oracle的SQL语句中同一函数执行的时间问题

【问题】请问以下两种SQL中：
select name, sum(price) from tbl order by sum(price);对sum(price)计算了两次？
select name, sum(price) t from tbl order by t;对sum(price)计算了一次？
如何证明？谢谢！

【测试】

1、建立自定义聚合函数

create or replace type string_sum_obj as object (
--聚合函数的实质就是一个对象
     sum_string varchar2(4000),
     static function ODCIAggregateInitialize(v_self in out string_sum_obj) return number,
     --对象初始化
     member function ODCIAggregateIterate(self in out string_sum_obj, value in varchar2) return number,
     --聚合函数的迭代方法(这是最重要的方法)
     member function ODCIAggregateMerge(self in out string_sum_obj, v_next in string_sum_obj) return number,
     --当查询语句并行运行时,才会使用该方法,可将多个并行运行的查询结果聚合
     
     member function ODCIAggregateTerminate(self in string_sum_obj, return_value out varchar2 ,v_flags in number) return number
     --终止聚集函数的处理,返回聚集函数处理的结果.
)
/
create or replace type body string_sum_obj is
     static function ODCIAggregateInitialize(v_self in out string_sum_obj) return number is
     begin
         v_self := string_sum_obj(null);
         return ODCICONST.Success;
     end;
     member function ODCIAggregateIterate(self in out string_sum_obj, value in varchar2) return number is
     begin
          /* 连接 */   
          self.sum_string := self.sum_string || value;
          return ODCICONST.Success;
          /* 最大值 */
          if self.sum_string<value then
              self.sum_string:=value;
          end if;
          /* 最小值 */
          if self.sum_string>value then
       self.sum_string:=value;          
          end if;
          
          return ODCICONST.Success;
     end;
     member function ODCIAggregateMerge(self in out string_sum_obj, v_next in string_sum_obj) return number is
     begin
          /* 连接 */   
          self.sum_string := self.sum_string || v_next.sum_string;
          return ODCICONST.Success;
          /* 最大值 */
          if self.sum_string<v_next.sum_string then
              self.sum_string:=v_next.sum_string;
          end if;

          /* 最小值 */
          if self.sum_string>v_next.sum_string then
              self.sum_string:=v_next.sum_string;          
          end if;
          
          return ODCICONST.Success;
     end;
     member function ODCIAggregateTerminate(self in string_sum_obj, return_value out varchar2 ,v_flags in number) return number is
     begin
          return_value:= self.sum_string ||to_char(systimestamp,'yyyy-mm-dd hh24:mi:ss:xFF');

           DBMS_LOCK.sleep(1);
          return ODCICONST.Success;
     end;
end;
/
create or replace function ConnStrSum(value Varchar2) return Varchar2
     parallel_enable aggregate using string_sum_obj;

2、在自定义聚合函数中加入了延时1s和最终处理时间

SQL> select connstrsum(C2),connstrsum(C2)  from test;

CONNSTRSUM(C2)                                                                   CONNSTRSUM(C2)
-------------------------------------------------------------------------------- --------------------------------------------------------------------------------
91939294 2010-11-02 18:48:07:.828000000                                          91939294 2010-11-02 18:48:07:.828000000

SQL> select connstrsum(C2), connstrsum(C2)  from test group by C3 order by connstrsum(C2);

CONNSTRSUM(C2)                                                                   CONNSTRSUM(C2)
-------------------------------------------------------------------------------- --------------------------------------------------------------------------------
91 2010-11-02 18:48:30:.093000000                                                91 2010-11-02 18:48:30:.093000000
92 2010-11-02 18:48:28:.093000000                                                92 2010-11-02 18:48:28:.093000000
93 2010-11-02 18:48:27:.093000000                                                93 2010-11-02 18:48:27:.093000000
94 2010-11-02 18:48:29:.093000000                                                94 2010-11-02 18:48:29:.093000000

SQL> select connstrsum(C2), connstrsum(C2)  from test group by C3 order by connstrsum(C2 ||1);

CONNSTRSUM(C2)                                                                   CONNSTRSUM(C2)
-------------------------------------------------------------------------------- --------------------------------------------------------------------------------
91 2010-11-02 18:51:03:.640000000                                                91 2010-11-02 18:51:03:.640000000
92 2010-11-02 18:50:59:.640000000                                                92 2010-11-02 18:50:59:.640000000
93 2010-11-02 18:50:57:.640000000                                                93 2010-11-02 18:50:57:.640000000
94 2010-11-02 18:51:01:.640000000                                                94 2010-11-02 18:51:01:.640000000

SQL>

【结论】

第一条说明两列执行时间相同，说明同时执行

第二条说明两列执行时间相同，并且结果中每条之差1S，则说明orderby 后面的SUM没有延误时间，即同时执行

第三条说明不同的SUM函数延误时间。

从上面SQL执行的时间中可以看到同样的聚合函数是一起执行的，即oracle在执行解析时和查询时已经优化过了