codope commented on pull request #4201: URL: https://github.com/apache/hudi/pull/4201#issuecomment-985380602
> > @YannByron Why revert it? I've added a fix for [HUDI-2909](https://issues.apache.org/jira/browse/HUDI-2909) : #4203 Can you please review? > > the key consideration is: keygen related change should ensure backward-compatible for existing user tables. The previous PR #3944 solves inconsistent behavior but may break for cases like reported in [HUDI-2909](https://issues.apache.org/jira/browse/HUDI-2909). > > The revert was meant to keep things unchanged for users. Inconsistency comes from enabling row-writer, which can be mitigated by disabling row-writer. We were thinking rolling out with a new flag for the future new row-writing feature to solve inconsistency. The future release notes can help migrate user tables if applicable. > > Looks like #4203 makes a further fix. is there any compatibility concern if rolled out ? If so, would it be easy for user to migrate and adapt to new keygen's behavior? and how to clarify it in the release notes. Just some points to consider. > > @codope @leesf @nsivabalan I see. #4203 builds on #3944 . I think it makes sense to revert for now. To address the compatibility issue would it be better to make row-writer path also produce long value of timestamp instead of having users to migrate tables in a future release? -- 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]
