adasari commented on code in PR #18334:
URL: https://github.com/apache/pinot/pull/18334#discussion_r3640133901
##########
pinot-core/src/main/java/org/apache/pinot/core/plan/AggregationPlanNode.java:
##########
@@ -115,17 +122,34 @@ public Operator<AggregationResultsBlock>
buildNonFilteredAggOperator() {
boolean hasNullValues = _queryContext.isNullHandlingEnabled() &&
hasNullValues(aggregationFunctions);
if (!hasNullValues) {
- // Priority 2: Check if non-scan based aggregation is feasible
- if (filterOperator.isResultMatchingAll() && isFitForNonScanBasedPlan()) {
+ // when the filter matches all documents, resolve as many functions as
possible from the column
+ // dictionary/metadata without scanning the segment. Eligibility is
evaluated once per function here
+ // and reused for both the fully non-scan path (all functions
resolvable) and
+ // the partial path (some functions resolvable).
+ if (filterOperator.isResultMatchingAll()) {
+ boolean[] metadataResolvable = new
boolean[aggregationFunctions.length];
DataSource[] dataSources = new DataSource[aggregationFunctions.length];
+ int numResolved = 0;
for (int i = 0; i < aggregationFunctions.length; i++) {
- List<?> inputExpressions =
aggregationFunctions[i].getInputExpressions();
- if (!inputExpressions.isEmpty()) {
- String column = ((ExpressionContext)
inputExpressions.get(0)).getIdentifier();
- dataSources[i] = _indexSegment.getDataSource(column,
_queryContext.getSchema());
+ DataSource dataSource =
getDataSourceForAggregationFunction(aggregationFunctions[i]);
+ if (isFitForNonScanBasedPlan(aggregationFunctions[i], dataSource)) {
+ metadataResolvable[i] = true;
+ dataSources[i] = dataSource;
+ numResolved++;
}
}
- return new NonScanBasedAggregationOperator(_queryContext, dataSources,
numTotalDocs);
+
+ if (numResolved == aggregationFunctions.length) {
+ // Priority 2: all functions can be resolved from
dictionary/metadata -> fully non-scan based execution
+ return new NonScanBasedAggregationOperator(_queryContext,
dataSources, numTotalDocs);
+ }
+ if (numResolved > 0) {
+ // some functions can be resolved from dictionary/metadata; the rest
fall back to scan-based
+ // execution in the AggregationOperator.
+ aggregationInfo =
AggregationFunctionUtils.buildAggregationInfoWithoutStarTree(_segmentContext,
_queryContext,
Review Comment:
I am forced to remove metadata resolved agg functions from projection to fix
NumEntriesScannedPostFilter execution state :)
So, its DONE now and ready for review.
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
