rails 使用after_commit 来替代 after_save
首先得明白after_save
和after_commit
的区别
1、after_save
将回调函数包含在同一个事务中,那么要么全部成功,要么全部失败。回调函数中的抛出的异常会导致整个操作失败。
2、after_commit
是在事务提交之后,再进行的处理,那么即使回调函数抛出异常,也不会导致之前的操作失败。
举个例子。
class ZZDeal
attr_accessor :title
after_save: record_to_redis
def record_to_redis
# TODO
rails Exception
end
end
# console
zz = ZZDeal.last
1] pry(main)> zz = ZZDeal.first
D, [2014-09-18T19:38:13.416572 #10908] DEBUG -- : ZZDeal Load (1.4ms) SELECT `zz_deals`.* FROM `zz_deals` ORDER BY `zz_deals`.`id` ASC LIMIT 1
=> #<ZZDeal id: 1, title: 'my_first_name'>
zz.title = "jjjj"
zz.save
D, [2014-09-18T19:38:22.943780 #10908] DEBUG -- : (0.5ms) BEGIN
D, [2014-09-18T19:38:22.989825 #10908] DEBUG -- : SQL (1.0ms) UPDATE `zz_deals` SET `title` = 'jjjj', `updated_at` = '2014-09-18 19:38:22' WHERE `zz_deals`.`id` = 1
D, [2014-09-18T19:38:22.990678 #10908] DEBUG -- : (0.5ms) ROLLBACK
[5] pry(main)> zz.reload.title
D, [2014-09-18T19:38:37.321616 #10908] DEBUG -- : ZZDeal Load (1.6ms) SELECT `zz_deals`.* FROM `zz_deals` WHERE `zz_deals`.`id` = 1 LIMIT 1
=> "my_first_name"
发现,title并没有改变。
若我将 after_save
变更为 after_commit
,便是另一种效果
class ZZDeal
attr_accessor :title
after_commit: record_to_redis, :on => [:create, :update]
def record_to_redis
# TODO
rails Exception
end
end
zz = ZZDeal.last
1] pry(main)> zz = ZZDeal.first
D, [2014-09-18T19:38:13.416572 #10908] DEBUG -- : ZZDeal Load (1.4ms) SELECT `zz_deals`.* FROM `zz_deals` ORDER BY `zz_deals`.`id` ASC LIMIT 1
=> #<ZZDeal id: 1, title: 'my_first_name'>
zz.title = "wahaha"
[4] pry(main)> zz.save
D, [2014-09-18T19:40:29.632423 #10926] DEBUG -- : (0.7ms) BEGIN
D, [2014-09-18T19:40:29.687067 #10926] DEBUG -- : SQL (0.8ms) UPDATE `zz_deals` SET `title` = 'wahaha', `updated_at` = '2014-09-18 19:40:29' WHERE `zz_deals`.`id` = 1
D, [2014-09-18T19:40:29.824718 #10926] DEBUG -- : (0.6ms) COMMIT
D, [2014-09-18T19:40:29.825749 #10926] DEBUG -- : (0.5ms) ROLLBACK
[6] pry(main)> zz.reload.title
D, [2014-09-18T19:40:39.389478 #10926] DEBUG -- : ZZDeal Load (8.4ms) SELECT `zz_deals`.* FROM `zz_deals` WHERE `zz_deals`.`id` = 1 LIMIT 1
=> "wahaha"
数据库先进行了COMMIT再进行了ROLLBACK,那么之前的数据肯定是保存成功了。
所以两种情况分别对待,如果你确实是需要同时成功,同时失败,那么after_save 是不错的选择,但是,如果不希望回调函数影响数据的保存,那么你需要after_commit。
另外,在after_save中,会出现脏数据的情况。
宣传语
历经两个半月的准备,三次大改版,十七次小改版。le1024终于要和大家见面了。
le1024每天推荐1~3段,有趣、有爱、有故事的视频。
为您工作、学习、生活之余增加一点快乐的感觉。程序员必看的快乐视频网站