Unexpected behaviour of non-persisted has_many through associations

[RKushnir]

Consider a basic hm:t association

class Host < ActiveRecord::Base
  has_many :links
  has_many :pages, through: :links
end

(Link and Page are trivial).

If you add and remove an item from links, it still thinks the item is magically present... in the empty collection:

host = Host.new
page = Page.new
host.pages.include? page
 => false
host.pages << page
host.pages.delete page
host.pages.include? page
 => true 
host.pages
 => #<ActiveRecord::Associations::CollectionProxy []>

Tried this on both 3.2.12 and 4.0.0.beta1.

PS. If you save the records, it works as expected.

[senny]

this looks strange indeed. I'll take a look and report back.

[senny]

I can confirm the problem. It's related to the join model that's built when you call << on a hmt association. I0ve written up a test-case and noticed that there could be more hidden errors:

  def test_include?
    ActiveRecord::Base.logger = Logger.new(STDOUT)
    person = Person.new(first_name: "Peter")
    post = Post.new(title: "Cats & Dogs", body: "are pets")

    person.posts << post
    # assert person.posts.include?(post)

    person.posts.delete(post)
    # assert_not person.posts.include?(post), "should not contain the post after deletion but did."

    person.save!
    # assert_not person.posts.include?(post)
  end

This leads to the following SQL statements:

D, [2013-03-05T13:39:19.470303 #5926] DEBUG -- :   SQL (0.2ms)  INSERT INTO "people" ("created_at", "first_name", "updated_at") VALUES (?, ?, ?)  [["created_at", 2013-03-05 12:39:19 UTC], ["first_name", "Peter"], ["updated_at", 2013-03-05 12:39:19 UTC]]
D, [2013-03-05T13:39:19.473836 #5926] DEBUG -- :   SQL (0.1ms)  INSERT INTO "posts" ("body", "title") VALUES (?, ?)  [["body", "are pets"], ["title", "Cats & Dogs"]]
D, [2013-03-05T13:39:19.474598 #5926] DEBUG -- :   SQL (0.1ms)  INSERT INTO "readers" ("person_id", "post_id") VALUES (?, ?)  [["person_id", 5], ["post_id", 12]]

The post was not inserted but the Reader that was built is. Is this the expected behavior?

@carlosantoniodasilva @rafaelfranca what do you think?

[rafaelfranca]

This issue has been automatically marked as stale because it has not been commented on for at least
three months.

The resources of the Rails team are limited, and so we are asking for your help.

If you can still reproduce this error on the 4-1-stable, 4-0-stable branches or on master,
please reply with all of the information you have about it in order to keep the issue open.

Thank you for all your contributions.

[nerdinand]

I tried to reproduce this on master. It still seems to be a problem:
https://gist.github.com/nerdinand/7665496e2b6d26619781

[nerdinand]

Proposed patch still works. Should I send a new PR?

[kamipo]

Fixed by #29593.