dawidwys commented on issue #9433: [FLINK-13708] [table-planner-blink] 
transformations should be cleared after execution in blink planner
URL: https://github.com/apache/flink/pull/9433#issuecomment-525695899
 
 
   I get that we should change sth. I am trying to understand if we need the 
whole buffering business in executors. I understand now that we need it in 
BatchExecutor. But I really see no reason why Blink StreamExecutor should 
behave differently from Flink's.
   
   Ad. 1 & 2 The problem is not in how the Executor is implemented (at least in 
the Flink planner), but because of the TableSource interface. Which always adds 
transformation to a StreamExecutionEnvironment. Generally speaking `explain` 
should be stateless.
   
   Just to sum it up a bit. I think changes in this PR are fine for now to 
improve the `BatchExecutor`. However, unless you have a good reason for it, and 
I haven't heard one so far, I would really prefer that the `StreamExecutor` 
does not buffer the transformations, because in the end I think we should aim 
for all the Executors to be stateless.
   
   
   

----------------------------------------------------------------
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]


With regards,
Apache Git Services

Reply via email to