zhengruifeng commented on PR #58165:
URL: https://github.com/apache/spark/pull/58165#issuecomment-5363727688

   Thanks, this is a good point. I think there are three different backport 
flows here:
   
   1. Happy path: we merge a PR to master and backport it in the same
      `merge_spark_pr.py` invocation. In this case the script sees all merged 
refs
      together and can set all Fix Versions correctly.
   
   2. The master PR was merged first, and later we decide to backport the same
      merged PR to more branches using `merge_spark_pr.py` backport mode. In 
this
      case the JIRA is already resolved, but the later backport merge still 
needs
      to add the newly inferred Fix Versions. This is the case fixed by
      84a309d91f702b2434e8525ff36e77d88f6a61cc / SPARK-58754.
   
   3. The PR cannot be backported cleanly while merging to master because of 
code
      conflicts, so we merge the master PR first and then open a separate PR
      targeting the release branch with the conflict resolution. This is 
becoming
      more common as branches diverge; I hit this twice this week. In that case
      the branch PR is a normal open PR, not the "already merged PR backport 
mode"
      from case 2, but the linked JIRA is already resolved from the master merge
      and we still need to add the release-branch Fix Version while merging the
      branch PR. PR #58147 -> #58184 is one example. Another is #58129 -> 
#58150:
      the master-side PR was merged first, then a separate release-branch PR was
      needed because the backport could not be completed directly from the 
original
      merge flow.
   
   So I agree we should not blindly update every resolved/closed JIRA. But 
checking
   for empty Fix Versions is too narrow, because cases 2 and 3 both need to add
   additional Fix Versions to an already resolved issue that may already have 
the
   master Fix Version.
   
   One related issue is that branch-4.x-only merges currently do not infer the 
4.x
   Fix Version unless master is also in `merge_branches`, so I may need to 
adjust
   the inference for separate release-branch PRs too.
   


-- 
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]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to