dennismdejong commented on code in PR #13668:
URL: https://github.com/apache/gravitino/pull/13668#discussion_r4215861400
##########
core/src/test/java/org/apache/gravitino/listener/TestTableEventDispatcher.java:
##########
@@ -127,6 +130,60 @@ void
testExtrasDoNotLeakIntoTheNextOperationOnTheSameThread() {
second.customInfo().isEmpty(), "Extras must not survive into a later
operation");
}
+ /**
+ * A failure-event listener must not be able to mask the original
create-table failure: when the
+ * dispatcher throws, the {@link CreateTableFailureEvent} listener exception
is swallowed by the
+ * {@link EventBus} so callers still see the original exception.
+ */
+ @Test
+ void
testCreateTableFailureEventShouldNotMaskOriginalExceptionWhenListenerThrows() {
Review Comment:
Correct. The production behavior is already fixed by #10665 (commit
85a8de09a), which added `EventBus.dispatchFailureEvent` to swallow listener
exceptions for all `FailureEvent` subclasses. #10597 reports the same symptom
on the create-table path.
This PR is test-only on purpose: it is the regression test pinned to the
exact call site from #10597. The existing
`TestEventBus.testDispatchEventSwallowsExceptionsForFailureEvents` covers
`EventBus` in isolation with a synthetic `FailureEvent`; this one drives
`TableEventDispatcher.createTable` end to end (dispatcher throws ->
`CreateTableFailureEvent` dispatched -> listener throws -> original exception
rethrown unchanged) and asserts it is the same instance. No production change
is needed.
--
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]
