hive HIVE-26496: Populating bucketId in the constructor instead of parse m…

What changes were proposed in this pull request?

Why are the changes needed?

Does this PR introduce any user-facing change?

How was this patch tested?

Sep 19 '22 13:09 difin

Is the jira reference correct; I don't understand wha'ts the connection with HIVE-26496.

@zabetak The connection to HIVE-26496 is that you raised the problem around assignment of bucketId in the previous PR for HIVE-26496 after the PR was merged and I couldn't fix it as part of that PR. Denys advised that the first PR was fine and the OrcSplit class was already badly designed with the way how bucketId was set and suggested to create a separate PR to fix it.

Sep 20 '22 14:09 difin

I completely forgot that I commented under the PR for HIVE-26496; sorry about that. Many thanks for the clarification and preparing the follow-up!

The fact that bucketId is not assigned in the constructor is not something that was caused by https://github.com/apache/hive/pull/3559; it was there before that so it is not really tight to HIVE-26496. This is a beneficial refactoring that makes sense with or without HIVE-26496 so it could have a JIRA on its own.

Other than that I think it is fine to commit under the same JIRA id although I don't see this pattern very often in the project.