Thesharing commented on PR #22686:
URL: https://github.com/apache/flink/pull/22686#issuecomment-1582122677

   > Thanks @huwh for creating this PR. I have a different idea:
   > 
   > We can create the `TaskDeploymentDescriptorFactory` outside ExecutionGraph 
and pass it in through the ctor of ExecutionGraph. 
The`TaskDeploymentDescriptorFactory` provides a 
`createDeploymentDescriptor(Execution execution, ...)` method to create the 
deployment descriptor for an execution.
   > 
   > In this way, the job/graph level configurations such as 
`partitionLocationConstraint` and `nonFinishedHybridPartitionShouldBeUnknown` 
should be passed into the `TaskDeploymentDescriptorFactory` instead of 
`ExecutionGraph`.
   > 
   > WDYT ? @Thesharing @reswqa
   
   It's a great idea. Currently `TaskDeploymentDescriptorFactory` is created 
for each execution. But it involves several job-level configurations. This is a 
bit confusing.
   
   However, this change would introduce a lot of code changes. Maybe we'll need 
a small design doc to show the scope affected by this refactoring job.


-- 
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.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to