public class SchemaInfer
extends Object
SchemaInfer extract row's type after query is executed. It is pretty rough version. Optimization
is on the way. The problem we have right now is that TiDB promote Sum to Decimal which is not
compatible with column's type. The solution we come up with right now is use record column's type
ad finalFieldType and build another list recording TiExpr's type as fieldType for row reading.
Once we finish row reading, we first check each element in fieldType and finalFieldType share the
same type or not. If yes, no need for casting. If no, casting is needed here.