如下圖， 計劃實現 ：按照 parent_code 分組， 取組中code最大值所在的整條記錄，如紅色部分。（類似hive中： row_number() over(partition by)）

select c.* from
(   select a.*, (@i := case when @key_i=parent_code then @i+1 else 1 end) as sort_num,(@key_i:=parent_code) as tmp 
        from my_test a, (SELECT  @i := 0,@key_i:='') b
        
 
order by parent_code,code desc) c 
where c.sort_num=1;

個人理解， mysql 執行順序：  from >>  where >> order by >> select .

以下部門應該有助於理解mysql中變數的概念。

1.=========================================================================================================

僅按照code排序,加序號sort_num效果如下：  

select (@i := @i + 1) as sort_num, a.*  
    from my_test a, (SELECT  @i := 0) b
    order by code;

2.=========================================================================================================

僅按parent_code分組，按照code排序，加序號.

select a.*, (@i := case when @key_i=parent_code then @i + 1 else 1 end) as sort_num,(@key_i:=parent_code) as tmp 
    from my_test a, (SELECT  @i := 0,@key_i:='') b
    order by parent_code,code desc;
 

參考連結 ：https://blog.csdn.net/zzhhoubin/article/details/79839610

https://www.cnblogs.com/CharlieLau/p/6737243.html

https://blog.csdn.net/u011728683/article/details/78594070