很多数据库对NOT IN的处理很慢,网上很多文章都建议使用NOT EXISTS来替代,而NOT EXISTS又可以通过left join来替代。例如:select * from t1 where not exists (select 1 from (select c1 from t2) zz where zz.c1=t1.c1)可以写成select t1.* from t1 left join t2 on t1.c1=t2.c1 and t2.c1 is null.但是,事实上NOT IN并不完全等价于NOT EXISTS。
NOT IN和NOT EXISTS的区别主要在于对null的处理。考虑一下SQL语句
NOT IN:select * from t1 where c1 not in (select c1 from t2);
NOT EXISTS:select * from t1 where not exists (select 1 from (select c1 from t2) zz where zz.c1=t1.c1);
对于表t2的c1有null的情况,NOT IN查询是没有结果的,而NOT EXISTS不受影响;
对于表t1的c1有null的情况,NOT IN会将null过滤掉,而NOT EXISTS则不会;
所以,NOT IN正确写成NOT EXISTS应该是:
select * from t1 where not exists (select 1 from (select c1 from t2) aa where aa.c1=t1.c1) and
not exists (select 1 from (select c1 from t2) bb where bb.c1 is null) and
t1.c1 is not null
对于t1或t2的c1明确为