lukecwik commented on a change in pull request #15634:
URL: https://github.com/apache/beam/pull/15634#discussion_r719864996
##########
File path: sdks/python/apache_beam/runners/worker/operations.py
##########
@@ -710,13 +710,12 @@ def start(self):
def process(self, o):
# type: (WindowedValue) -> None
with self.scoped_process_state:
- delayed_application = self.dofn_runner.process(o)
- if delayed_application:
+ delayed_applications = self.dofn_runner.process(o)
+ if delayed_applications:
assert self.execution_context is not None
- # TODO(BEAM-77746): there's disagreement between subclasses
- # of DoFnRunner over the return type annotations of process().
- self.execution_context.delayed_applications.append(
- (self, delayed_application)) # type: ignore[arg-type]
+ for delayed_application in delayed_applications:
Review comment:
I would have expected something that is being returned here would break
splitting since we would be returning appending a
`Tuple[operations.DoOperation, Iterable[common.SplitResultResidual]]` instead
of `Tuple[operations.DoOperation, common.SplitResultResidual]`. Maybe we are
unpacking the iterable somewhere inexplicably.
We have several SDF based validates runner tests, would be worthwhile to see
what they say for Dataflow.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
