Question

我们有两张桌子：

HUSBANDS
---------------
id
name
surname
age

WIFES
---------------
id
name
surname
age
husbandId                      --(with constrain -> HUSBANDS.ID)

并且假设我们需要编写删除60岁以上妻子及其丈夫的程序;）

使用典型的SQL语句执行此操作可能是完美的，但在Oracle中不可能从一个DELETE语句中的两个表中删除行，对吧？所以......

我们不能这样做：

PROCEDURE remove_old_wifes() IS
BEGIN
    DELETE FROM husbands WHERE id IN (SELECT husbandId FROM wifes WHERE age >= 60); 
    DELETE FROM wifes WHERE age >= 60; 
END;

因为约束。

另一方面，下面提到的解决方案也是错误的：

PROCEDURE remove_old_wifes() IS
BEGIN
    DELETE FROM wifes WHERE age >= 60; 
    DELETE FROM husbands WHERE id IN (SELECT husbandId FROM wifes WHERE age >= 60); 
END;

因为当我们首先移除妻子时，所有丈夫都不会被移除......

这种情况的典型解决方案是什么？

重要提示：我无法设置级联。

Answer 1

你可以先用一个没有妻子的指示来更新丈夫。对于您的数据，我们使用age = -1。然后从妻子[原文如此]中删除，然后从丈夫那里删除。

update husbands
    set age = -1
    where id in (select husbandId from wifes where age >= 60);

delete from wifes where age >= 60;

delete from husbands where age = -1;

Answer 2

如果桌子上不允许“未婚丈夫”，你可以先拆除妻子，然后再拆除所有不再有妻子的丈夫。如果临时不一致是一个问题，您可能希望将删除包装在事务中。

PROCEDURE remove_old_wifes() IS
BEGIN
    DELETE FROM wifes WHERE age >= 60; 
    DELETE FROM husbands WHERE id NOT IN (SELECT husbandId FROM wifes); 
END;

Answer 3

Oracle中有许多外键子句。例如，您可以使用＆＃34;删除级联＆＃34;你的外键中的子句：

alter table WIFES 
add foreign key (husbandId ) 
   references HUSBANDS(id)
on delete cascade;

使用此子句，您可以删除两个表的行，只删除一个。所以你的程序将是：

PROCEDURE remove_old_wifes() IS
BEGIN
    DELETE FROM husbands WHERE id IN (SELECT husbandId FROM wifes WHERE age >= 60); 
END;

如果您无法设置on delete delete cascade子句，则应使用全局临时表，例如：

create table gtt_id( id number) on commit delete rows;


PROCEDURE remove_old_wifes() IS
BEGIN
    insert into gtt_id SELECT husbandId FROM wifes WHERE age >= 60;
    delete FROM wifes WHERE age >= 60
    DELETE FROM husbands WHERE id IN (SELECT id FROM gtt_id); 
END;

从约束相关的两个表中删除行

3 个答案: