You are viewing a plain text version of this content. The canonical link for it is here.

Posted to github@beam.apache.org by GitBox <gi...@apache.org> on 2021/02/17 05:39:02 UTC

[GitHub] [beam] kennknowles commented on pull request #13998: [BEAM-11812] Fix ULR validate runner test failure caused by merging I…

kennknowles commented on pull request #13998:
URL: https://github.com/apache/beam/pull/13998#issuecomment-780313629


   See #14001 which might be related. CC @robertwb here too.
   
   In portability, I wonder if `InvalidWindows` is obsolete. The only reason for it was to crash in this case:
   
   1. User sets keys and chooses session windowing
   3. User groups/combines by session
   4. User maps the outputs to some new keys
   5. User groups again
   
   In the original model, the second grouping would do merging again, and probably be something random that they did not intend. Or anyhow that is how I remember the reasoning.
   
   Now I think `MergeStatus.ALREADY_MERGED` would allow better behavior. In the second group by key the windows would be treated as non-merging.
   
   Was there any discussion of `MergeStatus` that I may have missed?


----------------------------------------------------------------
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:
users@infra.apache.org