fqaiser94 commented on PR #174: URL: https://github.com/apache/flink-connector-kafka/pull/174#issuecomment-5873137190
Thanks for reviewing @MartijnVisser! I addressed your comments as separate commits to make the changes easier to review. Let me know once you're happy and I can squash into a single commit. > One behaviour change blocks it for me, with both new options off: a record with a null Kafka key is no longer skipped when the query selects no key column, so the row count depends on the projection. This is fixed now. There should be no behaviour change anymore. > Worth stating in the description and in a release note: every kafka and upsert-kafka table now advertises projection pushdown, so any query reading a subset of the columns gets a different plan. Where that drops the Calc above the source, every operator below it shifts one id, so a job with state downstream will not restore from its savepoint. The source keeps its id. Declared metadata that is not selected is no longer read either. PR description updated. I don't have permissions to update the JIRA ticket. I _might_ be able to make the change if you assign the [ticket](https://issues.apache.org/jira/browse/FLINK-32609) to me (username: fqaiser94) first? -- 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]
