Hi Brian, Passing the checkpoint ID to SplitEnumerator.snapshotState() seems reasonable to me. So +1 to the change.
BTW, typically for such API changes, we expect a FLIP wiki [1] to be included in the discussion thread. Thanks, Jiangjie (Becket) Qin [1] https://cwiki.apache.org/confluence/display/FLINK/Flink+Improvement+Proposals On Wed, Apr 14, 2021 at 3:42 PM Zhou, Brian <b.z...@dell.com> wrote: > Hi all, > > > > I'd like to start a discussion thread about adding a checkpointId > parameter in the SplitEnumerator:: snapshotState() method [1] . > > This is required for Pravega Flink connector[2] FLIP-27 source > implementation. We want to use the ExternallyInducedSourceReader, and it > requires the split enumerator to track the Flink checkpoint ID. I have > created an issue[3] for this. > > > > The change is simple, however, this changes the API of the interface which > will influence other FLIP-27 source implementations, so I’d like to start a > public discussion here about your ideas and concerns for this proposal. > > > > [1] > https://github.com/apache/flink/blob/a66487d4d2f77bb09b9961655cc5922c3e5c098c/flink-core/src/main/java/org/apache/flink/api/connector/source/SplitEnumerator.java#L76 > > [2] https://github.com/pravega/flink-connectors > > [3] https://issues.apache.org/jira/browse/FLINK-22133 > > >
