You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "Leonard Xu (Jira)" <ji...@apache.org> on 2022/05/13 06:32:00 UTC
[jira] [Updated] (FLINK-19830) Properly implements processing-time temporal table join
[ https://issues.apache.org/jira/browse/FLINK-19830?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Leonard Xu updated FLINK-19830:
-------------------------------
Fix Version/s: (was: 1.16.0)
> Properly implements processing-time temporal table join
> -------------------------------------------------------
>
> Key: FLINK-19830
> URL: https://issues.apache.org/jira/browse/FLINK-19830
> Project: Flink
> Issue Type: Improvement
> Components: Table SQL / Runtime
> Reporter: Leonard Xu
> Assignee: Leonard Xu
> Priority: Major
>
> The exsiting TemporalProcessTimeJoinOperator has already supported temporal table join.
> However, the semantic of this implementation is problematic, because the join processing for left stream doesn't wait for the complete snapshot of temporal table, this may mislead users in production environment.
> Under the processing time temporal join semantics, to get the complete snapshot of temporal table may need introduce new mechanism in FLINK SQL in the future.
> **Background** :
> * The reason why we turn off the switch[1] for `FOR SYSTEM_TIME AS OF` syntax for *temporal table join* is only the semantic consideration as above.
> * The reason why we turn on *temporal table function* is that it has been alive for a long time, thus although it exists same semantic problem, but we still support it from the perspective of compatibility.
> [1] [https://github.com/apache/flink/blob/4fe9f525a92319acc1e3434bebed601306f7a16f/flink-table/flink-table-planner/src/main/java/org/apache/flink/table/planner/plan/nodes/exec/stream/StreamExecTemporalJoin.java#L257]
--
This message was sent by Atlassian Jira
(v8.20.7#820007)