> How about this, we continue with the vote as is, and keep the discussion on the SourceFunction in Jira or a separate thread.
Sure, but I just want to mention two important things here before we switch over to [1]: >Given that eliminating the removal of SourceFunction was proposed 10 days - This is not the case - here is the original discussion thread from a year ago [2] - Here are the vote results [3] agreeing to do exactly the following: > This proposition implies marking the SourceFunction interface > itself as @Deprecated + redirecting to the FLIP-27 Source API > right away, without waiting for all the subtasks to be completed. This was all already discussed and agreed upon, the decision to revert the change comes a bit as a surprise. @Leonard If we follow the agreed-upon and voted path and do not revert [1], all the formalities get fulfilled. The rest of the discussion can happen in [1] [1] https://issues.apache.org/jira/browse/FLINK-28046 [2] https://lists.apache.org/thread/d6cwqw9b3105wcpdkwq7rr4s7x4ywqr9 [3] https://lists.apache.org/thread/hrpsddgz65hjvhjozhg72s0wsmxz145p Thanks, Alex On Fri, 21 Jul 2023 at 04:49, Leonard Xu <xbjt...@gmail.com> wrote: > Hi Alexander > > > I see your concerns regarding the complexity of migration, but we still > > have one year to address them. > > Not only the complexity of migration, but also we lack migration path for > now. We have to deprecate SourceFunction/SinkFunction in 1.18 which feature > freeze date is 2023/07/24 CEST If we want to remove > SourceFunction/SinkFunction in 2.0, remove a public API requires at least > two minor versions. There’re many subtasks to finish before we can mark > them as deprecated[1], I think we have no time to finish these subtasks > this week(hint: not this year). That’s why we suggested removing the > SourceFunction/SinkFunction item from must to have. > > Best, > Leonard > [1]https://issues.apache.org/jira/browse/FLINK-28045 > >
