[
https://issues.apache.org/jira/browse/BEAM-1630?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16275252#comment-16275252
]
ASF GitHub Bot commented on BEAM-1630:
--------------------------------------
chamikaramj commented on a change in pull request #4064: [BEAM-1630] Adds
support for processing Splittable DoFns using DirectRunner.
URL: https://github.com/apache/beam/pull/4064#discussion_r152439187
##########
File path: sdks/python/apache_beam/runners/direct/transform_evaluator.py
##########
@@ -529,16 +530,19 @@ def start_bundle(self):
self._counter_factory = counters.CounterFactory()
# TODO(aaltay): Consider storing the serialized form as an optimization.
- dofn = pickler.loads(pickler.dumps(transform.dofn))
+ dofn = transform.dofn
Review comment:
I ran into issues since we are piggybacking StepContext and SDF DoFnInvoker
objects in the DoFn. I added an option to disable this pickling only for the
SDF case.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
> Add Splittable DoFn to Python SDK
> ---------------------------------
>
> Key: BEAM-1630
> URL: https://issues.apache.org/jira/browse/BEAM-1630
> Project: Beam
> Issue Type: Improvement
> Components: sdk-py-core
> Reporter: Chamikara Jayalath
> Assignee: Chamikara Jayalath
>
> Splittable DoFn [1] is currently being implemented for Java SDK [2]. We
> should add this to Python SDK as well.
> Following document proposes an API for this.
> https://docs.google.com/document/d/1h_zprJrOilivK2xfvl4L42vaX4DMYGfH1YDmi-s_ozM/edit?usp=sharing
> [1] https://s.apache.org/splittable-do-fn
> [2]
> https://lists.apache.org/thread.html/0ce61ac162460a149d5c93cdface37cc383f8030fe86ca09e5699b18@%3Cdev.beam.apache.org%3E
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
