[
https://issues.apache.org/jira/browse/BEAM-11887?focusedWorklogId=559675&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-559675
]
ASF GitHub Bot logged work on BEAM-11887:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 02/Mar/21 00:57
Start Date: 02/Mar/21 00:57
Worklog Time Spent: 10m
Work Description: y1chi commented on pull request #14120:
URL: https://github.com/apache/beam/pull/14120#issuecomment-788465020
So IIUC, SortingFlinkCombineRunner merges IntevalWindow based on whether two
windows are intersecting with each other, and does not bother to invoke the
mergeWindows function from the WindowFn. This seems inconsistent with other
runners such as DirectRunner or DataflowRunner, where we rely on the merge
implementation in WindowFn without making an assumption. So the change I made
here is simply removing the usage of SortingFlinkCombineRunner since its
assumption seems unsafe.
----------------------------------------------------------------
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]
Issue Time Tracking
-------------------
Worklog Id: (was: 559675)
Time Spent: 1h 10m (was: 1h)
> testMergingCustomWindowsWithoutCustomWindowTypes failing on Flink VR
> --------------------------------------------------------------------
>
> Key: BEAM-11887
> URL: https://issues.apache.org/jira/browse/BEAM-11887
> Project: Beam
> Issue Type: Bug
> Components: test-failures
> Reporter: Kenneth Knowles
> Assignee: Yichi Zhang
> Priority: P1
> Time Spent: 1h 10m
> Remaining Estimate: 0h
>
> https://ci-beam.apache.org/view/PostCommit/job/beam_PostCommit_Java_ValidatesRunner_Flink/8618/testReport/junit/org.apache.beam.sdk.transforms.windowing/WindowTest/testMergingCustomWindowsWithoutCustomWindowTypes/
> Maybe the category of test needs to be sickbayed. But maybe the new test is
> not quite right? Better look into it.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)