Github user paul-rogers commented on a diff in the pull request:
https://github.com/apache/drill/pull/1059#discussion_r158593956
--- Diff:
exec/java-exec/src/main/java/org/apache/drill/exec/physical/impl/join/HashJoinBatch.java
---
@@ -305,11 +307,15 @@ public void executeBuildPhase() throws
SchemaChangeException, ClassTransformatio
//Setup the underlying hash table
// skip first batch if count is zero, as it may be an empty schema
batch
- if (right.getRecordCount() == 0) {
+ if (isFurtherProcessingRequired(rightUpstream) &&
right.getRecordCount() == 0) {
for (final VectorWrapper<?> w : right) {
w.clear();
}
rightUpstream = next(right);
+ if (isFurtherProcessingRequired(rightUpstream) &&
+ right.getRecordCount() > 0 && hashTable == null) {
+ setupHashTable();
--- End diff --
This handles an empty batch followed by a non-empty batch. Can we be sure
that there will only ever be a sequence of 0 or 1 empty batches? Might there be
a pathological scan that reads 20 (say) empty files, producing a series of 20
empty batches? In short, should the logic here be in a loop?
Did we create a test that checks for this case?
---
