dbos 1.3.0a5__tar.gz → 1.3.0a7__tar.gz

{dbos-1.3.0a5 → dbos-1.3.0a7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dbos
-Version: 1.3.0a5
+Version: 1.3.0a7
 Summary: Ultra-lightweight durable execution in Python
 Author-Email: "DBOS, Inc." <contact@dbos.dev>
 License: MIT

{dbos-1.3.0a5 → dbos-1.3.0a7}/dbos/_client.py

@@ -141,6 +141,11 @@ class DBOSClient:
             "priority": options.get("priority"),
         }
 
+        inputs: WorkflowInputs = {
+            "args": args,
+            "kwargs": kwargs,
+        }
+
         status: WorkflowStatusInternal = {
             "workflow_uuid": workflow_id,
             "status": WorkflowStatusString.ENQUEUED.value,
@@ -169,16 +174,11 @@ class DBOSClient:
                 if enqueue_options_internal["priority"] is not None
                 else 0
             ),
-        }
-
-        inputs: WorkflowInputs = {
-            "args": args,
-            "kwargs": kwargs,
+            "inputs": _serialization.serialize_args(inputs),
         }
 
         self._sys_db.init_workflow(
             status,
-            _serialization.serialize_args(inputs),
             max_recovery_attempts=None,
         )
         return workflow_id
@@ -237,6 +237,7 @@ class DBOSClient:
             "workflow_deadline_epoch_ms": None,
             "deduplication_id": None,
             "priority": 0,
+            "inputs": _serialization.serialize_args({"args": (), "kwargs": {}}),
         }
         with self._sys_db.engine.begin() as conn:
             self._sys_db._insert_workflow_status(

{dbos-1.3.0a5 → dbos-1.3.0a7}/dbos/_core.py

@@ -252,6 +252,10 @@ def _init_workflow(
             raise DBOSNonExistentWorkflowError(wfid)
         return get_status_result
 
+    # If we have a class name, the first arg is the instance and do not serialize
+    if class_name is not None:
+        inputs = {"args": inputs["args"][1:], "kwargs": inputs["kwargs"]}
+
     # Initialize a workflow status object from the context
     status: WorkflowStatusInternal = {
         "workflow_uuid": wfid,
@@ -291,16 +295,12 @@ def _init_workflow(
             if enqueue_options is not None
             else 0
         ),
+        "inputs": _serialization.serialize_args(inputs),
     }
 
-    # If we have a class name, the first arg is the instance and do not serialize
-    if class_name is not None:
-        inputs = {"args": inputs["args"][1:], "kwargs": inputs["kwargs"]}
-
     # Synchronously record the status and inputs for workflows
     wf_status, workflow_deadline_epoch_ms = dbos._sys_db.init_workflow(
         status,
-        _serialization.serialize_args(inputs),
         max_recovery_attempts=max_recovery_attempts,
     )
 
@@ -441,16 +441,13 @@ def execute_workflow_by_id(dbos: "DBOS", workflow_id: str) -> "WorkflowHandle[An
     status = dbos._sys_db.get_workflow_status(workflow_id)
     if not status:
         raise DBOSRecoveryError(workflow_id, "Workflow status not found")
-    inputs = dbos._sys_db.get_workflow_inputs(workflow_id)
-    if not inputs:
-        raise DBOSRecoveryError(workflow_id, "Workflow inputs not found")
+    inputs = _serialization.deserialize_args(status["inputs"])
     wf_func = dbos._registry.workflow_info_map.get(status["name"], None)
     if not wf_func:
         raise DBOSWorkflowFunctionNotFoundError(
             workflow_id, "Workflow function not found"
         )
     with DBOSContextEnsure():
-        ctx = assert_current_dbos_context()
         # If this function belongs to a configured class, add that class instance as its first argument
         if status["config_name"] is not None:
             config_name = status["config_name"]

{dbos-1.3.0a5 → dbos-1.3.0a7}/dbos/_dbos_config.py

@@ -91,7 +91,7 @@ class ConfigFile(TypedDict, total=False):
     Data structure containing the DBOS Configuration.
 
     This configuration data is typically loaded from `dbos-config.yaml`.
-    See `https://docs.dbos.dev/api-reference/configuration`_
+    See `https://docs.dbos.dev/python/reference/configuration#dbos-configuration-file`
 
     Attributes:
         name (str): Application name

dbos-1.3.0a7/dbos/_migrations/versions/d994145b47b6_consolidate_inputs.py

@@ -0,0 +1,30 @@
+"""consolidate_inputs
+
+Revision ID: d994145b47b6
+Revises: 66478e1b95e5
+Create Date: 2025-05-23 08:09:15.515009
+
+"""
+
+from typing import Sequence, Union
+
+import sqlalchemy as sa
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision: str = "d994145b47b6"
+down_revision: Union[str, None] = "66478e1b95e5"
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    op.add_column(
+        "workflow_status",
+        sa.Column("inputs", sa.Text(), nullable=True),
+        schema="dbos",
+    )
+
+
+def downgrade() -> None:
+    op.drop_column("workflow_status", "inputs", schema="dbos")

{dbos-1.3.0a5 → dbos-1.3.0a7}/dbos/_schemas/system_database.py

@@ -58,6 +58,7 @@ class SystemSchema:
         Column("workflow_deadline_epoch_ms", BigInteger, nullable=True),
         Column("started_at_epoch_ms", BigInteger(), nullable=True),
         Column("deduplication_id", Text(), nullable=True),
+        Column("inputs", Text()),
         Column("priority", Integer(), nullable=False, server_default=text("'0'::int")),
         Index("workflow_status_created_at_index", "created_at"),
         Index("workflow_status_executor_id_index", "executor_id"),
@@ -88,21 +89,6 @@ class SystemSchema:
         PrimaryKeyConstraint("workflow_uuid", "function_id"),
     )
 
-    workflow_inputs = Table(
-        "workflow_inputs",
-        metadata_obj,
-        Column(
-            "workflow_uuid",
-            Text,
-            ForeignKey(
-                "workflow_status.workflow_uuid", onupdate="CASCADE", ondelete="CASCADE"
-            ),
-            primary_key=True,
-            nullable=False,
-        ),
-        Column("inputs", Text, nullable=False),
-    )
-
     notifications = Table(
         "notifications",
         metadata_obj,

{dbos-1.3.0a5 → dbos-1.3.0a7}/dbos/_sys_db.py

@@ -140,6 +140,8 @@ class WorkflowStatusInternal(TypedDict):
     deduplication_id: Optional[str]
     # Priority of the workflow on the queue, starting from 1 ~ 2,147,483,647. Default 0 (highest priority).
     priority: int
+    # Serialized workflow inputs
+    inputs: str
 
 
 class EnqueueOptionsInternal(TypedDict):
@@ -462,6 +464,7 @@ class SystemDatabase:
                 workflow_deadline_epoch_ms=status["workflow_deadline_epoch_ms"],
                 deduplication_id=status["deduplication_id"],
                 priority=status["priority"],
+                inputs=status["inputs"],
             )
             .on_conflict_do_update(
                 index_elements=["workflow_uuid"],
@@ -642,9 +645,6 @@ class SystemDatabase:
         status = self.get_workflow_status(original_workflow_id)
         if status is None:
             raise Exception(f"Workflow {original_workflow_id} not found")
-        inputs = self.get_workflow_inputs(original_workflow_id)
-        if inputs is None:
-            raise Exception(f"Workflow {original_workflow_id} not found")
 
         with self.engine.begin() as c:
             # Create an entry for the forked workflow with the same
@@ -666,13 +666,7 @@ class SystemDatabase:
                     authenticated_roles=status["authenticated_roles"],
                     assumed_role=status["assumed_role"],
                     queue_name=INTERNAL_QUEUE_NAME,
-                )
-            )
-            # Copy the original workflow's inputs into the forked workflow
-            c.execute(
-                pg.insert(SystemSchema.workflow_inputs).values(
-                    workflow_uuid=forked_workflow_id,
-                    inputs=_serialization.serialize_args(inputs),
+                    inputs=status["inputs"],
                 )
             )
 
@@ -732,6 +726,7 @@ class SystemDatabase:
                     SystemSchema.workflow_status.c.workflow_timeout_ms,
                     SystemSchema.workflow_status.c.deduplication_id,
                     SystemSchema.workflow_status.c.priority,
+                    SystemSchema.workflow_status.c.inputs,
                 ).where(SystemSchema.workflow_status.c.workflow_uuid == workflow_uuid)
             ).fetchone()
             if row is None:
@@ -758,6 +753,7 @@ class SystemDatabase:
                 "workflow_timeout_ms": row[15],
                 "deduplication_id": row[16],
                 "priority": row[17],
+                "inputs": row[18],
             }
             return status
 
@@ -788,53 +784,6 @@ class SystemDatabase:
                     pass  # CB: I guess we're assuming the WF will show up eventually.
             time.sleep(1)
 
-    def _update_workflow_inputs(
-        self, workflow_uuid: str, inputs: str, conn: sa.Connection
-    ) -> None:
-        if self._debug_mode:
-            raise Exception("called update_workflow_inputs in debug mode")
-
-        cmd = (
-            pg.insert(SystemSchema.workflow_inputs)
-            .values(
-                workflow_uuid=workflow_uuid,
-                inputs=inputs,
-            )
-            .on_conflict_do_update(
-                index_elements=["workflow_uuid"],
-                set_=dict(workflow_uuid=SystemSchema.workflow_inputs.c.workflow_uuid),
-            )
-            .returning(SystemSchema.workflow_inputs.c.inputs)
-        )
-
-        row = conn.execute(cmd).fetchone()
-        if row is not None and row[0] != inputs:
-            # In a distributed environment, scheduled workflows are enqueued multiple times with slightly different timestamps
-            if not workflow_uuid.startswith("sched-"):
-                dbos_logger.warning(
-                    f"Workflow {workflow_uuid} has been called multiple times with different inputs"
-                )
-            # TODO: actually changing the input
-
-        return
-
-    @db_retry()
-    def get_workflow_inputs(
-        self, workflow_uuid: str
-    ) -> Optional[_serialization.WorkflowInputs]:
-        with self.engine.begin() as c:
-            row = c.execute(
-                sa.select(SystemSchema.workflow_inputs.c.inputs).where(
-                    SystemSchema.workflow_inputs.c.workflow_uuid == workflow_uuid
-                )
-            ).fetchone()
-            if row is None:
-                return None
-            inputs: _serialization.WorkflowInputs = _serialization.deserialize_args(
-                row[0]
-            )
-            return inputs
-
     def get_workflows(self, input: GetWorkflowsInput) -> List[WorkflowStatus]:
         """
         Retrieve a list of workflows result and inputs based on the input criteria. The result is a list of external-facing workflow status objects.
@@ -855,15 +804,11 @@ class SystemDatabase:
             SystemSchema.workflow_status.c.updated_at,
             SystemSchema.workflow_status.c.application_version,
             SystemSchema.workflow_status.c.application_id,
-            SystemSchema.workflow_inputs.c.inputs,
+            SystemSchema.workflow_status.c.inputs,
             SystemSchema.workflow_status.c.output,
             SystemSchema.workflow_status.c.error,
             SystemSchema.workflow_status.c.workflow_deadline_epoch_ms,
             SystemSchema.workflow_status.c.workflow_timeout_ms,
-        ).join(
-            SystemSchema.workflow_inputs,
-            SystemSchema.workflow_status.c.workflow_uuid
-            == SystemSchema.workflow_inputs.c.workflow_uuid,
         )
         if input.sort_desc:
             query = query.order_by(SystemSchema.workflow_status.c.created_at.desc())
@@ -953,39 +898,31 @@ class SystemDatabase:
         """
         Retrieve a list of queued workflows result and inputs based on the input criteria. The result is a list of external-facing workflow status objects.
         """
-        query = (
-            sa.select(
-                SystemSchema.workflow_status.c.workflow_uuid,
-                SystemSchema.workflow_status.c.status,
-                SystemSchema.workflow_status.c.name,
-                SystemSchema.workflow_status.c.recovery_attempts,
-                SystemSchema.workflow_status.c.config_name,
-                SystemSchema.workflow_status.c.class_name,
-                SystemSchema.workflow_status.c.authenticated_user,
-                SystemSchema.workflow_status.c.authenticated_roles,
-                SystemSchema.workflow_status.c.assumed_role,
-                SystemSchema.workflow_status.c.queue_name,
-                SystemSchema.workflow_status.c.executor_id,
-                SystemSchema.workflow_status.c.created_at,
-                SystemSchema.workflow_status.c.updated_at,
-                SystemSchema.workflow_status.c.application_version,
-                SystemSchema.workflow_status.c.application_id,
-                SystemSchema.workflow_inputs.c.inputs,
-                SystemSchema.workflow_status.c.output,
-                SystemSchema.workflow_status.c.error,
-                SystemSchema.workflow_status.c.workflow_deadline_epoch_ms,
-                SystemSchema.workflow_status.c.workflow_timeout_ms,
-            )
-            .join(
-                SystemSchema.workflow_inputs,
-                SystemSchema.workflow_status.c.workflow_uuid
-                == SystemSchema.workflow_inputs.c.workflow_uuid,
-            )
-            .where(
-                sa.and_(
-                    SystemSchema.workflow_status.c.queue_name.isnot(None),
-                    SystemSchema.workflow_status.c.status.in_(["ENQUEUED", "PENDING"]),
-                )
+        query = sa.select(
+            SystemSchema.workflow_status.c.workflow_uuid,
+            SystemSchema.workflow_status.c.status,
+            SystemSchema.workflow_status.c.name,
+            SystemSchema.workflow_status.c.recovery_attempts,
+            SystemSchema.workflow_status.c.config_name,
+            SystemSchema.workflow_status.c.class_name,
+            SystemSchema.workflow_status.c.authenticated_user,
+            SystemSchema.workflow_status.c.authenticated_roles,
+            SystemSchema.workflow_status.c.assumed_role,
+            SystemSchema.workflow_status.c.queue_name,
+            SystemSchema.workflow_status.c.executor_id,
+            SystemSchema.workflow_status.c.created_at,
+            SystemSchema.workflow_status.c.updated_at,
+            SystemSchema.workflow_status.c.application_version,
+            SystemSchema.workflow_status.c.application_id,
+            SystemSchema.workflow_status.c.inputs,
+            SystemSchema.workflow_status.c.output,
+            SystemSchema.workflow_status.c.error,
+            SystemSchema.workflow_status.c.workflow_deadline_epoch_ms,
+            SystemSchema.workflow_status.c.workflow_timeout_ms,
+        ).where(
+            sa.and_(
+                SystemSchema.workflow_status.c.queue_name.isnot(None),
+                SystemSchema.workflow_status.c.status.in_(["ENQUEUED", "PENDING"]),
             )
         )
         if input["sort_desc"]:
@@ -1895,7 +1832,6 @@ class SystemDatabase:
     def init_workflow(
         self,
         status: WorkflowStatusInternal,
-        inputs: str,
         *,
         max_recovery_attempts: Optional[int],
     ) -> tuple[WorkflowStatuses, Optional[int]]:
@@ -1906,7 +1842,6 @@ class SystemDatabase:
             wf_status, workflow_deadline_epoch_ms = self._insert_workflow_status(
                 status, conn, max_recovery_attempts=max_recovery_attempts
             )
-            self._update_workflow_inputs(status["workflow_uuid"], inputs, conn)
         return wf_status, workflow_deadline_epoch_ms
 
     def check_connection(self) -> None:

{dbos-1.3.0a5 → dbos-1.3.0a7}/pyproject.toml

@@ -27,7 +27,7 @@ dependencies = [
 ]
 requires-python = ">=3.9"
 readme = "README.md"
-version = "1.3.0a5"
+version = "1.3.0a7"
 
 [project.license]
 text = "MIT"

{dbos-1.3.0a5 → dbos-1.3.0a7}/tests/test_queue.py

@@ -574,7 +574,7 @@ def test_worker_concurrency_with_n_dbos_instances(dbos: DBOS) -> None:
 
 
 # Test error cases where we have duplicated workflows starting with the same workflow ID.
-def test_duplicate_workflow_id(dbos: DBOS, caplog: pytest.LogCaptureFixture) -> None:
+def test_duplicate_workflow_id(dbos: DBOS) -> None:
     wfid = str(uuid.uuid4())
 
     @DBOS.workflow()
@@ -606,17 +606,11 @@ def test_duplicate_workflow_id(dbos: DBOS, caplog: pytest.LogCaptureFixture) ->
             DBOS.sleep(0.1)
             return self.config_name + ":" + var1
 
-    original_propagate = logging.getLogger("dbos").propagate
-    caplog.set_level(logging.WARNING, "dbos")
-    logging.getLogger("dbos").propagate = True
-
     with SetWorkflowID(wfid):
         origHandle = DBOS.start_workflow(test_workflow, "abc")
         # The second one will generate a warning message but no error.
         test_dup_workflow()
 
-    assert "Multiple workflows started in the same SetWorkflowID block." in caplog.text
-
     # It's okay to call the same workflow with the same ID again.
     with SetWorkflowID(wfid):
         same_handle = DBOS.start_workflow(test_workflow, "abc")
@@ -661,25 +655,17 @@ def test_duplicate_workflow_id(dbos: DBOS, caplog: pytest.LogCaptureFixture) ->
     with SetWorkflowID(wfid):
         handle = queue.enqueue(test_workflow, "abc")
     assert handle.get_result() == "abc"
-    assert "Workflow already exists in queue" in caplog.text
 
-    # Call with a different input would generate a warning, but still use the recorded input.
+    # Call with a different input still uses the recorded input.
     with SetWorkflowID(wfid):
         res = test_workflow("def")
         # We want to see the warning message, but the result is non-deterministic
         # TODO: in the future, we may want to always use the recorded inputs.
         assert res == "abc" or res == "def"
-    assert (
-        f"Workflow {wfid} has been called multiple times with different inputs"
-        in caplog.text
-    )
 
     assert origHandle.get_result() == "abc"
     assert same_handle.get_result() == "abc"
 
-    # Reset logging
-    logging.getLogger("dbos").propagate = original_propagate
-
 
 def test_queue_recovery(dbos: DBOS) -> None:
     step_counter: int = 0

{dbos-1.3.0a5 → dbos-1.3.0a7}/tests/test_schema_migration.py

@@ -20,10 +20,6 @@ def test_systemdb_migration(dbos: DBOS) -> None:
         result = connection.execute(sql)
         assert result.fetchall() == []
 
-        sql = SystemSchema.workflow_inputs.select()
-        result = connection.execute(sql)
-        assert result.fetchall() == []
-
         sql = SystemSchema.operation_outputs.select()
         result = connection.execute(sql)
         assert result.fetchall() == []