query method #eager_loading? works incorrectly when order by sql contains table name with schema

[senid231]

Steps to reproduce

Here's bug report which reproduces the issue

# frozen_string_literal: true

require 'bundler/inline'

gemfile(true) do
  source 'https://rubygems.org'

  git_source(:github) { |repo| "https://github.com/#{repo}.git" }

  if ENV['RAILS_EDGE']
    gem 'rails', github: 'rails/rails', branch: 'main'
  else
    version = ENV['RAILS_VERSION'] || '~> 6.0'
    gem 'rails', version
  end
  gem 'pg'
end

require 'active_record'
require 'minitest/autorun'
require 'logger'

# Keep note: modify below three lines according to your postgresql server configurations
`psql -c "DROP DATABASE IF EXISTS test_rails_references_values;"`
`psql -c "CREATE DATABASE test_rails_references_values;"`
ActiveRecord::Base.establish_connection(adapter: 'postgresql', database: 'test_rails_references_values')
ActiveRecord::Base.logger = Logger.new($stdout)

ActiveRecord::Schema.define do
  create_schema 'shop'

  create_table 'posts', force: true do |t|
    t.string :title
    t.string :content
  end

  create_table 'post_details', force: true do |t|
    t.integer :post_id
    t.string :summary
  end

  create_table 'shop.posts', force: true do |t|
    t.string :title
    t.string :content
  end

  create_table 'shop.post_details', force: true do |t|
    t.integer :post_id
    t.string :summary
  end
end

class ApplicationRecord < ActiveRecord::Base
  self.abstract_class = true
end

class Post < ApplicationRecord
  self.table_name = 'posts'
  has_one :post_detail, class_name: 'PostDetail', dependent: :delete
end

class PostDetail < ApplicationRecord
  self.table_name = 'post_details'
  belongs_to :post
end

class ShopPost < ApplicationRecord
  self.table_name = 'shop.posts'
  has_one :post_detail, class_name: 'PostDetail', dependent: :delete
end

class ShopPostDetail < ApplicationRecord
  self.table_name = 'shop.post_details'
  belongs_to :post
end

class BugTest < Minitest::Test
  def test_references_values_without_schema_when_order
    query_reorder = Post.includes(:post_detail).order(Arel.sql('posts.id DESC'))

    refute query_reorder.eager_loading?
    refute query_reorder.send(:references_eager_loaded_tables?)

    assert_equal [Post.table_name], query_reorder.references_values
  end

  def test_references_values_with_schema_when_order
    query_reorder = ShopPost.includes(:post_detail).order(Arel.sql('shop.posts.id DESC'))

    refute query_reorder.eager_loading?
    refute query_reorder.send(:references_eager_loaded_tables?)

    assert_equal [ShopPost.table_name], query_reorder.references_values
  end
end

Expected behavior

for query like ShopPost.includes(:post_detail).order(Arel.sql('shop.posts.id DESC')) #eager_load? should return false.
As result includes should not eager loaded because when are not present in where/order clauses.

Actual behavior

for query like ShopPost.includes(:post_detail).order(Arel.sql('shop.posts.id DESC')) #eager_load? returns false.
As result includes always eager loaded.

System configuration

Rails version: 6.1.3.2, 5.2.6, edge 509f3ca

Ruby version: 2.7.3

PostgreSQL: 12

Additional information

This bug has big performance impact on calculations like

ShopPost.includes(:post_detail).order(Arel.sql('shop.posts.id DESC')).count

when many associations are included, because it transforms into SELECT COUNT(DISTINCT ...

In ActiveAdmin resources always apply sort as strings which leads to slow queries for all tables which have table schema.

[zzak]

I wonder if this is related to #41489 🤔 @interip thoughts?

[intrip]

@zzak thanks for pointing this out but I think this is a different issue, anyways I'll work on a PR to address this issue too.

[intrip]

#42602 should fix this

[rails-bot]

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 6-1-stable branch or on main, please reply with all of the information you have about it in order to keep the issue open.
Thank you for all your contributions.

[intrip]

still relevant, attached PR

[rails-bot]

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 6-1-stable branch or on main, please reply with all of the information you have about it in order to keep the issue open.
Thank you for all your contributions.

[intrip]

ping

[rails-bot]

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 7-0-stable branch or on main, please reply with all of the information you have about it in order to keep the issue open.
Thank you for all your contributions.

[intrip]

There's an attached PR, can anyone please reopen it?