ctubbsii commented on pull request #2293: URL: https://github.com/apache/accumulo/pull/2293#issuecomment-951376349
> Waiting to backport to 1.10 will delay getting it pushed to the clusters currently losing files. If the changes are so substantial that it will be that difficult to apply to 2.1.0-SNAPSHOT, and that difficulty is enough to cause a delay, then it is even *more* important that we get it right in 2.1 first, so we don't create churn by applying one implementation to 1.10, a more well-vetted different implementation in 2.1, and then needing to backport the final implementation back to 1.10. I'm not suggesting 2.1 be released first... only that we have a relatively concrete idea of what the destination looks like, so that we know the 1.10 implementation is moving in that direction and not away from it. What I'd like to avoid is something like the debacle with https://issues.apache.org/jira/browse/ACCUMULO-4086; there, we had a patch donated for an older version that wasn't thoroughly tested or code reviewed by us. They donated it to the project, where it got a lot of code review and polishing before being applied to 2.0. The implementation was improved through that process, but the end result was different enough that the original authors would have had a bit of a headache migrating on upgrade if they had be running using the original patch. So, we need to make sure that the implementation aligns with the direction the project is headed, rather than where it has been, to avoid those kinds of user headaches. That said, if a user wants to run with a pre-release patched version of 1.10, before we've polished it in 2.x, that's their prerogative. But, I would like to have some confidence we're not applying something in 1.10.2 that is radically different than the direction we're heading with 2.1. -- 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]
