[
https://issues.apache.org/jira/browse/BEAM-11715?focusedWorklogId=549983&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-549983
]
ASF GitHub Bot logged work on BEAM-11715:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 09/Feb/21 00:35
Start Date: 09/Feb/21 00:35
Worklog Time Spent: 10m
Work Description: robertwb commented on pull request #13877:
URL: https://github.com/apache/beam/pull/13877#issuecomment-775564361
> OK, still running into a bunch of issues:
>
> * FnApiRunner doesn't actually use the sub-transforms in CombinePerKey. So
I can't use it to test that the subtransforms in the packed CombinePerKey are
correct. Is there any runner that actually uses the subtransforms that I can
use for testing?
You could disable combiner lifting in th FnApiRunner to test things out.
> * DataflowRunner has a
[CombineValuesPTransformOverride](https://github.com/apache/beam/blob/0078bb35ba4aef9410d681d8b4e2c16d9f56433d/sdks/python/apache_beam/runners/dataflow/ptransform_overrides.py#L155)
that replaces the composite `CombineValues` with a primitive
`CombineValuesReplacement`. We need a _another_ code path to mimic this
replacement.
Ugh.
Maybe we could simplify things by only supporting portable job submission?
----------------------------------------------------------------
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: 549983)
Time Spent: 10h 10m (was: 10h)
> Combiner packing creates an incorrect proto
> -------------------------------------------
>
> Key: BEAM-11715
> URL: https://issues.apache.org/jira/browse/BEAM-11715
> Project: Beam
> Issue Type: Bug
> Components: sdk-py-core
> Reporter: Chamikara Madhusanka Jayalath
> Assignee: Yifan Mai
> Priority: P1
> Fix For: 2.28.0
>
> Time Spent: 10h 10m
> Remaining Estimate: 0h
>
> Seems like optimization creates a CombinePerKey transform that does not have
> any sub-transforms. We should fix this by preserving the old structure for
> CombinePerKey (GBK+CombineValues composite).
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
