[KYLIN-918] Calcite throws "java.lang.Float cannot be cast to java.lang.Double" error while executing SQL

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: v1.0
• Fix Version/s: v1.3.0, v1.4.0
• Component/s: Query Engine
• Assignee: Shao Feng Shi
• Reporter: Shao Feng Shi
• Created: 30/Jul/15 06:33
• Updated: 14/Mar/16 01:55
• Resolved: 07/Mar/16 03:16

Description

This is from an eBay internal deployment which deployed with a latest build from 0.7-staging (0.7.3-SNAPSHOT); User reported that running a SQL which got such ClassCastException while it has no problem before; The error trace is:

Caused by: java.lang.ClassCastException: java.lang.Float cannot be cast to java.lang.Double 
at Baz$4.apply(Unknown Source) 
at Baz$4.apply(Unknown Source) x
at Baz$4.apply(Unknown Source) 
at org.apache.calcite.linq4j.EnumerableDefaults.groupBy_(EnumerableDefaults.java:737) 
at org.apache.calcite.linq4j.EnumerableDefaults.groupBy(EnumerableDefaults.java:677) 
at org.apache.calcite.linq4j.DefaultEnumerable.groupBy(DefaultEnumerable.java:301) 
at Baz.bind(Unknown Source) 
at org.apache.calcite.jdbc.CalcitePrepare$CalciteSignature.enumerable(CalcitePrepare.java:281) 
at org.apache.calcite.jdbc.CalciteConnectionImpl.enumerable(CalciteConnectionImpl.java:235) 
at org.apache.calcite.jdbc.CalciteMetaImpl.createIterable(CalciteMetaImpl.java:476) 
at org.apache.calcite.avatica.AvaticaResultSet.execute(AvaticaResultSet.java:184) 
at org.apache.calcite.jdbc.CalciteResultSet.execute(CalciteResultSet.java:63) 
at org.apache.calcite.jdbc.CalciteResultSet.execute(CalciteResultSet.java:42) 
at org.apache.calcite.avatica.AvaticaConnection$1.execute(AvaticaConnection.java:473) 
at org.apache.calcite.jdbc.CalciteMetaImpl.prepareAndExecute(CalciteMetaImpl.java:508) 
at org.apache.calcite.avatica.AvaticaConnection.prepareAndExecuteInternal(AvaticaConnection.java:477) 
at org.apache.calcite.avatica.AvaticaStatement.executeInternal(AvaticaStatement.java:109) 
... 84 more 

The SQL is simple, like
select itemid, sum(column_a) from items group by itemid

I noticed the column "column_a" was declared as "float" in hive; After change the type to "double" in hive and sync it to Kylin, the same SQL can run successfully;