sohami commented on a change in pull request #1374: DRILL-6542 :
IndexOutOfBoundsException for multilevel lateral queries…
URL: https://github.com/apache/drill/pull/1374#discussion_r202218606
##########
File path:
exec/java-exec/src/main/java/org/apache/drill/exec/physical/impl/unnest/UnnestImpl.java
##########
@@ -97,8 +99,16 @@ public final int unnestRecords(final int recordCount) {
logger.debug("Unnest: currentRecord: {}, innerValueCount: {}, record
count: {}, output limit: {}", innerValueCount,
recordCount, outputLimit);
+ final SchemaChangeCallBack callBack = new SchemaChangeCallBack();
for (TransferPair t : transfers) {
t.splitAndTransfer(innerValueIndex, count);
+
+ // Get the corresponding ValueVector in output container and transfer
the data
+ final ValueVector vectorWithData = t.getTo();
+ final ValueVector outputVector =
outgoing.getContainer().addOrGet(vectorWithData.getField(), callBack);
+ Preconditions.checkState(!callBack.getSchemaChangedAndReset(), "Outgoing
container doesn't have " +
+ "expected ValueVector of type %s, present in TransferPair of unnest
field", vectorWithData.getClass());
+ vectorWithData.makeTransferPair(outputVector).transfer();
Review comment:
Yes we have to do 2 transfers since there is no api for
getTransferPairToSingleMap to take target vector. For New Schema there are 2
cases, in one we have to preserve the output container vector (when schema
change is for non-unnest column) but reset transfers whereas in other case we
have to create a new vector in the container (when schema change is on unnest
column). That's the reason it is done this way.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on 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
With regards,
Apache Git Services
