SQL optimization - in and not in, exist

WHY？

IN and NOT IN Is a more commonly used keyword , Why try to avoid it ？

1、 Low efficiency

There's a situation in the project ：

t1 surface and t2 surface   All are 150w Data ,600M The appearance of , It's not big .

But such a query ↓

select * from t1 where phone not in (select phone from t2)

Just fooled me ... Ten minutes , I checked  phone Index both tables , Field types are the same . original not in Can't hit the index ....

Change to NOT EXISTS Query after 20s , The efficiency is really poor, much better .

select * from t1 
where  not  EXISTS (select phone from t2  where t1.phone =t2.phone)

2、 Be prone to problems , Or the query result is incorrect （ No more serious shortcomings ）

With IN For example . Create two tables ：test1 and test2

create table test1 (id1 int)
create table test2 (id2 int)

insert into test1 (id1) values (1),(2),(3)
insert into test2 (id2) values (1),(2)

I want to check , stay test2 Existing in  test1 Medium id . Use IN The general way of writing is ：

select id1 from test1 
where id1 in (select id2 from test2)

The result is ：

OK Wood problems ！

But if I slip my hand for a moment , It has been written. ：

select id1 from test1 
where id1 in (select id1 from test2)

Don't be careful id2 It's written in id1 了 , What will happen ?

The result is ：

EXCUSE ME！ Why don't you make a mistake ？

Search separately  select id1 from test2  Yes, it will definitely report an error : news 207, Level 16, state 1, The first 11 That's ok Name 'id1' Invalid .

However, it uses IN The sub query of is so perfunctory , Find out directly 1 2 3

This is just an error prone situation , If you don't write it wrong, there's nothing wrong , So let's see NOT IN Find out the wrong result directly ：

to test2 Insert a null value ：

insert into test2 (id2) values (NULL)

I want to check , stay test2 There is no such thing as  test1 Medium id .

select id1 from test1 
where id1 not in (select id2 from test2)

The result is ：

blank ！ Obviously, this result is not what we want . We want to 3. Why is this so ？

as a result of ：NULL Not equal to any non empty value ！ If id2 Only 1 and 2, that 3<>1 And 3<>2 therefore 3 Output , however id2 Contains a null value , that 3 Also is not equal to NULL So it won't output .

Digress ： It's best not to allow null values when creating a table , Otherwise, there will be many problems .

be based on Spring Boot + MyBatis Plus + Vue & Element Implementation of the background management system + User applet , Support RBAC Dynamic permissions 、 multi-tenancy 、 Data access 、 workflow 、 Three party login 、 payment 、 SMS 、 Shopping malls and other functions .

Project address ：https://github.com/YunaiV/ruoyi-vue-pro

HOW？

1、 use EXISTS or NOT EXISTS Instead of

select *  from test1 
   where EXISTS (select * from test2  where id2 = id1 )

select *  FROM test1  
 where NOT EXISTS (select * from test2  where id2 = id1 )

2、 use JOIN Instead of

 select id1 from test1 
   INNER JOIN test2 ON id2 = id1 
   
 select id1 from test1 
   LEFT JOIN test2 ON id2 = id1 
   where id2 IS NULL

There's no problem ！

PS： Then we can't live or die IN and NOT IN Why? ？ did not , A great God once said , If it is a definite and finite set , have access to . Such as IN （0,1,2）.