提交 cdbd64d1 编写于 作者: R Rafael Mendonça França

Merge pull request #9252 from senny/8423_hmt_preloading_bug

don't cache invalid subsets when preloading hmt associations
## Rails 4.0.0 (unreleased) ##
* Preloading `has_many :through` associations with conditions won't
cache the `:through` association. This will prevent invalid
subsets to be cached.
Fixes #8423.
Example:
class User
has_many :posts
has_many :recent_comments, -> { where('created_at > ?', 1.week.ago) }, :through => :posts
end
a_user = User.includes(:recent_comments).first
# this is preloaded
a_user.recent_comments
# fetching the recent_comments through the posts association won't preload it.
a_user.posts
*Yves Senn*
* Don't run after_commit callback when creating through an association
if saving the record fails.
......
......@@ -31,7 +31,8 @@ def through_records_by_owner
through_records = Array.wrap(owner.send(through_reflection.name))
# Dont cache the association - we would only be caching a subset
if reflection.options[:source_type] && through_reflection.collection?
if (through_scope != through_reflection.klass.unscoped) ||
(reflection.options[:source_type] && through_reflection.collection?)
owner.association(through_reflection.name).reset
end
......
......@@ -1162,4 +1162,10 @@ def test_deep_including_through_habtm
Post.where('1 = 0').scoping { Comment.preload(:post).find(1).post }
)
end
test "preloading does not cache has many association subset when preloaded with a through association" do
author = Author.includes(:comments_with_order_and_conditions, :posts).first
assert_no_queries { assert_equal 2, author.comments_with_order_and_conditions.size }
assert_no_queries { assert_equal 5, author.posts.size, "should not cache a subset of the association" }
end
end
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册