Phenomenon of mysterious value of the position is skipped by one

[akicho8]

It will become as follows:

book.articles.create!.position # =>  2
book.articles.create!.position # =>  4
book.articles.create!.position # =>  6

Test code

If (A) or (B) or (C) comment out, position value is 1.

require "active_record"
require "acts_as_list"
require "acts_as_list/version"

ActiveRecord::VERSION::STRING     # => "4.2.2"
ActiveRecord::Acts::List::VERSION # => "0.7.2"

ActiveRecord::Base.logger = ActiveSupport::Logger.new(STDOUT)
ActiveSupport::LogSubscriber.colorize_logging = false
ActiveRecord::Migration.verbose = false
ActiveRecord::Base.establish_connection(:adapter => "sqlite3", :database => ":memory:")
ActiveRecord::Schema.define do
  create_table :books do |t|
    t.string :foo
  end
  create_table :articles do |t|
    t.belongs_to :book
    t.integer :position
  end
end

class Book < ActiveRecord::Base
  has_one :article              # (A)
  has_many :articles
  accepts_nested_attributes_for :articles # (B)
end

class Article < ActiveRecord::Base
  belongs_to :book
  acts_as_list scope: :book
  after_create do
    book.update!(:foo => 1)     # (C)
  end
end

book = Book.create!
article = book.articles.create!
article.position                # => 2
# >>    (0.4ms)  CREATE TABLE "books" ("id" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL, "foo" varchar) 
# >>    (0.1ms)  CREATE TABLE "articles" ("id" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL, "book_id" integer, "position" integer) 
# >>    (0.0ms)  begin transaction
# >>   SQL (0.1ms)  INSERT INTO "books" DEFAULT VALUES
# >>    (0.0ms)  commit transaction
# >>    (0.0ms)  begin transaction
# >>   Article Load (0.2ms)  SELECT  "articles".* FROM "articles" WHERE (articles.position IS NOT NULL) AND "articles"."book_id" = ?  ORDER BY articles.position DESC LIMIT 1  [["book_id", 1]]
# >>   SQL (0.1ms)  INSERT INTO "articles" ("book_id", "position") VALUES (?, ?)  [["book_id", 1], ["position", 1]]
# >>   Book Load (0.1ms)  SELECT  "books".* FROM "books" WHERE "books"."id" = ? LIMIT 1  [["id", 1]]
# >>   SQL (0.1ms)  UPDATE "books" SET "foo" = ? WHERE "books"."id" = ?  [["foo", "1"], ["id", 1]]
# >>   Article Load (0.1ms)  SELECT  "articles".* FROM "articles" WHERE (articles.position IS NOT NULL) AND "articles"."book_id" = ?  ORDER BY articles.position DESC LIMIT 1  [["book_id", 1]]
# >>   SQL (0.1ms)  UPDATE "articles" SET "book_id" = ?, "position" = ?, "id" = ? WHERE "articles"."id" = ?  [["book_id", 1], ["position", 2], ["id", 1], ["id", 1]]
# >>    (0.1ms)  SELECT COUNT(*) FROM "articles" WHERE "articles"."book_id" = ? AND (position = 2)  [["book_id", 1]]
# >>    (0.0ms)  commit transaction

[HannesBenson]

I've just hit the same issue and found that it was happening when save is called twice while the object is dirty. In our case save was called which fired a callback and the callback saved again. Since position was in the changes hash in both cases the position gets updated twice. Will see when I get some time to write a fix. I resolved it by adding clear_changes_information to the callback but it might be worthwhile removing the position key from the changes hash when the object gets saved with save(validate: false)

[brendon]

This should be fixed now. Feel free to let me know if not.