     server_info: ServerInfo | None = field(default=None)
     """Metadata injected by LangGraph Server. None when running open-source LangGraph without LangSmith deployments."""
 
-    control: RunControl | None = field(default=None)
-    """Run-scoped control plane for cooperative draining.
-
-    Populated automatically during graph runs. None outside an active
-    graph runtime.
-    """
-
     def merge(self, other: Runtime[ContextT]) -> Runtime[ContextT]:
         """Merge two runtimes together.
 
@@ -254,7 +217,6 @@ class Runtime(Generic[ContextT]):
             previous=self.previous if other.previous is None else other.previous,
             execution_info=other.execution_info or self.execution_info,
             server_info=other.server_info or self.server_info,
-            control=other.control or self.control,
         )
 
     def override(
@@ -273,14 +235,6 @@ class Runtime(Generic[ContextT]):
             execution_info=self.execution_info.patch(**overrides),
         )
 
-    @property
-    def drain_requested(self) -> bool:
-        return self.control.drain_requested if self.control is not None else False
-
-    @property
-    def drain_reason(self) -> str | None:
-        return self.control.drain_reason if self.control is not None else None
-
 
 DEFAULT_RUNTIME = Runtime(
     context=None,
@@ -289,7 +243,6 @@ DEFAULT_RUNTIME = Runtime(
     heartbeat=_no_op_heartbeat,
     previous=None,
     execution_info=None,
-    control=None,
 )
 
 

libs/langgraph/langgraph/stream/__init__.py

@@ -1,7 +1,7 @@
 """Streaming infrastructure for LangGraph.
 
-Compile a graph with `transformers=[...]` and call `graph.stream_events(version="v3")` /
-`graph.astream_events(version="v3")` to drive a transformer pipeline that projects the
+Compile a graph with `transformers=[...]` and call `graph.stream_v2()` /
+`graph.astream_v2()` to drive a transformer pipeline that projects the
 graph's raw events into ergonomic per-channel streams.
 """
 

libs/langgraph/langgraph/stream/_mux.py

@@ -91,11 +91,9 @@ class StreamMux:
         self._assign_seq = _assign_seq
         self._events: StreamChannel[ProtocolEvent] = StreamChannel()
         self._events._bind(is_async=is_async)
-        self._events._bind_mux(self)
         self._transformers: list[StreamTransformer] = []
         self._channels: list[StreamChannel[Any]] = []
         self._seq = 0
-        self._push_seq = 0
 
         self.extensions: dict[str, Any] = {}
         self.native_keys: set[str] = set()
@@ -126,10 +124,6 @@ class StreamMux:
         """Return the transformer that contributed `key` to the projection."""
         return self._transformer_by_key.get(key)
 
-    def _next_push_seq(self) -> int:
-        self._push_seq += 1
-        return self._push_seq
-
     # ------------------------------------------------------------------
     # Pump wiring + mini-mux nesting
     # ------------------------------------------------------------------
@@ -455,7 +449,6 @@ class StreamMux:
         for value in projection.values():
             if isinstance(value, StreamChannel):
                 value._bind(is_async=self.is_async)
-                value._bind_mux(self)
                 self._channels.append(value)
                 if value.name is not None:
                     method = value.name if native else f"custom:{value.name}"

libs/langgraph/langgraph/stream/_types.py

@@ -88,7 +88,7 @@ class StreamTransformer(ABC):
         required_stream_modes: Stream modes the graph must emit for
             this transformer to have anything to process. Computed as
             the union across all registered transformers to determine
-            which modes a `stream_events(version="v3")` run requests from the graph.
+            which modes a `stream_v2` run requests from the graph.
             Empty tuple means the transformer consumes only synthetic
             events (or is purely passive).
     """

libs/langgraph/langgraph/stream/run_stream.py

@@ -5,8 +5,6 @@ from collections.abc import AsyncIterator, Awaitable, Callable, Iterator, Mappin
 from types import MappingProxyType, TracebackType
 from typing import TYPE_CHECKING, Any
 
-from langchain_core._api import beta
-
 from langgraph.stream._convert import convert_to_protocol_event
 from langgraph.stream._mux import StreamMux
 from langgraph.stream._types import ProtocolEvent
@@ -27,7 +25,6 @@ async def _adrive_until_done(pump: Callable[[], Awaitable[bool]]) -> None:
         pass
 
 
-@beta(message="The v3 streaming protocol on Pregel is experimental.")
 class GraphRunStream:
     """Sync run stream with caller-driven pumping.
 
@@ -41,11 +38,6 @@ class GraphRunStream:
     All transformer projections live in `extensions`. Native transformer
     projections (those with `_native = True`) are also set as direct
     attributes on this instance (e.g. `run.values`, `run.messages`).
-
-    !!! warning
-
-        Returned by `Pregel.stream_events(version="v3")`, which is
-        experimental and may change.
     """
 
     def __init__(
@@ -193,25 +185,28 @@ class GraphRunStream:
         return iter(self._mux._events)
 
     def interleave(self, *names: str) -> Iterator[tuple[str, Any]]:
-        """Iterate multiple projections in arrival order, yielding ``(name, item)``.
+        """Iterate multiple projections round-robin, yielding ``(name, item)``.
 
-        Items are ordered by a monotonic push stamp assigned when each
-        transformer pushes into its `StreamChannel`. This gives strict
-        arrival ordering across projections, unlike round-robin.
+        Each turn advances one projection's cursor; when a cursor's buffer
+        is empty, pulling from it drives the pump once, which fans out to
+        every subscribed projection log. Projections whose items aren't
+        consumed on this turn sit in their own buffers only until the next
+        turn reaches them, bounding memory by the skew between projection
+        rates rather than letting any single log grow to the full run
+        length.
+
+        Projections are exhausted independently; a projection that finishes
+        early drops out of the rotation while others continue. The overall
+        iterator ends once all named projections are done.
 
         Args:
             *names: Projection keys to interleave. Must match keys in
                 ``extensions``.
 
         Yields:
-            ``(name, item)`` tuples in arrival order across the named
+            ``(name, item)`` tuples in round-robin order across the named
             projections.
 
-        Each named channel is locked for the duration of iteration and
-        released when the generator completes, is closed, or raises.
-        Channels cannot be subscribed concurrently — use `.tee(n)` if
-        you need fan-out.
-
         Raises:
             KeyError: If a name doesn't match a registered projection.
 
@@ -224,73 +219,22 @@ class GraphRunStream:
                     print("val:", item)
             ```
         """
-        from langgraph.stream.stream_channel import StreamChannel
-
-        channels: dict[str, StreamChannel[Any]] = {}
-        try:
-            for name in names:
-                ch = self.extensions[name]
-                if not isinstance(ch, StreamChannel):
-                    raise TypeError(
-                        f"interleave() requires StreamChannel projections, "
-                        f"got {type(ch).__name__} for {name!r}"
-                    )
-                if ch._is_async is None:
-                    raise TypeError(
-                        f"StreamChannel {name!r} has not been bound yet. "
-                        "Register the transformer with a StreamMux first."
-                    )
-                if ch._is_async:
-                    raise TypeError(
-                        f"StreamChannel {name!r} is bound to async mode — "
-                        "sync interleave() cannot consume async channels."
-                    )
-                if ch._subscribed:
-                    raise RuntimeError(
-                        f"StreamChannel {name!r} already has a subscriber; "
-                        "use .tee(n) for fan-out."
-                    )
-                ch._subscribed = True
-                channels[name] = ch
-
-            done: set[str] = set()
-
-            while len(done) < len(channels):
-                best: tuple[int, str] | None = None
-                for name, ch in channels.items():
-                    if name in done:
-                        continue
-                    if ch._closed and not ch._items:
-                        if ch._error is not None:
-                            raise ch._error
-                        done.add(name)
-                        continue
-                    if ch._items:
-                        stamp = ch._items[0][0]
-                        if best is None or stamp < best[0]:
-                            best = (stamp, name)
-
-                if best is not None:
-                    _stamp, item = channels[best[1]]._items.popleft()
-                    yield (best[1], item)
-                else:
-                    pump = self._mux._pump_fn
-                    if pump is None or not pump():
-                        before = len(done)
-                        for name, ch in channels.items():
-                            if name not in done and not ch._items:
-                                if ch._closed:
-                                    if ch._error is not None:
-                                        raise ch._error
-                                    done.add(name)
-                        if len(done) == before:
-                            break
-        finally:
-            for ch in channels.values():
-                ch._subscribed = False
+        cursors: dict[str, Iterator[Any]] = {
+            name: iter(self.extensions[name]) for name in names
+        }
+        done: set[str] = set()
+        while len(done) < len(cursors):
+            for name, cursor in cursors.items():
+                if name in done:
+                    continue
+                try:
+                    item = next(cursor)
+                except StopIteration:
+                    done.add(name)
+                    continue
+                yield (name, item)
 
 
-@beta(message="The v3 streaming protocol on Pregel is experimental.")
 class AsyncGraphRunStream:
     """Async run stream with caller-driven pumping.
 
@@ -312,11 +256,6 @@ class AsyncGraphRunStream:
         async for msg in run.messages:
             ...
     ```
-
-    !!! warning
-
-        Awaited from `Pregel.astream_events(version="v3")`, which is
-        experimental and may change.
     """
 
     def __init__(

libs/langgraph/langgraph/stream/stream_channel.py

@@ -3,10 +3,7 @@ from __future__ import annotations
 import asyncio
 from collections import deque
 from collections.abc import AsyncIterator, Awaitable, Callable, Iterator
-from typing import TYPE_CHECKING, Generic, TypeVar
-
-if TYPE_CHECKING:
-    from langgraph.stream._mux import StreamMux
+from typing import Generic, TypeVar
 
 T = TypeVar("T")
 
@@ -67,7 +64,7 @@ class StreamChannel(Generic[T]):
         if maxlen is not None and maxlen <= 0:
             raise ValueError("StreamChannel maxlen must be a positive int or None")
         self.name = name
-        self._items: deque[tuple[int, T]] = deque()
+        self._items: deque[T] = deque()
         self._maxlen: int | None = maxlen
         self._closed = False
         self._error: BaseException | None = None
@@ -80,15 +77,11 @@ class StreamChannel(Generic[T]):
         self._arequest_more: Callable[[], Awaitable[bool]] | None = None
 
         self._wire_fn: Callable[[T], None] | None = None
-        self._mux: StreamMux | None = None
 
     # ------------------------------------------------------------------
     # Binding
     # ------------------------------------------------------------------
 
-    def _bind_mux(self, mux: StreamMux) -> None:
-        self._mux = mux
-
     def _bind(self, *, is_async: bool) -> None:
         """Bind this channel to sync or async mode.
 
@@ -124,18 +117,13 @@ class StreamChannel(Generic[T]):
         registered, but auto-forwarding always fires so wired events
         reach the main event log regardless of subscription state.
 
-        Items are stored as `(stamp, item)` tuples where stamp is a
-        monotonic counter from the owning mux. Stamps are stripped by
-        the default cursors; raw stamped tuples are visible on `_items`.
-
         Raises:
             RuntimeError: If the channel is closed (and subscribed).
         """
         if self._subscribed:
             if self._closed:
                 raise RuntimeError("Cannot push to a closed StreamChannel")
-            stamp = self._mux._next_push_seq() if self._mux is not None else 0
-            self._items.append((stamp, item))
+            self._items.append(item)
         if self._wire_fn is not None:
             self._wire_fn(item)
 
@@ -182,8 +170,7 @@ class StreamChannel(Generic[T]):
     def _sync_cursor(self) -> Iterator[T]:
         while True:
             if self._items:
-                _stamp, item = self._items.popleft()
-                yield item
+                yield self._items.popleft()
             elif self._closed:
                 if self._error is not None:
                     raise self._error
@@ -225,8 +212,7 @@ class StreamChannel(Generic[T]):
     async def _async_cursor(self) -> AsyncIterator[T]:
         while True:
             if self._items:
-                _stamp, item = self._items.popleft()
-                yield item
+                yield self._items.popleft()
             elif self._closed:
                 if self._error is not None:
                     raise self._error

libs/langgraph/langgraph/stream/transformers.py

@@ -12,7 +12,7 @@ from langchain_core.messages import AIMessageChunk, BaseMessage
 from langchain_protocol.protocol import MessagesData
 from typing_extensions import NotRequired, TypedDict
 
-from langgraph.errors import GraphDrained, GraphInterrupt
+from langgraph.errors import GraphInterrupt
 from langgraph.stream._types import ProtocolEvent, StreamTransformer
 from langgraph.stream.run_stream import AsyncSubgraphRunStream, SubgraphRunStream
 from langgraph.stream.stream_channel import StreamChannel
@@ -38,8 +38,8 @@ class ValuesTransformer(StreamTransformer):
     Only values events at the run's own level are captured; snapshots
     from deeper subgraphs are left in the main event log but excluded
     from the projection. "Own level" is defined by `scope`, which
-    `stream_events(version="v3")` / `astream_events(version="v3")` populate from the caller's
-    checkpoint namespace so that a nested `stream_events(version="v3")` call still
+    `stream_v2` / `astream_v2` populate from the caller's
+    checkpoint namespace so that a nested `stream_v2` call still
     sees its own root snapshots.
     """
 
@@ -165,7 +165,7 @@ class MessagesTransformer(StreamTransformer):
     metadata)` from `StreamMessagesHandler`):
 
     1. Protocol event (dict with `"event"` key) — emitted by
-       `stream_events(version="v3")` / `astream_events(version="v3")` via the `on_stream_event`
+       `stream_v2()` / `astream_v2()` via the `on_stream_event`
        callback. Routed to an existing `ChatModelStream` by
        `metadata["run_id"]`. A `message-start` event creates a new
        stream; `message-finish` closes it.
@@ -177,15 +177,15 @@ class MessagesTransformer(StreamTransformer):
     V1 `AIMessageChunk` tuples (from `on_llm_new_token`) are not
     streamed into this projection: chat models that want to populate
     `run.messages` with content-block streaming must use
-    `stream_events(version="v3")` / `astream_events(version="v3")`. Models called via the legacy
+    `stream_v2()` / `astream_v2()`. Models called via the legacy
     `stream()` method still surface their final `AIMessage` via
     `on_chain_end` when a node returns it as state.
 
     Only events at the run's own level are projected; tokens from
     deeper subgraphs are left in the main event log but excluded from
     `.messages`. "Own level" is defined by `scope`, which
-    `stream_events(version="v3")` / `astream_events(version="v3")` populate from the caller's checkpoint
-    namespace so that a `stream_events(version="v3")` call inside a node still sees its
+    `stream_v2` / `astream_v2` populate from the caller's checkpoint
+    namespace so that a `stream_v2` call inside a node still sees its
     own root chat model streams on `.messages`. Consumers that need
     subgraph tokens should iterate the raw event stream or register a
     custom transformer.
@@ -281,7 +281,7 @@ class MessagesTransformer(StreamTransformer):
         ):
             self._route_whole_message(payload, node=node)
         # Legacy AIMessageChunk tuples (from on_llm_new_token) are ignored;
-        # v1 streaming callers must switch to stream_events(version="v3") to populate this
+        # v1 streaming callers must switch to stream_v2() to populate this
         # projection.
 
         return True
@@ -327,7 +327,7 @@ class MessagesTransformer(StreamTransformer):
         self._by_run.clear()
 
 
-SubgraphStatus = Literal["started", "completed", "failed", "interrupted", "drained"]
+SubgraphStatus = Literal["started", "completed", "failed", "interrupted"]
 
 
 def _parse_ns_segment(segment: str) -> tuple[str, str | None]:
@@ -472,8 +472,10 @@ class _TasksLifecycleBase(StreamTransformer):
         self._open.clear()
 
     def fail(self, err: BaseException) -> None:
-        """Emit terminal status for any tracked namespace still open."""
-        status, error_str = _status_from_exception(err)
+        """Emit `failed` / `interrupted` for any tracked namespace still open."""
+        is_interrupt = isinstance(err, GraphInterrupt)
+        status: SubgraphStatus = "interrupted" if is_interrupt else "failed"
+        error_str = None if is_interrupt else str(err)
         for ns in list(self._open):
             self._on_terminal(ns, status, error_str)
         self._open.clear()
@@ -481,8 +483,6 @@ class _TasksLifecycleBase(StreamTransformer):
 
 def _status_from_exception(err: BaseException) -> tuple[SubgraphStatus, str | None]:
     """Map a run exception to a subgraph terminal status and error string."""
-    if isinstance(err, GraphDrained):
-        return "drained", None
     if isinstance(err, GraphInterrupt):
         return "interrupted", None
     return "failed", str(err)

libs/langgraph/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "langgraph"
-version = "1.2.0a3"
+version = "1.1.10"
 description = "Building stateful, multi-actor applications with LLMs"
 authors = []
 requires-python = ">=3.10"
@@ -24,8 +24,8 @@ classifiers = [
     'Programming Language :: Python :: 3.13',
 ]
 dependencies = [
-    "langchain-core>=1.4.0a2,<2",
-    "langgraph-checkpoint>=4.1.0a3,<5.0.0",
+    "langchain-core>=1.3.2,<2",
+    "langgraph-checkpoint>=4.0.3,<5.0.0",
     "langgraph-sdk>=0.3.0,<0.4.0",
     "langgraph-prebuilt>=1.0.12,<1.1.0",
     "xxhash>=3.5.0",

libs/langgraph/tests/test_channels.py

@@ -371,7 +371,8 @@ def test_delta_channel_inmemory_saver_assembles_writes() -> None:
 
     saved = saver.get_tuple(config)
     assert saved is not None
-    assert "messages" not in saved.checkpoint["channel_values"]
+    assert "messages" in saved.checkpoint["channel_values"]
+    assert saved.checkpoint["channel_values"]["messages"] is DELTA_SENTINEL
 
     state = graph.get_state(config)
     assert len(state.values["messages"]) == 4  # 2 human + 2 AI
@@ -561,7 +562,7 @@ def test_delta_channel_dict_reducer_end_to_end_filesystem() -> None:
 
     saved = saver.get_tuple(config)
     assert saved is not None
-    assert "files" not in saved.checkpoint["channel_values"]
+    assert saved.checkpoint["channel_values"]["files"] is DELTA_SENTINEL
     state = graph.get_state(config)
     assert state.values["files"] == {
         "/doc_1.txt": "content for turn 1",

libs/langgraph/tests/test_interleave_arrival_order.py

@@ -1,359 +0,0 @@
-"""Tests for arrival-ordered interleave and push stamps."""
-
-from __future__ import annotations
-
-import operator
-from typing import Annotated, Any
-
-import pytest
-from typing_extensions import TypedDict
-
-from langgraph.constants import END, START
-from langgraph.graph import StateGraph
-from langgraph.stream import StreamChannel, StreamTransformer
-from langgraph.stream._mux import StreamMux
-from langgraph.stream._types import ProtocolEvent
-from langgraph.stream.run_stream import GraphRunStream
-from langgraph.stream.transformers import ValuesTransformer
-
-# ---------------------------------------------------------------------------
-# Helpers
-# ---------------------------------------------------------------------------
-
-
-class _TwoChannelTransformer(StreamTransformer):
-    """Transformer that exposes two named channels for testing interleave."""
-
-    _native = True
-
-    def __init__(self, scope: tuple[str, ...] = ()) -> None:
-        super().__init__(scope)
-        self._alpha: StreamChannel[str] = StreamChannel("alpha")
-        self._beta: StreamChannel[str] = StreamChannel("beta")
-
-    def init(self) -> dict[str, Any]:
-        return {"alpha": self._alpha, "beta": self._beta}
-
-    def process(self, event: ProtocolEvent) -> bool:
-        return True
-
-
-class SimpleState(TypedDict):
-    value: str
-    items: Annotated[list[str], operator.add]
-
-
-def _build_simple_graph():
-    def node_a(state: SimpleState) -> dict:
-        return {"value": state["value"] + "A", "items": ["a"]}
-
-    def node_b(state: SimpleState) -> dict:
-        return {"value": state["value"] + "B", "items": ["b"]}
-
-    builder = StateGraph(SimpleState)
-    builder.add_node("node_a", node_a)
-    builder.add_node("node_b", node_b)
-    builder.add_edge(START, "node_a")
-    builder.add_edge("node_a", "node_b")
-    builder.add_edge("node_b", END)
-    return builder.compile()
-
-
-# ---------------------------------------------------------------------------
-# Unit tests: push stamps on StreamChannel
-# ---------------------------------------------------------------------------
-
-
-class TestPushStamps:
-    def test_stamps_are_monotonic_across_channels(self) -> None:
-        mux = StreamMux(
-            factories=[ValuesTransformer, _TwoChannelTransformer],
-            is_async=False,
-        )
-        alpha = mux.extensions["alpha"]
-        beta = mux.extensions["beta"]
-
-        alpha._subscribed = True
-        beta._subscribed = True
-
-        alpha.push("a1")
-        beta.push("b1")
-        alpha.push("a2")
-        beta.push("b2")
-
-        all_stamped = list(alpha._items) + list(beta._items)
-        stamps = [s for s, _ in all_stamped]
-        assert len(set(stamps)) == 4
-        items_by_arrival = [item for _, item in sorted(all_stamped)]
-        assert items_by_arrival == ["a1", "b1", "a2", "b2"]
-
-    def test_regular_iter_strips_stamps(self) -> None:
-        mux = StreamMux(
-            factories=[ValuesTransformer, _TwoChannelTransformer],
-            is_async=False,
-        )
-        alpha = mux.extensions["alpha"]
-        it = iter(alpha)
-        alpha.push("a1")
-        alpha.push("a2")
-        alpha.close()
-        items = list(it)
-        assert items == ["a1", "a2"]
-        assert all(isinstance(item, str) for item in items)
-
-    def test_events_channel_gets_real_stamps(self) -> None:
-        mux = StreamMux(
-            factories=[ValuesTransformer, _TwoChannelTransformer],
-            is_async=False,
-        )
-        alpha = mux.extensions["alpha"]
-
-        alpha._subscribed = True
-        alpha.push("a1")
-
-        mux._events._subscribed = True
-        mux._events.push({"method": "test", "data": "x"})
-
-        alpha.push("a2")
-
-        all_stamps = [s for s, _ in alpha._items] + [s for s, _ in mux._events._items]
-        assert len(set(all_stamps)) == len(all_stamps), "all stamps should be unique"
-        assert all(s > 0 for s in all_stamps), "no stamp should be zero"
-
-    def test_channel_without_mux_gets_zero_stamp(self) -> None:
-        ch: StreamChannel[str] = StreamChannel()
-        ch._bind(is_async=False)
-        ch._subscribed = True
-        ch.push("x")
-        assert list(ch._items) == [(0, "x")]
-
-
-# ---------------------------------------------------------------------------
-# Unit tests: interleave arrival order
-# ---------------------------------------------------------------------------
-
-
-class TestInterleaveArrivalOrder:
-    def test_arrival_order_not_round_robin(self) -> None:
-        mux = StreamMux(
-            factories=[ValuesTransformer, _TwoChannelTransformer],
-            is_async=False,
-        )
-        alpha = mux.extensions["alpha"]
-        beta = mux.extensions["beta"]
-        run = GraphRunStream(None, mux, wire_pump=False)
-
-        # interleave() subscribes channels directly and reads _items
-        # for stamp-ordered iteration. We simulate the pump by wiring
-        # a custom callback that pushes items in a known order.
-        push_script = [
-            ("alpha", "a1"),
-            ("alpha", "a2"),
-            ("beta", "b1"),
-            ("alpha", "a3"),
-            ("beta", "b2"),
-        ]
-        push_iter = iter(push_script)
-        channels = {"alpha": alpha, "beta": beta}
-
-        def fake_pump() -> bool:
-            try:
-                name, item = next(push_iter)
-                channels[name].push(item)
-                return True
-            except StopIteration:
-                mux.close()
-                return False
-
-        mux.bind_pump(fake_pump)
-
-        result = list(run.interleave("alpha", "beta"))
-        names = [name for name, _ in result]
-        items = [item for _, item in result]
-
-        assert items == ["a1", "a2", "b1", "a3", "b2"]
-        assert names == ["alpha", "alpha", "beta", "alpha", "beta"]
-
-    def test_single_projection(self) -> None:
-        mux = StreamMux(
-            factories=[ValuesTransformer, _TwoChannelTransformer],
-            is_async=False,
-        )
-        alpha = mux.extensions["alpha"]
-        run = GraphRunStream(None, mux, wire_pump=False)
-
-        push_script = [("alpha", "a1"), ("alpha", "a2")]
-        push_iter = iter(push_script)
-
-        def fake_pump() -> bool:
-            try:
-                _, item = next(push_iter)
-                alpha.push(item)
-                return True
-            except StopIteration:
-                mux.close()
-                return False
-
-        mux.bind_pump(fake_pump)
-
-        result = list(run.interleave("alpha"))
-        assert result == [("alpha", "a1"), ("alpha", "a2")]
-
-    def test_empty_projection(self) -> None:
-        mux = StreamMux(
-            factories=[ValuesTransformer, _TwoChannelTransformer],
-            is_async=False,
-        )
-        alpha = mux.extensions["alpha"]
-        run = GraphRunStream(None, mux, wire_pump=False)
-
-        push_script = [("alpha", "a1"), ("alpha", "a2")]
-        push_iter = iter(push_script)
-        channels = {"alpha": alpha}
-
-        def fake_pump() -> bool:
-            try:
-                name, item = next(push_iter)
-                channels[name].push(item)
-                return True
-            except StopIteration:
-                mux.close()
-                return False
-
-        mux.bind_pump(fake_pump)
-
-        result = list(run.interleave("alpha", "beta"))
-        assert result == [("alpha", "a1"), ("alpha", "a2")]
-
-    def test_unknown_projection_raises(self) -> None:
-        mux = StreamMux(
-            factories=[ValuesTransformer, _TwoChannelTransformer],
-            is_async=False,
-        )
-        run = GraphRunStream(None, mux, wire_pump=False)
-        mux.close()
-        with pytest.raises((KeyError, AttributeError)):
-            list(run.interleave("alpha", "does_not_exist"))
-
-    def test_all_empty(self) -> None:
-        mux = StreamMux(
-            factories=[ValuesTransformer, _TwoChannelTransformer],
-            is_async=False,
-        )
-        run = GraphRunStream(None, mux, wire_pump=False)
-
-        def fake_pump() -> bool:
-            mux.close()
-            return False
-
-        mux.bind_pump(fake_pump)
-
-        result = list(run.interleave("alpha", "beta"))
-        assert result == []
-
-    def test_error_propagation(self) -> None:
-        mux = StreamMux(
-            factories=[ValuesTransformer, _TwoChannelTransformer],
-            is_async=False,
-        )
-        alpha = mux.extensions["alpha"]
-        beta = mux.extensions["beta"]
-        run = GraphRunStream(None, mux, wire_pump=False)
-
-        err = RuntimeError("boom")
-
-        push_script = [
-            ("alpha", "a1"),
-            ("beta", "b1"),
-        ]
-        push_iter = iter(push_script)
-        channels = {"alpha": alpha, "beta": beta}
-
-        def fake_pump() -> bool:
-            try:
-                name, item = next(push_iter)
-                channels[name].push(item)
-                return True
-            except StopIteration:
-                alpha.fail(err)
-                beta.close()
-                return False
-
-        mux.bind_pump(fake_pump)
-
-        collected = []
-        with pytest.raises(RuntimeError, match="boom"):
-            for pair in run.interleave("alpha", "beta"):
-                collected.append(pair)
-
-        assert ("alpha", "a1") in collected
-        assert ("beta", "b1") in collected
-
-
-# ---------------------------------------------------------------------------
-# Integration test: interleave with stream_events(version="v3")
-# ---------------------------------------------------------------------------
-
-
-class TestInterleaveIntegration:
-    def test_interleave_values_and_messages(self) -> None:
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
-        tagged = list(run.interleave("values", "messages"))
-        names = [name for name, _ in tagged]
-        assert set(names).issubset({"values", "messages"})
-        assert names.count("values") >= 1
-
-    def test_interleave_rejects_already_subscribed(self) -> None:
-        mux = StreamMux(
-            factories=[ValuesTransformer, _TwoChannelTransformer],
-            is_async=False,
-        )
-        alpha = mux.extensions["alpha"]
-        run = GraphRunStream(None, mux, wire_pump=False)
-
-        # Subscribe alpha via iter first
-        _ = iter(alpha)
-        mux.close()
-
-        with pytest.raises(RuntimeError, match="already has a subscriber"):
-            list(run.interleave("alpha"))
-
-    def test_interleave_releases_projections_on_completion(self) -> None:
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
-        list(run.interleave("values", "messages"))
-        # Subscriptions should be released after the generator completes,
-        # so the channels can be re-iterated (they'll be empty / closed).
-        assert run.extensions["values"]._subscribed is False
-        assert run.extensions["messages"]._subscribed is False
-
-    def test_interleave_releases_projections_on_early_break(self) -> None:
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
-        gen = run.interleave("values", "messages")
-        next(gen)
-        gen.close()
-        assert run.extensions["values"]._subscribed is False
-        assert run.extensions["messages"]._subscribed is False
-
-    def test_interleave_releases_projections_on_validation_failure(self) -> None:
-        mux = StreamMux(
-            factories=[ValuesTransformer, _TwoChannelTransformer],
-            is_async=False,
-        )
-        alpha = mux.extensions["alpha"]
-        # Pre-subscribe alpha so that interleave will fail validation when
-        # it gets to the second name. The first (already-validated) channel
-        # should still be released.
-        run = GraphRunStream(None, mux, wire_pump=False)
-        mux.close()
-        alpha._subscribed = True
-
-        with pytest.raises(RuntimeError, match="already has a subscriber"):
-            list(run.interleave("values", "alpha"))
-
-        assert mux.extensions["values"]._subscribed is False

libs/langgraph/tests/test_pregel.py

@@ -122,29 +122,6 @@ def test_graph_validation() -> None:
         graph.invoke({"hello": "there"})
 
 
-def test_request_drain_allows_inflight_call_scheduling(
-    sync_checkpointer: BaseCheckpointSaver,
-) -> None:
-    from langgraph.runtime import RunControl
-
-    @task
-    def child(x: int) -> int:
-        return x + 1
-
-    control = RunControl()
-
-    @entrypoint(checkpointer=sync_checkpointer)
-    def graph(x: int) -> int:
-        control.request_drain()
-        fut = child(x)
-        return fut.result()
-
-    config = {"configurable": {"thread_id": "drain-call-sync"}}
-
-    assert graph.invoke(1, config=config, control=control) == 2
-    assert control.drain_requested
-
-
 def test_invalid_checkpointer_type() -> None:
     class State(TypedDict):
         foo: str

libs/langgraph/tests/test_pregel_async.py

@@ -16,7 +16,6 @@ from typing import (
     Literal,
     Optional,
 )
-from unittest.mock import patch
 from uuid import UUID
 
 import pytest
@@ -49,7 +48,6 @@ from langgraph.channels.topic import Topic
 from langgraph.errors import (
     GraphRecursionError,
     InvalidUpdateError,
-    NodeError,
     ParentCommand,
 )
 from langgraph.func import entrypoint, task
@@ -217,30 +215,6 @@ async def test_checkpoint_errors() -> None:
             pass
 
 
-@NEEDS_CONTEXTVARS
-async def test_request_drain_allows_inflight_acall_scheduling(
-    async_checkpointer: BaseCheckpointSaver,
-) -> None:
-    from langgraph.runtime import RunControl
-
-    @task
-    async def child(x: int) -> int:
-        return x + 1
-
-    control = RunControl()
-
-    @entrypoint(checkpointer=async_checkpointer)
-    async def graph(x: int) -> int:
-        control.request_drain()
-        fut = child(x)
-        return await fut
-
-    config = {"configurable": {"thread_id": "drain-call-async"}}
-
-    assert await graph.ainvoke(1, config=config, control=control) == 2
-    assert control.drain_requested
-
-
 async def test_py_async_with_cancel_behavior() -> None:
     """This test confirms that in all versions of Python we support, __aexit__
     is not cancelled when the coroutine containing the async with block is cancelled."""
@@ -9765,126 +9739,3 @@ async def test_fork_does_not_apply_pending_writes(
 
     # 1 (input) + 20 (forked node_a) + 100 (node_b) = 121
     assert result == {"value": 121}
-
-
-async def test_graph_error_handler_async_runtime_info() -> None:
-    class State(TypedDict):
-        foo: str
-
-    attempts = 0
-    captured: dict[str, object] = {}
-
-    async def always_failing_node(state: State) -> State:
-        nonlocal attempts
-        attempts += 1
-        raise ValueError("Always fails async")
-
-    async def err_handler_node(state: State, error: NodeError) -> State:
-        captured["from_node_name"] = error.node
-        captured["from_node_error"] = error.error
-        return {"foo": "handled_async"}
-
-    graph = (
-        StateGraph(State)
-        .add_node(
-            "always_failing",
-            always_failing_node,
-            retry_policy=RetryPolicy(
-                max_attempts=2,
-                initial_interval=0.01,
-                jitter=False,
-                retry_on=ValueError,
-            ),
-            error_handler=err_handler_node,
-        )
-        .add_edge(START, "always_failing")
-        .compile()
-    )
-
-    with patch("asyncio.sleep"):
-        result = await graph.ainvoke({"foo": ""})
-
-    assert attempts == 2
-    assert result["foo"] == "handled_async"
-    assert captured["from_node_name"] == "always_failing"
-    assert isinstance(captured["from_node_error"], BaseException)
-
-
-@NEEDS_CONTEXTVARS
-async def test_graph_error_handler_does_not_swallow_interrupt_concurrent() -> None:
-    """When a graph error handler is configured and a node calls interrupt()
-    concurrently with other nodes, the interrupt must still be raised — not
-    silently swallowed."""
-
-    class State(TypedDict):
-        foo: str
-
-    async def node_a(state: State) -> State:
-        val = interrupt("need human input")
-        return {"foo": f"a_{val}"}
-
-    async def node_b(state: State) -> State:
-        return {}
-
-    async def err_handler(state: State) -> State:
-        return {"foo": "handled"}
-
-    checkpointer = InMemorySaver()
-    graph = (
-        StateGraph(State)
-        .add_node("node_a", node_a, error_handler=err_handler)
-        .add_node("node_b", node_b)
-        .add_edge(START, "node_a")
-        .add_edge(START, "node_b")
-        .compile(checkpointer=checkpointer)
-    )
-
-    config = {"configurable": {"thread_id": "test-interrupt-concurrent-async"}}
-
-    await graph.ainvoke({"foo": ""}, config)
-
-    state = await graph.aget_state(config)
-    assert len(state.tasks) > 0
-
-    interrupts = [t for t in state.tasks if hasattr(t, "interrupts") and t.interrupts]
-    assert len(interrupts) > 0, (
-        "GraphInterrupt was swallowed — interrupt() in node_a "
-        "should have paused execution"
-    )
-
-
-async def test_node_error_handler_handles_subgraph_internal_failure_async() -> None:
-    class SubState(TypedDict):
-        foo: str
-
-    class ParentState(TypedDict):
-        foo: str
-
-    captured: dict[str, object] = {}
-
-    async def sub_fail_node(state: SubState) -> SubState:
-        raise ValueError("async subgraph boom")
-
-    async def parent_handler(state: ParentState, error: NodeError) -> ParentState:
-        captured["from_node_name"] = error.node
-        captured["from_node_error"] = error.error
-        return {"foo": "handled_async_subgraph"}
-
-    subgraph = (
-        StateGraph(SubState)
-        .add_node("sub_fail_node", sub_fail_node)
-        .add_edge(START, "sub_fail_node")
-        .compile()
-    )
-
-    parent_graph = (
-        StateGraph(ParentState)
-        .add_node("subgraph_node", subgraph, error_handler=parent_handler)
-        .add_edge(START, "subgraph_node")
-        .compile()
-    )
-
-    result = await parent_graph.ainvoke({"foo": ""})
-    assert result["foo"] == "handled_async_subgraph"
-    assert captured["from_node_name"] == "subgraph_node"
-    assert isinstance(captured["from_node_error"], BaseException)

libs/langgraph/tests/test_pregel_stream_v2.py

@@ -1,4 +1,4 @@
-"""Tests for Pregel.stream_events(version="v3") / astream_events(version="v3") and the transformer pipeline."""
+"""Tests for Pregel.stream_v2 / astream_v2 and the transformer pipeline."""
 
 from __future__ import annotations
 
@@ -125,7 +125,7 @@ def _build_custom_stream_graph():
 class _CustomPassthroughTransformer(StreamTransformer):
     """Opts a run into the `custom` stream mode without building a projection.
 
-    `stream_events(version="v3")` requests only the modes that registered transformers
+    `stream_v2` requests only the modes that registered transformers
     declare via `required_stream_modes`. Custom events are raw user
     emissions from `StreamWriter`, so tests that want them visible on
     the main event log register this pass-through transformer.
@@ -390,31 +390,25 @@ class TestStreamChannelNamed:
 
 
 # ---------------------------------------------------------------------------
-# stream_events(version="v3") sync tests
+# stream_v2 sync tests
 # ---------------------------------------------------------------------------
 
 
 class TestStreamV2Sync:
     def test_values_projection(self) -> None:
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_simple_graph().stream_v2({"value": "x", "items": []})
         snapshots = list(run.values)
         assert len(snapshots) >= 1
         last = snapshots[-1]
         assert "A" in last["value"] and "B" in last["value"]
 
     def test_output(self) -> None:
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_simple_graph().stream_v2({"value": "x", "items": []})
         output = run.output
         assert output == {"value": "xAB", "items": ["a", "b"]}
 
     def test_raw_event_iteration(self) -> None:
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_simple_graph().stream_v2({"value": "x", "items": []})
         events = list(run)
         assert len(events) > 0
         for event in events:
@@ -424,27 +418,22 @@ class TestStreamV2Sync:
             assert isinstance(event["params"]["timestamp"], int)
 
     def test_extensions_has_native_keys(self) -> None:
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_simple_graph().stream_v2({"value": "x", "items": []})
         _ = run.output
         assert "values" in run.extensions and "messages" in run.extensions
         assert run.values is run.extensions["values"]
         assert run.messages is run.extensions["messages"]
 
     def test_extensions_is_read_only(self) -> None:
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_simple_graph().stream_v2({"value": "x", "items": []})
         with pytest.raises(TypeError):
             run.extensions["new_key"] = object()  # type: ignore[index]
         with pytest.raises(TypeError):
             del run.extensions["values"]  # type: ignore[attr-defined]
 
     def test_custom_stream_events(self) -> None:
-        run = _build_custom_stream_graph().stream_events(
+        run = _build_custom_stream_graph().stream_v2(
             {"value": "x", "items": []},
-            version="v3",
             transformers=[_CustomPassthroughTransformer],
         )
         custom_events = [e for e in run if e["method"] == "custom"]
@@ -455,34 +444,27 @@ class TestStreamV2Sync:
     def test_custom_events_suppressed_without_transformer(self) -> None:
         """Without a transformer declaring `"custom"`, no custom events flow.
 
-        `stream_events(version="v3")` asks the graph only for the modes that registered
+        `stream_v2` asks the graph only for the modes that registered
         transformers require. Built-ins cover `values` / `messages`;
         consumers that want raw custom events surface them by
         registering a transformer whose `required_stream_modes`
         includes `"custom"`.
         """
-        run = _build_custom_stream_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_custom_stream_graph().stream_v2({"value": "x", "items": []})
         custom_events = [e for e in run if e["method"] == "custom"]
         assert custom_events == []
 
     def test_interleave_values_and_messages(self) -> None:
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_simple_graph().stream_v2({"value": "x", "items": []})
         tagged = list(run.interleave("values", "messages"))
         names = [name for name, _ in tagged]
         assert set(names).issubset({"values", "messages"})
         assert names.count("values") >= 1
-        # interleave releases its subscription on completion.
-        assert run.extensions["values"]._subscribed is False
-        assert run.extensions["messages"]._subscribed is False
+        with pytest.raises(RuntimeError, match="already has a subscriber"):
+            list(run.values)
 
     def test_abort_marks_exhausted_and_closes_mux(self) -> None:
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_simple_graph().stream_v2({"value": "x", "items": []})
         values_iter = iter(run.values)
         _ = next(values_iter)
         run.abort()
@@ -491,63 +473,48 @@ class TestStreamV2Sync:
         run.abort()  # idempotent
 
     def test_context_manager_calls_abort_on_exit(self) -> None:
-        with _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        ) as run:
+        with _build_simple_graph().stream_v2({"value": "x", "items": []}) as run:
             _ = next(iter(run.values))
         assert run._exhausted is True
 
     def test_interleave_unknown_projection(self) -> None:
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_simple_graph().stream_v2({"value": "x", "items": []})
         with pytest.raises(KeyError):
             list(run.interleave("values", "does_not_exist"))
 
 
 class TestStreamV2SyncErrors:
     def test_error_propagation_output(self) -> None:
-        run = _build_error_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_error_graph().stream_v2({"value": "x", "items": []})
         with pytest.raises(ValueError, match="boom"):
             _ = run.output
 
     def test_error_propagation_values(self) -> None:
-        run = _build_error_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_error_graph().stream_v2({"value": "x", "items": []})
         with pytest.raises(ValueError, match="boom"):
             list(run.values)
 
     def test_error_propagation_raw_events(self) -> None:
-        run = _build_error_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_error_graph().stream_v2({"value": "x", "items": []})
         with pytest.raises(ValueError, match="boom"):
             list(run)
 
     def test_error_propagation_interrupted(self) -> None:
-        run = _build_error_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_error_graph().stream_v2({"value": "x", "items": []})
         with pytest.raises(ValueError, match="boom"):
             _ = run.interrupted
 
     def test_error_propagation_interrupts(self) -> None:
-        run = _build_error_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_error_graph().stream_v2({"value": "x", "items": []})
         with pytest.raises(ValueError, match="boom"):
             _ = run.interrupts
 
 
 class TestStreamV2SyncInterrupt:
     def test_interrupted(self) -> None:
-        run = _build_interrupt_graph().stream_events(
+        run = _build_interrupt_graph().stream_v2(
             {"value": "x", "items": []},
             {"configurable": {"thread_id": "t1"}},
-            version="v3",
         )
         _ = run.output
         assert run.interrupted is True
@@ -555,7 +522,7 @@ class TestStreamV2SyncInterrupt:
 
 
 # ---------------------------------------------------------------------------
-# astream_events(version="v3") async tests
+# astream_v2 async tests
 # ---------------------------------------------------------------------------
 
 
@@ -563,34 +530,26 @@ class TestStreamV2SyncInterrupt:
 @NEEDS_CONTEXTVARS
 class TestStreamV2Async:
     async def test_values_projection(self) -> None:
-        run = await _build_simple_graph().astream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = await _build_simple_graph().astream_v2({"value": "x", "items": []})
         snapshots = [s async for s in run.values]
         assert len(snapshots) >= 1
         last = snapshots[-1]
         assert "A" in last["value"] and "B" in last["value"]
 
     async def test_output(self) -> None:
-        run = await _build_simple_graph().astream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = await _build_simple_graph().astream_v2({"value": "x", "items": []})
         output = await run.output()
         assert output == {"value": "xAB", "items": ["a", "b"]}
 
     async def test_raw_event_iteration(self) -> None:
-        run = await _build_simple_graph().astream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = await _build_simple_graph().astream_v2({"value": "x", "items": []})
         events = [e async for e in run]
         assert len(events) > 0
         for event in events:
             assert event["type"] == "event"
 
     async def test_abort_marks_exhausted_and_closes_mux(self) -> None:
-        run = await _build_simple_graph().astream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = await _build_simple_graph().astream_v2({"value": "x", "items": []})
         values_iter = aiter(run.values)
         _ = await anext(values_iter)
         await run.abort()
@@ -600,26 +559,21 @@ class TestStreamV2Async:
         await run.abort()  # idempotent
 
     async def test_context_manager_calls_abort_on_exit(self) -> None:
-        run = await _build_simple_graph().astream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = await _build_simple_graph().astream_v2({"value": "x", "items": []})
         async with run:
             _ = await anext(aiter(run.values))
         assert run._exhausted is True
 
     async def test_extensions_has_native_keys(self) -> None:
-        run = await _build_simple_graph().astream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = await _build_simple_graph().astream_v2({"value": "x", "items": []})
         _ = await run.output()
         assert "values" in run.extensions and "messages" in run.extensions
         assert run.values is run.extensions["values"]
         assert run.messages is run.extensions["messages"]
 
     async def test_custom_stream_events(self) -> None:
-        run = await _build_custom_stream_graph().astream_events(
+        run = await _build_custom_stream_graph().astream_v2(
             {"value": "x", "items": []},
-            version="v3",
             transformers=[_CustomPassthroughTransformer],
         )
         events = [e async for e in run]
@@ -633,39 +587,29 @@ class TestStreamV2Async:
 @NEEDS_CONTEXTVARS
 class TestStreamV2AsyncErrors:
     async def test_error_propagation_output(self) -> None:
-        run = await _build_error_graph().astream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = await _build_error_graph().astream_v2({"value": "x", "items": []})
         with pytest.raises(ValueError, match="boom"):
             await run.output()
 
     async def test_error_propagation_values(self) -> None:
-        run = await _build_error_graph().astream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = await _build_error_graph().astream_v2({"value": "x", "items": []})
         with pytest.raises(ValueError, match="boom"):
             async for _ in run.values:
                 pass
 
     async def test_error_propagation_raw_events(self) -> None:
-        run = await _build_error_graph().astream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = await _build_error_graph().astream_v2({"value": "x", "items": []})
         with pytest.raises(ValueError, match="boom"):
             async for _ in run:
                 pass
 
     async def test_error_propagation_interrupted(self) -> None:
-        run = await _build_error_graph().astream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = await _build_error_graph().astream_v2({"value": "x", "items": []})
         with pytest.raises(ValueError, match="boom"):
             await run.interrupted()
 
     async def test_error_propagation_interrupts(self) -> None:
-        run = await _build_error_graph().astream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = await _build_error_graph().astream_v2({"value": "x", "items": []})
         with pytest.raises(ValueError, match="boom"):
             await run.interrupts()
 
@@ -674,10 +618,9 @@ class TestStreamV2AsyncErrors:
 @NEEDS_CONTEXTVARS
 class TestStreamV2AsyncInterrupt:
     async def test_interrupted(self) -> None:
-        run = await _build_interrupt_graph().astream_events(
+        run = await _build_interrupt_graph().astream_v2(
             {"value": "x", "items": []},
             {"configurable": {"thread_id": "t2"}},
-            version="v3",
         )
         _ = await run.output()
         assert await run.interrupted() is True
@@ -1041,8 +984,8 @@ class TestCustomTransformer:
                     self._channel.push(self._count)
                 return True
 
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3", transformers=[CounterTransformer]
+        run = _build_simple_graph().stream_v2(
+            {"value": "x", "items": []}, transformers=[CounterTransformer]
         )
         assert "counter" in run.extensions
         counter_iter = iter(run.extensions["counter"])
@@ -1067,15 +1010,15 @@ class TestCustomTransformer:
                     self._log.push("saw_values")
                 return True
 
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3", transformers=[FooTransformer]
+        run = _build_simple_graph().stream_v2(
+            {"value": "x", "items": []}, transformers=[FooTransformer]
         )
         foo_iter = iter(run.foo)
         _ = run.output
         assert "foo" in run.extensions and run.foo is run.extensions["foo"]
         assert "saw_values" in list(foo_iter)
 
-    def test_stream_events_v3_rejects_transformer_instances(self) -> None:
+    def test_stream_v2_rejects_transformer_instances(self) -> None:
         class InstanceTransformer(StreamTransformer):
             def init(self) -> dict[str, Any]:
                 return {}
@@ -1084,10 +1027,8 @@ class TestCustomTransformer:
                 return True
 
         with pytest.raises(TypeError, match="pre-built instance"):
-            _build_simple_graph().stream_events(
-                {"value": "x", "items": []},
-                version="v3",
-                transformers=[InstanceTransformer()],
+            _build_simple_graph().stream_v2(
+                {"value": "x", "items": []}, transformers=[InstanceTransformer()]
             )
 
     def test_stream_channel_auto_forward(self) -> None:
@@ -1106,8 +1047,8 @@ class TestCustomTransformer:
                     self._channel.push("emitted")
                 return True
 
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3", transformers=[EmitterTransformer]
+        run = _build_simple_graph().stream_v2(
+            {"value": "x", "items": []}, transformers=[EmitterTransformer]
         )
         custom_events = [e for e in run if e["method"] == "custom:emitter"]
         assert len(custom_events) > 0
@@ -1151,10 +1092,8 @@ class TestCustomTransformer:
                 return True
 
         with pytest.raises(ValueError, match=r"conflict.*'values'.*ValuesTransformer"):
-            _build_simple_graph().stream_events(
-                {"value": "x", "items": []},
-                version="v3",
-                transformers=[ConflictTransformer],
+            _build_simple_graph().stream_v2(
+                {"value": "x", "items": []}, transformers=[ConflictTransformer]
             )
 
 
@@ -1236,8 +1175,8 @@ class TestStreamChannelAutoLifecycle:
                     self._log.push("got_it")
                 return True
 
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3", transformers=[MinimalTransformer]
+        run = _build_simple_graph().stream_v2(
+            {"value": "x", "items": []}, transformers=[MinimalTransformer]
         )
         minimal_iter = iter(run.extensions["minimal"])
         _ = run.output
@@ -1497,8 +1436,8 @@ class TestAsyncTransformerLane:
             async def afinalize(self) -> None:
                 self._log.close()
 
-        run = await _build_simple_graph().astream_events(
-            {"value": "x", "items": []}, version="v3", transformers=[Scorer]
+        run = await _build_simple_graph().astream_v2(
+            {"value": "x", "items": []}, transformers=[Scorer]
         )
         scores_cursor = aiter(run.extensions["scores"])
         _ = await run.output()
@@ -1514,9 +1453,7 @@ class TestAsyncTransformerLane:
 @NEEDS_CONTEXTVARS
 class TestMemoryBounds:
     def test_sync_subscribed_buffer_stays_at_most_one_between_yields(self) -> None:
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_simple_graph().stream_v2({"value": "x", "items": []})
         events_iter = iter(run)
         max_buffered = 0
         count = 0
@@ -1529,9 +1466,7 @@ class TestMemoryBounds:
         )
 
     def test_unsubscribed_projections_never_accumulate(self) -> None:
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_simple_graph().stream_v2({"value": "x", "items": []})
         list(run)
         values_log = run.extensions["values"]
         messages_log = run.extensions["messages"]
@@ -1539,25 +1474,19 @@ class TestMemoryBounds:
         assert len(messages_log._items) == 0 and not messages_log._subscribed
 
     def test_output_path_does_not_retain_values(self) -> None:
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_simple_graph().stream_v2({"value": "x", "items": []})
         _ = run.output
         values_log = run.extensions["values"]
         assert len(values_log._items) == 0 and not values_log._subscribed
 
     def test_drained_subscriber_buffer_returns_to_empty(self) -> None:
-        run = _build_simple_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = _build_simple_graph().stream_v2({"value": "x", "items": []})
         list(run.values)
         assert len(run.extensions["values"]._items) == 0
 
     @pytest.mark.anyio
     async def test_async_single_consumer_buffer_stays_at_most_one(self) -> None:
-        run = await _build_simple_graph().astream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = await _build_simple_graph().astream_v2({"value": "x", "items": []})
         max_buffered = 0
         count = 0
         async for _ in run:
@@ -1568,9 +1497,7 @@ class TestMemoryBounds:
 
     @pytest.mark.anyio
     async def test_async_unsubscribed_projections_never_accumulate(self) -> None:
-        run = await _build_simple_graph().astream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run = await _build_simple_graph().astream_v2({"value": "x", "items": []})
         _ = await run.output()
         values_log = run.extensions["values"]
         messages_log = run.extensions["messages"]

libs/langgraph/tests/test_retry.py

@@ -1,5 +1,4 @@
 import asyncio
-import operator
 import sys
 import threading
 import time
@@ -16,7 +15,7 @@ from langchain_core.language_models.fake_chat_models import GenericFakeChatModel
 from langchain_core.messages import AIMessage, AIMessageChunk, BaseMessage, HumanMessage
 from langchain_core.outputs import ChatGeneration, ChatGenerationChunk, ChatResult
 from langchain_core.runnables import RunnableLambda, RunnableParallel
-from langgraph.checkpoint.memory import InMemorySaver, MemorySaver
+from langgraph.checkpoint.memory import MemorySaver
 from langgraph.checkpoint.serde.jsonplus import JsonPlusSerializer
 from typing_extensions import TypedDict
 
@@ -35,7 +34,7 @@ from langgraph._internal._runnable import RunnableCallable
 from langgraph._internal._timeout import coerce_timeout_policy
 from langgraph.channels.ephemeral_value import EphemeralValue
 from langgraph.channels.last_value import LastValue
-from langgraph.errors import GraphInterrupt, NodeError, NodeTimeoutError, ParentCommand
+from langgraph.errors import GraphInterrupt, NodeTimeoutError, ParentCommand
 from langgraph.func import entrypoint, task
 from langgraph.graph import END, START, StateGraph, add_messages
 from langgraph.pregel import NodeBuilder, Pregel
@@ -210,15 +209,6 @@ def test_should_retry_default_retry_on():
     req_error_no_resp.response = None
     assert _should_retry_on(policy, req_error_no_resp) is True
 
-    # NodeTimeoutError should be retryable by default
-    assert (
-        _should_retry_on(
-            policy,
-            NodeTimeoutError("node", 1.0, kind="run", run_timeout=0.5),
-        )
-        is True
-    )
-
     # Should retry on other exceptions by default
     class CustomException(Exception):
         pass
@@ -1465,14 +1455,14 @@ async def test_state_graph_add_node_timeout_composes_with_retry():
     async def flaky(state: _TimeoutState) -> _TimeoutState:
         attempts.append(len(attempts))
         if len(attempts) < 2:
-            await asyncio.sleep(1.0)
+            await asyncio.sleep(0.5)
         return {"x": state["x"] + 1}
 
     builder = StateGraph(_TimeoutState)
     builder.add_node(
         "flaky",
         flaky,
-        timeout=TimeoutPolicy(idle_timeout=0.3),
+        timeout=TimeoutPolicy(idle_timeout=0.1),
         retry_policy=RetryPolicy(
             max_attempts=3,
             initial_interval=0.0,
@@ -1765,505 +1755,3 @@ async def test_arun_with_retry_timeout_observer_treats_bubble_up_as_non_error():
     assert finish.status == "success"
     assert finish.error_type is None
     assert finish.error_message is None
-
-
-# ---------------------------------------------------------------------------
-# Watcher invariant: any timeout that retry/error_handler can recover from
-# MUST emit `finish=error` BEFORE the in-process recovery work happens. The
-# external watchdog (langgraph-api) relies on this so it only kills a worker
-# when no `finish` arrives within the deadline. The tests below pin down the
-# three recovery paths so a refactor that moves `_finish_timed_attempt` past
-# an `await` (or past the final `raise`) trips CI.
-# ---------------------------------------------------------------------------
-
-
-@pytest.mark.anyio
-async def test_arun_with_retry_observer_emits_finish_before_retry_backoff():
-    """`finish=error` of attempt N must arrive before the retry backoff sleep."""
-    timeline: list[tuple[float, Any]] = []
-
-    class TimingOutOnceProc:
-        def __init__(self) -> None:
-            self.calls = 0
-
-        async def ainvoke(self, input, config):
-            self.calls += 1
-            if self.calls == 1:
-                await asyncio.sleep(1.0)
-            return "ok"
-
-    backoff = 0.25
-    policy = RetryPolicy(
-        max_attempts=2,
-        initial_interval=backoff,
-        backoff_factor=1.0,
-        max_interval=backoff,
-        jitter=False,
-        retry_on=NodeTimeoutError,
-    )
-    task = _make_task(
-        TimingOutOnceProc(),
-        timeout=_idle_timeout(0.05),
-        retry_policy=(policy,),
-        name="backoff_watcher",
-    )
-    task.config[CONF][CONFIG_KEY_TIMED_ATTEMPT_OBSERVER] = lambda ev: timeline.append(
-        (time.monotonic(), ev)
-    )
-
-    assert await arun_with_retry(task, retry_policy=None) == "ok"
-
-    starts = [(t, ev) for t, ev in timeline if ev.event == "start"]
-    finishes = [(t, ev) for t, ev in timeline if ev.event == "finish"]
-    assert [ev.context.attempt for _, ev in starts] == [1, 2]
-    assert [ev.status for _, ev in finishes] == ["error", "success"]
-
-    first_finish_t = finishes[0][0]
-    second_start_t = starts[1][0]
-
-    # The watcher relies on this gap: `finish=error` for attempt 1 must arrive
-    # before `arun_with_retry` enters `await asyncio.sleep(backoff)`. We give a
-    # generous slack to keep this stable on slow CI; the structural invariant
-    # is "finish lands first", not "the gap equals exactly backoff".
-    assert second_start_t - first_finish_t >= backoff * 0.5, (
-        f"finish=error appears to be emitted after retry backoff sleep; "
-        f"gap was {second_start_t - first_finish_t:.3f}s, expected >= {backoff * 0.5:.3f}s"
-    )
-
-
-@pytest.mark.anyio
-async def test_state_graph_observer_emits_finish_before_error_handler_start():
-    """Original task's `finish=error` must arrive before the error_handler task's `start`."""
-
-    class State(TypedDict):
-        foo: str
-
-    async def slow_node(state: State) -> State:
-        await asyncio.sleep(1.0)
-        return {"foo": "should-not-happen"}
-
-    async def handler_node(state: State, error: NodeError) -> State:
-        return {"foo": "handled"}
-
-    events: list = []
-    graph = (
-        StateGraph(State)
-        .add_node(
-            "slow",
-            slow_node,
-            timeout=TimeoutPolicy(idle_timeout=0.05),
-            error_handler=handler_node,
-        )
-        .add_edge(START, "slow")
-        .compile()
-    )
-
-    result = await graph.ainvoke(
-        {"foo": ""},
-        config={
-            "configurable": {CONFIG_KEY_TIMED_ATTEMPT_OBSERVER: events.append},
-        },
-    )
-    assert result["foo"] == "handled"
-
-    # Filter to events from the failing node only — the handler node has no
-    # timeout configured here, so it doesn't appear in the observer stream.
-    slow_events = [ev for ev in events if ev.context.task_name == "slow"]
-    starts = [ev for ev in slow_events if ev.event == "start"]
-    finishes = [ev for ev in slow_events if ev.event == "finish"]
-    assert len(starts) == 1
-    assert len(finishes) == 1
-    assert finishes[0].status == "error"
-    assert finishes[0].error_type == "NodeTimeoutError"
-    # The slow task's finish-error event must precede every event for any
-    # follow-up task in the same observer stream.
-    slow_finish_index = events.index(finishes[0])
-    for ev in events[slow_finish_index + 1 :]:
-        assert ev.context.task_name == "slow" or ev.event == "start", (
-            f"unexpected event {ev.event} for {ev.context.task_name} "
-            f"after slow's finish=error"
-        )
-
-
-@pytest.mark.anyio
-async def test_arun_with_retry_observer_emits_finish_before_final_raise_on_exhaustion():
-    """When retry exhausts and the timeout propagates, the final `finish=error` must
-    be emitted before `arun_with_retry` re-raises."""
-    events: list = []
-
-    class AlwaysTimingOutProc:
-        async def ainvoke(self, input, config):
-            await asyncio.sleep(1.0)
-            return "never"
-
-    policy = RetryPolicy(
-        max_attempts=2,
-        initial_interval=0.0,
-        jitter=False,
-        retry_on=NodeTimeoutError,
-    )
-    task = _make_task(
-        AlwaysTimingOutProc(),
-        timeout=_idle_timeout(0.05),
-        retry_policy=(policy,),
-        name="never_succeeds",
-    )
-    task.config[CONF][CONFIG_KEY_TIMED_ATTEMPT_OBSERVER] = events.append
-
-    with pytest.raises(NodeTimeoutError):
-        await arun_with_retry(task, retry_policy=None)
-
-    starts = [ev for ev in events if ev.event == "start"]
-    finishes = [ev for ev in events if ev.event == "finish"]
-    assert [ev.context.attempt for ev in starts] == [1, 2]
-    assert [ev.context.attempt for ev in finishes] == [1, 2]
-    assert [ev.status for ev in finishes] == ["error", "error"]
-    assert all(ev.error_type == "NodeTimeoutError" for ev in finishes)
-    # Both finish events were observed BEFORE arun_with_retry raised, otherwise
-    # the `with pytest.raises` block would have exited before `events` got
-    # populated with the second finish.
-
-
-@pytest.mark.anyio
-async def test_sync_sleep_in_async_node_bypasses_timeout_and_emits_finish_success():
-    """Sync `time.sleep` inside an async node blocks the event loop so the
-    in-process watchdog cannot fire. We document the resulting behavior here:
-
-    1. `NodeTimeoutError` is NOT raised, even though the sync sleep exceeds
-       `idle_timeout`.
-    2. The node's normal return value flows through.
-    3. `finish=success` is emitted to the observer.
-
-    This is the canonical case where the in-process timeout is defeated and
-    the only safety net is the external watcher (langgraph-api), which
-    SIGKILLs the worker when no `finish` arrives within its deadline. The
-    catch is that with a *short* sync sleep the event loop unblocks before
-    the watcher's deadline expires, so the watcher legitimately does not
-    kill — meaning the configured `idle_timeout` is silently honored at the
-    process level only when the block is long enough to outlast the
-    watcher's grace.
-
-    This is the documented "Cooperative cancellation" caveat on
-    `TimeoutPolicy`. The test pins the behavior so any future change that
-    starts raising `NodeTimeoutError` for sync-blocked async nodes (or stops
-    emitting `finish=success`) is caught.
-    """
-    events: list = []
-
-    class SyncSleepingProc:
-        async def ainvoke(self, input, config):
-            time.sleep(0.1)
-            return "completed_despite_timeout"
-
-    task = _make_task(
-        SyncSleepingProc(),
-        timeout=_idle_timeout(0.05),
-        name="sync_sleeper",
-    )
-    task.config[CONF][CONFIG_KEY_TIMED_ATTEMPT_OBSERVER] = events.append
-
-    result = await arun_with_retry(task, retry_policy=None)
-
-    assert result == "completed_despite_timeout"
-    starts = [ev for ev in events if ev.event == "start"]
-    finishes = [ev for ev in events if ev.event == "finish"]
-    assert len(starts) == 1
-    assert len(finishes) == 1
-    assert finishes[0].status == "success"
-    assert finishes[0].error_type is None
-
-
-def test_graph_error_handler_runs_after_retry_exhaustion():
-    class State(TypedDict):
-        foo: str
-
-    attempts = 0
-    captured: dict[str, object] = {}
-
-    def always_failing_node(state: State) -> State:
-        nonlocal attempts
-        attempts += 1
-        raise ValueError("Always fails")
-
-    def err_handler_node(state: State, error: NodeError) -> Command:
-        captured["from_node_name"] = error.node
-        captured["from_node_error"] = error.error
-        return Command(update={"foo": "handled"}, goto="after_handler")
-
-    def after_handler(state: State) -> State:
-        return {"foo": f"{state['foo']}_after"}
-
-    retry_policy = RetryPolicy(
-        max_attempts=2,
-        initial_interval=0.01,
-        jitter=False,
-        retry_on=ValueError,
-    )
-
-    graph = (
-        StateGraph(State)
-        .add_node(
-            "always_failing",
-            always_failing_node,
-            retry_policy=retry_policy,
-            error_handler=err_handler_node,
-        )
-        .add_node("after_handler", after_handler)
-        .add_edge(START, "always_failing")
-        .compile()
-    )
-
-    with patch("time.sleep"):
-        result = graph.invoke({"foo": ""})
-
-    assert attempts == 2
-    assert result["foo"] == "handled_after"
-    assert captured["from_node_name"] == "always_failing"
-    assert isinstance(captured["from_node_error"], BaseException)
-
-
-def test_graph_error_handler_can_route_with_command():
-    class State(TypedDict):
-        foo: str
-
-    attempts = 0
-
-    def always_failing_node(state: State) -> State:
-        nonlocal attempts
-        attempts += 1
-        raise ValueError("Always fails")
-
-    def err_handler_node(state: State) -> Command:
-        return Command(update={"foo": "handled"}, goto="next_node")
-
-    def next_node(state: State) -> State:
-        return {"foo": f"{state['foo']}_next"}
-
-    retry_policy = RetryPolicy(
-        max_attempts=1,
-        initial_interval=0.01,
-        jitter=False,
-        retry_on=ValueError,
-    )
-
-    graph = (
-        StateGraph(State)
-        .add_node(
-            "always_failing",
-            always_failing_node,
-            retry_policy=retry_policy,
-            error_handler=err_handler_node,
-        )
-        .add_node("next_node", next_node)
-        .add_edge(START, "always_failing")
-        .compile()
-    )
-
-    result = graph.invoke({"foo": ""})
-    assert attempts == 1
-    assert result["foo"] == "handled_next"
-
-
-def test_graph_error_handler_failure_fails_run():
-    class State(TypedDict):
-        foo: str
-
-    def always_failing_node(state: State) -> State:
-        raise ValueError("Always fails")
-
-    def err_handler_node(state: State) -> State:
-        raise RuntimeError("handler failed")
-
-    graph = (
-        StateGraph(State)
-        .add_node("always_failing", always_failing_node, error_handler=err_handler_node)
-        .add_edge(START, "always_failing")
-        .compile()
-    )
-
-    with pytest.raises(RuntimeError, match="handler failed"):
-        graph.invoke({"foo": ""})
-
-
-def test_graph_error_handler_handles_subgraph_internal_failure():
-    class SubState(TypedDict):
-        foo: str
-
-    class ParentState(TypedDict):
-        foo: str
-
-    parent_handler_called = False
-    captured: dict[str, object] = {}
-
-    def sub_fail_node(state: SubState) -> SubState:
-        raise ValueError("subgraph boom")
-
-    def parent_handler(state: ParentState, error: NodeError) -> ParentState:
-        nonlocal parent_handler_called
-        parent_handler_called = True
-        captured["from_node_name"] = error.node
-        captured["from_node_error"] = error.error
-        return {"foo": "handled_by_parent"}
-
-    subgraph = (
-        StateGraph(SubState)
-        .add_node("sub_fail_node", sub_fail_node)
-        .add_edge(START, "sub_fail_node")
-        .compile()
-    )
-
-    parent_graph = (
-        StateGraph(ParentState)
-        .add_node("subgraph_node", subgraph, error_handler=parent_handler)
-        .add_edge(START, "subgraph_node")
-        .compile()
-    )
-
-    result = parent_graph.invoke({"foo": ""})
-    assert result["foo"] == "handled_by_parent"
-    assert parent_handler_called is True
-    assert captured["from_node_name"] == "subgraph_node"
-    assert isinstance(captured["from_node_error"], BaseException)
-
-
-def test_graph_error_handler_error_context_survives_checkpoint_resume():
-    class State(TypedDict):
-        foo: str
-
-    captured: dict[str, object] = {}
-
-    def always_failing_node(state: State) -> State:
-        raise RuntimeError("failed before handler")
-
-    def err_handler_node(state: State, error: NodeError) -> State:
-        captured["from_node_name"] = error.node
-        captured["from_node_error"] = error.error
-        return {"foo": "handled_after_resume"}
-
-    checkpointer = InMemorySaver()
-    config = {"configurable": {"thread_id": "graph-error-resume"}}
-    graph = (
-        StateGraph(State)
-        .add_node("always_failing", always_failing_node, error_handler=err_handler_node)
-        .add_edge(START, "always_failing")
-        .compile(
-            checkpointer=checkpointer,
-            interrupt_before=["__error_handler__always_failing"],
-        )
-    )
-
-    # First run pauses before handler, after failure context is checkpointed.
-    graph.invoke({"foo": ""}, config)
-    # Resume should execute handler and recover serialized error context.
-    result = graph.invoke(None, config)
-
-    assert result["foo"] == "handled_after_resume"
-    assert captured["from_node_name"] == "always_failing"
-    assert isinstance(captured["from_node_error"], BaseException)
-
-
-def test_graph_error_handler_does_not_swallow_interrupt_concurrent():
-    """When a graph error handler is configured and a node calls interrupt()
-    concurrently with other nodes, the interrupt must still be raised — not
-    silently swallowed."""
-    from langgraph.types import interrupt
-
-    class State(TypedDict):
-        foo: str
-
-    def node_a(state: State) -> State:
-        # This node uses interrupt() which raises GraphInterrupt
-        val = interrupt("need human input")
-        return {"foo": f"a_{val}"}
-
-    def node_b(state: State) -> State:
-        return {}
-
-    def err_handler(state: State) -> State:
-        return {"foo": "handled"}
-
-    checkpointer = InMemorySaver()
-    graph = (
-        StateGraph(State)
-        .add_node("node_a", node_a, error_handler=err_handler)
-        .add_node("node_b", node_b)
-        # Fan-out: both node_a and node_b run concurrently
-        .add_edge(START, "node_a")
-        .add_edge(START, "node_b")
-        .compile(checkpointer=checkpointer)
-    )
-
-    config = {"configurable": {"thread_id": "test-interrupt-concurrent"}}
-
-    # First invoke should pause at the interrupt, not silently complete
-    graph.invoke({"foo": ""}, config)
-
-    # The graph should have an interrupt pending
-    state = graph.get_state(config)
-    assert len(state.tasks) > 0
-
-    # There should be a pending interrupt from node_a
-    interrupts = [t for t in state.tasks if hasattr(t, "interrupts") and t.interrupts]
-    assert len(interrupts) > 0, (
-        "GraphInterrupt was swallowed — interrupt() in node_a "
-        "should have paused execution"
-    )
-
-
-def test_node_error_handlers_route_to_matching_handler():
-    class State(TypedDict):
-        route: str
-        foo: Annotated[list[str], operator.add]
-
-    def route_node(state: State) -> State:
-        return {"foo": []}
-
-    def choose_node(state: State) -> str:
-        return state["route"]
-
-    def fail_a(state: State) -> State:
-        raise ValueError("a failed")
-
-    def fail_b(state: State) -> State:
-        raise RuntimeError("b failed")
-
-    def handler_a(state: State, error: NodeError) -> State:
-        assert error.node == "fail_a"
-        return {"foo": ["handled_a"]}
-
-    def handler_b(state: State, error: NodeError) -> State:
-        assert error.node == "fail_b"
-        return {"foo": ["handled_b"]}
-
-    graph = (
-        StateGraph(State)
-        .add_node("route_node", route_node)
-        .add_node("fail_a", fail_a, error_handler=handler_a)
-        .add_node("fail_b", fail_b, error_handler=handler_b)
-        .add_edge(START, "route_node")
-        .add_conditional_edges("route_node", choose_node, path_map=["fail_a", "fail_b"])
-        .compile()
-    )
-
-    result_a = graph.invoke({"route": "fail_a", "foo": []})
-    result_b = graph.invoke({"route": "fail_b", "foo": []})
-    assert result_a["foo"] == ["handled_a"]
-    assert result_b["foo"] == ["handled_b"]
-
-
-def test_node_without_error_handler_still_fails_run():
-    class State(TypedDict):
-        foo: str
-
-    def fail_without_handler(state: State) -> State:
-        raise ValueError("no handler")
-
-    graph = (
-        StateGraph(State)
-        .add_node("fail_without_handler", fail_without_handler)
-        .add_edge(START, "fail_without_handler")
-        .compile()
-    )
-
-    with pytest.raises(ValueError, match="no handler"):
-        graph.invoke({"foo": ""})

libs/langgraph/tests/test_runtime.py

@@ -1,6 +1,3 @@
-import asyncio
-import threading
-import time
 from dataclasses import dataclass
 from typing import Any
 
@@ -9,15 +6,8 @@ from langgraph.checkpoint.memory import MemorySaver
 from pydantic import BaseModel, ValidationError
 from typing_extensions import TypedDict
 
-from langgraph.errors import GraphDrained
 from langgraph.graph import END, START, StateGraph
-from langgraph.runtime import (
-    ExecutionInfo,
-    RunControl,
-    Runtime,
-    ServerInfo,
-    get_runtime,
-)
+from langgraph.runtime import ExecutionInfo, Runtime, ServerInfo, get_runtime
 
 
 def test_injected_runtime() -> None:
@@ -89,183 +79,6 @@ def test_merge_runtime() -> None:
     assert runtime1.merge(runtime3).context.api_key == "abc"  # type: ignore
 
 
-def test_merge_runtime_preserves_run_control() -> None:
-    control = RunControl()
-    runtime1 = Runtime(control=control)
-    runtime2 = Runtime(context=None)
-
-    assert runtime1.merge(runtime2).control is control
-
-
-def test_run_control_request_drain_stops_future_steps() -> None:
-    class State(TypedDict, total=False):
-        first: str
-        second: str
-
-    control = RunControl()
-
-    def first_node(state: State) -> dict[str, str]:
-        control.request_drain()
-        return {"first": "done"}
-
-    def second_node(state: State) -> dict[str, str]:
-        return {"second": "should-not-run"}
-
-    graph = StateGraph(State)
-    graph.add_node("first", first_node)
-    graph.add_node("second", second_node)
-    graph.add_edge(START, "first")
-    graph.add_edge("first", "second")
-    graph.add_edge("second", END)
-
-    with pytest.raises(GraphDrained, match="shutdown"):
-        graph.compile().invoke({}, control=control)
-
-
-@pytest.mark.anyio
-async def test_run_control_request_drain_stops_future_steps_async() -> None:
-    class State(TypedDict, total=False):
-        first: str
-        second: str
-
-    control = RunControl()
-
-    async def first_node(state: State) -> dict[str, str]:
-        control.request_drain()
-        return {"first": "done"}
-
-    async def second_node(state: State) -> dict[str, str]:
-        return {"second": "should-not-run"}
-
-    graph = StateGraph(State)
-    graph.add_node("first", first_node)
-    graph.add_node("second", second_node)
-    graph.add_edge(START, "first")
-    graph.add_edge("first", "second")
-    graph.add_edge("second", END)
-
-    with pytest.raises(GraphDrained, match="shutdown"):
-        await graph.compile().ainvoke({}, control=control)
-
-
-def test_drain_requested_in_terminal_step_finishes_normally() -> None:
-    class State(TypedDict, total=False):
-        value: str
-
-    control = RunControl()
-
-    def node(state: State) -> dict[str, str]:
-        control.request_drain()
-        return {"value": "done"}
-
-    graph = StateGraph(State)
-    graph.add_node("node", node)
-    graph.add_edge(START, "node")
-    graph.add_edge("node", END)
-
-    assert graph.compile().invoke({}, control=control) == {"value": "done"}
-    assert control.drain_requested
-
-
-def test_drain_with_exit_durability_persists_resume_checkpoint() -> None:
-    class State(TypedDict, total=False):
-        first: str
-        second: str
-
-    control = RunControl()
-
-    def first_node(state: State) -> dict[str, str]:
-        control.request_drain("sigterm")
-        return {"first": "done"}
-
-    def second_node(state: State) -> dict[str, str]:
-        return {"second": "done"}
-
-    graph = StateGraph(State)
-    graph.add_node("first", first_node)
-    graph.add_node("second", second_node)
-    graph.add_edge(START, "first")
-    graph.add_edge("first", "second")
-    graph.add_edge("second", END)
-
-    compiled = graph.compile(checkpointer=MemorySaver())
-    config = {"configurable": {"thread_id": "drain-exit"}}
-
-    with pytest.raises(GraphDrained, match="sigterm"):
-        compiled.invoke({}, config, durability="exit", control=control)
-
-    assert compiled.invoke(None, config, durability="exit") == {
-        "first": "done",
-        "second": "done",
-    }
-
-
-def test_drain_from_subgraph_can_resume_parent() -> None:
-    class State(TypedDict, total=False):
-        child_first: str
-        child_second: str
-        parent_second: str
-
-    control = RunControl()
-
-    def child_first(state: State) -> dict[str, str]:
-        control.request_drain("sigterm")
-        return {"child_first": "done"}
-
-    def child_second(state: State) -> dict[str, str]:
-        return {"child_second": "done"}
-
-    child_builder = StateGraph(State)
-    child_builder.add_node("child_first", child_first)
-    child_builder.add_node("child_second", child_second)
-    child_builder.add_edge(START, "child_first")
-    child_builder.add_edge("child_first", "child_second")
-    child_builder.add_edge("child_second", END)
-    child_graph = child_builder.compile(checkpointer=True)
-
-    def parent_second(state: State) -> dict[str, str]:
-        return {"parent_second": "done"}
-
-    parent_builder = StateGraph(State)
-    parent_builder.add_node("child", child_graph)
-    parent_builder.add_node("parent_second", parent_second)
-    parent_builder.add_edge(START, "child")
-    parent_builder.add_edge("child", "parent_second")
-    parent_builder.add_edge("parent_second", END)
-
-    compiled = parent_builder.compile(checkpointer=MemorySaver())
-    config = {"configurable": {"thread_id": "drain-subgraph"}}
-
-    with pytest.raises(GraphDrained, match="sigterm"):
-        compiled.invoke({}, config, control=control)
-
-    assert compiled.invoke(None, config) == {
-        "child_first": "done",
-        "child_second": "done",
-        "parent_second": "done",
-    }
-
-
-@pytest.mark.anyio
-async def test_drain_requested_in_terminal_step_finishes_normally_async() -> None:
-    class State(TypedDict, total=False):
-        value: str
-
-    control = RunControl()
-
-    async def node(state: State) -> dict[str, str]:
-        control.request_drain()
-        return {"value": "done"}
-
-    graph = StateGraph(State)
-    graph.add_node("node", node)
-    graph.add_edge(START, "node")
-    graph.add_edge("node", END)
-
-    assert await graph.compile().ainvoke({}, control=control) == {"value": "done"}
-    assert control.drain_requested
-
-
 def test_runtime_propogated_to_subgraph() -> None:
     @dataclass
     class Context:
@@ -579,334 +392,6 @@ def test_context_coercion_pydantic_validation_errors() -> None:
         )
 
 
-def test_external_drain_concurrent_sync() -> None:
-    """External thread calls request_drain() while graph is mid-execution."""
-
-    class State(TypedDict, total=False):
-        first: str
-        second: str
-
-    started = threading.Event()
-
-    def first_node(state: State) -> dict[str, str]:
-        started.set()
-        time.sleep(0.05)
-        return {"first": "done"}
-
-    def second_node(state: State) -> dict[str, str]:
-        return {"second": "should-not-run"}
-
-    graph = StateGraph(State)
-    graph.add_node("first", first_node)
-    graph.add_node("second", second_node)
-    graph.add_edge(START, "first")
-    graph.add_edge("first", "second")
-    graph.add_edge("second", END)
-
-    control = RunControl()
-    compiled = graph.compile()
-
-    exc_holder: list[BaseException | None] = [None]
-
-    def run_graph() -> None:
-        try:
-            compiled.invoke({}, control=control)
-        except GraphDrained as e:
-            exc_holder[0] = e
-
-    t = threading.Thread(target=run_graph)
-    t.start()
-
-    started.wait(timeout=5)
-    control.request_drain("sigterm")
-
-    t.join(timeout=10)
-
-    exc = exc_holder[0]
-    assert isinstance(exc, GraphDrained)
-    assert exc.reason == "sigterm"
-
-
-@pytest.mark.anyio
-async def test_external_drain_concurrent_async() -> None:
-    """External task calls request_drain() while graph is mid-execution."""
-
-    class State(TypedDict, total=False):
-        first: str
-        second: str
-
-    started = asyncio.Event()
-
-    async def first_node(state: State) -> dict[str, str]:
-        started.set()
-        await asyncio.sleep(0.05)
-        return {"first": "done"}
-
-    async def second_node(state: State) -> dict[str, str]:
-        return {"second": "should-not-run"}
-
-    graph = StateGraph(State)
-    graph.add_node("first", first_node)
-    graph.add_node("second", second_node)
-    graph.add_edge(START, "first")
-    graph.add_edge("first", "second")
-    graph.add_edge("second", END)
-
-    control = RunControl()
-    compiled = graph.compile()
-
-    async def drain_after_start() -> None:
-        await started.wait()
-        control.request_drain("sigterm")
-
-    drain_task = asyncio.create_task(drain_after_start())
-
-    with pytest.raises(GraphDrained, match="sigterm"):
-        await compiled.ainvoke({}, control=control)
-
-    await drain_task
-
-
-@pytest.mark.anyio
-async def test_drain_then_cancel_after_graceful_timeout() -> None:
-    """Simulate: drain requested -> node still running -> graceful timeout -> cancel.
-
-    This shows what happens when a long-running node doesn't finish within
-    the graceful period after drain is requested.
-    """
-
-    class State(TypedDict, total=False):
-        first: str
-        second: str
-
-    node_started = asyncio.Event()
-    node_cancelled = asyncio.Event()
-    node_finished = asyncio.Event()
-
-    async def slow_node(state: State) -> dict[str, str]:
-        node_started.set()
-        try:
-            await asyncio.sleep(30)  # very long operation
-        except asyncio.CancelledError:
-            node_cancelled.set()
-            raise
-        node_finished.set()
-        return {"first": "done"}
-
-    async def second_node(state: State) -> dict[str, str]:
-        return {"second": "should-not-run"}
-
-    graph = StateGraph(State)
-    graph.add_node("first", slow_node)
-    graph.add_node("second", second_node)
-    graph.add_edge(START, "first")
-    graph.add_edge("first", "second")
-    graph.add_edge("second", END)
-
-    control = RunControl()
-    compiled = graph.compile()
-
-    # Phase 1: start graph
-    graph_task = asyncio.create_task(compiled.ainvoke({}, control=control))
-
-    # Phase 2: wait for node to start, then request drain
-    await node_started.wait()
-    control.request_drain("sigterm")
-
-    # Phase 3: graceful timeout — node is still running, cancel after 1s
-    graceful_timeout = 1.0
-    await asyncio.sleep(graceful_timeout)
-
-    assert not node_finished.is_set(), "node should still be running"
-    assert not node_cancelled.is_set(), "node should not be cancelled yet"
-
-    # Phase 4: force cancel
-    graph_task.cancel()
-
-    with pytest.raises(asyncio.CancelledError):
-        await graph_task
-
-    # The node received CancelledError at the await point
-    assert node_cancelled.is_set(), "node should have received CancelledError"
-    assert not node_finished.is_set(), "node should NOT have finished normally"
-
-
-@pytest.mark.anyio
-async def test_cancel_ainvoke_with_async_node() -> None:
-    """Cancel ainvoke running an async node: CancelledError is delivered
-    at the await point and the node stops immediately."""
-
-    class State(TypedDict, total=False):
-        first: str
-        second: str
-
-    timeline: list[str] = []
-    node_started = asyncio.Event()
-
-    async def slow_async_node(state: State) -> dict[str, str]:
-        timeline.append(f"async_node:start thread={threading.current_thread().name}")
-        node_started.set()
-        try:
-            await asyncio.sleep(30)
-        except asyncio.CancelledError:
-            timeline.append("async_node:cancelled")
-            raise
-        timeline.append("async_node:finished")
-        return {"first": "done"}
-
-    async def second_node(state: State) -> dict[str, str]:
-        timeline.append("second_node:run")
-        return {"second": "should-not-run"}
-
-    graph = StateGraph(State)
-    graph.add_node("first", slow_async_node)
-    graph.add_node("second", second_node)
-    graph.add_edge(START, "first")
-    graph.add_edge("first", "second")
-    graph.add_edge("second", END)
-
-    compiled = graph.compile()
-    graph_task = asyncio.create_task(compiled.ainvoke({}))
-
-    await node_started.wait()
-    timeline.append("test:cancel")
-    graph_task.cancel()
-
-    with pytest.raises(asyncio.CancelledError):
-        await graph_task
-    timeline.append("test:done")
-
-    # async node runs on the event loop thread (MainThread)
-    assert any("MainThread" in e for e in timeline if "async_node:start" in e)
-    # CancelledError was delivered at the await point — node stopped
-    assert "async_node:cancelled" in timeline
-    # Node did NOT run to completion
-    assert "async_node:finished" not in timeline
-    # Second node never ran
-    assert "second_node:run" not in timeline
-
-
-@pytest.mark.anyio
-async def test_cancel_ainvoke_with_sync_node() -> None:
-    """Cancel ainvoke running a sync node.
-
-    Sync nodes in ainvoke run on a separate thread (via run_in_executor),
-    NOT on the event loop thread. Cancelling the asyncio task disconnects
-    from the thread future, but the thread keeps running as an orphan and
-    completes on its own.
-
-    Key difference from async nodes:
-    - async node: CancelledError stops the coroutine at an await point
-    - sync node: cancel only disconnects asyncio; the thread runs to completion
-
-    In shutdown case, we will ignore this because the instance will be destroyed soon.
-    """
-
-    class State(TypedDict, total=False):
-        first: str
-        second: str
-
-    timeline: list[str] = []
-    node_started = threading.Event()
-    node_finished = threading.Event()
-
-    def slow_sync_node(state: State) -> dict[str, str]:
-        timeline.append(f"sync_node:start thread={threading.current_thread().name}")
-        node_started.set()
-        time.sleep(1)
-        timeline.append("sync_node:after_sleep")
-        node_finished.set()
-        return {"first": "done"}
-
-    def second_node(state: State) -> dict[str, str]:
-        timeline.append("second_node:run")
-        return {"second": "should-not-run"}
-
-    graph = StateGraph(State)
-    graph.add_node("first", slow_sync_node)
-    graph.add_node("second", second_node)
-    graph.add_edge(START, "first")
-    graph.add_edge("first", "second")
-    graph.add_edge("second", END)
-
-    control = RunControl()
-    compiled = graph.compile()
-
-    timeline.append(f"test:main thread={threading.current_thread().name}")
-    graph_task = asyncio.create_task(compiled.ainvoke({}, control=control))
-
-    loop = asyncio.get_event_loop()
-    await loop.run_in_executor(None, node_started.wait, 5)
-
-    timeline.append("test:cancel+drain")
-    graph_task.cancel()
-    control.request_drain("sigterm")
-
-    with pytest.raises(asyncio.CancelledError):
-        await graph_task
-    timeline.append("test:exc=CancelledError")
-
-    # Sync node runs on a background thread (asyncio_*), NOT MainThread
-    sync_start = next(e for e in timeline if "sync_node:start" in e)
-    assert "MainThread" not in sync_start, (
-        "sync node should run on a background thread, not the event loop thread"
-    )
-
-    # At this point, the asyncio task is done but the thread is orphaned.
-    # The sync node has NOT finished yet — cancel only disconnected asyncio.
-    assert not node_finished.is_set(), (
-        "sync node should still be running in its background thread"
-    )
-
-    # Wait for the orphaned thread to complete on its own.
-    await loop.run_in_executor(None, node_finished.wait, 5)
-    assert node_finished.is_set()
-
-    # After the orphaned thread finishes, the full timeline looks like:
-    #   test:main thread=MainThread
-    #   sync_node:start thread=asyncio_N    <- background thread
-    #   test:cancel+drain                   <- cancel + drain fired
-    #   test:exc=CancelledError             <- asyncio disconnected
-    #   sync_node:after_sleep               <- thread ran to completion anyway
-    assert "sync_node:after_sleep" in timeline
-    # Second node never ran
-    assert "second_node:run" not in timeline
-
-    # Verify timeline ordering: cancel happened before node finished
-    cancel_idx = timeline.index("test:cancel+drain")
-    sleep_idx = timeline.index("sync_node:after_sleep")
-    assert cancel_idx < sleep_idx, (
-        "cancel was issued while the sync node was still sleeping"
-    )
-
-
-def test_drain_with_control_parameter_sync() -> None:
-    """Control parameter is wired through invoke -> stream."""
-
-    class State(TypedDict, total=False):
-        value: str
-
-    ran = False
-
-    def node(state: State) -> dict[str, str]:
-        nonlocal ran
-        ran = True
-        return {"value": "done"}
-
-    graph = StateGraph(State)
-    graph.add_node("node", node)
-    graph.add_edge(START, "node")
-    graph.add_edge("node", END)
-
-    # Pre-drained control stops before executing the first pending task.
-    control = RunControl()
-    control.request_drain("pre-drained")
-
-    with pytest.raises(GraphDrained, match="pre-drained"):
-        graph.compile().invoke({}, control=control)
-    assert not ran
-
-
 # --- ExecutionInfo unit tests ---
 
 

libs/langgraph/tests/test_stream_data_transformers.py

@@ -4,7 +4,7 @@ These transformers capture raw protocol events for their respective stream
 modes and expose them as native projections on the run stream (run.custom,
 run.updates, run.checkpoints, run.debug, run.tasks). Tests dispatch synthetic
 protocol events through a StreamMux to isolate transformer logic; the final
-group exercises real graphs through stream_events(version="v3").
+group exercises real graphs through stream_v2.
 """
 
 from __future__ import annotations
@@ -77,13 +77,8 @@ def _arm(mux: StreamMux, transformer: Any) -> None:
     transformer._log._subscribed = True
 
 
-def _unstamped(items):
-    """Strip push stamps from a StreamChannel's internal buffer."""
-    return [item for _stamp, item in items]
-
-
 def _drain(transformer: Any) -> list[Any]:
-    return _unstamped(transformer._log._items)
+    return list(transformer._log._items)
 
 
 # ---------------------------------------------------------------------------
@@ -145,7 +140,7 @@ def test_custom_does_not_suppress_from_main_log() -> None:
 
     mux.push(_custom_event([], "data"))
 
-    methods = [evt["method"] for evt in _unstamped(mux._events._items)]
+    methods = [evt["method"] for evt in mux._events._items]
     assert "custom" in methods
 
 
@@ -224,7 +219,7 @@ def test_checkpoints_does_not_suppress_from_main_log() -> None:
 
     mux.push(_checkpoints_event([], {"values": {}}))
 
-    methods = [evt["method"] for evt in _unstamped(mux._events._items)]
+    methods = [evt["method"] for evt in mux._events._items]
     assert "checkpoints" in methods
 
 
@@ -289,7 +284,7 @@ def test_debug_does_not_suppress_from_main_log() -> None:
 
     mux.push(_debug_event([], {"step": 0}))
 
-    methods = [evt["method"] for evt in _unstamped(mux._events._items)]
+    methods = [evt["method"] for evt in mux._events._items]
     assert "debug" in methods
 
 
@@ -365,7 +360,7 @@ def test_tasks_does_not_suppress_from_main_log() -> None:
 
     mux.push(_tasks_event([], {"id": "t1"}))
 
-    methods = [evt["method"] for evt in _unstamped(mux._events._items)]
+    methods = [evt["method"] for evt in mux._events._items]
     assert "tasks" in methods
 
 
@@ -434,7 +429,7 @@ def test_updates_does_not_suppress_from_main_log() -> None:
 
     mux.push(_updates_event([], {"n": {}}))
 
-    methods = [evt["method"] for evt in _unstamped(mux._events._items)]
+    methods = [evt["method"] for evt in mux._events._items]
     assert "updates" in methods
 
 
@@ -474,11 +469,11 @@ def test_unrelated_events_ignored_by_all() -> None:
     )
 
     for t in transformers:
-        assert _unstamped(t._log._items) == []
+        assert list(t._log._items) == []
 
 
 # ---------------------------------------------------------------------------
-# End-to-end: real graphs through stream_events(version="v3")
+# End-to-end: real graphs through stream_v2
 # ---------------------------------------------------------------------------
 
 
@@ -503,11 +498,11 @@ def _make_simple_graph() -> Any:
     return builder.compile()
 
 
-def test_stream_events_v3_custom_projection_opt_in() -> None:
+def test_stream_v2_custom_projection_opt_in() -> None:
     """run.custom surfaces get_stream_writer() payloads when opted in."""
     graph = _make_simple_graph()
-    run = graph.stream_events(
-        {"value": "hello", "items": []}, version="v3", transformers=[CustomTransformer]
+    run = graph.stream_v2(
+        {"value": "hello", "items": []}, transformers=[CustomTransformer]
     )
 
     custom_events = list(run.custom)
@@ -515,11 +510,11 @@ def test_stream_events_v3_custom_projection_opt_in() -> None:
     assert any(e.get("status") == "working" for e in custom_events)
 
 
-def test_stream_events_v3_custom_and_values_coexist() -> None:
+def test_stream_v2_custom_and_values_coexist() -> None:
     """Both run.custom and run.values work in the same run."""
     graph = _make_simple_graph()
-    run = graph.stream_events(
-        {"value": "hello", "items": []}, version="v3", transformers=[CustomTransformer]
+    run = graph.stream_v2(
+        {"value": "hello", "items": []}, transformers=[CustomTransformer]
     )
 
     custom_events = list(run.custom)
@@ -528,12 +523,10 @@ def test_stream_events_v3_custom_and_values_coexist() -> None:
     assert len(custom_events) >= 1
 
 
-def test_stream_events_v3_tasks_projection_opt_in() -> None:
+def test_stream_v2_tasks_projection_opt_in() -> None:
     """run.tasks surfaces raw task events when opted in via transformers=."""
     graph = _make_simple_graph()
-    run = graph.stream_events(
-        {"value": "x", "items": []}, transformers=[TasksTransformer], version="v3"
-    )
+    run = graph.stream_v2({"value": "x", "items": []}, transformers=[TasksTransformer])
 
     tasks_events = list(run.tasks)
     assert len(tasks_events) >= 1
@@ -541,12 +534,10 @@ def test_stream_events_v3_tasks_projection_opt_in() -> None:
     assert "my_node" in names
 
 
-def test_stream_events_v3_debug_projection_opt_in() -> None:
+def test_stream_v2_debug_projection_opt_in() -> None:
     """run.debug surfaces debug events when opted in via transformers=."""
     graph = _make_simple_graph()
-    run = graph.stream_events(
-        {"value": "x", "items": []}, transformers=[DebugTransformer], version="v3"
-    )
+    run = graph.stream_v2({"value": "x", "items": []}, transformers=[DebugTransformer])
 
     debug_events = list(run.debug)
     assert len(debug_events) >= 1
@@ -554,11 +545,11 @@ def test_stream_events_v3_debug_projection_opt_in() -> None:
     assert types & {"checkpoint", "task", "task_result"}
 
 
-def test_stream_events_v3_updates_projection_opt_in() -> None:
+def test_stream_v2_updates_projection_opt_in() -> None:
     """run.updates surfaces node output dicts when opted in via transformers=."""
     graph = _make_simple_graph()
-    run = graph.stream_events(
-        {"value": "x", "items": []}, version="v3", transformers=[UpdatesTransformer]
+    run = graph.stream_v2(
+        {"value": "x", "items": []}, transformers=[UpdatesTransformer]
     )
 
     updates = list(run.updates)
@@ -567,12 +558,11 @@ def test_stream_events_v3_updates_projection_opt_in() -> None:
     assert "my_node" in node_names
 
 
-def test_stream_events_v3_all_transformers_interleaved() -> None:
+def test_stream_v2_all_transformers_interleaved() -> None:
     """All five transformers registered together, consumed via interleave."""
     graph = _make_simple_graph()
-    run = graph.stream_events(
+    run = graph.stream_v2(
         {"value": "x", "items": []},
-        version="v3",
         transformers=[
             CustomTransformer,
             UpdatesTransformer,
@@ -604,7 +594,7 @@ def test_stream_events_v3_all_transformers_interleaved() -> None:
     assert run.output["value"] == "x!"
 
 
-def test_stream_events_v3_all_transformers_with_checkpointer() -> None:
+def test_stream_v2_all_transformers_with_checkpointer() -> None:
     """All transformers with a checkpointer — run.checkpoints populated."""
     from langgraph.checkpoint.memory import InMemorySaver
 
@@ -614,9 +604,8 @@ def test_stream_events_v3_all_transformers_with_checkpointer() -> None:
     builder.add_edge("my_node", END)
     graph = builder.compile(checkpointer=InMemorySaver())
 
-    run = graph.stream_events(
+    run = graph.stream_v2(
         {"value": "x", "items": []},
-        version="v3",
         config={"configurable": {"thread_id": "test-all"}},
         transformers=[
             CustomTransformer,
@@ -643,7 +632,7 @@ def test_stream_events_v3_all_transformers_with_checkpointer() -> None:
     assert len(collected["custom"]) >= 1
 
 
-def test_stream_events_v3_checkpoints_projection_opt_in() -> None:
+def test_stream_v2_checkpoints_projection_opt_in() -> None:
     """run.checkpoints surfaces checkpoint data when opted in with a checkpointer."""
     from langgraph.checkpoint.memory import InMemorySaver
 
@@ -653,9 +642,8 @@ def test_stream_events_v3_checkpoints_projection_opt_in() -> None:
     builder.add_edge("my_node", END)
     graph = builder.compile(checkpointer=InMemorySaver())
 
-    run = graph.stream_events(
+    run = graph.stream_v2(
         {"value": "x", "items": []},
-        version="v3",
         config={"configurable": {"thread_id": "test-ckpt-standalone"}},
         transformers=[CheckpointsTransformer],
     )
@@ -686,7 +674,7 @@ def test_tasks_and_lifecycle_coregistration() -> None:
 
     assert _drain(tasks) == [task_data]
 
-    methods = [evt["method"] for evt in _unstamped(mux._events._items)]
+    methods = [evt["method"] for evt in mux._events._items]
     assert "tasks" not in methods
 
 
@@ -695,9 +683,8 @@ def test_tasks_and_lifecycle_coregistration_e2e() -> None:
     is present and suppressing them from the main log.
     """
     graph = _make_simple_graph()
-    run = graph.stream_events(
+    run = graph.stream_v2(
         {"value": "x", "items": []},
-        version="v3",
         transformers=[TasksTransformer],
     )
 

libs/langgraph/tests/test_stream_lifecycle_transformer.py

@@ -6,7 +6,7 @@ on the `lifecycle` channel for both in-process iteration via
 events. Most tests dispatch synthetic protocol events through a
 `StreamMux` to keep the inference logic isolated; the end-of-file
 group exercises the path through real graphs (multi-depth
-discovery, nested `stream_events(version="v3")` calls with non-empty `parent_ns`).
+discovery, nested `stream_v2` calls with non-empty `parent_ns`).
 """
 
 from __future__ import annotations
@@ -92,16 +92,11 @@ def _arm(mux: StreamMux) -> None:
             transformer._channel._subscribed = True
 
 
-def _unstamped(items):
-    """Strip push stamps from a StreamChannel's internal buffer."""
-    return [item for _stamp, item in items]
-
-
 def _drain_lifecycle(mux: StreamMux) -> list[LifecyclePayload]:
     """Snapshot the lifecycle channel's buffer."""
     transformer = mux.transformer_by_key("lifecycle")
     assert isinstance(transformer, LifecycleTransformer)
-    return _unstamped(transformer._channel._items)
+    return list(transformer._channel._items)
 
 
 def _build_lifecycle_mux(*, scope: tuple[str, ...] = ()) -> StreamMux:
@@ -306,7 +301,7 @@ def test_protocol_event_method_is_native() -> None:
     mux = _build_lifecycle_mux()
     mux.push(_tasks_start(["agent:abc"], task_id="t1", name="tool"))
 
-    methods = {evt["method"] for evt in _unstamped(mux._events._items)}
+    methods = {evt["method"] for evt in mux._events._items}
     assert "lifecycle" in methods
     assert "custom:lifecycle" not in methods
 
@@ -317,14 +312,14 @@ def test_tasks_events_suppressed_from_main_log() -> None:
     mux.push(_tasks_start(["agent:abc"], task_id="t1", name="tool"))
     mux.push(_tasks_result([], task_id="abc", name="agent"))
 
-    methods = [evt["method"] for evt in _unstamped(mux._events._items)]
+    methods = [evt["method"] for evt in mux._events._items]
     assert "tasks" not in methods
     # Lifecycle events did make it through, though.
     assert "lifecycle" in methods
 
 
 # ---------------------------------------------------------------------------
-# End-to-end: real graphs through stream_events(version="v3")
+# End-to-end: real graphs through stream_v2
 # ---------------------------------------------------------------------------
 
 
@@ -358,10 +353,10 @@ def _make_two_level_nested() -> Any:
     return outer_b.compile()
 
 
-def test_stream_events_v3_real_graph_emits_lifecycle_at_each_depth() -> None:
+def test_stream_v2_real_graph_emits_lifecycle_at_each_depth() -> None:
     """Outer graph with two nested subgraphs surfaces lifecycle for both."""
     graph = _make_two_level_nested()
-    run = graph.stream_events({"value": "x", "items": []}, version="v3")
+    run = graph.stream_v2({"value": "x", "items": []})
 
     # Iterating the projection drives the pump and drains synthesized
     # lifecycle events at the same time.
@@ -384,17 +379,17 @@ def test_stream_events_v3_real_graph_emits_lifecycle_at_each_depth() -> None:
         )
 
 
-def test_stream_events_v3_with_nested_parent_ns_scopes_lifecycle() -> None:
-    """When `stream_events(version="v3")` is called with a non-empty checkpoint_ns in config,
+def test_stream_v2_with_nested_parent_ns_scopes_lifecycle() -> None:
+    """When `stream_v2` is called with a non-empty checkpoint_ns in config,
     `_resolve_parent_ns` returns that namespace and the registered
     `LifecycleTransformer` is constructed with `scope=parent_ns`. This
-    exercises the path that exists today purely for nested-stream_events(version="v3")
+    exercises the path that exists today purely for nested-stream_v2
     callers; the test simulates such a caller by injecting a
     checkpoint_ns into the config.
     """
     graph = _make_two_level_nested()
     config = {CONF: {CONFIG_KEY_CHECKPOINT_NS: "outer:abc"}}
-    run = graph.stream_events({"value": "x", "items": []}, config=config, version="v3")
+    run = graph.stream_v2({"value": "x", "items": []}, config=config)
 
     payloads = list(run.lifecycle)
     # Every emitted lifecycle namespace must extend the caller's scope —

libs/langgraph/tests/test_stream_messages_transformer.py

@@ -1,5 +1,5 @@
 """Tests for MessagesTransformer: protocol event routing, whole-message fallback,
-legacy v1 chunk filtering, and end-to-end via stream_events(version="v3") / astream_events(version="v3")."""
+legacy v1 chunk filtering, and end-to-end via stream_v2 / astream_v2."""
 
 from __future__ import annotations
 
@@ -26,11 +26,6 @@ from langgraph.stream.transformers import MessagesTransformer, ValuesTransformer
 TS = int(time.time() * 1000)
 
 
-def _unstamped(items):
-    """Strip push stamps from a StreamChannel's internal buffer."""
-    return [item for _stamp, item in items]
-
-
 # ---------------------------------------------------------------------------
 # Helpers
 # ---------------------------------------------------------------------------
@@ -152,7 +147,7 @@ def _lifecycle(
 def _simple_graph():
     def call_model(state: MessagesState) -> dict[str, Any]:
         model = GenericFakeChatModel(messages=iter(["hello world"]))
-        stream = model.stream_events(state["messages"], version="v3")
+        stream = model.stream_v2(state["messages"])
         return {"messages": stream.output}
 
     return (
@@ -179,7 +174,7 @@ class TestProtocolEventRouting:
             )
         )
         log.close()
-        (stream,) = _unstamped(log._items)
+        (stream,) = list(log._items)
         assert isinstance(stream, ChatModelStream)
         assert stream.message_id == "run-1"
 
@@ -188,7 +183,7 @@ class TestProtocolEventRouting:
         for evt in _lifecycle(text="hello world"):
             t.process(_proto_event(evt, run_id="run-1"))
         log.close()
-        (stream,) = _unstamped(log._items)
+        (stream,) = list(log._items)
         assert stream.done
         assert stream.output.text == "hello world"
 
@@ -211,7 +206,7 @@ class TestProtocolEventRouting:
             )
         )
         log.close()
-        assert _unstamped(log._items) == []
+        assert list(log._items) == []
 
     def test_concurrent_streams_routed_by_run_id(self) -> None:
         t, log = _make_sync_transformer()
@@ -221,7 +216,7 @@ class TestProtocolEventRouting:
             t.process(_proto_event(a, run_id="run-a"))
             t.process(_proto_event(b, run_id="run-b"))
         log.close()
-        streams = _unstamped(log._items)
+        streams = list(log._items)
         assert len(streams) == 2
         by_id = {s.message_id: s for s in streams}
         assert by_id["run-a"].output.text == "aaaa"
@@ -232,7 +227,7 @@ class TestProtocolEventRouting:
         for evt in _lifecycle(text="abcdef"):
             t.process(_proto_event(evt))
         log.close()
-        (stream,) = _unstamped(log._items)
+        (stream,) = list(log._items)
         assert "".join(stream._text_proj._deltas) == "abcdef"
 
     def test_stream_pushed_on_message_start_not_finish(self) -> None:
@@ -255,7 +250,7 @@ class TestProtocolEventRouting:
                 node="my_llm",
             )
         )
-        (stream,) = _unstamped(log._items)
+        (stream,) = list(log._items)
         assert stream.node == "my_llm"
 
 
@@ -269,7 +264,7 @@ class TestWholeMessageFallback:
         t, log = _make_sync_transformer()
         t.process(_whole_msg("the full answer"))
         log.close()
-        (stream,) = _unstamped(log._items)
+        (stream,) = list(log._items)
         assert stream.done
         assert stream.output.text == "the full answer"
 
@@ -277,7 +272,7 @@ class TestWholeMessageFallback:
         t, log = _make_sync_transformer()
         t.process(_whole_msg("full"))
         log.close()
-        (stream,) = _unstamped(log._items)
+        (stream,) = list(log._items)
         assert [e["event"] for e in stream._events] == [
             "message-start",
             "content-block-start",
@@ -323,16 +318,16 @@ class TestFiltering:
             }
         )
         log.close()
-        assert _unstamped(log._items) == []
+        assert list(log._items) == []
 
     def test_legacy_v1_chunks_ignored(self) -> None:
         # v1 AIMessageChunk tuples (from on_llm_new_token) are not streamed
-        # into this projection; callers must migrate to stream_events(version="v3").
+        # into this projection; callers must migrate to stream_v2.
         t, log = _make_sync_transformer()
         t.process(_v1_chunk("hello"))
         t.process(_v1_chunk(" world", finish=True))
         log.close()
-        assert _unstamped(log._items) == []
+        assert list(log._items) == []
 
 
 # ---------------------------------------------------------------------------
@@ -348,7 +343,7 @@ class TestLifecycle:
                 {"event": "message-start", "message_id": "run-1"}, run_id="run-1"
             )
         )
-        streams = _unstamped(log._items)
+        streams = list(log._items)
         err = RuntimeError("graph died")
         t.fail(err)
         assert t._by_run == {}
@@ -376,14 +371,14 @@ class TestAsyncMode:
         t, log = _make_async_transformer()
         for evt in _lifecycle(text="async stream"):
             t.process(_proto_event(evt))
-        assert isinstance(_unstamped(log._items)[0], AsyncChatModelStream)
+        assert isinstance(list(log._items)[0], AsyncChatModelStream)
 
     @pytest.mark.anyio
     async def test_text_projection_yields_deltas(self) -> None:
         t, log = _make_async_transformer()
         for evt in _lifecycle(text="hello world"):
             t.process(_proto_event(evt))
-        (stream,) = _unstamped(log._items)
+        (stream,) = list(log._items)
         assert isinstance(stream, AsyncChatModelStream)
         assert "".join([d async for d in stream.text]) == "hello world"
 
@@ -392,7 +387,7 @@ class TestAsyncMode:
         t, log = _make_async_transformer()
         for evt in _lifecycle(text="async"):
             t.process(_proto_event(evt))
-        (stream,) = _unstamped(log._items)
+        (stream,) = list(log._items)
         assert (await stream.output).text == "async"
 
 
@@ -424,7 +419,7 @@ class TestWireRequestMore:
         for evt in _lifecycle():
             messages_t.process(_proto_event(evt))
 
-        (stream,) = _unstamped(log._items)
+        (stream,) = list(log._items)
         assert stream._request_more is messages_t._pump_fn
 
 
@@ -450,14 +445,14 @@ class TestViaMux:
         for evt in _lifecycle(text="mux stream"):
             mux.push(_proto_event(evt))
         mux.close()
-        (stream,) = _unstamped(log._items)
+        (stream,) = list(log._items)
         assert stream.output.text == "mux stream"
 
     def test_whole_message_via_mux(self) -> None:
         t, mux, log = self._make_mux()
         mux.push(_whole_msg("result"))
         mux.close()
-        (stream,) = _unstamped(log._items)
+        (stream,) = list(log._items)
         assert stream.output.text == "result"
 
     @pytest.mark.anyio
@@ -471,24 +466,24 @@ class TestViaMux:
         for evt in _lifecycle(text="async mux"):
             await mux.apush(_proto_event(evt))
 
-        (stream,) = _unstamped(log._items)
+        (stream,) = list(log._items)
         assert (await stream.output).text == "async mux"
         await mux.aclose()
 
 
 # ---------------------------------------------------------------------------
-# End-to-end: graph → stream_events(version="v3") → run.messages (node calls stream_events)
+# End-to-end: graph → stream_v2 → run.messages (node calls stream_v2)
 # ---------------------------------------------------------------------------
 
 
 class TestEndToEnd:
-    """stream_events(version="v3") path: node calls model.stream_events() explicitly."""
+    """stream_v2 path: node calls model.stream_v2() explicitly."""
 
     def test_node_calling_stream_v2_populates_messages(self) -> None:
         model = GenericFakeChatModel(messages=iter(["hello world"]))
 
         def call_model(state: MessagesState) -> dict[str, Any]:
-            stream = model.stream_events(state["messages"], version="v3")
+            stream = model.stream_v2(state["messages"])
             return {"messages": stream.output}
 
         graph = (
@@ -499,7 +494,7 @@ class TestEndToEnd:
             .compile()
         )
 
-        run = graph.stream_events({"messages": "hi"}, version="v3")
+        run = graph.stream_v2({"messages": "hi"})
         (stream,) = list(run.messages)
         assert isinstance(stream, ChatModelStream)
         assert stream.output.text == "hello world"
@@ -509,7 +504,7 @@ class TestEndToEnd:
         model = GenericFakeChatModel(messages=iter(["streamed answer"]))
 
         def call_model(state: MessagesState) -> dict[str, Any]:
-            stream = model.stream_events(state["messages"], version="v3")
+            stream = model.stream_v2(state["messages"])
             return {"messages": stream.output}
 
         graph = (
@@ -520,7 +515,7 @@ class TestEndToEnd:
             .compile()
         )
 
-        run = graph.stream_events({"messages": "go"}, version="v3")
+        run = graph.stream_v2({"messages": "go"})
         (stream,) = list(run.messages)
         assert "".join(stream.text) == "streamed answer"
 
@@ -538,7 +533,7 @@ class TestEndToEnd:
             .compile()
         )
 
-        run = graph.stream_events({"messages": "hi"}, version="v3")
+        run = graph.stream_v2({"messages": "hi"})
         (stream,) = list(run.messages)
         assert stream.output.text == "hardcoded"
 
@@ -547,7 +542,7 @@ class TestEndToEnd:
         model = GenericFakeChatModel(messages=iter(["async answer"]))
 
         async def call_model(state: MessagesState) -> dict[str, Any]:
-            stream = await model.astream_events(state["messages"], version="v3")
+            stream = await model.astream_v2(state["messages"])
             return {"messages": await stream}
 
         graph = (
@@ -558,7 +553,7 @@ class TestEndToEnd:
             .compile()
         )
 
-        run = await graph.astream_events({"messages": "hi"}, version="v3")
+        run = await graph.astream_v2({"messages": "hi"})
         streams = [s async for s in run.messages]
         assert len(streams) == 1
         assert isinstance(streams[0], AsyncChatModelStream)
@@ -572,7 +567,7 @@ class TestEndToEnd:
         model = GenericFakeChatModel(messages=iter(["hello world"]))
 
         async def call_model(state: MessagesState) -> dict[str, Any]:
-            stream = await model.astream_events(state["messages"], version="v3")
+            stream = await model.astream_v2(state["messages"])
             return {"messages": await stream}
 
         graph = (
@@ -583,7 +578,7 @@ class TestEndToEnd:
             .compile()
         )
 
-        run = await graph.astream_events({"messages": "hi"}, version="v3")
+        run = await graph.astream_v2({"messages": "hi"})
 
         async def consume() -> list[str]:
             collected: list[str] = []
@@ -596,12 +591,12 @@ class TestEndToEnd:
 
 
 # ---------------------------------------------------------------------------
-# End-to-end: graph → stream_events(version="v3") → run.messages (node calls invoke)
+# End-to-end: graph → stream_v2 → run.messages (node calls invoke)
 # ---------------------------------------------------------------------------
 
 
 class TestEndToEndV2Invoke:
-    """Auto-routing path: stream_events(version="v3") injects CONFIG_KEY_STREAM_MESSAGES_V2,
+    """Auto-routing path: stream_v2 injects CONFIG_KEY_STREAM_MESSAGES_V2,
     causing BaseChatModel to drive the v2 protocol event generator even for
     model.invoke()."""
 
@@ -620,7 +615,7 @@ class TestEndToEndV2Invoke:
     def test_invoke_populates_messages(self) -> None:
         run = self._graph(
             GenericFakeChatModel(messages=iter(["hello world"]))
-        ).stream_events({"messages": "hi"}, version="v3")
+        ).stream_v2({"messages": "hi"})
         (stream,) = list(run.messages)
         assert isinstance(stream, ChatModelStream)
         assert stream.output.text == "hello world"
@@ -629,7 +624,7 @@ class TestEndToEndV2Invoke:
         """Iterating the stream yields the full v2 lifecycle, not v1 chunks."""
         run = self._graph(
             GenericFakeChatModel(messages=iter(["streamed answer"]))
-        ).stream_events({"messages": "go"}, version="v3")
+        ).stream_v2({"messages": "go"})
         (stream,) = list(run.messages)
 
         events = list(stream)
@@ -650,7 +645,7 @@ class TestEndToEndV2Invoke:
     def test_invoke_text_deltas_iterate(self) -> None:
         run = self._graph(
             GenericFakeChatModel(messages=iter(["delta streaming works"]))
-        ).stream_events({"messages": "hi"}, version="v3")
+        ).stream_v2({"messages": "hi"})
         (stream,) = list(run.messages)
         assert "".join(stream.text) == "delta streaming works"
 
@@ -674,7 +669,7 @@ class TestEndToEndV2Invoke:
             .compile()
         )
 
-        streams = list(graph.stream_events({"messages": "hi"}, version="v3").messages)
+        streams = list(graph.stream_v2({"messages": "hi"}).messages)
         assert len(streams) == 2
         assert {s.output.text for s in streams} == {"alpha", "beta"}
 
@@ -698,7 +693,7 @@ class TestEndToEndV2Invoke:
             .compile()
         )
 
-        run = graph.stream_events({"messages": "hi"}, version="v3")
+        run = graph.stream_v2({"messages": "hi"})
         streams = list(run.messages)
         assert len(streams) == 2
         assert streams[0].node == "streaming_node"
@@ -722,7 +717,7 @@ class TestEndToEndV2Invoke:
             .compile()
         )
 
-        run = await graph.astream_events({"messages": "hi"}, version="v3")
+        run = await graph.astream_v2({"messages": "hi"})
         streams = [s async for s in run.messages]
         assert len(streams) == 1
         assert isinstance(streams[0], AsyncChatModelStream)
@@ -737,7 +732,7 @@ class TestEndToEndV2Invoke:
 class TestDirectMessagesModeStaysV1:
     def test_direct_graph_stream_messages_yields_ai_message_chunks(self) -> None:
         """graph.stream(stream_mode="messages") must not leak v2 event dicts —
-        the v2 flag is only injected by stream_events(version="v3") / astream_events(version="v3")."""
+        the v2 flag is only injected by stream_v2 / astream_v2."""
         model = GenericFakeChatModel(messages=iter(["legacy path"]))
 
         def call_model(state: MessagesState) -> dict[str, Any]:
@@ -760,10 +755,8 @@ class TestDirectMessagesModeStaysV1:
             == "legacy path"
         )
 
-    def test_nested_graph_stream_messages_stays_v1_under_outer_stream_events_v3(
-        self,
-    ) -> None:
-        """An outer `stream_events(version="v3")` run must not flip an inner direct
+    def test_nested_graph_stream_messages_stays_v1_under_outer_stream_v2(self) -> None:
+        """An outer `stream_v2()` run must not flip an inner direct
         `stream_mode="messages"` call onto the v2 event protocol."""
         model = GenericFakeChatModel(messages=iter(["nested legacy path"]))
 
@@ -814,7 +807,7 @@ class TestDirectMessagesModeStaysV1:
             .compile()
         )
 
-        result = outer.stream_events({}, version="v3").output
+        result = outer.stream_v2({}).output
 
         assert result is not None
         assert result["saw_only_chunks"] is True

libs/langgraph/tests/test_stream_subgraph_transformer.py

@@ -4,7 +4,7 @@ Subscribes to `tasks` events and produces in-process `SubgraphRunStream`
 handles backed by mini-muxes (built via `StreamMux._make_child`). The
 synthetic-event tests isolate the inference / mini-mux wiring; the
 real-graph tests exercise the end-to-end navigation path through
-`stream_events(version="v3")`.
+`stream_v2`.
 """
 
 from __future__ import annotations
@@ -93,7 +93,7 @@ def _tasks_result(
 
 
 def _native_factories() -> list[Any]:
-    """Mirror the factory list `Pregel.stream_events(version="v3")` registers."""
+    """Mirror the factory list `Pregel.stream_v2` registers."""
     return [
         ValuesTransformer,
         MessagesTransformer,
@@ -159,13 +159,8 @@ def _subgraph_transformer(mux: StreamMux) -> SubgraphTransformer:
     return transformer
 
 
-def _unstamped(items):
-    """Strip push stamps from a StreamChannel's internal buffer."""
-    return [item for _stamp, item in items]
-
-
 def _drain_subgraphs(mux: StreamMux) -> list[SubgraphRunStream]:
-    return _unstamped(_subgraph_transformer(mux)._log._items)
+    return list(_subgraph_transformer(mux)._log._items)
 
 
 def _child_mux(handle: SubgraphRunStream | AsyncSubgraphRunStream) -> StreamMux:
@@ -174,13 +169,13 @@ def _child_mux(handle: SubgraphRunStream | AsyncSubgraphRunStream) -> StreamMux:
 
 
 def _event_items(mux: StreamMux) -> list[ProtocolEvent]:
-    return _unstamped(mux._events._items)
+    return list(mux._events._items)
 
 
 def _lifecycle_payloads(mux: StreamMux) -> list[dict[str, Any]]:
     lifecycle_t = mux.transformer_by_key("lifecycle")
     assert isinstance(lifecycle_t, LifecycleTransformer)
-    return _unstamped(lifecycle_t._channel._items)
+    return list(lifecycle_t._channel._items)
 
 
 # ---------------------------------------------------------------------------
@@ -252,7 +247,7 @@ def test_grandchild_discovered_via_child_mini_mux() -> None:
     [child_handle] = _drain_subgraphs(mux)
     assert child_handle.path == ("agent:abc",)
     # The grandchild appears on the CHILD'S subgraphs projection.
-    grandchildren = _unstamped(child_handle.subgraphs._items)
+    grandchildren = list(child_handle.subgraphs._items)
     assert len(grandchildren) == 1
     assert grandchildren[0].path == ("agent:abc", "tool:def")
 
@@ -761,10 +756,10 @@ def _make_failing_nested() -> Any:
     return outer_b.compile()
 
 
-def test_stream_events_v3_real_graph_yields_subgraph_handles() -> None:
+def test_stream_v2_real_graph_yields_subgraph_handles() -> None:
     """Iterating `run.subgraphs` yields handles for direct-child subgraphs."""
     graph = _make_two_level_nested()
-    run = graph.stream_events({"value": "x", "items": []}, version="v3")
+    run = graph.stream_v2({"value": "x", "items": []})
 
     handle_paths: list[tuple[str, ...]] = []
     final_status: dict[tuple[str, ...], str] = {}
@@ -780,10 +775,10 @@ def test_stream_events_v3_real_graph_yields_subgraph_handles() -> None:
     assert final_status[handle_paths[0]] == "completed"
 
 
-def test_stream_events_v3_grandchild_visible_on_child_handle() -> None:
+def test_stream_v2_grandchild_visible_on_child_handle() -> None:
     """Drilling into `handle.subgraphs` surfaces nested grandchildren."""
     graph = _make_two_level_nested()
-    run = graph.stream_events({"value": "x", "items": []}, version="v3")
+    run = graph.stream_v2({"value": "x", "items": []})
 
     grandchild_paths: list[tuple[str, ...]] = []
     middle_path: tuple[str, ...] | None = None
@@ -810,7 +805,7 @@ def test_subgraph_output_stops_at_own_terminal_without_draining_siblings() -> No
     inside the loop body misses its events.
     """
     graph = _make_two_sibling_subgraphs()
-    run = graph.stream_events({"value": "x", "items": []}, version="v3")
+    run = graph.stream_v2({"value": "x", "items": []})
 
     paths: list[tuple[str, ...]] = []
     second_values: list[dict[str, Any]] = []
@@ -829,7 +824,7 @@ def test_subgraph_output_stops_at_own_terminal_without_draining_siblings() -> No
 
 def test_aborted_subgraph_handle_does_not_fail_parent_forwarding() -> None:
     graph = _make_two_sibling_subgraphs()
-    run = graph.stream_events({"value": "x", "items": []}, version="v3")
+    run = graph.stream_v2({"value": "x", "items": []})
 
     seen: list[str | None] = []
     for handle in run.subgraphs:
@@ -847,7 +842,7 @@ def test_aborted_subgraph_handle_does_not_fail_parent_forwarding() -> None:
 
 def test_failed_subgraph_output_raises_terminal_error() -> None:
     graph = _make_failing_nested()
-    run = graph.stream_events({"value": "x", "items": []}, version="v3")
+    run = graph.stream_v2({"value": "x", "items": []})
 
     handle = next(iter(run.subgraphs))
     with pytest.raises(RuntimeError, match="child boom"):

libs/langgraph/tests/test_stream_v2.py

@@ -19,11 +19,9 @@ from typing_extensions import TypedDict, assert_type
 
 from langgraph._internal._constants import INTERRUPT
 from langgraph.constants import END, START
-from langgraph.errors import GraphDrained
 from langgraph.func import entrypoint
 from langgraph.graph import StateGraph
 from langgraph.graph.message import MessagesState
-from langgraph.runtime import RunControl
 from langgraph.types import (
     CheckpointPayload,
     CheckpointStreamPart,
@@ -231,32 +229,6 @@ class TestV2Stream:
         for c in chunks:
             _assert_stream_part_shape(c)
 
-    def test_stream_events_v3_accepts_control_for_drain(self) -> None:
-        class DrainState(TypedDict, total=False):
-            value: str
-            skipped: str
-
-        control = RunControl()
-
-        def first_node(state: DrainState) -> dict[str, str]:
-            control.request_drain("sigterm")
-            return {"value": "done"}
-
-        def second_node(state: DrainState) -> dict[str, str]:
-            return {"skipped": "nope"}
-
-        builder = StateGraph(DrainState)
-        builder.add_node("first", first_node)
-        builder.add_node("second", second_node)
-        builder.add_edge(START, "first")
-        builder.add_edge("first", "second")
-        builder.add_edge("second", END)
-        graph = builder.compile()
-
-        run = graph.stream_events({}, control=control, version="v3")
-        with pytest.raises(GraphDrained, match="sigterm"):
-            list(run.values)
-
     def test_subgraphs_ns(self) -> None:
         outer = _make_subgraph()
         chunks = list(
@@ -1124,7 +1096,7 @@ class TestV2ValidationErrors:
 
     _INVALID_INPUT: dict[str, Any] = {"value": [1, 2, 3], "items": []}
 
-    def test_stream_events_v3_pydantic_validation_error(self) -> None:
+    def test_stream_v2_pydantic_validation_error(self) -> None:
         """Invalid input to stream with v2 + pydantic state raises ValidationError."""
         graph = _make_pydantic_graph()
         with pytest.raises(ValidationError):

libs/langgraph/tests/test_stream_v2_e2e.py

@@ -1,9 +1,9 @@
-"""End-to-end tests exercising all stream_events(version="v3") projections together.
+"""End-to-end tests exercising all stream_v2 projections together.
 
 Each test builds a realistic graph (subgraphs, LLM calls, custom writers,
 interrupts) and verifies that every projection — values, messages, lifecycle,
 subgraphs, raw events, output, interleave — produces correct, consistent
-results through a single stream_events(version="v3") / astream_events(version="v3") run.
+results through a single stream_v2 / astream_v2 run.
 """
 
 from __future__ import annotations
@@ -218,9 +218,9 @@ class _CounterTransformer(StreamTransformer):
 
 class TestStreamV2E2ESync:
     def test_all_projections_nested_graph(self) -> None:
-        """Run a nested graph through stream_events(version="v3") and verify values + lifecycle."""
+        """Run a nested graph through stream_v2 and verify values + lifecycle."""
         graph = _make_nested_graph()
-        run = graph.stream_events({"value": "x", "items": []}, version="v3")
+        run = graph.stream_v2({"value": "x", "items": []})
 
         values_snapshots: list[dict[str, Any]] = []
         lifecycle_events: list[dict[str, Any]] = []
@@ -246,7 +246,7 @@ class TestStreamV2E2ESync:
     def test_subgraph_handles_with_drill_down(self) -> None:
         """Subgraph handles yield and support values drill-down."""
         graph = _make_nested_graph()
-        run = graph.stream_events({"value": "x", "items": []}, version="v3")
+        run = graph.stream_v2({"value": "x", "items": []})
 
         handles = []
         for handle in run.subgraphs:
@@ -270,7 +270,7 @@ class TestStreamV2E2ESync:
     def test_raw_events_have_monotonic_seq(self) -> None:
         """Raw protocol events have monotonically increasing seq numbers."""
         graph = _make_nested_graph()
-        run = graph.stream_events({"value": "x", "items": []}, version="v3")
+        run = graph.stream_v2({"value": "x", "items": []})
         events = list(run)
         assert len(events) > 0
 
@@ -285,15 +285,11 @@ class TestStreamV2E2ESync:
 
     def test_output_matches_final_values_snapshot(self) -> None:
         """output property returns the same state as the last values snapshot."""
-        run1 = _make_nested_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run1 = _make_nested_graph().stream_v2({"value": "x", "items": []})
         snapshots = list(run1.values)
         final_via_values = snapshots[-1]
 
-        run2 = _make_nested_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run2 = _make_nested_graph().stream_v2({"value": "x", "items": []})
         final_via_output = run2.output
 
         assert final_via_values == final_via_output
@@ -301,7 +297,7 @@ class TestStreamV2E2ESync:
     def test_context_manager_and_abort(self) -> None:
         """Context manager calls abort, marking the stream exhausted."""
         graph = _make_nested_graph()
-        with graph.stream_events({"value": "x", "items": []}, version="v3") as run:
+        with graph.stream_v2({"value": "x", "items": []}) as run:
             first_val = next(iter(run.values))
             assert isinstance(first_val, dict)
         assert run._exhausted is True
@@ -309,7 +305,7 @@ class TestStreamV2E2ESync:
     def test_extensions_has_all_native_keys(self) -> None:
         """Extensions dict exposes all native projection keys."""
         graph = _make_nested_graph()
-        run = graph.stream_events({"value": "x", "items": []}, version="v3")
+        run = graph.stream_v2({"value": "x", "items": []})
         _ = run.output
 
         assert "values" in run.extensions
@@ -331,7 +327,7 @@ class TestStreamV2E2EMessages:
     def test_messages_projection_from_invoke(self) -> None:
         """Messages projection captures LLM calls via model.invoke() auto-routing."""
         graph = _make_messages_graph()
-        run = graph.stream_events({"messages": "hi"}, version="v3")
+        run = graph.stream_v2({"messages": "hi"})
         streams = list(run.messages)
 
         assert len(streams) >= 1
@@ -354,7 +350,7 @@ class TestStreamV2E2EMessages:
             .compile()
         )
 
-        run = graph.stream_events({"messages": "go"}, version="v3")
+        run = graph.stream_v2({"messages": "go"})
         (stream,) = list(run.messages)
         assert "".join(stream.text) == "streamed answer"
 
@@ -372,7 +368,7 @@ class TestStreamV2E2EMessages:
             .compile()
         )
 
-        run = graph.stream_events({"messages": "hi"}, version="v3")
+        run = graph.stream_v2({"messages": "hi"})
         (stream,) = list(run.messages)
         assert stream.output.text == "hardcoded"
         assert stream.message_id == "msg-1"
@@ -380,7 +376,7 @@ class TestStreamV2E2EMessages:
     def test_root_messages_only_shows_root_scope(self) -> None:
         """Root messages projection doesn't surface subgraph-scoped messages."""
         graph = _make_messages_subgraph()
-        run = graph.stream_events({"messages": ["hi"], "done": False}, version="v3")
+        run = graph.stream_v2({"messages": ["hi"], "done": False})
         root_streams = list(run.messages)
         # The message is emitted inside the subgraph, so the root
         # messages projection (scoped to root namespace) doesn't see it.
@@ -389,7 +385,7 @@ class TestStreamV2E2EMessages:
     def test_subgraph_handle_messages_drill_down(self) -> None:
         """Drilling into subgraph handle's messages surfaces subgraph messages."""
         graph = _make_messages_subgraph()
-        run = graph.stream_events({"messages": ["hi"], "done": False}, version="v3")
+        run = graph.stream_v2({"messages": ["hi"], "done": False})
 
         found_messages = False
         for handle in run.subgraphs:
@@ -411,9 +407,8 @@ class TestStreamV2E2ECustom:
         """Custom StreamWriter events appear on the main log when a
         transformer declares the custom mode."""
         graph = _make_custom_writer_graph()
-        run = graph.stream_events(
+        run = graph.stream_v2(
             {"value": "x", "items": []},
-            version="v3",
             transformers=[_CustomPassthroughTransformer],
         )
         events = list(run)
@@ -425,7 +420,7 @@ class TestStreamV2E2ECustom:
     def test_custom_events_suppressed_without_transformer(self) -> None:
         """Without a custom-mode transformer, custom events don't flow."""
         graph = _make_custom_writer_graph()
-        run = graph.stream_events({"value": "x", "items": []}, version="v3")
+        run = graph.stream_v2({"value": "x", "items": []})
         events = list(run)
         custom = [e for e in events if e["method"] == "custom"]
         assert custom == []
@@ -433,9 +428,8 @@ class TestStreamV2E2ECustom:
     def test_custom_transformer_with_stream_channel(self) -> None:
         """A custom transformer with a StreamChannel produces extension data."""
         graph = _make_nested_graph()
-        run = graph.stream_events(
+        run = graph.stream_v2(
             {"value": "x", "items": []},
-            version="v3",
             transformers=[_CounterTransformer],
         )
 
@@ -450,9 +444,8 @@ class TestStreamV2E2ECustom:
     def test_custom_channel_events_on_main_log(self) -> None:
         """StreamChannel auto-forward injects custom:<name> events into the main log."""
         graph = _make_nested_graph()
-        run = graph.stream_events(
+        run = graph.stream_v2(
             {"value": "x", "items": []},
-            version="v3",
             transformers=[_CounterTransformer],
         )
         events = list(run)
@@ -471,7 +464,7 @@ class TestStreamV2E2EInterrupt:
         """Interrupted run has correct flags and interrupt payloads."""
         graph = _make_interrupt_graph()
         config: dict[str, Any] = {"configurable": {"thread_id": "int-1"}}
-        run = graph.stream_events({"value": "x", "items": []}, config, version="v3")
+        run = graph.stream_v2({"value": "x", "items": []}, config)
 
         output = run.output
         assert output is not None
@@ -484,7 +477,7 @@ class TestStreamV2E2EInterrupt:
         """Values snapshots captured before the interrupt reflect partial state."""
         graph = _make_interrupt_graph()
         config: dict[str, Any] = {"configurable": {"thread_id": "int-2"}}
-        run = graph.stream_events({"value": "x", "items": []}, config, version="v3")
+        run = graph.stream_v2({"value": "x", "items": []}, config)
 
         snapshots = list(run.values)
         assert len(snapshots) >= 1
@@ -501,14 +494,14 @@ class TestStreamV2E2EErrors:
     def test_subgraph_error_propagates_through_output(self) -> None:
         """Error in a subgraph propagates through output."""
         graph = _make_error_subgraph()
-        run = graph.stream_events({"value": "x", "items": []}, version="v3")
+        run = graph.stream_v2({"value": "x", "items": []})
 
         with pytest.raises(ValueError, match="subgraph explosion"):
             _ = run.output
 
     def test_subgraph_error_propagates_through_raw_events(self) -> None:
         graph = _make_error_subgraph()
-        run = graph.stream_events({"value": "x", "items": []}, version="v3")
+        run = graph.stream_v2({"value": "x", "items": []})
 
         with pytest.raises(ValueError, match="subgraph explosion"):
             list(run)
@@ -516,7 +509,7 @@ class TestStreamV2E2EErrors:
     def test_error_subgraph_handle_status(self) -> None:
         """Subgraph handle surfaces the error status."""
         graph = _make_error_subgraph()
-        run = graph.stream_events({"value": "x", "items": []}, version="v3")
+        run = graph.stream_v2({"value": "x", "items": []})
 
         handle = next(iter(run.subgraphs))
         with pytest.raises(RuntimeError, match="subgraph explosion"):
@@ -536,7 +529,7 @@ class TestStreamV2E2EAsync:
     async def test_all_projections_async(self) -> None:
         """Async run exercises values projection."""
         graph = _make_nested_graph()
-        run = await graph.astream_events({"value": "x", "items": []}, version="v3")
+        run = await graph.astream_v2({"value": "x", "items": []})
 
         values_snapshots = [s async for s in run.values]
         assert len(values_snapshots) >= 1
@@ -547,7 +540,7 @@ class TestStreamV2E2EAsync:
     async def test_async_output(self) -> None:
         """Async output returns the final state."""
         graph = _make_nested_graph()
-        run = await graph.astream_events({"value": "x", "items": []}, version="v3")
+        run = await graph.astream_v2({"value": "x", "items": []})
         output = await run.output()
         assert output is not None
         assert output["value"] == "x_routed_processed"
@@ -557,7 +550,7 @@ class TestStreamV2E2EAsync:
     async def test_async_raw_events(self) -> None:
         """Async raw event iteration yields well-formed ProtocolEvents."""
         graph = _make_nested_graph()
-        run = await graph.astream_events({"value": "x", "items": []}, version="v3")
+        run = await graph.astream_v2({"value": "x", "items": []})
         events = [e async for e in run]
         assert len(events) > 0
         seqs = [e["seq"] for e in events]
@@ -579,7 +572,7 @@ class TestStreamV2E2EAsync:
             .compile()
         )
 
-        run = await graph.astream_events({"messages": "hi"}, version="v3")
+        run = await graph.astream_v2({"messages": "hi"})
         streams = [s async for s in run.messages]
         assert len(streams) >= 1
         for s in streams:
@@ -590,9 +583,7 @@ class TestStreamV2E2EAsync:
         """Async interrupted run has correct flags."""
         graph = _make_interrupt_graph()
         config: dict[str, Any] = {"configurable": {"thread_id": "async-int-1"}}
-        run = await graph.astream_events(
-            {"value": "x", "items": []}, config, version="v3"
-        )
+        run = await graph.astream_v2({"value": "x", "items": []}, config)
 
         output = await run.output()
         assert output is not None
@@ -602,14 +593,14 @@ class TestStreamV2E2EAsync:
     async def test_async_error_propagation(self) -> None:
         """Async error from subgraph propagates through output."""
         graph = _make_error_subgraph()
-        run = await graph.astream_events({"value": "x", "items": []}, version="v3")
+        run = await graph.astream_v2({"value": "x", "items": []})
         with pytest.raises(ValueError, match="subgraph explosion"):
             await run.output()
 
     async def test_async_context_manager(self) -> None:
         """Async context manager calls abort on exit."""
         graph = _make_nested_graph()
-        run = await graph.astream_events({"value": "x", "items": []}, version="v3")
+        run = await graph.astream_v2({"value": "x", "items": []})
         async with run:
             _ = await anext(aiter(run.values))
         assert run._exhausted is True
@@ -617,7 +608,7 @@ class TestStreamV2E2EAsync:
     async def test_async_extensions_present(self) -> None:
         """Async run has all native extensions."""
         graph = _make_nested_graph()
-        run = await graph.astream_events({"value": "x", "items": []}, version="v3")
+        run = await graph.astream_v2({"value": "x", "items": []})
         _ = await run.output()
         assert "values" in run.extensions
         assert "messages" in run.extensions
@@ -627,9 +618,8 @@ class TestStreamV2E2EAsync:
     async def test_async_custom_transformer(self) -> None:
         """Async custom transformer with StreamChannel works."""
         graph = _make_nested_graph()
-        run = await graph.astream_events(
+        run = await graph.astream_v2(
             {"value": "x", "items": []},
-            version="v3",
             transformers=[_CounterTransformer],
         )
         assert "counter" in run.extensions
@@ -649,7 +639,7 @@ class TestStreamV2E2ECombined:
     def test_interleave_all_native_projections(self) -> None:
         """Interleave values + messages + lifecycle without deadlock."""
         graph = _make_nested_graph()
-        run = graph.stream_events({"value": "x", "items": []}, version="v3")
+        run = graph.stream_v2({"value": "x", "items": []})
 
         seen_names: set[str] = set()
         for name, _item in run.interleave("values", "messages", "lifecycle"):
@@ -677,9 +667,8 @@ class TestStreamV2E2ECombined:
                 return True
 
         graph = _make_nested_graph()
-        run = graph.stream_events(
+        run = graph.stream_v2(
             {"value": "x", "items": []},
-            version="v3",
             transformers=[_CounterTransformer, TagTransformer],
         )
 
@@ -733,7 +722,7 @@ class TestStreamV2E2ECombined:
             .compile()
         )
 
-        run = outer.stream_events({"items": []}, version="v3")
+        run = outer.stream_v2({"items": []})
         handles = []
         for handle in run.subgraphs:
             list(handle.values)
@@ -751,17 +740,13 @@ class TestStreamV2E2ECombined:
 
     def test_lifecycle_matches_subgraph_handles(self) -> None:
         """Lifecycle events and subgraph handles agree on discovered subgraphs."""
-        run1 = _make_nested_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run1 = _make_nested_graph().stream_v2({"value": "x", "items": []})
         handle_paths: list[tuple[str, ...]] = []
         for handle in run1.subgraphs:
             list(handle.values)
             handle_paths.append(handle.path)
 
-        run2 = _make_nested_graph().stream_events(
-            {"value": "x", "items": []}, version="v3"
-        )
+        run2 = _make_nested_graph().stream_v2({"value": "x", "items": []})
         lifecycle = list(run2.lifecycle)
 
         started_ns = [
@@ -788,9 +773,8 @@ class TestStreamV2E2ECombined:
             .compile()
         )
 
-        run = graph.stream_events(
+        run = graph.stream_v2(
             {"messages": "hi"},
-            version="v3",
             transformers=[_CounterTransformer],
         )
 

libs/langgraph/uv.lock

@@ -1348,7 +1348,7 @@ wheels = [
 
 [[package]]
 name = "langchain-core"
-version = "1.4.0a2"
+version = "1.3.2"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "jsonpatch" },
@@ -1361,26 +1361,26 @@ dependencies = [
     { name = "typing-extensions" },
     { name = "uuid-utils" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/3c/93/68bafa047f8e1770d0cf0f61d6c70889f1dec42ef6bd263540d916c421b9/langchain_core-1.4.0a2.tar.gz", hash = "sha256:b723c7961b615c7f2180ce2bcf352fdad8247bc51a60adecd3d97088235c120d", size = 916486, upload-time = "2026-05-01T15:02:19.029Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/a8/03/7219502e8ca728d65eb44d7a3eb60239230742a70dbfc9241b9bfd61c4ab/langchain_core-1.3.2.tar.gz", hash = "sha256:fd7a50b2f28ba561fd9d7f5d2760bc9e06cf00cdf820a3ccafe88a94ffa8d5b7", size = 911813, upload-time = "2026-04-24T15:49:23.699Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/4e/8e/933e0ba7ba0430ce264e36b178d581b255239ad45093872483142d93478c/langchain_core-1.4.0a2-py3-none-any.whl", hash = "sha256:a5c689f8404357df797120c012da7704144a953b2ae18f258df263301e7badd5", size = 546297, upload-time = "2026-05-01T15:02:17.731Z" },
+    { url = "https://files.pythonhosted.org/packages/7d/d5/8fa4431007cbb7cfed7590f4d6a5dea3ad724f4174d248f6642ef5ce7d05/langchain_core-1.3.2-py3-none-any.whl", hash = "sha256:d44a66127f9f8db735bdfd0ab9661bccb47a97113cfd3f2d89c74864422b7274", size = 542390, upload-time = "2026-04-24T15:49:21.991Z" },
 ]
 
 [[package]]
 name = "langchain-protocol"
-version = "0.0.14"
+version = "0.0.12"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "typing-extensions" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/05/bf/efb5e2ed832e4d6d45590e25a9e5191986b291b543bc6a807b48bee070b0/langchain_protocol-0.0.14.tar.gz", hash = "sha256:bc1e8553122e6ede310280462d5813023a172ff2785ccbbdec54d43f3a15e5f2", size = 5862, upload-time = "2026-04-29T16:40:18.657Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/5c/51/1157009b6f94e6e58be58fa8b620187d657909a8b36a6bf5b0c52a2711f6/langchain_protocol-0.0.12.tar.gz", hash = "sha256:5e14c434290a705c9510fdb1a83ecf7561a5e6e0dfd053930ade80dba069269f", size = 6408, upload-time = "2026-04-25T01:05:01.489Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/c2/e9/06c47ecb2aff08f83dfa30058da3bf86be64862c19569043ed5331bbeecd/langchain_protocol-0.0.14-py3-none-any.whl", hash = "sha256:ffc35089779bd8ca217015180cef5e660fc3b074efdaa0f2e95df73583f1a047", size = 6984, upload-time = "2026-04-29T16:40:17.841Z" },
+    { url = "https://files.pythonhosted.org/packages/95/82/3431e3061c917439589fa88a6b23c9bc0e154cba0f05d2e895a68c76ff74/langchain_protocol-0.0.12-py3-none-any.whl", hash = "sha256:402b61f42d4139692528cf37226c367bb6efc8ff8165b29380accb0abfece7b2", size = 6639, upload-time = "2026-04-25T01:05:00.487Z" },
 ]
 
 [[package]]
 name = "langgraph"
-version = "1.2.0a3"
+version = "1.1.10"
 source = { editable = "." }
 dependencies = [
     { name = "langchain-core" },
@@ -1452,7 +1452,7 @@ test = [
 
 [package.metadata]
 requires-dist = [
-    { name = "langchain-core", specifier = ">=1.4.0a2,<2" },
+    { name = "langchain-core", specifier = ">=1.3.2,<2" },
     { name = "langgraph-checkpoint", editable = "../checkpoint" },
     { name = "langgraph-prebuilt", editable = "../prebuilt" },
     { name = "langgraph-sdk", editable = "../sdk-py" },
@@ -1561,7 +1561,7 @@ wheels = [
 
 [[package]]
 name = "langgraph-checkpoint"
-version = "4.1.0a3"
+version = "4.0.3"
 source = { editable = "../checkpoint" }
 dependencies = [
     { name = "langchain-core" },
@@ -1609,7 +1609,7 @@ test = [
 
 [[package]]
 name = "langgraph-checkpoint-postgres"
-version = "3.1.0a3"
+version = "3.0.5"
 source = { editable = "../checkpoint-postgres" }
 dependencies = [
     { name = "langgraph-checkpoint" },

libs/prebuilt/langgraph/prebuilt/tool_node.py

@@ -44,7 +44,7 @@ import inspect
 import json
 from collections.abc import Awaitable, Callable
 from copy import copy, deepcopy
-from dataclasses import dataclass, field, replace
+from dataclasses import dataclass, replace
 from types import UnionType
 from typing import (
     TYPE_CHECKING,
@@ -1723,9 +1723,9 @@ class ToolRuntime(_DirectlyInjectedToolArg, Generic[ContextT, StateT]):
     context: ContextT
     config: RunnableConfig
     stream_writer: StreamWriter
+    tools: list[BaseTool]
     tool_call_id: str | None
     store: BaseStore | None
-    tools: list[BaseTool] = field(default_factory=list)
     execution_info: ExecutionInfo | None = None
     server_info: ServerInfo | None = None
 

libs/prebuilt/tests/test_tool_call_transformer.py

@@ -30,11 +30,6 @@ from langgraph.prebuilt._tool_call_stream import ToolCallStream
 TS = int(time.time() * 1000)
 
 
-def _unstamped(items):
-    """Strip push stamps from a StreamChannel's internal buffer."""
-    return [item for _stamp, item in items]
-
-
 def _tool_event(
     event: str,
     tool_call_id: str,
@@ -100,7 +95,7 @@ class TestToolCallTransformerUnit:
                 input={"text": "hi"},
             )
         )
-        handles = _unstamped(transformer._log._items)
+        handles = list(transformer._log._items)
         assert len(handles) == 1
         h = handles[0]
         assert isinstance(h, ToolCallStream)
@@ -116,7 +111,7 @@ class TestToolCallTransformerUnit:
         mux.push(_tool_event("tool-output-delta", "tc1", delta="a"))
         mux.push(_tool_event("tool-output-delta", "tc1", delta="b"))
         stream = transformer._active["tc1"]
-        assert _unstamped(stream._output_deltas._items) == ["a", "b"]
+        assert list(stream._output_deltas._items) == ["a", "b"]
 
     def test_finish_closes_stream(self) -> None:
         mux, transformer = _mux()
@@ -147,17 +142,14 @@ class TestToolCallTransformerUnit:
         mux.push(_tool_event("tool-output-delta", "a", delta="A1"))
         mux.push(_tool_event("tool-output-delta", "b", delta="B1"))
         mux.push(_tool_event("tool-output-delta", "a", delta="A2"))
-        assert _unstamped(transformer._active["a"]._output_deltas._items) == [
-            "A1",
-            "A2",
-        ]
-        assert _unstamped(transformer._active["b"]._output_deltas._items) == ["B1"]
+        assert list(transformer._active["a"]._output_deltas._items) == ["A1", "A2"]
+        assert list(transformer._active["b"]._output_deltas._items) == ["B1"]
 
     def test_tools_event_passes_through_main_log(self) -> None:
         mux, transformer = _mux()
         _subscribe(mux._events)
         mux.push(_tool_event("tool-started", "tc1", tool_name="echo"))
-        kept = [e for e in _unstamped(mux._events._items) if e["method"] == "tools"]
+        kept = [e for e in mux._events._items if e["method"] == "tools"]
         assert len(kept) == 1
 
 
@@ -202,9 +194,7 @@ class TestToolCallTransformerEndToEnd:
             }
 
         graph = _build_graph(caller, [streamer])
-        run = graph.stream_events(
-            {"messages": []}, transformers=[ToolCallTransformer], version="v3"
-        )
+        run = graph.stream_v2({"messages": []}, transformers=[ToolCallTransformer])
 
         tool_calls: list[ToolCallStream] = []
         for tc in run.tool_calls:
@@ -240,13 +230,11 @@ class TestToolCallTransformerEndToEnd:
         # Without ToolCallTransformer, no tool_calls projection is
         # exposed and no `tools` events flow through (required_stream_modes
         # omits it).
-        run_no_tc = graph.stream_events({"messages": []}, version="v3")
+        run_no_tc = graph.stream_v2({"messages": []})
         assert "tool_calls" not in run_no_tc._mux.extensions  # type: ignore[attr-defined]
 
         # With ToolCallTransformer, the projection is present.
-        run = graph.stream_events(
-            {"messages": []}, transformers=[ToolCallTransformer], version="v3"
-        )
+        run = graph.stream_v2({"messages": []}, transformers=[ToolCallTransformer])
         assert "tool_calls" in run._mux.extensions  # type: ignore[attr-defined]
         # Drain so the run closes cleanly.
         list(run.tool_calls)
@@ -273,8 +261,8 @@ class TestToolCallTransformerEndToEnd:
             }
 
         graph = _build_graph(caller, [astreamer])
-        run = await graph.astream_events(
-            {"messages": []}, version="v3", transformers=[ToolCallTransformer]
+        run = await graph.astream_v2(
+            {"messages": []}, transformers=[ToolCallTransformer]
         )
 
         collected: list[ToolCallStream] = []
@@ -303,9 +291,7 @@ class TestToolCallTransformerEndToEnd:
             }
 
         graph = _build_graph(caller, [boom])
-        run = graph.stream_events(
-            {"messages": []}, transformers=[ToolCallTransformer], version="v3"
-        )
+        run = graph.stream_v2({"messages": []}, transformers=[ToolCallTransformer])
 
         collected: list[ToolCallStream] = []
         with pytest.raises(ValueError, match="nope"):

libs/prebuilt/tests/test_tool_node.py

@@ -2016,19 +2016,6 @@ async def test_tool_node_inject_runtime_dynamic_tool_via_wrap_tool_call_async()
     assert tool_message.tool_call_id == "call_dynamic_2"
 
 
-def test_tool_runtime_defaults_tools_to_empty_list() -> None:
-    runtime = ToolRuntime(
-        state={},
-        context=None,
-        config={},
-        stream_writer=lambda *args, **kwargs: None,
-        tool_call_id=None,
-        store=None,
-    )
-
-    assert runtime.tools == []
-
-
 def test_tool_runtime_forwards_execution_info_server_info_and_tools() -> None:
     """Test that execution_info, server_info, and tools are forwarded from Runtime to ToolRuntime."""
     from langgraph.runtime import ExecutionInfo, ServerInfo

libs/prebuilt/uv.lock

@@ -249,7 +249,7 @@ wheels = [
 
 [[package]]
 name = "langchain-core"
-version = "1.4.0a2"
+version = "1.3.2"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "jsonpatch" },
@@ -262,26 +262,26 @@ dependencies = [
     { name = "typing-extensions" },
     { name = "uuid-utils" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/3c/93/68bafa047f8e1770d0cf0f61d6c70889f1dec42ef6bd263540d916c421b9/langchain_core-1.4.0a2.tar.gz", hash = "sha256:b723c7961b615c7f2180ce2bcf352fdad8247bc51a60adecd3d97088235c120d", size = 916486, upload-time = "2026-05-01T15:02:19.029Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/a8/03/7219502e8ca728d65eb44d7a3eb60239230742a70dbfc9241b9bfd61c4ab/langchain_core-1.3.2.tar.gz", hash = "sha256:fd7a50b2f28ba561fd9d7f5d2760bc9e06cf00cdf820a3ccafe88a94ffa8d5b7", size = 911813, upload-time = "2026-04-24T15:49:23.699Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/4e/8e/933e0ba7ba0430ce264e36b178d581b255239ad45093872483142d93478c/langchain_core-1.4.0a2-py3-none-any.whl", hash = "sha256:a5c689f8404357df797120c012da7704144a953b2ae18f258df263301e7badd5", size = 546297, upload-time = "2026-05-01T15:02:17.731Z" },
+    { url = "https://files.pythonhosted.org/packages/7d/d5/8fa4431007cbb7cfed7590f4d6a5dea3ad724f4174d248f6642ef5ce7d05/langchain_core-1.3.2-py3-none-any.whl", hash = "sha256:d44a66127f9f8db735bdfd0ab9661bccb47a97113cfd3f2d89c74864422b7274", size = 542390, upload-time = "2026-04-24T15:49:21.991Z" },
 ]
 
 [[package]]
 name = "langchain-protocol"
-version = "0.0.14"
+version = "0.0.12"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "typing-extensions" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/05/bf/efb5e2ed832e4d6d45590e25a9e5191986b291b543bc6a807b48bee070b0/langchain_protocol-0.0.14.tar.gz", hash = "sha256:bc1e8553122e6ede310280462d5813023a172ff2785ccbbdec54d43f3a15e5f2", size = 5862, upload-time = "2026-04-29T16:40:18.657Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/5c/51/1157009b6f94e6e58be58fa8b620187d657909a8b36a6bf5b0c52a2711f6/langchain_protocol-0.0.12.tar.gz", hash = "sha256:5e14c434290a705c9510fdb1a83ecf7561a5e6e0dfd053930ade80dba069269f", size = 6408, upload-time = "2026-04-25T01:05:01.489Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/c2/e9/06c47ecb2aff08f83dfa30058da3bf86be64862c19569043ed5331bbeecd/langchain_protocol-0.0.14-py3-none-any.whl", hash = "sha256:ffc35089779bd8ca217015180cef5e660fc3b074efdaa0f2e95df73583f1a047", size = 6984, upload-time = "2026-04-29T16:40:17.841Z" },
+    { url = "https://files.pythonhosted.org/packages/95/82/3431e3061c917439589fa88a6b23c9bc0e154cba0f05d2e895a68c76ff74/langchain_protocol-0.0.12-py3-none-any.whl", hash = "sha256:402b61f42d4139692528cf37226c367bb6efc8ff8165b29380accb0abfece7b2", size = 6639, upload-time = "2026-04-25T01:05:00.487Z" },
 ]
 
 [[package]]
 name = "langgraph"
-version = "1.2.0a3"
+version = "1.1.10"
 source = { editable = "../langgraph" }
 dependencies = [
     { name = "langchain-core" },
@@ -294,7 +294,7 @@ dependencies = [
 
 [package.metadata]
 requires-dist = [
-    { name = "langchain-core", specifier = ">=1.4.0a2,<2" },
+    { name = "langchain-core", specifier = ">=1.3.2,<2" },
     { name = "langgraph-checkpoint", editable = "../checkpoint" },
     { name = "langgraph-prebuilt", editable = "." },
     { name = "langgraph-sdk", editable = "../sdk-py" },
@@ -365,7 +365,7 @@ test = [
 
 [[package]]
 name = "langgraph-checkpoint"
-version = "4.1.0a3"
+version = "4.0.3"
 source = { editable = "../checkpoint" }
 dependencies = [
     { name = "langchain-core" },
@@ -413,7 +413,7 @@ test = [
 
 [[package]]
 name = "langgraph-checkpoint-postgres"
-version = "3.1.0a3"
+version = "3.0.5"
 source = { editable = "../checkpoint-postgres" }
 dependencies = [
     { name = "langgraph-checkpoint" },

libs/sdk-py/uv.lock

@@ -266,7 +266,7 @@ wheels = [
 
 [[package]]
 name = "langchain-core"
-version = "1.4.0a2"
+version = "1.3.2"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "jsonpatch" },
@@ -279,26 +279,26 @@ dependencies = [
     { name = "typing-extensions" },
     { name = "uuid-utils" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/3c/93/68bafa047f8e1770d0cf0f61d6c70889f1dec42ef6bd263540d916c421b9/langchain_core-1.4.0a2.tar.gz", hash = "sha256:b723c7961b615c7f2180ce2bcf352fdad8247bc51a60adecd3d97088235c120d", size = 916486, upload-time = "2026-05-01T15:02:19.029Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/a8/03/7219502e8ca728d65eb44d7a3eb60239230742a70dbfc9241b9bfd61c4ab/langchain_core-1.3.2.tar.gz", hash = "sha256:fd7a50b2f28ba561fd9d7f5d2760bc9e06cf00cdf820a3ccafe88a94ffa8d5b7", size = 911813, upload-time = "2026-04-24T15:49:23.699Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/4e/8e/933e0ba7ba0430ce264e36b178d581b255239ad45093872483142d93478c/langchain_core-1.4.0a2-py3-none-any.whl", hash = "sha256:a5c689f8404357df797120c012da7704144a953b2ae18f258df263301e7badd5", size = 546297, upload-time = "2026-05-01T15:02:17.731Z" },
+    { url = "https://files.pythonhosted.org/packages/7d/d5/8fa4431007cbb7cfed7590f4d6a5dea3ad724f4174d248f6642ef5ce7d05/langchain_core-1.3.2-py3-none-any.whl", hash = "sha256:d44a66127f9f8db735bdfd0ab9661bccb47a97113cfd3f2d89c74864422b7274", size = 542390, upload-time = "2026-04-24T15:49:21.991Z" },
 ]
 
 [[package]]
 name = "langchain-protocol"
-version = "0.0.14"
+version = "0.0.12"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "typing-extensions" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/05/bf/efb5e2ed832e4d6d45590e25a9e5191986b291b543bc6a807b48bee070b0/langchain_protocol-0.0.14.tar.gz", hash = "sha256:bc1e8553122e6ede310280462d5813023a172ff2785ccbbdec54d43f3a15e5f2", size = 5862, upload-time = "2026-04-29T16:40:18.657Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/5c/51/1157009b6f94e6e58be58fa8b620187d657909a8b36a6bf5b0c52a2711f6/langchain_protocol-0.0.12.tar.gz", hash = "sha256:5e14c434290a705c9510fdb1a83ecf7561a5e6e0dfd053930ade80dba069269f", size = 6408, upload-time = "2026-04-25T01:05:01.489Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/c2/e9/06c47ecb2aff08f83dfa30058da3bf86be64862c19569043ed5331bbeecd/langchain_protocol-0.0.14-py3-none-any.whl", hash = "sha256:ffc35089779bd8ca217015180cef5e660fc3b074efdaa0f2e95df73583f1a047", size = 6984, upload-time = "2026-04-29T16:40:17.841Z" },
+    { url = "https://files.pythonhosted.org/packages/95/82/3431e3061c917439589fa88a6b23c9bc0e154cba0f05d2e895a68c76ff74/langchain_protocol-0.0.12-py3-none-any.whl", hash = "sha256:402b61f42d4139692528cf37226c367bb6efc8ff8165b29380accb0abfece7b2", size = 6639, upload-time = "2026-04-25T01:05:00.487Z" },
 ]
 
 [[package]]
 name = "langgraph"
-version = "1.2.0a3"
+version = "1.1.10"
 source = { editable = "../langgraph" }
 dependencies = [
     { name = "langchain-core" },
@@ -311,7 +311,7 @@ dependencies = [
 
 [package.metadata]
 requires-dist = [
-    { name = "langchain-core", specifier = ">=1.4.0a2,<2" },
+    { name = "langchain-core", specifier = ">=1.3.2,<2" },
     { name = "langgraph-checkpoint", editable = "../checkpoint" },
     { name = "langgraph-prebuilt", editable = "../prebuilt" },
     { name = "langgraph-sdk", editable = "." },
@@ -382,7 +382,7 @@ test = [
 
 [[package]]
 name = "langgraph-checkpoint"
-version = "4.1.0a3"
+version = "4.0.3"
 source = { editable = "../checkpoint" }
 dependencies = [
     { name = "langchain-core" },