[
https://issues.apache.org/jira/browse/FLINK-34551?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17831657#comment-17831657
]
Matthias Pohl commented on FLINK-34551:
---------------------------------------
The intention of this ticket came from FLINK-34227 where I wanted to add logic
for retrying forever. I managed to split the
{{retrySuccessfulOperationWithDelay}} in FLINK-34227 in a way now that I didn't
generate too much additional redundant code. I created FLINK-34551 as a
follow-up anyway because I noticed that {{retrySuccessfulOperationWithDelay}}
and {{retryOperation}} share some common logic and that we could improve the
way how these methods decide on which executor to run the {{operation}} on
(scheduledExecutor vs calling thread).
Your current proposal has still redundant code. We would need to iterate over
the change a bit more and discuss the contract of these methods in more detail.
But unfortunately, I am gone for quite a bit soon. So, I would not be able to
help you. Additionally, it's not a high-priority task right. I'm wondering
whether we should unassign the task again. I want to avoid that you spend time
on it and then get stuck because of missing feedback from my side.
I should have considered it yesterday already. Sorry for that.
> Align retry mechanisms of FutureUtils
> -------------------------------------
>
> Key: FLINK-34551
> URL: https://issues.apache.org/jira/browse/FLINK-34551
> Project: Flink
> Issue Type: Technical Debt
> Components: API / Core
> Affects Versions: 1.20.0
> Reporter: Matthias Pohl
> Assignee: Kumar Mallikarjuna
> Priority: Major
> Labels: pull-request-available
>
> The retry mechanisms of FutureUtils include quite a bit of redundant code
> which makes it hard to understand and to extend. The logic should be aligned
> properly.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
