Rails sql延迟加载和自带缓存
Rails sql延迟加载和自带缓存Ruby代码
color_lot_manuallies = color_lot.color_lot_manuallies
if color_lot_manuallies.size == 1
end
引用SELECT count(*) AS count_all FROM `color_lot_manuallies` WHERE (`color_lot_manuallies`.color_lot_id = 237)当有A has many B时，a.bs.size并不是检索出所有a.bs再求出size，而是直接用select count(*)来计算结果。这应该是延迟加载，有点类似named_scope,但是并不是named_scope,named_scope的结果类是ActiveRecord::NamedScope::Scope，但是A.bs.class的结果是Array，这部分看了下源码但是并没有看懂。
所以当我这样写的时候：
Ruby代码color_lot_manuallies = color_lot.color_lot_manuallies
if color_lot_manuallies.size == 1
end
color_lot.color_lot_manuallies.each do |i|
puts i.id
end
引用SQL (0.6ms) SELECT count(*) AS count_all FROM `color_lot_manuallies` WHERE (`color_lot_manuallies`.color_lot_id = 200)
ColorLotManually Load (0.1ms) SELECT * FROM `color_lot_manuallies` WHERE (`color_lot_manuallies`.color_lot_id = 200)
这样的话rails使用了延迟加载造成两个sql不同，所以后面。color_lot.color_lot_manuallies再次使用时，不会使用缓存的sql。
改成这样：
Ruby代码color_lot_manuallies = color_lot.color_lot_manuallies.all
if color_lot_manuallies.size == 1
end
color_lot.color_lot_manuallies.each do |i|
puts i.id
end引用ColorLotManually Load (0.1ms) SELECT * FROM `color_lot_manuallies` WHERE (`color_lot_manuallies`.color_lot_id = 198)
CACHE (0.0ms) SELECT * FROM `color_lot_manuallies` WHERE (`color_lot_manuallies`.color_lot_id = 198)
加一个all，color_lot.color_lot_manuallies.all，这样的话，后面再次调用时，会使用rails的sql缓存。
Ruby代码#PurchaseOrder
has_one :purchase_order_marketing, :dependent => :destroy
has_many :purchase_invoices, :dependent => :destroy,:through => :purchase_order_marketing
#PurchaseOrderMarketing
has_many :purchase_invoices,:dependent => :destroy
PurchaseOrder.first.purchase_invoices.all(:select => 'purchase_invoices.id') 引用PurchaseOrder Load (0.2ms) SELECT * FROM `purchase_orders` LIMIT 1
PurchaseInvoice Load (0.8ms) SELECT purchase_invoices.id FROM `purchase_invoices` INNER JOIN `purchase_order_marketings` ON `purchase_invoices`.purchase_order_marketing_id = `purchase_order_marketings`.id WHERE ((`purchase_order_marketings`.purchase_order_id = 13)) 一条sql就完成了，真强大。
Ruby代码#article
f sizes
sizes = []
art = self
sf = art.article_secondary_feature
size_group = SizeGroup.find_by_id(sf.size_groupid) if sf
sizes = size_group.sizes if size_group
return sizes
endRuby代码ArticleMarketing.first.article.sizes.all(:select => 'id')引用ArticleMarketing Load (0.3ms) SELECT * FROM `article_marketings` LIMIT 1
Article Load (0.8ms) SELECT * FROM `articles` WHERE (`articles`.`id` = 43)
ArticleSecondaryFeature Load (0.7ms) SELECT `article_secondary_features`.* FROM `article_secondary_features` WHERE (`article_secondary_features`.article_id = 43)
SizeGroup Load (0.6ms) SELECT * FROM `size_groups` WHERE (`size_groups`.`id` = 2) LIMIT 1
Size Load (1.0ms) SELECT id FROM `sizes` WHERE (`sizes`.size_group_id = 2)
find和find_by_id
Ruby代码params[:root].classify.constantize.find_by_id(params[:id])
params[:root].classify.constantize.find(params[:id])引用Opportunity Load (0.1ms) SELECT * FROM `opportunities` WHERE (`opportunities`.`id` = '126') LIMIT 1
Opportunity Load (0.1ms) SELECT * FROM `opportunities` WHERE (`opportunities`.`id` = 126)
find_by_id的SQL多了一行limit 1，对于sql语句而言，有一点性能上的提高。不过有时候我们需要find来捕捉异常。如果能用find_by_id最好了。
都改成find_by_id
引用Opportunity Load (0.1ms) SELECT * FROM `opportunities` WHERE (`opportunities`.`id` = '126') LIMIT 1
CACHE (0.0ms) SELECT * FROM `opportunities` WHERE (`opportunities`.`id` = '126') LIMIT 1这样都有缓存了。rails的自带的缓存是很脆弱的，B.find_by_id(a.id)方法变成a.b时，这个缓存就不会用上。同样前面的例子里改成
Ruby代码color_lot_manuallies = color_lot.color_lot_manuallies.all
ColorLotManually.find_all_by_color_lot_id(color_lot.id)rails自带的缓存也不会用上。
关于||=缓存，参考
http://fuliang.javaeye.com/blog/827321
http://www.javaeye.com/topic/810957
但是有一点，||=不会自动清除或者更新，所以使用的时候还是要注意点，可能会引起取值错误，而且不会报错。
Ruby代码class PortOfDischage < ActiveRecord::Base
def _name
@_name ||= self.city
end
endRuby代码Reloading...
=> true
>> p=PortOfDischage.first
SQL (0.2ms) SET SQL_AUTO_IS_NULL=0
PortOfDischage Load (21.2ms) SELECT * FROM `port_of_dischages` LIMIT 1
PortOfDischage Columns (1.7ms) SHOW FIELDS FROM `port_of_dischages`
+----+------------+-----------------+---------+---------+--------------------------------+--------------------------------+
| id | ap_list_id | ap_marketing_id | city | country | created_at | updated_at |
+----+------------+-----------------+---------+---------+--------------------------------+--------------------------------+
| 1| 2 | 2 | Piraeus | Greece| Thu Oct 07 07:10:21 +0800 2010 | Thu Oct 07 07:10:21 +0800 2010 |
+----+------------+-----------------+---------+---------+--------------------------------+--------------------------------+
1 row in set
>> p._name
=> "Piraeus"
>> p.update_attribute(:city,'p')
SQL (0.2ms) BEGIN
ApMarketing Columns (43.0ms) SHOW FIELDS FROM `ap_marketings`
ApMarketing Load (18.6ms) SELECT * FROM `ap_marketings` WHERE (`ap_marketings`.`id` = 2)
PortOfDischage Update (45.5ms) UPDATE `port_of_dischages` SET `updated_at` = '2010-12-30 13:08:47', `city` = 'p' WHERE `id` = 1
SQL (55.5ms) COMMIT
=> true
>> p._name
=> "Piraeus"
>> p.city
=> "p"
>> p.reload
PortOfDischage Load (0.7ms) SELECT * FROM `port_of_dischages` WHERE (`port_of_dischages`.`id` = 1)
+----+------------+-----------------+------+---------+--------------------------------+--------------------------------+
| id | ap_list_id | ap_marketing_id | city | country | created_at | updated_at |
+----+------------+-----------------+------+---------+--------------------------------+--------------------------------+
| 1| 2 | 2 | p | Greece| Thu Oct 07 07:10:21 +0800 2010 | Thu Dec 30 13:08:47 +0800 2010 |
+----+------------+-----------------+------+---------+--------------------------------+--------------------------------+
1 row in set
>> p._name
=> "Piraeus"
谢谢！ 谢谢！
2gua 发表于 2011-01-11 13:35 http://bbs.chinaunix.net/images/common/back.gif
？？？为啥要谢啊 回复 3# feiyang10086
感谢你为Ruby版块做贡献啊。:emn10: 回复 4# 2gua
客气了！ 呵呵，感谢支持本版啊。
页:
[1]