vvcephei commented on a change in pull request #8181:
URL: https://github.com/apache/kafka/pull/8181#discussion_r494369894
##########
File path:
streams/src/main/java/org/apache/kafka/streams/processor/internals/StreamTask.java
##########
@@ -581,8 +580,8 @@ public void punctuate(final ProcessorNode node, final long
timestamp, final Punc
if (processorContext.currentNode() != null) {
throw new IllegalStateException(format("%sCurrent node is not
null", logPrefix));
}
-
- updateProcessorContext(new StampedRecord(DUMMY_RECORD, timestamp),
node);
+
+ updateProcessorContext(new StampedRecord(new
ConsumerRecord<>(ProcessorContextImpl.NONEXIST_TOPIC, -1, -1L, null, null),
timestamp), node);
Review comment:
Yes, that's more or less what I was concerned about. We are trying to
achieve a specific result here (and brand-new 'Header' instance each time), but
it's only achieved via a side effect.
On the other hand, I just took a closer look at the code, and I see that
there's a better reason to keep using `ConsumerRecord`, namely that
`updateProcessorContext` is used both for regular processing (with "real"
records) and for punctuation here with this dummy record. It seems like a good
idea to prevent those code paths from diverging, so I'm +1 with keeping this
change as-is.
----------------------------------------------------------------
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:
us...@infra.apache.org
