This is an automated email from the ASF dual-hosted git repository.
mayanks pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-pinot.git
The following commit(s) were added to refs/heads/master by this push:
new 753dd34 Temporary work-around to bypass function arguments being
treated as expressions with columns. (#5382)
753dd34 is described below
commit 753dd34a6fd3fd4c52fd7d0dc8123ed979ecf842
Author: Mayank Shrivastava <[email protected]>
AuthorDate: Wed May 13 20:28:07 2020 -0700
Temporary work-around to bypass function arguments being treated as
expressions with columns. (#5382)
Only aggregation functions have the knowledge on how to interpret their
arguments.
However, the assumption within ServerQueryRequest is that all expression
arguments will contain
columns to be collected. A clean fix would be to create aggregation
functions upfront and let
them interpret the arguments. However, until that happens, working-around
by assuming that first
argument contains all the columns (to be used by DataSchemaPruner), which
is true for all functions.
A side effect would be that DataSchemaPruner won't work for
DistinctCountThetaSketchAggregatinoFunction.
But since this pruner is only for schema mis-matches, and not for
performance, there should not be
an impact.
---
.../apache/pinot/core/query/request/ServerQueryRequest.java | 11 +++++++++--
1 file changed, 9 insertions(+), 2 deletions(-)
diff --git
a/pinot-core/src/main/java/org/apache/pinot/core/query/request/ServerQueryRequest.java
b/pinot-core/src/main/java/org/apache/pinot/core/query/request/ServerQueryRequest.java
index bf2fa07..32e1fc3 100644
---
a/pinot-core/src/main/java/org/apache/pinot/core/query/request/ServerQueryRequest.java
+++
b/pinot-core/src/main/java/org/apache/pinot/core/query/request/ServerQueryRequest.java
@@ -94,7 +94,15 @@ public class ServerQueryRequest {
if (aggregationsInfo != null) {
_aggregationExpressions = new HashSet<>();
for (AggregationInfo aggregationInfo : aggregationsInfo) {
- if
(!aggregationInfo.getAggregationType().equalsIgnoreCase(AggregationFunctionType.COUNT.getName()))
{
+ String aggregationType = aggregationInfo.getAggregationType();
+
+ // TODO: Remove special casing of theta-sketch.
+ // This is needed as a work-around because only aggregation functions
know how to interpret the expressions.
+ // The code below assumes that all expressions have columns in them,
which may not be true. But only Aggregation
+ // functions have that knowledge, and have not been instantiated yet.
A clean fix would be to instantiate
+ // aggregation functions upfront to interpret the expressions.
+ if
(!aggregationType.equalsIgnoreCase(AggregationFunctionType.COUNT.getName()) &&
!aggregationType
+
.equalsIgnoreCase(AggregationFunctionType.DISTINCTCOUNTTHETASKETCH.getName())) {
for (String expressions :
AggregationFunctionUtils.getArguments(aggregationInfo)) {
_aggregationExpressions.add(TransformExpressionTree.compileToExpressionTree(expressions));
}
@@ -131,7 +139,6 @@ public class ServerQueryRequest {
}
_selectionColumns =
RequestUtils.extractColumnsFromExpressions(_selectionExpressions);
_allColumns.addAll(_selectionColumns);
-
} else {
_selectionColumns = null;
_selectionExpressions = null;
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
