the module name was changed in 8e814a0a

Its already doc'ed in
activerecord/lib/active_record/connection_adapters/abstract/schema_statements.rb

`exec_query` create `ActiveRecord::Result` instance. It is better to use
`select_value` instead of `exec_query` for performance.

`select_one` create `ActiveRecord::Result` instance. It is better to use
`select_rows` instead of `select_one` for performance.

MySQL does not support partial index. And, the create index algorithm in
create table can not be specified.

Example:

    create_table :foos do |t|
      t.string :string_en, collation: 'en_US.UTF-8'
      t.text   :text_ja,   collation: 'ja_JP.UTF-8'
    end

Some databases like MySQL allow defining collation charset for specific
columns.

Follow up #18914.

If the adapter supports indexes in create table, generated SQL is
slightly more efficient.

Its nodoc'ed for the other implementations, and doc'ed in the base
class, just like the other change_column* methods.

The microseconds handling was already moved to `Quoting#quoted_date`.

`change_column_null` is doc'ed only in
ActiveRecord::ConnectionAdapters::SchemaStatements, so it would make
sense to :nodoc: it elsewhere.

When `AR::Base.save!` or `AR::Base.destroy!` is called and an exception
is raised, the exception doesn't have any error message or has a weird
message like `#<FailedBulb:0x0000000907b4b8>`. Give a better message so
we can easily understand why it's failing to save/destroy.

 - This method is used only by adapters to validate length of new index names.

 - Added documentation for index_name_exists? and rename_index.
 - Also changed rails to \Rails in documentation of
   allowed_index_name_length.

The `db:schema:cache:dump` rake task dumps the database schema structure
to `db/schema_cache.dump`. If this file is present, the schema details
are loaded into the currently checked out connection by a railtie while
Rails is booting, to avoid having to query the database for its schema.

The schema cache dump is only applied to the initial connection used to
boot the application though; other connections from the same pool are
created with an empty schema cache, and still have to load the structure
of each table directly from the database.

With this change, a copy of the schema cache is associated with the
connection pool and applied to connections as they are created.

This reverts commit 08ff4ccb, reversing
changes made to 6c9ed6db.

Caused by #17920.

Closes #19545.

This patch introduced regressions because initializers were no longer
loaded. Specifically missing inflections result in broken restores of
the database.

Also added reference for to_date, as it belongs historically.

follow up to 107526e8

In 1f006c an option was added called :class to allow passing anonymous
classes to association definitions. Since using :class instead of
:class_name is a fairly common typo even amongst experienced developers
this can result in hard to debug errors arising in raise_on_type_mismatch?

To fix this we're renaming the option from :class to :anonymous_class as
that is a more correct description of what the option is for. Since this
was an internal, undocumented option there is no need for a deprecation.

Fixes #19659

This will avoid the indirection of having calling id since we already
know which is the primary key column.

Also this will make explicit the behavior since it is not clear that id
gets the right primary key value and not just the value of the "id"
column.

Fixes #19776

change test variable names and use more verbose on method

less verbose

use _read_attribute instead of send

To me it seems like this should only be the case if `autosave: true` is
set on the association. However, when implemented that way, it caused
issues with has many associations, where we have explicit tests stating
that child records are updated when the parent is new, even if autosave
is not set (presumably to update the parent id, but other changed
attributes would be persisted as well).

It's quirky, but at least we should be consistently quirky. This
constitutes a minor but subtle change in behavior, and therefore should
not be backported to 4.2 and earlier.

Fixes #19782

IMO we shouldn't encourage users to use methods they shouldn't need to know about. As Song (in this example) inherits from ActiveRecord, we can use super here instead to get the same effect with the bonus of not knowing how active record actually implements these methods.

The `index` option used with `timestamps` should be passed to both
`column` definitions for `created_at` and `updated_at` rather than just
the first.

This was happening because `Hash#delete` is used to extract the `index`
option passed to `timestamps`, thereby mutating the `options` hash
in-place. Now take a copy of the `options` before deleting so that the
original is not modified.

This reverts commit 524d4059, reversing
changes made to 34d3a609.

Reasoning behind the revert are in the PR discussion:
https://github.com/rails/rails/pull/19755

- This means that types can no longer cast to/from `Set`, and reasonably
  work with `where` (we already have this problem for `array`/`json`
  types on pg)
- This adds precedent for every other `Enumerable`, and we can't target
  `Enumerable` directly.
- Calling `to_a` on a `Set` is reasonable.

Previously `#where` used to treat `Set`objects as nil, but now it treats
them as an array:

  set = Set.new([1, 2])
  Author.where(:id => set)
  # => SELECT "authors".* FROM "authors" WHERE "authors"."id" IN (1, 2)