Github user jackylk commented on a diff in the pull request:
https://github.com/apache/carbondata/pull/1402#discussion_r144205561
--- Diff:
core/src/main/java/org/apache/carbondata/core/scan/collector/impl/AbstractScannedResultCollector.java
---
@@ -89,24 +90,23 @@ protected void fillMeasureData(Object[] msrValues, int
offset,
protected Object getMeasureData(ColumnPage dataChunk, int index,
CarbonMeasure carbonMeasure) {
if (!dataChunk.getNullBits().get(index)) {
- switch (carbonMeasure.getDataType()) {
- case SHORT:
- return (short)dataChunk.getLong(index);
- case INT:
- return (int)dataChunk.getLong(index);
- case LONG:
- return dataChunk.getLong(index);
- case DECIMAL:
- BigDecimal bigDecimalMsrValue =
- dataChunk.getDecimal(index);
- if (null != bigDecimalMsrValue && carbonMeasure.getScale() >
bigDecimalMsrValue.scale()) {
- bigDecimalMsrValue =
- bigDecimalMsrValue.setScale(carbonMeasure.getScale(),
RoundingMode.HALF_UP);
- }
- // convert data type as per the computing engine
- return
DataTypeUtil.getDataTypeConverter().convertToDecimal(bigDecimalMsrValue);
- default:
- return dataChunk.getDouble(index);
+ DataType dataType = carbonMeasure.getDataType();
+ if (dataType == DataTypes.SHORT) {
--- End diff --
For primitive type and String type, we can compare object address, it is
simpler.
Only one place need to compare using id, `fromWrapperToExternalDataType`
method. It needs it because GSON breaks the singleton pattern
---
