A key point
MYSQL queries double error arises, there are two key points:
Generates temporary tables (1) SQL statement using the GROUP BY statement;
(2) RAND () is generated and stored in the query may be different random numbers.
Added: ================================================ ===========================================
(. 1) Group by Chang and min (), max (), COUNT (), SUM (), AVG () like aggregate function in conjunction with, such as: SELECT name , COUNT ( * ) from user Group by name; this statement based on user table name packet fields, the same values will be recorded in the name field is divided into the same group, and COUNT ( * ) will be calculated in each group,
the number of records, when we run the above statements will be: + --- ----- + - ---------- + | name | COUNT ( * ) | + -------- + - ---------- + | andy | 1 | | nana | 1 | | Mike | 1 | + -------- + - + ---------- name field in the user table has three values: mike, anna, andy, so that the SQL statement will be executed after the three packets, the number of records of each packet is 1; us to the user
add a record in the table: INSERT INTO User value ( . 4 , ' Mike ' ); this time table has two people with the same name, Mike call, then when the SQL statement is executed, the result will be as follows: + ---- + ---- - ---------- + | name | COUNT ( *) | + -------- + - ---------- +
| Andy | . 1 |
| Nana | . 1 |
| mike | 2 |
+ -------- + - ---------- +
packet field value name mike, it is possible to see 2 becomes the number of records.
(2) RAND () function to generate random numbers (except for 0 and 1) between 0 and 1
==================== ================================================== ==========================
Second, the principle explore
1. Review the error messages generated SQL statement using the built environment in the "MYSQL double Query Error 1":
SELECT COUNT ( * ), the concat (( SELECT Database ()), Floor ( RAND () * 2 )) AS A from User Group by A;
Description: SQL statement, rand () * 2 randomly generated between 0 and 2 (not including 0 and 2) the random number, and the floor (rand () * 2) is randomly produce one of the two numbers 0 and 1
When the code executes, during a session, the database maintains a session for the current master key is a temporary table, each field is as follows:
a COUNT ( * )
Additional information: In fact, the order of the fields should count (*) a primary key we just put on a front
For each of a value generated by our SQL statement, the database will first query the temporary table has not the same as a field value, and if so, the appropriate count (*) character
Segment value by 1, otherwise it will be a value into a temporary table in a field, the corresponding count (*) field value is set to 1. Here, we first and patted two questions:
(1) value for a newly created, temporary tables have inquiring whether the same error occurs when a field value?
(2) the value for a newly created, temporary table is not inquiring whether the same error occurs when a field value?
First to solve the first problem the SQL statement is modified as follows:
SELECT COUNT ( * ), the concat (( SELECT Database ()), . 1 ) AS A from user Group by A;
changes: for each record in the user table, a is the value of the SQL statements are generated injection1
Repeatedly implementation of the new SQL statement is: there is no query error occurs, the results of each have been as follows:
Why not produce a query mistake? During this period, understanding how information is stored in a temporary table changes will understand. First of all, the beginning of the temporary table has no data.
For the first record in the user table, a new SQL statement generated a value injection1, database query in a temporary table to a field does not have any value and just produce
A same value, decided: the a value into a field .
For the second record user table, a value of the new SQL statements generated for the same injection1, database query in a temporary table to have a new field values and
As the a value decided: the corresponding count (*) field value incremented by one .
user table has four records, count the final temporary table (*) field will change to 4. It can be seen during a session of the new SQL statement is executed, no query error occurred.
情况逐渐明了了，查询错误的产生可能与第一个问题无关. 接下来我们讨论第二个问题.
2. 为了便于叙述，将新的SQL语句称为SQL2，而另一个称为SQL1吧. 执行SQL1，当产生的a值与临时表中的各个a字段值不
同时（查询阶段），会将新的a值插入临时表中的a字段（插入阶段），并将对应的count(*)字段值置为1. 在这个过程中，错误
就发生在插入阶段. 在这个过程中，有两个关键的时点，一是查询时，二时插入时，在这两个时间点，rand()都会产生随机数，
随机数的产生使得两个时间点的a值有可能不一致，导致在插入阶段可能发生错误. 简单点说就是：在查询阶段和插入阶段的a
值有可能不同，这就是造成错误的原因. 接下来，我们从1条记录开始慢慢增加user表中的记录数，每一次都执行SQL1，看看
会发生什么.
（1）记录数为1时（只保留id为1的记录）
情况是：无论执行多少次，都不会发生错误，为什么呢？首先临时表是空的，而user表中只有一条记录，对应
要统计到临时表中的a值也只有一个，错误自然不会发生.
（2）记录数为2时（增加id为2，name为anna的记录）
可以看到，第三次执行SQL1的时候，发生了错误. 第一次为什么没错误呢？发生的情况是这样的.
i）针对user表中的第一条记录，产生的a值为injection0，查询阶段的a值就是injection0，查询到临时表中a字段值没
有injection0，于是数据库决定：插入a值. 前面我们说过，由于随机数的关系，在插入阶段a值有可能会变化，有可能变为
injection1，或者还是injection0.不过这时候由于临时表是空的，所以都不会有错误发生，在插入阶段a为injection0时，就
插入injection0，否则插入injection1.显然，这里最终插入的是injection0.
ii）针对user表中第二条记录，产生的a值必然是injection0，在查询阶段数据库检查到临时表中a字段值中有injection0，
于是数据库决定：将对应的count(*)字段值加1. 这时候，不论在插入阶段a值变为什么，数据库都不会再执行检查，而是直
接将count(*)字段值加1. 这就是第一次执行SQL1时发生的情况. 那么对于第一次执行SQL1，在什么情况下会发生错误呢？
iii）在ii中，针对user表中第二条记录产生的a值若为injection1，在查询阶段数据库检查到临时表中所有a字段值中没有
injection1，于是决定：将injection1插入到临时表中. 不巧，在插入阶段，a值变为了injection0，由于表中本来就存在
injection0了，于是就会报错了. 同样可以这样去分析第二次第三次执行SQL1发生的情况.
（3）记录数为3及以上，都会引发错误.
3. 经过以上分析，我们已经抓到了错误最关键的点了：在查询阶段和插入阶段的主键值不一致导致了错误的产生.
三、结论
在将数据统计到临时表中时（查询阶段和插入阶段的数据不一致），由于主键具有唯一性，如果某主键值在临时表中已经存
在，而在插入阶段还试图插入相同的主键值时，数据库便会报错.
====================================== 参考 ========================================
https://www.cnblogs.com/laoxiajiadeyun/p/10278512.html