Before IN clause optimization 70ddb8a7, Active Record had generated an
SQL with binds when `prepared_statements: true`:

```ruby
# prepared_statements: true
#
#   SELECT `authors`.* FROM `authors` WHERE `authors`.`id` IN (?, ?, ?)
#
# prepared_statements: false
#
#   SELECT `authors`.* FROM `authors` WHERE `authors`.`id` IN (1, 2, 3)
#
Author.where(id: [1, 2, 3]).to_a
```

But now, binds in IN clause is substituted regardless of whether
`prepared_statements: true` or not:

```ruby
# prepared_statements: true
#
#   SELECT `authors`.* FROM `authors` WHERE `authors`.`id`IN (1,2,3)
#
# prepared_statements: false
#
#   SELECT `authors`.* FROM `authors` WHERE `authors`.`id`IN (1,2,3)
#
Author.where(id: [1, 2, 3]).to_a
```

I suppose that is considered as a regression for the context:

> While I would prefer that we fix/avoid the too-many-parameters
problem, but I don't like the idea of globally ditching bind params for
this edge case... we're getting to the point where I'd almost consider
anything that doesn't use a bind to be a bug.

https://github.com/rails/rails/pull/33844#issuecomment-421000003

This makes binds consider whether `prepared_statements: true` or not
(i.e. restore the original behavior as before), but still gain that
optimization when need the substitute binds (`prepared_statements: false`,
`relation.to_sql`). Even when `prepared_statements: true`, it still
much faster than before by optimized (bind node less) binds generation.

```ruby
class Post < ActiveRecord::Base
end

ids = (1..1000).each.map do |n|
  Post.create!.id
end

puts "prepared_statements: #{Post.connection.prepared_statements.inspect}"

Benchmark.ips do |x|
  x.report("where with ids") do
    Post.where(id: ids).to_a
  end
end
```

* Before (200058b0)

`prepared_statements: true`:

```
Warming up --------------------------------------
      where with ids     6.000  i/100ms
Calculating -------------------------------------
      where with ids     63.806  (± 7.8%) i/s -    318.000  in   5.015903s
```

`prepared_statements: false`:

```
Warming up --------------------------------------
      where with ids     7.000  i/100ms
Calculating -------------------------------------
      where with ids     73.550  (± 8.2%) i/s -    371.000  in   5.085672s
```

* Now with this change

`prepared_statements: true`:

```
Warming up --------------------------------------
      where with ids     9.000  i/100ms
Calculating -------------------------------------
      where with ids     91.992  (± 7.6%) i/s -    459.000  in   5.020817s
```

`prepared_statements: false`:

```
Warming up --------------------------------------
      where with ids    10.000  i/100ms
Calculating -------------------------------------
      where with ids    104.335  (± 8.6%) i/s -    520.000  in   5.026425s
```

Eager generate relation methods if a method is on `Kernel`

Follow up of #34122.

Relation method call is relying on method_missing, but if `Kernel` has
the same named method (e.g. `open`, etc), it will invoke Kernel's method
since method_missing is not happened.

To prevent that, eager generate relation methods if a method is the same
name on `Kernel`.

Fixes #39195.

dfa439ee
6708f3a5

`singleton_class` is in Ruby 1.9.2, and there is no use singleton class
eval in the files.

Refactor index creation to use index definition visitor

Current SQL generation code is hard to maintenance, I've found a bug
that create index comment in bulk change table when I'm refactoring
that.

I've found the bug when I'm refactoring index creation code in #39203.

Add tests for to_s(:inspect)

Association callbacks work with abort instead of exceptions [ci skip]

To prevent an object from being added/removed from an association an
abort must be thrown, not an Exception.

Support kwargs for named scopes

We fixed `generate_relation_method` to address kwargs warnings at
#38038, but I missed generated named scopes also need the same fix.

Test case has picked from #39196.
Co-authored-by: NJohn Hawthorn <john@hawthorn.email>

We now always use `arel_attribute` for attribute alias resolution.

Positional hash argument should not be dup-ed.

Make NameError#missing_name work even for real Ruby NameError

On constant missing Ruby call `#inspect` on the receiver to build
the error message.

For instance, the error message for `Foo::Bar` will be `"#{Foo.inspect}::Bar"`.

And since Active Record override the model classes inspect method, this
breaks `missing_name` assumptions.

Until now it worked because missing_name was only called on errors
raised by the classic autoloader, and the classic autoloader calls
`#name` to build its error message.

Add `config.generators.after_generate` for processing to generated files

`remove_foreign_key` doesn't care `:validate` option if database has no feature

Register a callback that will get called right after generators has
finished.

This is useful if users want to process generated files.
For example, can execute an autocorrect of RuboCop for generated files
as like following.

```ruby
config.generators.after_generate do |files|
  system("bundle exec rubocop --auto-correct " + files.join(" "), exception: true)
end
```

Fixes #39170.

#39170 is a regression caused by 362348b5 to maintain kwargs flag to
address kwargs warnings.

Support procs for assert_{enqueued,performed}_with

Recognize Windows-style paths in test runner

PR #33995 added support for specifying the `args` argument of
`assert_enqueued_with` and `assert_performed_with` as a matcher proc.
In doing so, it added undocumented support for specifying the other
arguments as matcher procs as well.  This commit officially documents
that support, and adds tests to ensure the behavior.

Fix assert_{enqueued,performed}_with doc examples [ci skip]

Add a link between config.cache_store in configuration.md and cache_store documentation

Co-authored-by: NJonathan Hefner <jonathan@hefner.pro>

The example tests in the method docs for `assert_enqueued_with` and
`assert_performed_with` previously specified `queue` args in their
assertions but not in their setups.  Thus, the example tests would not
pass as written.  This commit fixes the examples, and properly
demonstrates the `queue` arg.

Load selenium/webdriver only if needed

Currently, 'selenium-webdriver' gem is required to use system tests even if a non-selenium driver is used (such as Poltergeist, Cuprite, etc.). We should require it only if Selenium is used by a user

Clarify that belongs_to is not always a one-to-one relation [ci skip]

"it's" => "its"

Fix the result of aggregations to maintain duplicated "group by" fields