lukecwik commented on a change in pull request #12880:
URL: https://github.com/apache/beam/pull/12880#discussion_r492353774
##########
File path: sdks/python/apache_beam/runners/dataflow/dataflow_runner.py
##########
@@ -430,6 +430,11 @@ def _check_for_unsupported_fnapi_features(self,
pipeline_proto):
components.coders[windowing_strategy.window_coder_id].spec.urn,
windowing_strategy.window_fn.urn))
+ def _adjust_types_for_dataflow(self, pipeline):
Review comment:
_adjust_pipeline_for_dataflow_v2?
##########
File path: sdks/python/apache_beam/runners/dataflow/dataflow_runner.py
##########
@@ -488,12 +497,15 @@ def run_pipeline(self, pipeline, options):
# Cross language transform require using a pipeline object constructed
# from the full pipeline proto to make sure that expanded version of
# external transforms are reflected in the Pipeline job graph.
+ # TODO(chamikara): remove following pipeline and pipeline proto
recreation
+ # after portable job submission path is fully in place.
from apache_beam import Pipeline
pipeline = Pipeline.from_runner_api(
self.proto_pipeline,
pipeline.runner,
options,
allow_proto_holders=True)
+ self._adjust_types_for_dataflow(pipeline)
Review comment:
Why do the type encodings not survive the round trip (pipeline -> proto
-> pipeline)?
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
