gaoyunhaii commented on pull request #14820:
URL: https://github.com/apache/flink/pull/14820#issuecomment-859508853


   Hi Dawid, very thanks for the thoughts! 
   
   I think the analysis on the `EndOfDataEvent` is exactly right, currently 
when a task get triggered, all the user records must be processed and there are 
only possible barriers remaining. 
   
   For the current approach, I think it avoids the source / non-source split 
since it unified the logic of triggering checkpoints by distinguish the 
different situation via `getNotFinishedChannels()`, right ? And 
`getCheckpointBarrierHandler()` seems need to be an abstract method that source 
return null while non-source returns the actual `CheckpointBarrier`? 
   
   I think I am also agree with this method. I'll update the PR according to 
this approach. very thanks for the suggestions!


-- 
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]


Reply via email to