SEZ9 commented on issue #12966:
URL: https://github.com/apache/seatunnel/issues/12966#issuecomment-6431986227
Thanks for the correction, @SEPURI-SAI-KRISHNA — extending
`KafkaFormatIT.testFormatCompatibleCheck` rather than adding a greenfield conf
is the right call, and reusing the `jdbc_source_record` topic, the
`compatible_data.txt` envelope shape and the existing `checkCompatibleFormat()`
Postgres assertion keeps the follow-up small and reviewable. No objection to
the plan.
On the mutation argument: yes, that is exactly the bar I want. Removing `if
(msg.value() == null) { return; }` in
`CompatibleKafkaConnectDeserializationSchema` must fail both the `8ad46124f`
unit test and the new IT, and restoring it must bring both back. Please paste
the before/after results in the follow-up PR description so a reviewer does not
have to re-run the mutation to trust it.
On the fixture loop obstacle: agreed that `while ((line = br.readLine()) !=
null)` cannot express a null value, so an explicit `ProducerRecord<>(topic,
key, null)` next to the file-driven records is the way to go. Please keep that
addition narrowly scoped to the compatible-format test path so the canal, ogg,
maxwell and debezium lanes that share the plumbing are not touched. If it has
to go into a shared helper, call that out in the PR so it gets a close look.
Two things I still need from you:
1. Keep the earlier commitment to drive both `format_error_handle_way`
values and to assert the offset moved past the tombstone, not only that the
3-row count still holds — that is what catches a replay after restart.
2. Post the follow-up PR number here once it is open so this issue links
#12967 and the IT together, as you said.
Understood that you cannot run the Kafka IT locally; CI will be the first
signal and that is fine. If it comes back red on the IT rather than the unit
test, please report the failure here rather than adjusting the fixture until it
passes.
<!-- streview-comment:2057 -->
--
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]
