dawidwys commented on a change in pull request #14078:
URL: https://github.com/apache/flink/pull/14078#discussion_r524365037
##########
File path:
flink-streaming-java/src/test/java/org/apache/flink/streaming/runtime/tasks/SourceOperatorStreamTaskTest.java
##########
@@ -99,6 +102,29 @@ public void testSnapshotAndAdvanceToEndOfEventTime() throws
Exception {
}
}
+ @Test
+ public void testEmittingMaxWatermarkAfterReadingAllRecords() throws
Exception {
+ try (StreamTaskMailboxTestHarness<Integer> testHarness =
createTestHarness()) {
+ testHarness.processAll();
Review comment:
Honestly it is not needed here. I just thought this way it better
resembles the "real" process, where you processAll records and then finish the
stream task.
In the second case the processAll could be misleading as one could say that
after processingAll records it might have been finished already and cancelling
makes no sense.
To sum up. It does not make a real difference if we have the `processAll` or
not, but I thought it slightly better describes what is happening in reality.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
