junaiddshaukat commented on PR #38957:
URL: https://github.com/apache/beam/pull/38957#issuecomment-4708629546
Thanks for the review addressed all of it in the latest commit:
1. HOLD_MILLIS constant dropped it. Using BoundedWindow.TIMESTAMP_MIN_VALUE
directly now so there's no clash with the Beam "watermark hold" concept.
2. "partitions feeding this stage" right, these are the parent stage's
output / repartition-topic partitions that feed this stage. Reworded the
docs to say that explicitly.
3. Partition-count change now clears the whole per-partition map instead of
pruning only out-of-range partitions, so completeness and min() are always
computed over one consistent partitioning.
4. isComplete() — renamed to isReady() (agree "complete" could read as the
watermark having reached TIMESTAMP_MAX_VALUE).
5. Instant + the empty==MIN question witched the whole class from long
millis to org.joda.time.Instant. advance() now returns an Instant, and a
holding stage returns TIMESTAMP_MIN_VALUE instead of an empty Optional, so
that ambiguity is gone.
6. The two inspection getters — made them package-private +
@VisibleForTesting.
:runners:kafka-streams:check is green (10 unit tests). Let me know if
anything
else needs tweaking.
--
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]
