#EXPLAIN
# 通用描述
EXPLAIN用于展示一条SQL语句的执行计划（Execution Plan）。
当SQL语句的执行计划已经在在plan cache中存在时，直接重读该计划并输出；否则，由优化器进行试算生成计划到plan cache后再读取并输出。
只能对DML类的SQL语句展示执行计划，同时需满足执行该SQL语句的各项约束要求。
# 语句定义
explain::=
# plan for
用于兼容，可省略。EXPLAIN PLAN FOR=EXPLAIN。
# sql_statement
要展示执行计划的SQL语句，长度不超过2M。
示例
--单机部署中的SQL语句
EXPLAIN
WITH q_area(ano,aname) AS
(SELECT area_no ano,area_name aname FROM area WHERE area_no IN ('01','04'))
SELECT bno,bname,aname FROM (
SELECT a.branch_no bno,a.branch_name bname,b.aname aname FROM branches a,q_area b WHERE a.area_no=b.ano AND a.area_no='01'
UNION
SELECT a.branch_no bno,a.branch_name bname,b.aname aname FROM branches a,q_area b WHERE a.area_no=b.ano AND a.area_no='04'
);
--分布式部署中的SQL语句
EXPLAIN
SELECT *
FROM (SELECT * FROM area WHERE area_no IN ('01','02')
UNION
SELECT * FROM area WHERE area_no IN ('01','03')
);
# 输出定义
输出1.单机HEAP表输出的执行计划为：
PLAN_DESCRIPTION
----------------------------------------------------------------
SQL hash value: 3782799764
Optimizer: ADOPT_C
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
| Id | Operation type | Name | Owner | Rows | Cost(%CPU) | Partition info |
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
| 0 | SELECT STATEMENT | | | | | |
| 1 | VIEW | | | 1000| 1652( 0)| |
| 2 | HASH DISTINCT | | | 1000| 1652( 0)| |
| 3 | VIEW | | | 2512000| 1576( 0)| |
| 4 | UNION ALL | | | 2512000| 1446( 0)| |
| 5 | NESTED LOOPS INNER | | | 1256000| 687( 0)| |
|* 6 | TABLE ACCESS FULL | BRANCHES | SALES | 4000| 445( 0)| |
|* 7 | VIEW | | | 314| 151( 0)| |
| 8 | TABLE ACCESS BY INDEX ROWID| AREA | SALES | | | |
|* 9 | INDEX RANGE SCAN | SYS_C_13 | SALES | 7840| 151( 0)| |
| 10 | NESTED LOOPS INNER | | | 1256000| 687( 0)| |
|*11 | TABLE ACCESS FULL | BRANCHES | SALES | 4000| 445( 0)| |
|*12 | VIEW | | | 314| 151( 0)| |
| 13 | TABLE ACCESS BY INDEX ROWID| AREA | SALES | | | |
|*14 | INDEX RANGE SCAN | SYS_C_13 | SALES | 7840| 151( 0)| |
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
Operation Information (identified by operation id):
---------------------------------------------------
6 - Predicate : filter("A"."AREA_NO" = '01')
7 - Predicate : filter("B"."ANO" = '01')
9 - Predicate : access("AREA"."AREA_NO" IN ('01', '04'))
11 - Predicate : filter("A"."AREA_NO" = '04')
12 - Predicate : filter("B"."ANO" = '04')
14 - Predicate : access("AREA"."AREA_NO" IN ('01', '04'))
输出2.单机TAC表输出的执行计划为：
PLAN_DESCRIPTION
----------------------------------------------------------------
SQL hash value: 3782799764
Optimizer: ADOPT_C
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
| Id | Operation type | Name | Owner | Rows | Cost(%CPU) | Partition info |
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
| 0 | SELECT STATEMENT | | | | | |
| 1 | COL TO ROW | | | | | |
| 2 | RESULT | | | 1000| 1652( 0)| |
| 3 | HASH GROUP | | | 1000| 1652( 0)| |
| 4 | RESULT | | | 2512000| 1576( 0)| |
| 5 | UNION ALL | | | 2512000| 1446( 0)| |
| 6 | NESTED LOOPS INNER | | | 1256000| 687( 0)| |
|* 7 | TABLE ACCESS FULL | BRANCHES | SALES | 4000| 445( 0)| |
|* 8 | RESULT | | | 314| 151( 0)| |
| 9 | TABLE ACCESS BY INDEX ROWID| AREA | SALES | | | |
|*10 | INDEX RANGE SCAN | SYS_C_58 | SALES | 7840| 151( 0)| |
| 11 | NESTED LOOPS INNER | | | 1256000| 687( 0)| |
|*12 | TABLE ACCESS FULL | BRANCHES | SALES | 4000| 445( 0)| |
|*13 | RESULT | | | 314| 151( 0)| |
| 14 | TABLE ACCESS BY INDEX ROWID| AREA | SALES | | | |
|*15 | INDEX RANGE SCAN | SYS_C_58 | SALES | 7840| 151( 0)| |
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
Operation Information (identified by operation id):
---------------------------------------------------
1 - Projection: RemoteTable[1][CHAR, 4], RemoteTable[1][VARCHAR, 200], RemoteTable[1][VARCHAR, 60]
2 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200], Tuple[0, 2][VARCHAR, 60]
3 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200], Tuple[0, 2][VARCHAR, 60]
4 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200], Tuple[0, 2][VARCHAR, 60]
5 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200], Tuple[0, 2][VARCHAR, 60]
6 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200], Tuple[1, 0][VARCHAR, 60]
7 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200]
Predicate : access("A"."AREA_NO" = '01')
8 - Projection: Tuple[0, 1][VARCHAR, 60]
Predicate : filter(Tuple[0, 0] = '01')
10 - Projection: Tuple[0, 0][CHAR, 2], Tuple[0, 1][VARCHAR, 60]
Predicate : access("AREA"."AREA_NO" IN ('01', '04'))
11 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200], Tuple[1, 0][VARCHAR, 60]
12 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200]
Predicate : access("A"."AREA_NO" = '04')
13 - Projection: Tuple[0, 1][VARCHAR, 60]
Predicate : filter(Tuple[0, 0] = '04')
15 - Projection: Tuple[0, 0][CHAR, 2], Tuple[0, 1][VARCHAR, 60]
Predicate : access("AREA"."AREA_NO" IN ('01', '04'))
输出3.单机LSC表输出的执行计划为：
PLAN_DESCRIPTION
----------------------------------------------------------------
SQL hash value: 3782799764
Optimizer: ADOPT_C
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
| Id | Operation type | Name | Owner | Rows | Cost(%CPU) | Partition info |
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
| 0 | SELECT STATEMENT | | | | | |
| 1 | COL TO ROW | | | | | |
| 2 | RESULT | | | 1000| 2252( 0)| |
| 3 | HASH GROUP | | | 1000| 2252( 0)| |
| 4 | RESULT | | | 2512000| 2176( 0)| |
| 5 | UNION ALL | | | 2512000| 2046( 0)| |
| 6 | NESTED LOOPS INNER | | | 1256000| 987( 0)| |
|* 7 | TABLE ACCESS FULL | BRANCHES | SALES | 4000| 445( 0)| |
|* 8 | RESULT | | | 314| 451( 0)| |
|* 9 | TABLE ACCESS FULL | AREA | SALES | 7840| 451( 0)| |
| 10 | NESTED LOOPS INNER | | | 1256000| 987( 0)| |
|*11 | TABLE ACCESS FULL | BRANCHES | SALES | 4000| 445( 0)| |
|*12 | RESULT | | | 314| 451( 0)| |
|*13 | TABLE ACCESS FULL | AREA | SALES | 7840| 451( 0)| |
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
Operation Information (identified by operation id):
---------------------------------------------------
1 - Projection: RemoteTable[1][CHAR, 4], RemoteTable[1][VARCHAR, 200], RemoteTable[1][VARCHAR, 60]
2 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200], Tuple[0, 2][VARCHAR, 60]
3 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200], Tuple[0, 2][VARCHAR, 60]
4 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200], Tuple[0, 2][VARCHAR, 60]
5 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200], Tuple[0, 2][VARCHAR, 60]
6 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200], Tuple[1, 0][VARCHAR, 60]
7 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200]
Predicate : access("A"."AREA_NO" = '01')
8 - Projection: Tuple[0, 1][VARCHAR, 60]
Predicate : filter(Tuple[0, 0] = '01')
9 - Projection: Tuple[0, 0][CHAR, 2], Tuple[0, 1][VARCHAR, 60]
Predicate : access("AREA"."AREA_NO" IN ('01', '04'))
10 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200], Tuple[1, 0][VARCHAR, 60]
11 - Projection: Tuple[0, 0][CHAR, 4], Tuple[0, 1][VARCHAR, 200]
Predicate : access("A"."AREA_NO" = '04')
12 - Projection: Tuple[0, 1][VARCHAR, 60]
Predicate : filter(Tuple[0, 0] = '04')
13 - Projection: Tuple[0, 0][CHAR, 2], Tuple[0, 1][VARCHAR, 60]
Predicate : access("AREA"."AREA_NO" IN ('01', '04'))
输出4.分布式TAC/LSC表输出的执行计划为：
PLAN_DESCRIPTION
----------------------------------------------------------------
SQL hash value: 3958538062
Optimizer: ADOPT_C
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
| Id | Operation type | Name | Owner | Rows | Cost(%CPU) | Partition info |
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
| 0 | SELECT STATEMENT | | | | | |
| 1 | DISTRIBUTED COORDINATOR | | | | | |
| 2 | COL TO ROW | | | | | |
| 3 | RESULT | | | 1000| 959( 0)| |
| 4 | HASH DISTINCT | | | 1000| 959( 0)| |
| 5 | RESULT | | | 15680| 958( 0)| |
| 6 | UNION ALL | | | 15680| 957( 0)| |
| 7 | PX N2I REMOTE | QUEUE_0 | | 7840| 478( 0)| |
|* 8 | TABLE ACCESS FULL | AREA | SALES | 7840| 451( 0)| |
| 9 | PX N2I REMOTE | QUEUE_1 | | 7840| 478( 0)| |
|*10 | TABLE ACCESS FULL | AREA | SALES | 7840| 451( 0)| |
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
Operation Information (identified by operation id):
---------------------------------------------------
2 - Projection: RemoteTable[1][CHAR, 2]
3 - Projection: Tuple[0, 0][CHAR, 2]
4 - Projection: Tuple[0, 0][CHAR, 2]
5 - Projection: Tuple[0, 0][CHAR, 2]
6 - Projection: Tuple[0, 0][CHAR, 2]
7 - Projection: Tuple[0, 0][CHAR, 2]
PX RemoteInfo: (RANDOM SENDER -> RANDOM RECEIVER : 3->1 [3][4][5]->[2])
8 - Projection: Tuple[0, 0][CHAR, 2]
Predicate : access("AREA"."AREA_NO" IN ('01', '02'))
9 - Projection: Tuple[0, 0][CHAR, 2]
PX RemoteInfo: (RANDOM SENDER -> RANDOM RECEIVER : 3->1 [3][4][5]->[2])
10 - Projection: Tuple[0, 0][CHAR, 2]
Predicate : access("AREA"."AREA_NO" IN ('01', '03'))
# sql hash value
SQL语句的哈希号，即V$SQL视图里的hash_value字段值。
# optimizer
YashanDB使用的优化器名称， ADOPT_C表示CBO。
# id
执行步骤号，按二叉树结构从底部向上逆序执行。
如下为一个执行二叉树示例：
# operation type
执行算子，定义了执行计划每一步的操作。
如上例输出1和2中的INDEX RANGE SCAN表示执行索引范围扫描，检索到ROWID，TABLE ACCESS BY INDEX ROWID则根据这个ROWID读取行数据，然后继续向上直到0步的算子。
上例输出3和4中的TABLE ACCESS FULL表示执行全表扫描，检索到记录RESULT后，对两个RESULT进行UNION ALL算法，然后继续向上直到0步的算子。
其中，PX算子可分为BROADCAST、HASH、RANDOM等类型，表示数据在节点间的重分布方式。在每个PX算子后面会列出节点发送情况，N2I表示多个节点将数据发往一个节点，具体发送和接收节点可在投影信息中查看。
# name
算子操作的数据库对象名称，如扫描表、索引时对应的表名、索引名等。
对于PX算子，Name列示的是该算子在通讯过程中的唯一标识号。
# owner
操作对象的属主。如上例输出1中的area、branches表，及SYS_C_16索引，其Owner为SALES用户。
上例输出2中的area、branches表，其Owner为SALES_LSC用户。
上例输出3中的area表，其Owner为SALES用户。
# rows
算子扫描的记录行数。
# cost(%cpu)
估算执行此算子产生的CPU代价。
# partition info
对于分区表，此列展示算子扫描到的分区。
示例1（单机HEAP表）
--创建sales表为一张范围分区表，且在year字段上创建local索引。
CREATE TABLE sales
(year CHAR(4) NOT NULL,
month CHAR(2) NOT NULL,
branch CHAR(4),
product CHAR(5),
quantity NUMBER DEFAULT 0 NOT NULL,
amount NUMBER(10,2) DEFAULT 0 NOT NULL,
salsperson CHAR(10))
PARTITION BY RANGE(year)
(PARTITION p_sales_1 VALUES LESS THAN ('2001'),
PARTITION p_sales_2 VALUES LESS THAN ('2011'),
PARTITION p_sales_3 VALUES LESS THAN (2022));
CREATE INDEX idx_sales ON sales(year) LOCAL;
INSERT INTO sales VALUES ('2001','01','0201','11001',30,500,'0201010011');
INSERT INTO sales VALUES ('2015','11','0101','11001',20,300,'');
INSERT INTO sales VALUES ('2021','10','0101','11001',20,300,'');
INSERT INTO sales VALUES ('2000','12','0102','11001',20,300,'');
INSERT INTO sales VALUES ('2015','03','0102','11001',20,300,'');
INSERT INTO sales VALUES ('2020','05','0101','11001',40,600,'');
COMMIT;
--查看执行计划
EXPLAIN
SELECT a.area_name,
b.branch_name,
s.product,
s.amount
FROM branches b,
area a,
(SELECT branch,
product,SUM(amount) amount
FROM sales
WHERE year>'2005'
GROUP BY branch,product) s
WHERE s.branch=b.branch_no
AND b.area_no=a.area_no;
上例的输出结果（[1,2]表示扫描第1到第2个分区）为：
--以HEAP表为例，TAC表将会多出一个COL TO ROW算子
PLAN_DESCRIPTION
----------------------------------------------------------------
SQL hash value: 3362858118
Optimizer: ADOPT_C
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
| Id | Operation type | Name | Owner | Rows | Cost(%CPU) | Partition info |
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
| 0 | SELECT STATEMENT | | | | | |
| 1 | NESTED LOOPS INNER | | | 1000| 461( 0)| |
| 2 | NESTED LOOPS INNER | | | 1000| 311( 0)| |
| 3 | VIEW | | | 1000| 161( 0)| |
|* 4 | HASH GROUP | | | 1000| 161( 0)| |
| 5 | PART SCAN ITERATOR | | | 33000| 160( 0)| [1,2] |
| 6 | TABLE ACCESS BY INDEX ROWID| SALES | SALES | | | |
|* 7 | INDEX RANGE SCAN | IDX_SALES | SALES | 33000| 160( 0)| |
| 8 | TABLE ACCESS BY INDEX ROWID | BRANCHES | SALES | | | |
|* 9 | INDEX UNIQUE SCAN | SYS_C_125 | SALES | 1| 149( 0)| |
| 10 | TABLE ACCESS BY INDEX ROWID | AREA | SALES | | | |
|*11 | INDEX UNIQUE SCAN | SYS_C_123 | SALES | 1| 149( 0)| |
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
Operation Information (identified by operation id):
---------------------------------------------------
4 - Predicate : group expressions("SALES"."BRANCH", "SALES"."PRODUCT")
7 - Predicate : access("SALES"."YEAR" > '2005')
9 - Predicate : access("B"."BRANCH_NO" = "S"."BRANCH")
11 - Predicate : access("A"."AREA_NO" = "B"."AREA_NO")
示例2（分布式LSC表）
--sales表为一张范围分区表
CREATE TABLE sales
(year CHAR(4) NOT NULL,
month CHAR(2) NOT NULL,
branch CHAR(4),
product CHAR(5),
quantity NUMBER DEFAULT 0 NOT NULL,
amount NUMBER(10,2) DEFAULT 0 NOT NULL,
salsperson CHAR(10))
PARTITION BY RANGE(year)
(PARTITION p_sales_1 VALUES LESS THAN ('2001'),
PARTITION p_sales_2 VALUES LESS THAN ('2011'),
PARTITION p_sales_3 VALUES LESS THAN (2022));
INSERT INTO sales VALUES ('2001','01','0201','11001',30,500,'0201010011');
INSERT INTO sales VALUES ('2015','11','0101','11001',20,300,'');
INSERT INTO sales VALUES ('2021','10','0101','11001',20,300,'');
INSERT INTO sales VALUES ('2000','12','0102','11001',20,300,'');
INSERT INTO sales VALUES ('2015','03','0102','11001',20,300,'');
INSERT INTO sales VALUES ('2020','05','0101','11001',40,600,'');
COMMIT;
--查看如下语句执行计划，[1,2]表示扫描到第1到第2个分区
EXPLAIN
SELECT a.area_name,
b.branch_name,
s.product,
s.amount
FROM branches b,
area a,
(SELECT branch,
product,SUM(amount) amount
FROM sales
WHERE year>'2005'
GROUP BY branch,product) s
WHERE s.branch=b.branch_no
AND b.area_no=a.area_no;
PLAN_DESCRIPTION
----------------------------------------------------------------
SQL hash value: 3362858118
Optimizer: ADOPT_C
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
| Id | Operation type | Name | Owner | Rows | Cost(%CPU) | Partition info |
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
| 0 | SELECT STATEMENT | | | | | |
| 1 | DISTRIBUTED COORDINATOR | | | | | |
| 2 | COL TO ROW | | | | | |
| 3 | PX N2I REMOTE | QUEUE_0 | | 1000| 1473( 0)| |
|* 4 | HASH JOIN INNER | | | 1000| 1469( 0)| |
| 5 | TABLE ACCESS FULL | AREA | SALES | 100000| 442( 0)| |
|* 6 | HASH JOIN INNER | | | 1000| 959( 0)| |
| 7 | TABLE ACCESS FULL | BRANCHES | SALES | 100000| 442( 0)| |
| 8 | RESULT | | | 1000| 448( 0)| |
| 9 | PX N2N REMOTE | QUEUE_1 | | 1000| 448( 0)| |
| 10 | HASH GROUP | | | 1000| 433( 0)| |
|*11 | PX N2N REMOTE | QUEUE_2 | | 1000| 433( 0)| |
| 12 | HASH GROUP | | | 1000| 431( 0)| |
| 13 | PART SCAN ITERATOR | | | 33000| 431( 0)| [1,2] |
|*14 | TABLE ACCESS FULL | SALES | SALES | 33000| 431( 0)| |
+----+--------------------------------+----------------------+------------+----------+-------------+--------------------------------+
Operation Information (identified by operation id):
---------------------------------------------------
2 - Projection: RemoteTable[3][CHAR, 2], RemoteTable[3][CHAR, 2], RemoteTable[3][CHAR, 5], RemoteTable[3][NUMBER]
3 - Projection: Tuple[0, 0][CHAR, 2], Tuple[0, 1][CHAR, 2], Tuple[0, 2][CHAR, 5], Tuple[0, 3][NUMBER]
PX RemoteInfo: (RANDOM SENDER -> RANDOM RECEIVER : 3->1 [3][4][5]->[2])
4 - Projection: Tuple[0, 1][CHAR, 2], Tuple[1, 1][CHAR, 2], Tuple[1, 2][CHAR, 5], Tuple[1, 3][NUMBER]
Predicate : access(Tuple[0, 0] = Tuple[1, 0])
5 - Projection: Tuple[0, 0][INTEGER], Tuple[0, 1][CHAR, 2]
6 - Projection: Tuple[0, 1][INTEGER], Tuple[0, 2][CHAR, 2], Tuple[1, 1][CHAR, 5], Tuple[1, 2][NUMBER]
Predicate : access(Tuple[0, 0] = Tuple[1, 0])
7 - Projection: Tuple[0, 1][CHAR, 4], Tuple[0, 0][INTEGER], Tuple[0, 2][CHAR, 2]
8 - Projection: Tuple[0, 1][CHAR, 4], Tuple[0, 2][CHAR, 5], Tuple[0, 0][NUMBER]
9 - Projection: Tuple[0, 0][NUMBER], Tuple[0, 1][CHAR, 4], Tuple[0, 2][CHAR, 5]
PX RemoteInfo: (BROADCAST SENDER -> RANDOM RECEIVER : 3->3 [3][4][5]->[3][4][5])
10 - Projection: SUM(Tuple[0, 0])[NUMBER], Tuple[0, 1][CHAR, 4], Tuple[0, 2][CHAR, 5]
Group Expression: (Tuple[0, 1], Tuple[0, 2])
11 - Projection: Tuple[0, 0][NUMBER], Tuple[0, 1][CHAR, 4], Tuple[0, 2][CHAR, 5]
PX RemoteInfo: (HASH SENDER -> RANDOM RECEIVER : 3->3 [3][4][5]->[3][4][5])
access(Tuple[0, 1], Tuple[0, 2])
12 - Projection: SUM(Tuple[0, 0])[NUMBER], Tuple[0, 1][CHAR, 4], Tuple[0, 2][CHAR, 5]
Group Expression: (Tuple[0, 1], Tuple[0, 2])
13 - Projection: Tuple[0, 0][NUMBER, (10, 2)], Tuple[0, 1][CHAR, 4], Tuple[0, 2][CHAR, 5]
14 - Projection: Tuple[0, 5][NUMBER, (10, 2)], Tuple[0, 2][CHAR, 4], Tuple[0, 3][CHAR, 5]
Predicate : access("SALES"."YEAR" > '2005')
# predicate
谓词信息，包括：
access：限定谓词，用于减少结果集的大小，如索引限定谓词、HashJoin限定谓词。
filter：过滤谓词，对结果集中的每条记录进行过滤运算。
projection：投影信息，对下层算子的引用。其中Tuple[sourceId, attrId]表示下层算子的第sourceId个数据源的第attrId个投影，两者的序号都是从0开始。
下载文档
复制链接