Github user ChengXiangLi commented on a diff in the pull request: https://github.com/apache/flink/pull/1469#discussion_r50218331 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/operators/hash/MutableHashTable.java --- @@ -486,11 +518,43 @@ protected boolean processProbeIter() throws IOException{ } } // -------------- partition done --------------- - + return false; } + protected boolean processUnmatchedBuildIter() throws IOException { + if (this.unmatchedBuildVisited) { + return false; + } + + this.probeMatchedPhase = false; + UnmatchedBuildIterator<BT, PT> unmatchedBuildIter = new UnmatchedBuildIterator<>(this.buildSideSerializer, this.numBuckets, + this.bucketsPerSegmentBits, this.bucketsPerSegmentMask, this.buckets, this.partitionsBeingBuilt, probedSet); + this.unmatchedBuildIterator = unmatchedBuildIter; + + // There maybe none unmatched build element, so we add a verification here to make sure we do not return (null, null) to user. + if (unmatchedBuildIter.next() == null) { --- End diff -- Yes, it's better to have `hasNext()` method. Just UnmatchedBuildIterator is quite complicated, redesign is risky, and the `back()` method would only be called once, it does not introduce extra effort, so i would just leave it here.
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. ---