Hi, I split the blockers [1] from the nice-to-haves [2] and added some missing items. >From [1], the one about support for the ExternallyInducedSource [3] is debatable - AFAIK, it is only used by Pravega, which is not an officially-supported connector. This can, arguably, be something we could hope for them to contribute to the 2.x release line.
[1] https://issues.apache.org/jira/browse/FLINK-28045 [2] https://issues.apache.org/jira/browse/FLINK-32692 [3] https://issues.apache.org/jira/browse/FLINK-28051 On Wed, 26 Jul 2023 at 17:59, Jing Ge <j...@ververica.com.invalid> wrote: > Hi > > I agree with Konstantin that we should have a todo list to provide a clear > picture of when and how to deprecate the SinkFunction. Please let me check > if I can prepare a dedicated thread for it, since I got some feedback and > hints from previous discussions. > > Best regards, > Jing > > > On Wed, Jul 26, 2023 at 3:26 PM Konstantin Knauf <kna...@apache.org> > wrote: > > > Hi everyone, > > > > I'd just like to add that we also said, that we would continue the > > discussion to come up and agree on a list of concrete blockers for the > > removal of SourceFunction, so that don't need to have the same discussion > > again in half a year. And while we are add it, we should do the same > thing > > for SinkFunction. > > > > Best, > > > > Konstantin > > > > Am Mi., 26. Juli 2023 um 03:35 Uhr schrieb Xintong Song < > > tonysong...@gmail.com>: > > > > > Thanks Leonard for driving this, and thanks everyone for the > discussion. > > > The back-and-force reflects the importance and complexity around this > > > topic. Glad to see we finally reached consensus. > > > > > > Best, > > > > > > Xintong > > > > > > > > > > > > On Wed, Jul 26, 2023 at 12:42 AM Jing Ge <j...@ververica.com> wrote: > > > > > > > Thanks Leonard for driving it. We are now on the same page. > > > > > > > > Best regards, > > > > Jing > > > > > > > > On Tue, Jul 25, 2023 at 9:19 PM Leonard Xu <xbjt...@gmail.com> > wrote: > > > > > > > >> We’ve detailed offline discussions with @Alexander and @Jingsong, > > about > > > >> “Remove SourceFunction” item, we’ve reached a consensus as > following: > > > >> > > > >> 1. Deprecate SourceFunction in 1.18 and implement following > > improvement > > > >> subtasks of FLINK-28045[1] later is reasonable for all of us. > > > >> > > > >> 2. Deleting SourceFunction API depends on future’s work progress, > thus > > > >> “Remove SourceFunction APIs” should be a nice to have item. > Alexander > > > has > > > >> volunteered to take these subtasks and would try to finish them > next, > > > >> thanks again. > > > >> > > > >> 3. As a nice to have item, and its READY status depends on future’s > > > work > > > >> progress, this won't block release 2.0 must-have item vote. > > > >> > > > >> Thanks again @Alexander, @Jingsong and @Xintong for driving these > > > things > > > >> forward. > > > >> > > > >> Also CC RMs for 1.18 @QingSheng @Jing @Martijn @Konstantin, I’ve > > > >> communicated with Alexander and would like to help review the > > > deprecation > > > >> PR again. > > > >> > > > >> Best, > > > >> Leonard > > > >> > > > >> [1] https://issues.apache.org/jira/browse/FLINK-28045 > > > >> > > > >> > > > >> On Jul 21, 2023, at 6:09 PM, Chesnay Schepler <ches...@apache.org> > > > wrote: > > > >> > > > >> On 21/07/2023 11:45, Leonard Xu wrote: > > > >> > > > >> In this way, the user will see the deprecated API firstly but they > can > > > >> not find a candidate if we can not finish all tasks in one minor > > > version . > > > >> > > > >> > > > >> i'm not convinced that this matters. There will be a whole bunch of > > APIs > > > >> deprecated in 1.18 (that will remain in 1.x!) without a replacement > so > > > we > > > >> can remove them in 2.0. > > > >> We already accepted this scenario. > > > >> > > > >> > > > >> > > > > > > > > > -- > > https://twitter.com/snntrable > > https://github.com/knaufk > > >
