sql基础之from和ansi join一起使用（parse）总结

create table temp1
(
  aid       VARCHAR(5) not null,
  car       VARCHAR(10) not null
); 
 
create table temp2
(
  bid       VARCHAR(5) not null,
  username       VARCHAR(10) not null
);
 
create table temp3
(
  cid        VARCHAR(5) not null,
  dogname       VARCHAR(10) not null
);  
 
insert into temp1(aid,car) values('001','benz');           
insert into temp1(aid,car) values('001','BMW');
insert into temp1(aid,car) values('001','ford');
insert into temp1(aid,car) values('001','jeep');
insert into temp1(aid,car) values('002','jeep');
insert into temp1(aid,car) values('003','hongqi');
 
insert into temp2(bid,username) values('001','mayun');           
 
insert into temp3(cid,dogname) values('001','lily'); 
insert into temp3(cid,dogname) values('001','lucy');
insert into temp3(cid,dogname) values('002','xiaohua');

SQL的逻辑顺序如下：

1. FROM
　　1.1. JOIN
　　1.2. ON
2. WHERE
3. GROUP BY
4. WITH CUBE or WITH ROLLUP
5. HAVING
6. SELECT
7. DISTINCT
8. ORDER BY
11. TOP

在实际中，ANSI JOIN的优先级高于FROM平铺连接的优先级，对于外关联，通常但不绝对会先生成子查询。JOIN中NATURAL，USING，ON只能三者选一，而且是语法解析器控制，对于NATURAL，语法分析器会自动加上对应的USING。有了USING后，就会填充qual等值过滤条件。如果写了USING，则写了啥就是啥、和WHERE一样。

示例，

select * from temp2 a2,temp1 a1 left join temp3 a3 on a1.aid=a3.cid where a1.aid=a3.cid;

通过on过滤，left join就近连a1表，之后生成临时表，和a2一起参与where。所以，结果为：