ASF GitHub Bot logged work on BEAM-4549:

                Author: ASF GitHub Bot
            Created on: 12/Jun/18 23:24
            Start Date: 12/Jun/18 23:24
    Worklog Time Spent: 10m 
      Work Description: robertwb commented on a change in pull request #5623: 
[BEAM-4549] Use per-pipeline unique ids for side inputs in DataflowRunner
URL: https://github.com/apache/beam/pull/5623#discussion_r194917486

 File path: sdks/python/apache_beam/runners/dataflow/dataflow_runner.py
 @@ -587,10 +592,10 @@ def run_ParDo(self, transform_node):
     si_labels = {}
     full_label_counts = defaultdict(int)
     lookup_label = lambda side_pval: si_labels[side_pval]
-    for ix, side_pval in enumerate(transform_node.side_inputs):
+    for side_pval in transform_node.side_inputs:
       assert isinstance(side_pval, AsSideInput)
       step_name = 'SideInput-' + self._get_unique_step_name()
-      si_label = 'side%d' % ix
+      si_label = self._get_unique_side_input_name()
 Review comment:
   As I mentioned, it'd be cleaner to let this be step name + ix rather than 
introducing global state. 

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:

Issue Time Tracking

    Worklog Id:     (was: 111335)
    Time Spent: 0.5h  (was: 20m)

> Streaming pipelines with multiple side inputs fail on DataflowRunner
> --------------------------------------------------------------------
>                 Key: BEAM-4549
>                 URL: https://issues.apache.org/jira/browse/BEAM-4549
>             Project: Beam
>          Issue Type: Bug
>          Components: sdk-py-core
>            Reporter: Charles Chen
>            Assignee: Charles Chen
>            Priority: Major
>          Time Spent: 0.5h
>  Remaining Estimate: 0h
> Streaming pipelines with multiple side inputs currently fail on 
> DataflowRunner.

This message was sent by Atlassian JIRA

Reply via email to