activerecord 7.0.7.2 → 7.1.0.beta1

data/CHANGELOG.md

@@ -1,2317 +1,2178 @@
-## Rails 7.0.7.2 (August 22, 2023) ##
+## Rails 7.1.0.beta1 (September 13, 2023) ##
 
-*   No changes.
+*   Encryption now supports `support_unencrypted_data` being set per-attribute.
 
+    You can now opt out of `support_unencrypted_data` on a specific encrypted attribute.
+    This only has an effect if `ActiveRecord::Encryption.config.support_unencrypted_data == true`.
 
-## Rails 7.0.7.1 (August 22, 2023) ##
-
-*   No changes.
-
-
-## Rails 7.0.7 (August 09, 2023) ##
-
-*   Restores functionality to the missing method when using enums and fixes.
+    ```ruby
+    class User < ActiveRecord::Base
+      encrypts :name, deterministic: true, support_unencrypted_data: false
+      encrypts :email, deterministic: true
+    end
+    ```
 
-    *paulreece*
+    *Alex Ghiculescu*
 
-*   Fix `StatementCache::Substitute` with serialized type.
+*   Add instrumentation for Active Record transactions
 
-    *ywenc*
+    Allows subscribing to transaction events for tracking/instrumentation. The event payload contains the connection, as well as timing details.
 
-*   Fix `:db_runtime` on notification payload when application have multiple databases.
+    ```ruby
+    ActiveSupport::Notifications.subscribe("transaction.active_record") do |event|
+      puts "Transaction event occurred!"
+      connection = event.payload[:connection]
+      puts "Connection: #{connection.inspect}"
+    end
+    ```
 
-    *Eileen M. Uchitelle*
+    *Daniel Colson*, *Ian Candy*
 
-*   Correctly dump check constraints for MySQL 8.0.16+.
+*   Support composite foreign keys via migration helpers.
 
-    *Steve Hill*
+    ```ruby
+    # Assuming "carts" table has "(shop_id, user_id)" as a primary key.
 
-*   Fix `ActiveRecord::QueryMethods#in_order_of` to include `nil`s, to match the
-    behavior of `Enumerable#in_order_of`.
+    add_foreign_key(:orders, :carts, primary_key: [:shop_id, :user_id])
 
-    For example, `Post.in_order_of(:title, [nil, "foo"])` will now include posts
-    with `nil` titles, the same as `Post.all.to_a.in_order_of(:title, [nil, "foo"])`.
+    remove_foreign_key(:orders, :carts, primary_key: [:shop_id, :user_id])
+    foreign_key_exists?(:orders, :carts, primary_key: [:shop_id, :user_id])
+    ```
 
     *fatkodima*
 
-*   Revert "Fix autosave associations with validations added on `:base` of the associated objects."
-
-    This change intended to remove the :base attribute from the message,
-    but broke many assumptions which key these errors were stored.
-
-    *zzak*
-
-*   Fix `#previously_new_record?` to return true for destroyed records.
+*   Adds support for `if_not_exists` when adding a check constraint.
 
-    Before, if a record was created and then destroyed, `#previously_new_record?` would return true.
-    Now, any UPDATE or DELETE to a record is considered a change, and will result in `#previously_new_record?`
-    returning false.
-
-    *Adrianna Chang*
-
-*   Revert breaking changes to `has_one` relationship deleting the old record before the new one is validated.
-
-    *zzak*
-
-*   Fix support for Active Record instances being uses in queries.
-
-    As of `7.0.5`, query arguments were deep duped to avoid mutations impacting
-    the query cache, but this had the adverse effect to clearing the primary key when
-    the query argument contained an `ActiveRecord::Base` instance.
-
-    This broke the `noticed` gem.
-
-    *Jean Boussier*
-
-
-## Rails 7.0.6 (June 29, 2023) ##
-
-*   Fix autosave associations with validations added on `:base` of the associated objects.
-
-    *fatkodima*
+    ```ruby
+    add_check_constraint :posts, "post_type IN ('blog', 'comment', 'share')", if_not_exists: true
+    ```
 
-*   Fix result with anonymous PostgreSQL columns of different type from json.
+    *Cody Cutrer*
 
-    *Oleksandr Avoiants*
+*   Raise an `ArgumentError` when `#accepts_nested_attributes_for` is declared more than once for an association in
+    the same class. Previously, the last declaration would silently override the previous one. Overriding in a subclass
+    is still allowed.
 
-*   Preserve timestamp when setting an `ActiveSupport::TimeWithZone` value to `timestamptz` attribute.
+    *Joshua Young*
 
-    *fatkodima*
+*   Deprecate `rewhere` argument on `#merge`.
 
-*   Fix assignment into an `has_one` relationship deleting the old record before the new one is validated.
+    The `rewhere` argument on `#merge`is deprecated without replacement and
+    will be removed in Rails 7.2.
 
-    *Jean Boussier*
+    *Adam Hess*
 
-*   Fix where on association with has_one/has_many polymorphic relations.
+*   Fix unscope is not working in specific case
 
     Before:
     ```ruby
-    Treasure.where(price_estimates: PriceEstimate.all)
-    #=> SELECT (...) WHERE "treasures"."id" IN (SELECT "price_estimates"."estimate_of_id" FROM "price_estimates")
+    Post.where(id: 1...3).unscope(where: :id).to_sql # "SELECT `posts`.* FROM `posts` WHERE `posts`.`id` >= 1 AND `posts`.`id` < 3"
+
     ```
 
-    Later:
+    After:
     ```ruby
-    Treasure.where(price_estimates: PriceEstimate.all)
-    #=> SELECT (...) WHERE "treasures"."id" IN (SELECT "price_estimates"."estimate_of_id" FROM "price_estimates" WHERE "price_estimates"."estimate_of_type" = 'Treasure')
+    Post.where(id: 1...3).unscope(where: :id).to_sql # "SELECT `posts`.* FROM `posts`"
     ```
 
-    *Lázaro Nixon*
-
-*   Fix decrementing counter caches on optimistically locked record deletion
-
-    *fatkodima*
-
-*   Ensure binary-destined values have binary encoding during type cast.
-
-    *Matthew Draper*
-
-*   Preserve existing column default functions when altering table in SQLite.
-
-    *fatkodima*
-
-*   Remove table alias added when using `where.missing` or `where.associated`.
-
-    *fatkodima*
+    Fixes #48094.
 
-*   Fix `Enumerable#in_order_of` to only flatten first level to preserve nesting.
+    *Kazuya Hatanaka*
 
-    *Miha Rekar*
+*   Change `has_secure_token` default to `on: :initialize`
 
+    Change the new default value from `on: :create` to `on: :initialize`
 
-## Rails 7.0.5.1 (June 26, 2023) ##
+    Can be controlled by the `config.active_record.generate_secure_token_on`
+    configuration:
 
-*   No changes.
-
-## Rails 7.0.5 (May 24, 2023) ##
-
-*   Type cast `#attribute_changed?` `:from` and `:to` options.
-
-    *Andrew Novoselac*
-
-*   Fix `index_exists?` when column is an array.
-
-    *Eileen M. Uchitelle*
-
-*   Handle `Date` objects for PostgreSQL `timestamptz` columns.
+    ```ruby
+    config.active_record.generate_secure_token_on = :create
+    ```
 
-    *Alex Ghiculescu*
+    *Sean Doyle*
 
-*   Fix collation for changing column to non-string.
+*   Fix `change_column` not setting `precision: 6` on `datetime` columns when
+    using 7.0+ Migrations and SQLite.
 
     *Hartley McGuire*
 
-*   Map through subtype in `PostgreSQL::OID::Array`.
-
-    *Jonathan Hefner*
-
-*   Store correct environment in `internal_metadata` when run rails `db:prepare`.
-
-    *fatkodima*
-
-*   Make sure `ActiveRecord::Relation#sum` works with objects that implement `#coerce` without deprecation.
-
-    *Alex Ghiculescu*
-
-*   Fix retrieving foreign keys referencing tables named like keywords in PostgreSQL and MySQL.
-
-    *fatkodima*
-
-*   Support UUIDs in Disable Joins.
-
-    *Samuel Cochran*
-
-*   Fix Active Record's explain for queries starting with comments.
-
-    *fatkodima*
-
-*   Fix incorrectly preloading through association records when middle association has been loaded.
-
-    *Joshua Young*
-
-*   Fix where.missing and where.associated for parent/child associations.
-
-    *fatkodima*
-
-*   Fix Enumerable#in_order_of to preserve duplicates.
-
-    *fatkodima*
-
-*   Fix autoincrement on primary key for mysql.
+*   Support composite identifiers in `to_key`
 
-    *Eileen M. Uchitelle*
+    `to_key` avoids wrapping `#id` value into an `Array` if `#id` already an array
 
-*   Restore ability to redefine column in `create_table` for Rails 5.2 migrations.
-
-    *fatkodima*
-
-*   Fix schema cache dumping of virtual columns.
-
-    *fatkodima*
-
-*   Fix Active Record grouped calculations on joined tables on column present in both tables.
-
-    *fatkodima*
+    *Nikita Vasilevsky*
 
-*   Fix mutation detection for serialized attributes backed by binary columns.
+*   Add validation option for `enum`
 
-    *Jean Boussier*
-
-*   Fix a bug where using groups and counts with long table names would return incorrect results.
+    ```ruby
+    class Contract < ApplicationRecord
+      enum :status, %w[in_progress completed], validate: true
+    end
+    Contract.new(status: "unknown").valid? # => false
+    Contract.new(status: nil).valid? # => false
+    Contract.new(status: "completed").valid? # => true
 
-    *Shota Toguchi*, *Yusaku Ono*
+    class Contract < ApplicationRecord
+      enum :status, %w[in_progress completed], validate: { allow_nil: true }
+    end
+    Contract.new(status: "unknown").valid? # => false
+    Contract.new(status: nil).valid? # => true
+    Contract.new(status: "completed").valid? # => true
+    ```
 
-*   Fix erroneous nil default precision on virtual datetime columns.
+    *Edem Topuzov*, *Ryuta Kamizono*
 
-    Prior to this change, virtual datetime columns did not have the same
-    default precision as regular datetime columns, resulting in the following
-    being erroneously equivalent:
+*   Allow batching methods to use already loaded relation if available
 
-        t.virtual :name, type: datetime,                 as: "expression"
-        t.virtual :name, type: datetime, precision: nil, as: "expression"
+    Calling batch methods on already loaded relations will use the records previously loaded instead of retrieving
+    them from the database again.
 
-    This change fixes the default precision lookup, so virtual and regular
-    datetime column default precisions match.
+    *Adam Hess*
 
-    *Sam Bostock*
+*   Deprecate `read_attribute(:id)` returning the primary key if the primary key is not `:id`.
 
-*   Fix a case where the query cache can return wrong values. See #46044
+    Starting in Rails 7.2, `read_attribute(:id)` will return the value of the id column, regardless of the model's
+    primary key. To retrieve the value of the primary key, use `#id` instead. `read_attribute(:id)` for composite
+    primary key models will now return the value of the id column.
 
-    *Aaron Patterson*
+    *Adrianna Chang*
 
+*   Fix `change_table` setting datetime precision for 6.1 Migrations
 
-## Rails 7.0.4.3 (March 13, 2023) ##
+    *Hartley McGuire*
 
-*   No changes.
+*   Fix change_column setting datetime precision for 6.1 Migrations
 
+    *Hartley McGuire*
 
-## Rails 7.0.4.2 (January 24, 2023) ##
+*   Add `ActiveRecord::Base#id_value` alias to access the raw value of a record's id column.
 
-*   No changes.
+    This alias is only provided for models that declare an `:id` column.
 
+    *Adrianna Chang*
 
-## Rails 7.0.4.1 (January 17, 2023) ##
+*   Fix previous change tracking for `ActiveRecord::Store` when using a column with JSON structured database type
 
-*   Make sanitize_as_sql_comment more strict
+    Before, the methods to access the changes made during the last save `#saved_change_to_key?`, `#saved_change_to_key`, and `#key_before_last_save` did not work if the store was defined as a `store_accessor` on a column with a JSON structured database type
 
-    Though this method was likely never meant to take user input, it was
-    attempting sanitization. That sanitization could be bypassed with
-    carefully crafted input.
+    *Robert DiMartino*
 
-    This commit makes the sanitization more robust by replacing any
-    occurrences of "/*" or "*/" with "/ *" or "* /". It also performs a
-    first pass to remove one surrounding comment to avoid compatibility
-    issues for users relying on the existing removal.
+*   Fully support `NULLS [NOT] DISTINCT` for PostgreSQL 15+ indexes.
 
-    This also clarifies in the documentation of annotate that it should not
-    be provided user input.
+    Previous work was done to allow the index to be created in a migration, but it was not
+    supported in schema.rb. Additionally, the matching for `NULLS [NOT] DISTINCT` was not
+    in the correct order, which could have resulted in inconsistent schema detection.
 
-    [CVE-2023-22794]
+    *Gregory Jones*
 
-*   Added integer width check to PostgreSQL::Quoting
+*   Allow escaping of literal colon characters in `sanitize_sql_*` methods when named bind variables are used
 
-    Given a value outside the range for a 64bit signed integer type
-    PostgreSQL will treat the column type as numeric. Comparing
-    integer values against numeric values can result in a slow
-    sequential scan.
+    *Justin Bull*
 
-    This behavior is configurable via
-    ActiveRecord::Base.raise_int_wider_than_64bit which defaults to true.
+*   Fix `#previously_new_record?` to return true for destroyed records.
 
-    [CVE-2022-44566]
+    Before, if a record was created and then destroyed, `#previously_new_record?` would return true.
+    Now, any UPDATE or DELETE to a record is considered a change, and will result in `#previously_new_record?`
+    returning false.
 
+    *Adrianna Chang*
 
-## Rails 7.0.4 (September 09, 2022) ##
+*   Specify callback in `has_secure_token`
 
-*   Symbol is allowed by default for YAML columns
+    ```ruby
+    class User < ApplicationRecord
+      has_secure_token on: :initialize
+    end
 
-    *Étienne Barrié*
+    User.new.token # => "abc123...."
+    ```
 
-*   Fix `ActiveRecord::Store` to serialize as a regular Hash
+    *Sean Doyle*
 
-    Previously it would serialize as an `ActiveSupport::HashWithIndifferentAccess`
-    which is wasteful and cause problem with YAML safe_load.
+*   Fix incrementation of in memory counter caches when associations overlap
 
-    *Jean Boussier*
+    When two associations had a similarly named counter cache column, Active Record
+    could sometime increment the wrong one.
 
-*   Add `timestamptz` as a time zone aware type for PostgreSQL
+    *Jacopo Beschi*, *Jean Boussier*
 
-    This is required for correctly parsing `timestamp with time zone` values in your database.
+*   Don't show secrets for Active Record's `Cipher::Aes256Gcm#inspect`.
 
-    If you don't want this, you can opt out by adding this initializer:
+    Before:
 
     ```ruby
-    ActiveRecord::Base.time_zone_aware_types -= [:timestamptz]
+    ActiveRecord::Encryption::Cipher::Aes256Gcm.new(secret).inspect
+    "#<ActiveRecord::Encryption::Cipher::Aes256Gcm:0x0000000104888038 ... @secret=\"\\xAF\\bFh]LV}q\\nl\\xB2U\\xB3 ... >"
     ```
 
-    *Alex Ghiculescu*
-
-*   Fix supporting timezone awareness for `tsrange` and `tstzrange` array columns.
+    After:
 
     ```ruby
-    # In database migrations
-    add_column :shops, :open_hours, :tsrange, array: true
-    # In app config
-    ActiveRecord::Base.time_zone_aware_types += [:tsrange]
-    # In the code times are properly converted to app time zone
-    Shop.create!(open_hours: [Time.current..8.hour.from_now])
+    ActiveRecord::Encryption::Cipher::Aes256Gcm(secret).inspect
+    "#<ActiveRecord::Encryption::Cipher::Aes256Gcm:0x0000000104888038>"
     ```
 
-    *Wojciech Wnętrzak*
-
-*   Resolve issue where a relation cache_version could be left stale.
-
-    Previously, when `reset` was called on a relation object it did not reset the cache_versions
-    ivar. This led to a confusing situation where despite having the correct data the relation
-    still reported a stale cache_version.
+    *Petrik de Heus*
 
-    Usage:
+*   Bring back the historical behavior of committing transaction on non-local return.
 
     ```ruby
-    developers = Developer.all
-    developers.cache_version
-
-    Developer.update_all(updated_at: Time.now.utc + 1.second)
-
-    developers.cache_version # Stale cache_version
-    developers.reset
-    developers.cache_version # Returns the current correct cache_version
+    Model.transaction do
+      model.save
+      return
+      other_model.save # not executed
+    end
     ```
 
-    Fixes #45341.
+    Historically only raised errors would trigger a rollback, but in Ruby `2.3`, the `timeout` library
+    started using `throw` to interrupt execution which had the adverse effect of committing open transactions.
 
-    *Austen Madden*
+    To solve this, in Active Record 6.1 the behavior was changed to instead rollback the transaction as it was safer
+    than to potentially commit an incomplete transaction.
 
-*   Fix `load_async` when called on an association proxy.
+    Using `return`, `break` or `throw` inside a `transaction` block was essentially deprecated from Rails 6.1 onwards.
 
-    Calling `load_async` directly an association would schedule
-    a query but never use it.
+    However with the release of `timeout 0.4.0`, `Timeout.timeout` now raises an error again, and Active Record is able
+    to return to its original, less surprising, behavior.
+
+    This historical behavior can now be opt-ed in via:
 
-    ```ruby
-    comments = post.comments.load_async # schedule a query
-    comments.to_a # perform an entirely new sync query
+    ```
+    Rails.application.config.active_record.commit_transaction_on_non_local_return = true
     ```
 
-    Now it does use the async query, however note that it doesn't
-    cause the association to be loaded.
+    And is the default for new applications created in Rails 7.1.
 
     *Jean Boussier*
 
-*   Fix eager loading for models without primary keys.
+*   Deprecate `name` argument on `#remove_connection`.
 
-    *Anmol Chopra*, *Matt Lawrence*, and *Jonathan Hefner*
+    The `name` argument is deprecated on `#remove_connection` without replacement. `#remove_connection` should be called directly on the class that established the connection.
 
-*   `rails db:schema:{dump,load}` now checks `ENV["SCHEMA_FORMAT"]` before config
-
-    Since `rails db:structure:{dump,load}` was deprecated there wasn't a simple
-    way to dump a schema to both SQL and Ruby formats. You can now do this with
-    an environment variable. For example:
-
-    ```
-    SCHEMA_FORMAT=sql rake db:schema:dump
-    ```
-
-    *Alex Ghiculescu*
-
-*   Fix Hstore deserialize regression.
-
-    *edsharp*
+    *Eileen M. Uchitelle*
 
+*   Fix has_one through singular building with inverse.
 
-## Rails 7.0.3.1 (July 12, 2022) ##
+    Allows building of records from an association with a has_one through a
+    singular association with inverse. For belongs_to through associations,
+    linking the foreign key to the primary key model isn't needed.
+    For has_one, we cannot build records due to the association not being mutable.
 
-*   Change ActiveRecord::Coders::YAMLColumn default to safe_load
+    *Gannon McGibbon*
 
-    This adds two new configuration options The configuration options are as
-    follows:
+*   Disable database prepared statements when query logs are enabled
 
-    * `config.active_record.use_yaml_unsafe_load`
+    Prepared Statements and Query Logs are incompatible features due to query logs making every query unique.
 
-    When set to true, this configuration option tells Rails to use the old
-    "unsafe" YAML loading strategy, maintaining the existing behavior but leaving
-    the possible escalation vulnerability in place.  Setting this option to true
-    is *not* recommended, but can aid in upgrading.
+    *zzak, Jean Boussier*
 
-    * `config.active_record.yaml_column_permitted_classes`
+*   Support decrypting data encrypted non-deterministically with a SHA1 hash digest.
 
-    The "safe YAML" loading method does not allow all classes to be deserialized
-    by default.  This option allows you to specify classes deemed "safe" in your
-    application.  For example, if your application uses Symbol and Time in
-    serialized data, you can add Symbol and Time to the allowed list as follows:
+    This adds a new Active Record encryption option to support decrypting data encrypted
+    non-deterministically with a SHA1 hash digest:
 
     ```
-    config.active_record.yaml_column_permitted_classes = [Symbol, Date, Time]
+    Rails.application.config.active_record.encryption.support_sha1_for_non_deterministic_encryption = true
     ```
 
-    [CVE-2022-32224]
-
-
-## Rails 7.0.3 (May 09, 2022) ##
-
-*   Some internal housekeeping on reloads could break custom `respond_to?`
-    methods in class objects that referenced reloadable constants. See
-    [#44125](https://github.com/rails/rails/issues/44125) for details.
-
-    *Xavier Noria*
-
-*   Fixed MariaDB default function support.
-
-    Defaults would be written wrong in "db/schema.rb" and not work correctly
-    if using `db:schema:load`. Further more the function name would be
-    added as string content when saving new records.
-
-    *kaspernj*
+    The new option addresses a problem when upgrading from 7.0 to 7.1. Due to a bug in how Active Record
+    Encryption was getting initialized, the key provider used for non-deterministic encryption were using
+    SHA-1 as its digest class, instead of the one configured globally by Rails via
+    `Rails.application.config.active_support.key_generator_hash_digest_class`.
 
-*   Fix `remove_foreign_key` with `:if_exists` option when foreign key actually exists.
+    *Cadu Ribeiro and Jorge Manrubia*
 
-    *fatkodima*
+*   Added PostgreSQL migration commands for enum rename, add value, and rename value.
 
-*   Remove `--no-comments` flag in structure dumps for PostgreSQL
-
-    This broke some apps that used custom schema comments. If you don't want
-    comments in your structure dump, you can use:
+    `rename_enum` and `rename_enum_value` are reversible. Due to Postgres
+    limitation, `add_enum_value` is not reversible since you cannot delete enum
+    values. As an alternative you should drop and recreate the enum entirely.
 
     ```ruby
-    ActiveRecord::Tasks::DatabaseTasks.structure_dump_flags = ['--no-comments']
+    rename_enum :article_status, to: :article_state
     ```
 
-    *Alex Ghiculescu*
+    ```ruby
+    add_enum_value :article_state, "archived" # will be at the end of existing values
+    add_enum_value :article_state, "in review", before: "published"
+    add_enum_value :article_state, "approved", after: "in review"
+    ```
 
-*   Use the model name as a prefix when filtering encrypted attributes from logs.
+    ```ruby
+    rename_enum_value :article_state, from: "archived", to: "deleted"
+    ```
 
-    For example, when encrypting `Person#name` it will add `person.name` as a filter
-    parameter, instead of just `name`. This prevents unintended filtering of parameters
-    with a matching name in other models.
+    *Ray Faddis*
 
-    *Jorge Manrubia*
+*   Allow composite primary key to be derived from schema
 
-*   Fix quoting of `ActiveSupport::Duration` and `Rational` numbers in the MySQL adapter.
+    Booting an application with a schema that contains composite primary keys
+    will not issue warning and won't `nil`ify the `ActiveRecord::Base#primary_key` value anymore.
 
-    *Kevin McPhillips*
+    Given a `travel_routes` table definition and a `TravelRoute` model like:
+    ```ruby
+    create_table :travel_routes, primary_key: [:origin, :destination], force: true do |t|
+      t.string :origin
+      t.string :destination
+    end
 
-*   Fix `change_column_comment` to preserve column's AUTO_INCREMENT in the MySQL adapter
+    class TravelRoute < ActiveRecord::Base; end
+    ```
+    The `TravelRoute.primary_key` value will be automatically derived to `["origin", "destination"]`
 
-    *fatkodima*
+    *Nikita Vasilevsky*
 
-## Rails 7.0.2.4 (April 26, 2022) ##
+*   Include the `connection_pool` with exceptions raised from an adapter.
 
-*   No changes.
+    The `connection_pool` provides added context such as the connection used
+    that led to the exception as well as which role and shard.
 
+    *Luan Vieira*
 
-## Rails 7.0.2.3 (March 08, 2022) ##
+*   Support multiple column ordering for `find_each`, `find_in_batches` and `in_batches`.
 
-*   No changes.
+    When find_each/find_in_batches/in_batches are performed on a table with composite primary keys, ascending or descending order can be selected for each key.
 
+    ```ruby
+    Person.find_each(order: [:desc, :asc]) do |person|
+      person.party_all_night!
+    end
+    ```
 
-## Rails 7.0.2.2 (February 11, 2022) ##
+    *Takuya Kurimoto*
 
-*   No changes.
+*   Fix where on association with has_one/has_many polymorphic relations.
 
+    Before:
+    ```ruby
+    Treasure.where(price_estimates: PriceEstimate.all)
+    #=> SELECT (...) WHERE "treasures"."id" IN (SELECT "price_estimates"."estimate_of_id" FROM "price_estimates")
+    ```
 
-## Rails 7.0.2.1 (February 11, 2022) ##
+    Later:
+    ```ruby
+    Treasure.where(price_estimates: PriceEstimate.all)
+    #=> SELECT (...) WHERE "treasures"."id" IN (SELECT "price_estimates"."estimate_of_id" FROM "price_estimates" WHERE "price_estimates"."estimate_of_type" = 'Treasure')
+    ```
 
-*   No changes.
+    *Lázaro Nixon*
 
+*   Assign auto populated columns on Active Record record creation.
 
-## Rails 7.0.2 (February 08, 2022) ##
+    Changes record creation logic to allow for the `auto_increment` column to be assigned
+    immediately after creation regardless of it's relation to the model's primary key.
 
-*   Fix `PG.connect` keyword arguments deprecation warning on ruby 2.7.
+    The PostgreSQL adapter benefits the most from the change allowing for any number of auto-populated
+    columns to be assigned on the object immediately after row insertion utilizing the `RETURNING` statement.
 
     *Nikita Vasilevsky*
 
-*   Fix the ability to exclude encryption params from being autofiltered.
-
-    *Mark Gangl*
-
-*   Dump the precision for datetime columns following the new defaults.
-
-    *Rafael Mendonça França*
-
-*   Make sure encrypted attributes are not being filtered twice.
+*   Use the first key in the `shards` hash from `connected_to` for the `default_shard`.
 
-    *Nikita Vasilevsky*
-
-*   Dump the database schema containing the current Rails version.
+    Some applications may not want to use `:default` as a shard name in their connection model. Unfortunately Active Record expects there to be a `:default` shard because it must assume a shard to get the right connection from the pool manager. Rather than force applications to manually set this, `connects_to` can infer the default shard name from the hash of shards and will now assume that the first shard is your default.
 
-    Since https://github.com/rails/rails/pull/42297, Rails now generate datetime columns
-    with a default precision of 6. This means that users upgrading to Rails 7.0 from 6.1,
-    when loading the database schema, would get the new precision value, which would not match
-    the production schema.
+    For example if your model looked like this:
 
-    To avoid this the schema dumper will generate the new format which will include the Rails
-    version and will look like this:
+    ```ruby
+    class ShardRecord < ApplicationRecord
+      self.abstract_class = true
 
-    ```
-    ActiveRecord::Schema[7.0].define
+      connects_to shards: {
+        shard_one: { writing: :shard_one },
+        shard_two: { writing: :shard_two }
+      }
     ```
 
-    When upgrading from Rails 6.1 to Rails 7.0, you can run the `rails app:update` task that will
-    set the current schema version to 6.1.
+    Then the `default_shard` for this class would be set to `shard_one`.
 
-    *Rafael Mendonça França*
+    Fixes: #45390
 
-*   Fix parsing expression for PostgreSQL generated column.
-
-    *fatkodima*
+    *Eileen M. Uchitelle*
 
-*   Fix `Mysql2::Error: Commands out of sync; you can't run this command now`
-    when bulk-inserting fixtures that exceed `max_allowed_packet` configuration.
+*   Fix mutation detection for serialized attributes backed by binary columns.
 
-    *Nikita Vasilevsky*
+    *Jean Boussier*
 
-*   Fix error when saving an association with a relation named `record`.
+*   Add `ActiveRecord.disconnect_all!` method to immediately close all connections from all pools.
 
-    *Dorian Marié*
+    *Jean Boussier*
 
-*   Fix `MySQL::SchemaDumper` behavior about datetime precision value.
+*   Discard connections which may have been left in a transaction.
 
-    *y0t4*
+    There are cases where, due to an error, `within_new_transaction` may unexpectedly leave a connection in an open transaction. In these cases the connection may be reused, and the following may occur:
+    - Writes appear to fail when they actually succeed.
+    - Writes appear to succeed when they actually fail.
+    - Reads return stale or uncommitted data.
 
-*   Improve associated with no reflection error.
+    Previously, the following case was detected:
+    - An error is encountered during the transaction, then another error is encountered while attempting to roll it back.
 
-    *Nikolai*
+    Now, the following additional cases are detected:
+    - An error is encountered just after successfully beginning a transaction.
+    - An error is encountered while committing a transaction, then another error is encountered while attempting to roll it back.
+    - An error is encountered while rolling back a transaction.
 
-*   Fix PG.connect keyword arguments deprecation warning on ruby 2.7.
+    *Nick Dower*
 
-    Fixes #44307.
+*   Active Record query cache now evicts least recently used entries
 
-    *Nikita Vasilevsky*
+    By default it only keeps the `100` most recently used queries.
 
-*   Fix passing options to `check_constraint` from `change_table`.
+    The cache size can be configured via `database.yml`
 
-    *Frederick Cheung*
+    ```yaml
+    development:
+      adapter: mysql2
+      query_cache: 200
+    ```
 
+    It can also be entirely disabled:
 
-## Rails 7.0.1 (January 06, 2022) ##
+    ```yaml
+    development:
+      adapter: mysql2
+      query_cache: false
+    ```
 
+    *Jean Boussier*
 
-*   Change `QueryMethods#in_order_of` to drop records not listed in values.
+*   Deprecate `check_pending!` in favor of `check_all_pending!`.
 
-    `in_order_of` now filters down to the values provided, to match the behavior of the `Enumerable` version.
+    `check_pending!` will only check for pending migrations on the current database connection or the one passed in. This has been deprecated in favor of `check_all_pending!` which will find all pending migrations for the database configurations in a given environment.
 
-    *Kevin Newton*
-
-*   Allow named expression indexes to be revertible.
+    *Eileen M. Uchitelle*
 
-    Previously, the following code would raise an error in a reversible migration executed while rolling back, due to the index name not being used in the index removal.
+*   Make `increment_counter`/`decrement_counter` accept an amount argument
 
     ```ruby
-    add_index(:settings, "(data->'property')", using: :gin, name: :index_settings_data_property)
+    Post.increment_counter(:comments_count, 5, by: 3)
     ```
 
-    Fixes #43331.
+    *fatkodima*
 
-    *Oliver Günther*
+*   Add support for `Array#intersect?` to `ActiveRecord::Relation`.
 
-*   Better error messages when association name is invalid in the argument of `ActiveRecord::QueryMethods::WhereChain#missing`.
+    `Array#intersect?` is only available on Ruby 3.1 or later.
 
-    *ykpythemind*
+    This allows the Rubocop `Style/ArrayIntersect` cop to work with `ActiveRecord::Relation` objects.
 
-*   Fix ordered migrations for single db in multi db environment.
+    *John Harry Kelly*
 
-    *Himanshu*
+*   The deferrable foreign key can be passed to `t.references`.
 
-*   Extract `on update CURRENT_TIMESTAMP` for mysql2 adapter.
+    *Hiroyuki Ishii*
 
-    *Kazuhiro Masuda*
+*   Deprecate `deferrable: true` option of `add_foreign_key`.
 
-*   Fix incorrect argument in PostgreSQL structure dump tasks.
+    `deferrable: true` is deprecated in favor of `deferrable: :immediate`, and
+    will be removed in Rails 7.2.
 
-    Updating the `--no-comment` argument added in Rails 7 to the correct `--no-comments` argument.
+    Because `deferrable: true` and `deferrable: :deferred` are hard to understand.
+    Both true and :deferred are truthy values.
+    This behavior is the same as the deferrable option of the add_unique_key method, added in #46192.
 
-    *Alex Dent*
-
-*   Fix schema dumping column default SQL values for sqlite3.
+    *Hiroyuki Ishii*
 
-    *fatkodima*
+*   `AbstractAdapter#execute` and `#exec_query` now clear the query cache
 
-*   Correctly parse complex check constraint expressions for PostgreSQL.
+    If you need to perform a read only SQL query without clearing the query
+    cache, use `AbstractAdapter#select_all`.
 
-    *fatkodima*
+    *Jean Boussier*
 
-*   Fix `timestamptz` attributes on PostgreSQL handle blank inputs.
+*   Make `.joins` / `.left_outer_joins` work with CTEs.
 
-    *Alex Ghiculescu*
+    For example:
 
-*   Fix migration compatibility to create SQLite references/belongs_to column as integer when migration version is 6.0.
+    ```ruby
+    Post
+     .with(commented_posts: Comment.select(:post_id).distinct)
+     .joins(:commented_posts)
+    #=> WITH (...) SELECT ... INNER JOIN commented_posts on posts.id = commented_posts.post_id
+    ```
 
-    Reference/belongs_to in migrations with version 6.0 were creating columns as
-    bigint instead of integer for the SQLite Adapter.
+    *Vladimir Dementyev*
 
-    *Marcelo Lauxen*
+*   Add a load hook for `ActiveRecord::ConnectionAdapters::Mysql2Adapter`
+    (named `active_record_mysql2adapter`) to allow for overriding aspects of the
+    `ActiveRecord::ConnectionAdapters::Mysql2Adapter` class. This makes `Mysql2Adapter`
+    consistent with `PostgreSQLAdapter` and `SQLite3Adapter` that already have load hooks.
 
-*   Fix joining through a polymorphic association.
+    *fatkodima*
 
-    *Alexandre Ruban*
+*   Introduce adapter for Trilogy database client
 
-*   Fix `QueryMethods#in_order_of` to handle empty order list.
+    Trilogy is a MySQL-compatible database client. Rails applications can use Trilogy
+    by configuring their `config/database.yml`:
 
-    ```ruby
-    Post.in_order_of(:id, []).to_a
+    ```yaml
+    development:
+    adapter: trilogy
+    database: blog_development
+    pool: 5
     ```
 
-    Also more explicitly set the column as secondary order, so that any other
-    value is still ordered.
-
-    *Jean Boussier*
-
-*   Fix `rails dbconsole` for 3-tier config.
+    Or by using the `DATABASE_URL` environment variable:
 
-    *Eileen M. Uchitelle*
+    ```ruby
+    ENV['DATABASE_URL'] # => "trilogy://localhost/blog_development?pool=5"
+    ```
 
-*   Fix quoting of column aliases generated by calculation methods.
+    *Adrianna Chang*
 
-    Since the alias is derived from the table name, we can't assume the result
-    is a valid identifier.
+*   `after_commit` callbacks defined on models now execute in the correct order.
 
     ```ruby
-    class Test < ActiveRecord::Base
-      self.table_name = '1abc'
+    class User < ActiveRecord::Base
+      after_commit { puts("this gets called first") }
+      after_commit { puts("this gets called second") }
     end
-    Test.group(:id).count
-    # syntax error at or near "1" (ActiveRecord::StatementInvalid)
-    # LINE 1: SELECT COUNT(*) AS count_all, "1abc"."id" AS 1abc_id FROM "1...
     ```
 
-    *Jean Boussier*
+    Previously, the callbacks executed in the reverse order. To opt in to the new behaviour:
 
+    ```ruby
+    config.active_record.run_after_transaction_callbacks_in_order_defined = true
+    ```
+
+    This is the default for new apps.
 
-## Rails 7.0.0 (December 15, 2021) ##
+    *Alex Ghiculescu*
 
-*   Better handle SQL queries with invalid encoding.
+*   Infer `foreign_key` when `inverse_of` is present on `has_one` and `has_many` associations.
 
     ```ruby
-    Post.create(name: "broken \xC8 UTF-8")
+    has_many :citations, foreign_key: "book1_id", inverse_of: :book
     ```
 
-    Would cause all adapters to fail in a non controlled way in the code
-    responsible to detect write queries.
+    can be simplified to
 
-    The query is now properly passed to the database connection, which might or might
-    not be able to handle it, but will either succeed or failed in a more correct way.
-
-    *Jean Boussier*
+    ```ruby
+    has_many :citations, inverse_of: :book
+    ```
 
-*   Move database and shard selection config options to a generator.
+    and the foreign_key will be read from the corresponding `belongs_to` association.
 
-    Rather than generating the config options in `production.rb` when applications are created, applications can now run a generator to create an initializer and uncomment / update options as needed. All multi-db configuration can be implemented in this initializer.
+    *Daniel Whitney*
 
-    *Eileen M. Uchitelle*
+*   Limit max length of auto generated index names
 
+    Auto generated index names are now limited to 62 bytes, which fits within
+    the default index name length limits for MySQL, Postgres and SQLite.
 
-## Rails 7.0.0.rc3 (December 14, 2021) ##
+    Any index name over the limit will fallback to the new short format.
 
-*   No changes.
+    Before (too long):
+    ```
+    index_testings_on_foo_and_bar_and_first_name_and_last_name_and_administrator
+    ```
 
+    After (short format):
+    ```
+    idx_on_foo_bar_first_name_last_name_administrator_5939248142
+    ```
 
-## Rails 7.0.0.rc2 (December 14, 2021) ##
+    The short format includes a hash to ensure the name is unique database-wide.
 
-*   No changes.
+    *Mike Coutermarsh*
 
+*   Introduce a more stable and optimized Marshal serializer for Active Record models.
 
-## Rails 7.0.0.rc1 (December 06, 2021) ##
+    Can be enabled with `config.active_record.marshalling_format_version = 7.1`.
 
-*   Remove deprecated `ActiveRecord::DatabaseConfigurations::DatabaseConfig#spec_name`.
+    *Jean Boussier*
 
-    *Rafael Mendonça França*
+*   Allow specifying where clauses with column-tuple syntax.
 
-*   Remove deprecated `ActiveRecord::Connection#in_clause_length`.
+    Querying through `#where` now accepts a new tuple-syntax which accepts, as
+    a key, an array of columns and, as a value, an array of corresponding tuples.
+    The key specifies a list of columns, while the value is an array of
+    ordered-tuples that conform to the column list.
 
-    *Rafael Mendonça França*
+    For instance:
 
-*   Remove deprecated `ActiveRecord::Connection#allowed_index_name_length`.
+    ```ruby
+    # Cpk::Book => Cpk::Book(author_id: integer, number: integer, title: string, revision: integer)
+    # Cpk::Book.primary_key => ["author_id", "number"]
 
-    *Rafael Mendonça França*
+    book = Cpk::Book.create!(author_id: 1, number: 1)
+    Cpk::Book.where(Cpk::Book.primary_key => [[1, 2]]) # => [book]
 
-*   Remove deprecated `ActiveRecord::Base#remove_connection`.
+    # Topic => Topic(id: integer, title: string, author_name: string...)
 
-    *Rafael Mendonça França*
+    Topic.where([:title, :author_name] => [["The Alchemist", "Paul Coelho"], ["Harry Potter", "J.K Rowling"]])
+    ```
 
-*   Load STI Models in fixtures
+    *Paarth Madan*
 
-    Data from Fixtures now loads based on the specific class for models with
-    Single Table Inheritance. This affects enums defined in subclasses, previously
-    the value of these fields was not parsed and remained `nil`
+*   Allow warning codes to be ignore when reporting SQL warnings.
 
-    *Andres Howard*
+    Active Record config that can ignore warning codes
 
-*   `#authenticate` returns false when the password is blank instead of raising an error.
+    ```ruby
+    # Configure allowlist of warnings that should always be ignored
+    config.active_record.db_warnings_ignore = [
+      "1062", # MySQL Error 1062: Duplicate entry
+    ]
+    ```
 
-    *Muhammad Muhammad Ibrahim*
+    This is supported for the MySQL and PostgreSQL adapters.
 
-*   Fix `ActiveRecord::QueryMethods#in_order_of` behavior for integer enums.
+    *Nick Borromeo*
 
-    `ActiveRecord::QueryMethods#in_order_of` didn't work as expected for enums stored as integers in the database when passing an array of strings or symbols as the order argument. This unexpected behavior occurred because the string or symbol values were not casted to match the integers in the database.
+*   Introduce `:active_record_fixtures` lazy load hook.
 
-    The following example now works as expected:
+    Hooks defined with this name will be run whenever `TestFixtures` is included
+    in a class.
 
     ```ruby
-    class Book < ApplicationRecord
-      enum status: [:proposed, :written, :published]
+    ActiveSupport.on_load(:active_record_fixtures) do
+      self.fixture_paths << "test/fixtures"
     end
 
-    Book.in_order_of(:status, %w[written published proposed])
-    ```
-
-    *Alexandre Ruban*
+    klass = Class.new
+    klass.include(ActiveRecord::TestFixtures)
 
-*   Ignore persisted in-memory records when merging target lists.
+    klass.fixture_paths # => ["test/fixtures"]
+    ```
 
-    *Kevin Sjöberg*
+    *Andrew Novoselac*
 
-*   Add a new option `:update_only` to `upsert_all` to configure the list of columns to update in case of conflict.
+*   Introduce `TestFixtures#fixture_paths`.
 
-    Before, you could only customize the update SQL sentence via `:on_duplicate`. There is now a new option `:update_only` that lets you provide a list of columns to update in case of conflict:
+    Multiple fixture paths can now be specified using the `#fixture_paths` accessor.
+    Apps will continue to have `test/fixtures` as their one fixture path by default,
+    but additional fixture paths can be specified.
 
     ```ruby
-    Commodity.upsert_all(
-      [
-        { id: 2, name: "Copper", price: 4.84 },
-        { id: 4, name: "Gold", price: 1380.87 },
-        { id: 6, name: "Aluminium", price: 0.35 }
-      ],
-      update_only: [:price] # Only prices will be updated
-    )
+    ActiveSupport::TestCase.fixture_paths << "component1/test/fixtures"
+    ActiveSupport::TestCase.fixture_paths << "component2/test/fixtures"
     ```
 
-    *Jorge Manrubia*
+    `TestFixtures#fixture_path` is now deprecated.
 
-*   Remove deprecated `ActiveRecord::Result#map!` and `ActiveRecord::Result#collect!`.
-
-    *Rafael Mendonça França*
+    *Andrew Novoselac*
 
-*   Remove deprecated `ActiveRecord::Base.configurations.to_h`.
+*   Adds support for deferrable exclude constraints in PostgreSQL.
 
-    *Rafael Mendonça França*
+    By default, exclude constraints in PostgreSQL are checked after each statement.
+    This works for most use cases, but becomes a major limitation when replacing
+    records with overlapping ranges by using multiple statements.
 
-*   Remove deprecated `ActiveRecord::Base.configurations.default_hash`.
+    ```ruby
+    exclusion_constraint :users, "daterange(valid_from, valid_to) WITH &&", deferrable: :immediate
+    ```
 
-    *Rafael Mendonça França*
+    Passing `deferrable: :immediate` checks constraint after each statement,
+    but allows manually deferring the check using `SET CONSTRAINTS ALL DEFERRED`
+    within a transaction. This will cause the excludes to be checked after the transaction.
 
-*   Remove deprecated `ActiveRecord::Base.arel_attribute`.
+    It's also possible to change the default behavior from an immediate check
+    (after the statement), to a deferred check (after the transaction):
 
-    *Rafael Mendonça França*
+    ```ruby
+    exclusion_constraint :users, "daterange(valid_from, valid_to) WITH &&", deferrable: :deferred
+    ```
 
-*   Remove deprecated `ActiveRecord::Base.connection_config`.
+    *Hiroyuki Ishii*
 
-    *Rafael Mendonça França*
+*   Respect `foreign_type` option to `delegated_type` for `{role}_class` method.
 
-*   Filter attributes in SQL logs
+    Usage of `delegated_type` with non-conventional `{role}_type` column names can now be specified with `foreign_type` option.
+    This option is the same as `foreign_type` as forwarded to the underlying `belongs_to` association that `delegated_type` wraps.
 
-    Previously, SQL queries in logs containing `ActiveRecord::Base.filter_attributes` were not filtered.
+    *Jason Karns*
 
-    Now, the filter attributes will be masked `[FILTERED]` in the logs when `prepared_statement` is enabled.
+*   Add support for unique constraints (PostgreSQL-only).
 
+    ```ruby
+    add_unique_key :sections, [:position], deferrable: :deferred, name: "unique_section_position"
+    remove_unique_key :sections, name: "unique_section_position"
     ```
-    # Before:
-      Foo Load (0.2ms)  SELECT "foos".* FROM "foos" WHERE "foos"."passw" = ? LIMIT ?  [["passw", "hello"], ["LIMIT", 1]]
 
-    # After:
-      Foo Load (0.5ms)  SELECT "foos".* FROM "foos" WHERE "foos"."passw" = ? LIMIT ?  [["passw", "[FILTERED]"], ["LIMIT", 1]]
+    See PostgreSQL's [Unique Constraints](https://www.postgresql.org/docs/current/ddl-constraints.html#DDL-CONSTRAINTS-UNIQUE-CONSTRAINTS) documentation for more on unique constraints.
+
+    By default, unique constraints in PostgreSQL are checked after each statement.
+    This works for most use cases, but becomes a major limitation when replacing
+    records with unique column by using multiple statements.
+
+    An example of swapping unique columns between records.
+
+    ```ruby
+    # position is unique column
+    old_item = Item.create!(position: 1)
+    new_item = Item.create!(position: 2)
+
+    Item.transaction do
+      old_item.update!(position: 2)
+      new_item.update!(position: 1)
+    end
     ```
 
-    *Aishwarya Subramanian*
+    Using the default behavior, the transaction would fail when executing the
+    first `UPDATE` statement.
 
-*   Remove deprecated `Tasks::DatabaseTasks.spec`.
+    By passing the `:deferrable` option to the `add_unique_key` statement in
+    migrations, it's possible to defer this check.
 
-    *Rafael Mendonça França*
+    ```ruby
+    add_unique_key :items, [:position], deferrable: :immediate
+    ```
 
-*   Remove deprecated `Tasks::DatabaseTasks.current_config`.
+    Passing `deferrable: :immediate` does not change the behaviour of the previous example,
+    but allows manually deferring the check using `SET CONSTRAINTS ALL DEFERRED` within a transaction.
+    This will cause the unique constraints to be checked after the transaction.
 
-    *Rafael Mendonça França*
+    It's also possible to adjust the default behavior from an immediate
+    check (after the statement), to a deferred check (after the transaction):
 
-*   Deprecate `Tasks::DatabaseTasks.schema_file_type`.
+    ```ruby
+    add_unique_key :items, [:position], deferrable: :deferred
+    ```
 
-    *Rafael Mendonça França*
+    If you want to change an existing unique index to deferrable, you can use :using_index
+    to create deferrable unique constraints.
 
-*   Remove deprecated `Tasks::DatabaseTasks.dump_filename`.
+    ```ruby
+    add_unique_key :items, deferrable: :deferred, using_index: "index_items_on_position"
+    ```
 
-    *Rafael Mendonça França*
+    *Hiroyuki Ishii*
 
-*   Remove deprecated `Tasks::DatabaseTasks.schema_file`.
+*   Remove deprecated `Tasks::DatabaseTasks.schema_file_type`.
 
     *Rafael Mendonça França*
 
-*   Remove deprecated `environment` and `name` arguments from `Tasks::DatabaseTasks.schema_up_to_date?`.
+*   Remove deprecated `config.active_record.partial_writes`.
 
     *Rafael Mendonça França*
 
-*   Merging conditions on the same column no longer maintain both conditions,
-    and will be consistently replaced by the latter condition.
-
-    ```ruby
-    # Rails 6.1 (IN clause is replaced by merger side equality condition)
-    Author.where(id: [david.id, mary.id]).merge(Author.where(id: bob)) # => [bob]
-    # Rails 6.1 (both conflict conditions exists, deprecated)
-    Author.where(id: david.id..mary.id).merge(Author.where(id: bob)) # => []
-    # Rails 6.1 with rewhere to migrate to Rails 7.0's behavior
-    Author.where(id: david.id..mary.id).merge(Author.where(id: bob), rewhere: true) # => [bob]
-    # Rails 7.0 (same behavior with IN clause, mergee side condition is consistently replaced)
-    Author.where(id: [david.id, mary.id]).merge(Author.where(id: bob)) # => [bob]
-    Author.where(id: david.id..mary.id).merge(Author.where(id: bob)) # => [bob]
+*   Remove deprecated `ActiveRecord::Base` config accessors.
 
     *Rafael Mendonça França*
 
-*   Remove deprecated support to `Model.reorder(nil).first` to search using non-deterministic order.
+*   Remove the `:include_replicas` argument from `configs_for`. Use `:include_hidden` argument instead.
 
-    *Rafael Mendonça França*
+    *Eileen M. Uchitelle*
 
-*   Remove deprecated rake tasks:
+*   Allow applications to lookup a config via a custom hash key.
 
-    * `db:schema:load_if_ruby`
-    * `db:structure:dump`
-    * `db:structure:load`
-    * `db:structure:load_if_sql`
-    * `db:structure:dump:#{name}`
-    * `db:structure:load:#{name}`
-    * `db:test:load_structure`
-    * `db:test:load_structure:#{name}`
+    If you have registered a custom config or want to find configs where the hash matches a specific key, now you can pass `config_key` to `configs_for`. For example if you have a `db_config` with the key `vitess` you can look up a database configuration hash by  matching that key.
 
-    *Rafael Mendonça França*
+    ```ruby
+    ActiveRecord::Base.configurations.configs_for(env_name: "development", name: "primary", config_key: :vitess)
+    ActiveRecord::Base.configurations.configs_for(env_name: "development", config_key: :vitess)
+    ```
+
+    *Eileen M. Uchitelle*
 
-*   Remove deprecated `DatabaseConfig#config` method.
+*   Allow applications to register a custom database configuration handler.
 
-    *Rafael Mendonça França*
+    Adds a mechanism for registering a custom handler for cases where you want database configurations to respond to custom methods. This is useful for non-Rails database adapters or tools like Vitess that you may want to configure differently from a standard `HashConfig` or `UrlConfig`.
 
-*   Rollback transactions when the block returns earlier than expected.
+    Given the following database YAML we want the `animals` db to create a `CustomConfig` object instead while the `primary` database will be a `UrlConfig`:
 
-    Before this change, when a transaction block returned early, the transaction would be committed.
+    ```yaml
+    development:
+      primary:
+        url: postgres://localhost/primary
+      animals:
+        url: postgres://localhost/animals
+        custom_config:
+          sharded: 1
+    ```
 
-    The problem is that timeouts triggered inside the transaction block was also making the incomplete transaction
-    to be committed, so in order to avoid this mistake, the transaction block is rolled back.
+    To register a custom handler first make a class that has your custom methods:
 
-    *Rafael Mendonça França*
+    ```ruby
+    class CustomConfig < ActiveRecord::DatabaseConfigurations::UrlConfig
+      def sharded?
+        custom_config.fetch("sharded", false)
+      end
 
-*   Add middleware for automatic shard swapping.
+      private
+        def custom_config
+          configuration_hash.fetch(:custom_config)
+        end
+    end
+    ```
 
-    Provides a basic middleware to perform automatic shard swapping. Applications will provide a resolver which will determine for an individual request which shard should be used. Example:
+    Then register the config in an initializer:
 
     ```ruby
-    config.active_record.shard_resolver = ->(request) {
-      subdomain = request.subdomain
-      tenant = Tenant.find_by_subdomain!(subdomain)
-      tenant.shard
-    }
+    ActiveRecord::DatabaseConfigurations.register_db_config_handler do |env_name, name, url, config|
+      next unless config.key?(:custom_config)
+      CustomConfig.new(env_name, name, url, config)
+    end
     ```
 
-    See guides for more details.
+    When the application is booted, configuration hashes with the `:custom_config` key will be `CustomConfig` objects and respond to `sharded?`. Applications must handle the condition in which Active Record should use their custom handler.
 
-    *Eileen M. Uchitelle*, *John Crepezzi*
+    *Eileen M. Uchitelle and John Crepezzi*
 
-*   Remove deprecated support to pass a column to `type_cast`.
+*   `ActiveRecord::Base.serialize` no longer uses YAML by default.
 
-    *Rafael Mendonça França*
+    YAML isn't particularly performant and can lead to security issues
+    if not used carefully.
 
-*   Remove deprecated support to type cast to database values `ActiveRecord::Base` objects.
+    Unfortunately there isn't really any good serializers in Ruby's stdlib
+    to replace it.
 
-    *Rafael Mendonça França*
+    The obvious choice would be JSON, which is a fine format for this use case,
+    however the JSON serializer in Ruby's stdlib isn't strict enough, as it fallback
+    to casting unknown types to strings, which could lead to corrupted data.
 
-*   Remove deprecated support to quote `ActiveRecord::Base` objects.
+    Some third party JSON libraries like `Oj` have a suitable strict mode.
 
-    *Rafael Mendonça França*
+    So it's preferable that users choose a serializer based on their own constraints.
 
-*   Remove deprecacated support to resolve connection using `"primary"` as connection specification name.
+    The original default can be restored by setting `config.active_record.default_column_serializer = YAML`.
 
-    *Rafael Mendonça França*
+    *Jean Boussier*
 
-*   Remove deprecation warning when using `:interval` column is used in PostgreSQL database.
+*   `ActiveRecord::Base.serialize` signature changed.
 
-    Now, interval columns will return `ActiveSupport::Duration` objects instead of strings.
+    Rather than a single positional argument that accepts two possible
+    types of values, `serialize` now accepts two distinct keyword arguments.
 
-    To keep the old behavior, you can add this line to your model:
+    Before:
 
     ```ruby
-    attribute :column, :string
+    serialize :content, JSON
+    serialize :backtrace, Array
     ```
 
-    *Rafael Mendonça França*
-
-*   Remove deprecated support to YAML load `ActiveRecord::Base` instance in the Rails 4.2 and 4.1 formats.
+    After:
 
-    *Rafael Mendonça França*
+    ```ruby
+    serialize :content, coder: JSON
+    serialize :backtrace, type: Array
+    ```
 
-*   Remove deprecated option `:spec_name` in the `configs_for` method.
+    *Jean Boussier*
 
-    *Rafael Mendonça França*
+*   YAML columns use `YAML.safe_dump` is available.
 
-*   Remove deprecated  `ActiveRecord::Base.allow_unsafe_raw_sql`.
+    As of `psych 5.1.0`, `YAML.safe_dump` can now apply the same permitted
+    types restrictions than `YAML.safe_load`.
 
-    *Rafael Mendonça França*
+    It's preferable to ensure the payload only use allowed types when we first
+    try to serialize it, otherwise you may end up with invalid records in the
+    database.
 
-*   Fix regression bug that caused ignoring additional conditions for preloading has_many-through relations.
+    *Jean Boussier*
 
-    Fixes #43132
+*   `ActiveRecord::QueryLogs` better handle broken encoding.
 
-    *Alexander Pauly*
+    It's not uncommon when building queries with BLOB fields to contain
+    binary data. Unless the call carefully encode the string in ASCII-8BIT
+    it generally end up being encoded in `UTF-8`, and `QueryLogs` would
+    end up failing on it.
 
-*   Fix `has_many` inversing recursion on models with recursive associations.
+    `ActiveRecord::QueryLogs` no longer depend on the query to be properly encoded.
 
-    *Gannon McGibbon*
+    *Jean Boussier*
 
-*   Add `accepts_nested_attributes_for` support for `delegated_type`
+*   Fix a bug where `ActiveRecord::Generators::ModelGenerator` would not respect create_table_migration template overrides.
 
-    ```ruby
-    class Entry < ApplicationRecord
-      delegated_type :entryable, types: %w[ Message Comment ]
-      accepts_nested_attributes_for :entryable
-    end
+    ```
+    rails g model create_books title:string content:text
+    ```
+    will now read from the create_table_migration.rb.tt template in the following locations in order:
+    ```
+    lib/templates/active_record/model/create_table_migration.rb
+    lib/templates/active_record/migration/create_table_migration.rb
+    ```
 
-    entry = Entry.create(entryable_type: 'Message', entryable_attributes: { content: 'Hello world' })
-    # => #<Entry:0x00>
-    # id: 1
-    # entryable_id: 1,
-    # entryable_type: 'Message'
-    # ...>
+    *Spencer Neste*
 
-    entry.entryable
-    # => #<Message:0x01>
-    # id: 1
-    # content: 'Hello world'
-    # ...>
-    ```
+*   `ActiveRecord::Relation#explain` now accepts options.
 
-    Previously it would raise an error:
+    For databases and adapters which support them (currently PostgreSQL
+    and MySQL), options can be passed to `explain` to provide more
+    detailed query plan analysis:
 
     ```ruby
-    Entry.create(entryable_type: 'Message', entryable_attributes: { content: 'Hello world' })
-    # ArgumentError: Cannot build association `entryable'. Are you trying to build a polymorphic one-to-one association?
+    Customer.where(id: 1).joins(:orders).explain(:analyze, :verbose)
     ```
 
-    *Sjors Baltus*
-
-*   Use subquery for DELETE with GROUP_BY and HAVING clauses.
+    *Reid Lynch*
 
-    Prior to this change, deletes with GROUP_BY and HAVING were returning an error.
+*   Multiple `Arel::Nodes::SqlLiteral` nodes can now be added together to
+    form `Arel::Nodes::Fragments` nodes. This allows joining several pieces
+    of SQL.
 
-    After this change, GROUP_BY and HAVING are valid clauses in DELETE queries, generating the following query:
+    *Matthew Draper*, *Ole Friis*
 
-    ```sql
-    DELETE FROM "posts" WHERE "posts"."id" IN (
-        SELECT "posts"."id" FROM "posts" INNER JOIN "comments" ON "comments"."post_id" = "posts"."id" GROUP BY "posts"."id" HAVING (count(comments.id) >= 2))
-    )  [["flagged", "t"]]
-    ```
+*   `ActiveRecord::Base#signed_id` raises if called on a new record.
 
-    *Ignacio Chiazzo Cardarello*
+    Previously it would return an ID that was not usable, since it was based on `id = nil`.
 
-*   Use subquery for UPDATE with GROUP_BY and HAVING clauses.
+    *Alex Ghiculescu*
 
-    Prior to this change, updates with GROUP_BY and HAVING were being ignored, generating a SQL like this:
+*   Allow SQL warnings to be reported.
 
-    ```sql
-    UPDATE "posts" SET "flagged" = ? WHERE "posts"."id" IN (
-        SELECT "posts"."id" FROM "posts" INNER JOIN "comments" ON "comments"."post_id" = "posts"."id"
-    )  [["flagged", "t"]]
-    ```
+    Active Record configs can be set to enable SQL warning reporting.
 
-    After this change, GROUP_BY and HAVING clauses are used as a subquery in updates, like this:
+    ```ruby
+    # Configure action to take when SQL query produces warning
+    config.active_record.db_warnings_action = :raise
 
-    ```sql
-    UPDATE "posts" SET "flagged" = ? WHERE "posts"."id" IN (
-        SELECT "posts"."id" FROM "posts" INNER JOIN "comments" ON "comments"."post_id" = "posts"."id"
-        GROUP BY posts.id HAVING (count(comments.id) >= 2)
-    )  [["flagged", "t"]]
+    # Configure allowlist of warnings that should always be ignored
+    config.active_record.db_warnings_ignore = [
+      /Invalid utf8mb4 character string/,
+      "An exact warning message",
+    ]
     ```
 
-    *Ignacio Chiazzo Cardarello*
+    This is supported for the MySQL and PostgreSQL adapters.
 
-*   Add support for setting the filename of the schema or structure dump in the database config.
+    *Adrianna Chang*, *Paarth Madan*
 
-    Applications may now set their the filename or path of the schema / structure dump file in their database configuration.
+*   Add `#regroup` query method as a short-hand for `.unscope(:group).group(fields)`
 
-    ```yaml
-    production:
-      primary:
-        database: my_db
-        schema_dump: my_schema_dump_filename.rb
-      animals:
-        database: animals_db
-        schema_dump: false
+    Example:
+
+    ```ruby
+    Post.group(:title).regroup(:author)
+    # SELECT `posts`.`*` FROM `posts` GROUP BY `posts`.`author`
     ```
 
-    The filename set in `schema_dump` will be used by the application. If set to `false` the schema will not be dumped. The database tasks are responsible for adding the database directory to the filename. If a full path is provided, the Rails tasks will use that instead of `ActiveRecord::DatabaseTasks.db_dir`.
+    *Danielius Visockas*
 
-    *Eileen M. Uchitelle*, *Ryan Kerr*
+*   PostgreSQL adapter method `enable_extension` now allows parameter to be `[schema_name.]<extension_name>`
+    if the extension must be installed on another schema.
 
-*   Add `ActiveRecord::Base.prohibit_shard_swapping` to prevent attempts to change the shard within a block.
+    Example: `enable_extension('heroku_ext.hstore')`
 
-    *John Crepezzi*, *Eileen M. Uchitelle*
+    *Leonardo Luarte*
 
-*   Filter unchanged attributes with default function from insert query when `partial_inserts` is disabled.
+*   Add `:include` option to `add_index`.
 
-    *Akshay Birajdar*, *Jacopo Beschi*
+    Add support for including non-key columns in indexes for PostgreSQL
+    with the `INCLUDE` parameter.
 
-*   Add support for FILTER clause (SQL:2003) to Arel.
+    ```ruby
+    add_index(:users, :email, include: [:id, :created_at])
+    ```
 
-    Currently supported by PostgreSQL 9.4+ and SQLite 3.30+.
+    will result in:
 
-    *Andrey Novikov*
+    ```sql
+    CREATE INDEX index_users_on_email USING btree (email) INCLUDE (id, created_at)
+    ```
 
-*   Automatically set timestamps on record creation during bulk insert/upsert
+    *Steve Abrams*
 
-    Prior to this change, only updates during an upsert operation (e.g. `upsert_all`) would touch timestamps (`updated_{at,on}`). Now, record creations also touch timestamp columns (`{created,updated}_{at,on}`).
+*   `ActiveRecord::Relation`’s `#any?`, `#none?`, and `#one?` methods take an optional pattern
+    argument, more closely matching their `Enumerable` equivalents.
 
-    This behaviour is controlled by the `<model>.record_timestamps` config, matching the behaviour of `create`, `update`, etc. It can also be overridden by using the `record_timestamps:` keyword argument.
+    *George Claghorn*
 
-    Note that this means `upsert_all` on models with `record_timestamps = false` will no longer touch `updated_{at,on}` automatically.
+*   Add `ActiveRecord::Base.normalizes` for declaring attribute normalizations.
 
-    *Sam Bostock*
+    An attribute normalization is applied when the attribute is assigned or
+    updated, and the normalized value will be persisted to the database.  The
+    normalization is also applied to the corresponding keyword argument of query
+    methods, allowing records to be queried using unnormalized values.
 
-*   Don't require `role` when passing `shard` to `connected_to`.
+    For example:
 
-    `connected_to` can now be called with a `shard` only. Note that `role` is still inherited if `connected_to` calls are nested.
+      ```ruby
+      class User < ActiveRecord::Base
+        normalizes :email, with: -> email { email.strip.downcase }
+        normalizes :phone, with: -> phone { phone.delete("^0-9").delete_prefix("1") }
+      end
 
-    *Eileen M. Uchitelle*
+      user = User.create(email: " CRUISE-CONTROL@EXAMPLE.COM\n")
+      user.email                  # => "cruise-control@example.com"
 
-*   Add option to lazily load the schema cache on the connection.
+      user = User.find_by(email: "\tCRUISE-CONTROL@EXAMPLE.COM ")
+      user.email                  # => "cruise-control@example.com"
+      user.email_before_type_cast # => "cruise-control@example.com"
 
-    Previously, the only way to load the schema cache in Active Record was through the Railtie on boot. This option provides the ability to load the schema cache on the connection after it's been established. Loading the cache lazily on the connection can be beneficial for Rails applications that use multiple databases because it will load the cache at the time the connection is established. Currently Railties doesn't have access to the connections before boot.
+      User.where(email: "\tCRUISE-CONTROL@EXAMPLE.COM ").count         # => 1
+      User.where(["email = ?", "\tCRUISE-CONTROL@EXAMPLE.COM "]).count # => 0
 
-    To use the cache, set `config.active_record.lazily_load_schema_cache = true` in your application configuration. In addition a `schema_cache_path` should be set in your database configuration if you don't want to use the default "db/schema_cache.yml" path.
+      User.exists?(email: "\tCRUISE-CONTROL@EXAMPLE.COM ")         # => true
+      User.exists?(["email = ?", "\tCRUISE-CONTROL@EXAMPLE.COM "]) # => false
 
-    *Eileen M. Uchitelle*
+      User.normalize_value_for(:phone, "+1 (555) 867-5309") # => "5558675309"
+      ```
 
-*   Allow automatic `inverse_of` detection for associations with scopes.
+    *Jonathan Hefner*
 
-    Automatic `inverse_of` detection now works for associations with scopes. For
-    example, the `comments` association here now automatically detects
-    `inverse_of: :post`, so we don't need to pass that option:
+*   Hide changes to before_committed! callback behaviour behind flag.
 
-    ```ruby
-    class Post < ActiveRecord::Base
-      has_many :comments, -> { visible }
-    end
+    In #46525, behavior around before_committed! callbacks was changed so that callbacks
+    would run on every enrolled record in a transaction, not just the first copy of a record.
+    This change in behavior is now controlled by a configuration option,
+    `config.active_record.before_committed_on_all_records`. It will be enabled by default on Rails 7.1.
 
-    class Comment < ActiveRecord::Base
-      belongs_to :post
-    end
-    ```
+    *Adrianna Chang*
 
-    Note that the automatic detection still won't work if the inverse
-    association has a scope. In this example a scope on the `post` association
-    would still prevent Rails from finding the inverse for the `comments`
-    association.
+*   The `namespaced_controller` Query Log tag now matches the `controller` format
 
-    This will be the default for new apps in Rails 7. To opt in:
+    For example, a request processed by `NameSpaced::UsersController` will now log as:
 
-    ```ruby
-    config.active_record.automatic_scope_inversing = true
+    ```
+    :controller # "users"
+    :namespaced_controller # "name_spaced/users"
     ```
 
-    *Daniel Colson*, *Chris Bloom*
-
-*   Accept optional transaction args to `ActiveRecord::Locking::Pessimistic#with_lock`
-
-    `#with_lock` now accepts transaction options like `requires_new:`,
-    `isolation:`, and `joinable:`
-
-    *John Mileham*
+    *Alex Ghiculescu*
 
-*   Adds support for deferrable foreign key constraints in PostgreSQL.
+*   Return only unique ids from ActiveRecord::Calculations#ids
 
-    By default, foreign key constraints in PostgreSQL are checked after each statement. This works for most use cases,
-    but becomes a major limitation when creating related records before the parent record is inserted into the database.
-    One example of this is looking up / creating a person via one or more unique alias.
+    Updated ActiveRecord::Calculations#ids to only return the unique ids of the base model
+    when using eager_load, preload and includes.
 
     ```ruby
-    Person.transaction do
-      alias = Alias
-        .create_with(user_id: SecureRandom.uuid)
-        .create_or_find_by(name: "DHH")
-
-      person = Person
-        .create_with(name: "David Heinemeier Hansson")
-        .create_or_find_by(id: alias.user_id)
-    end
+    Post.find_by(id: 1).comments.count
+    # => 5
+    Post.includes(:comments).where(id: 1).pluck(:id)
+    # => [1, 1, 1, 1, 1]
+    Post.includes(:comments).where(id: 1).ids
+    # => [1]
     ```
 
-    Using the default behavior, the transaction would fail when executing the first `INSERT` statement.
-
-    By passing the `:deferrable` option to the `add_foreign_key` statement in migrations, it's possible to defer this
-    check.
+    *Joshua Young*
 
-    ```ruby
-    add_foreign_key :aliases, :person, deferrable: true
-    ```
+*   Stop using `LOWER()` for case-insensitive queries on `citext` columns
 
-    Passing `deferrable: true` doesn't change the default behavior, but allows manually deferring the check using
-    `SET CONSTRAINTS ALL DEFERRED` within a transaction. This will cause the foreign keys to be checked after the
-    transaction.
+    Previously, `LOWER()` was added for e.g. uniqueness validations with
+    `case_sensitive: false`.
+    It wasn't mentioned in the documentation that the index without `LOWER()`
+    wouldn't be used in this case.
 
-    It's also possible to adjust the default behavior from an immediate check (after the statement), to a deferred check
-    (after the transaction):
+    *Phil Pirozhkov*
 
-    ```ruby
-    add_foreign_key :aliases, :person, deferrable: :deferred
-    ```
+*   Extract `#sync_timezone_changes` method in AbstractMysqlAdapter to enable subclasses
+    to sync database timezone changes without overriding `#raw_execute`.
 
-    *Benedikt Deicke*
+    *Adrianna Chang*, *Paarth Madan*
 
-*   Allow configuring Postgres password through the socket URL.
+*   Do not write additional new lines when dumping sql migration versions
 
-    For example:
-    ```ruby
-    ActiveRecord::DatabaseConfigurations::UrlConfig.new(
-      :production, :production, 'postgres:///?user=user&password=secret&dbname=app', {}
-    ).configuration_hash
-    ```
+    This change updates the `insert_versions_sql` function so that the database insert string containing the current database migration versions does not end with two additional new lines.
 
-    will now return,
+    *Misha Schwartz*
 
-    ```ruby
-    { :user=>"user", :password=>"secret", :dbname=>"app", :adapter=>"postgresql" }
-    ```
+*   Fix `composed_of` value freezing and duplication.
 
-    *Abeid Ahmed*
+    Previously composite values exhibited two confusing behaviors:
 
-*   PostgreSQL: support custom enum types
+    - When reading a compositve value it'd _NOT_ be frozen, allowing it to get out of sync with its underlying database
+      columns.
+    - When writing a compositve value the argument would be frozen, potentially confusing the caller.
 
-    In migrations, use `create_enum` to add a new enum type, and `t.enum` to add a column.
+    Currently, composite values instantiated based on database columns are frozen (addressing the first issue) and
+    assigned compositve values are duplicated and the duplicate is frozen (addressing the second issue).
 
-    ```ruby
-    def up
-      create_enum :mood, ["happy", "sad"]
+    *Greg Navis*
 
-      change_table :cats do |t|
-        t.enum :current_mood, enum_type: "mood", default: "happy", null: false
-      end
-    end
-    ```
+*   Fix redundant updates to the column insensitivity cache
 
-    Enums will be presented correctly in `schema.rb`. Note that this is only supported by
-    the PostgreSQL adapter.
+    Fixed redundant queries checking column capability for insensitive
+    comparison.
 
-    *Alex Ghiculescu*
+    *Phil Pirozhkov*
 
-*   Avoid COMMENT statements in PostgreSQL structure dumps
+*   Allow disabling methods generated by `ActiveRecord.enum`.
 
-    COMMENT statements are now omitted from the output of `db:structure:dump` when using PostgreSQL >= 11.
-    This allows loading the dump without a pgsql superuser account.
+    *Alfred Dominic*
 
-    Fixes #36816, #43107.
+*   Avoid validating `belongs_to` association if it has not changed.
 
-    *Janosch Müller*
+    Previously, when updating a record, Active Record will perform an extra query to check for the presence of
+    `belongs_to` associations (if the presence is configured to be mandatory), even if that attribute hasn't changed.
 
-*   Add support for generated columns in PostgreSQL adapter
+    Currently, only `belongs_to`-related columns are checked for presence. It is possible to have orphaned records with
+    this approach. To avoid this problem, you need to use a foreign key.
 
-    Generated columns are supported since version 12.0 of PostgreSQL. This adds
-    support of those to the PostgreSQL adapter.
+    This behavior can be controlled by configuration:
 
     ```ruby
-    create_table :users do |t|
-      t.string :name
-      t.virtual :name_upcased, type: :string, as: 'upper(name)', stored: true
-    end
+    config.active_record.belongs_to_required_validates_foreign_key = false
     ```
 
-    *Michał Begejowicz*
+    and will be disabled by default with `config.load_defaults 7.1`.
 
+    *fatkodima*
 
-## Rails 7.0.0.alpha2 (September 15, 2021) ##
+*   `has_one` and `belongs_to` associations now define a `reset_association` method
+    on the owner model (where `association` is the name of the association). This
+    method unloads the cached associate record, if any, and causes the next access
+    to query it from the database.
 
-*   No changes.
+    *George Claghorn*
 
+*   Allow per attribute setting of YAML permitted classes (safe load) and unsafe load.
 
-## Rails 7.0.0.alpha1 (September 15, 2021) ##
+    *Carlos Palhares*
 
-*   Remove warning when overwriting existing scopes
+*   Add a build persistence method
 
-    Removes the following unnecessary warning message that appeared when overwriting existing scopes
+    Provides a wrapper for `new`, to provide feature parity with `create`s
+    ability to create multiple records from an array of hashes, using the
+    same notation as the `build` method on associations.
 
-    ```
-    Creating scope :my_scope_name. Overwriting existing method "MyClass.my_scope_name" when overwriting existing scopes
-    ```
+    *Sean Denny*
 
-     *Weston Ganger*
+*   Raise on assignment to readonly attributes
 
-*   Use full precision for `updated_at` in `insert_all`/`upsert_all`
+    ```ruby
+    class Post < ActiveRecord::Base
+      attr_readonly :content
+    end
+    Post.create!(content: "cannot be updated")
+    post.content # "cannot be updated"
+    post.content = "something else" # => ActiveRecord::ReadonlyAttributeError
+    ```
 
-    `CURRENT_TIMESTAMP` provides differing precision depending on the database,
-    and not all databases support explicitly specifying additional precision.
+    Previously, assignment would succeed but silently not write to the database.
 
-    Instead, we delegate to the new `connection.high_precision_current_timestamp`
-    for the SQL to produce a high precision timestamp on the current database.
+    This behavior can be controlled by configuration:
 
-    Fixes #42992
+    ```ruby
+    config.active_record.raise_on_assign_to_attr_readonly = true
+    ```
 
-    *Sam Bostock*
+    and will be enabled by default with `config.load_defaults 7.1`.
 
-*   Add ssl support for postgresql database tasks
+    *Alex Ghiculescu*, *Hartley McGuire*
 
-    Add `PGSSLMODE`, `PGSSLCERT`, `PGSSLKEY` and `PGSSLROOTCERT` to pg_env from database config
-    when running postgresql database tasks.
+*   Allow unscoping of preload and eager_load associations
 
-    ```yaml
-    # config/database.yml
+    Added the ability to unscope preload and eager_load associations just like
+    includes, joins, etc. See ActiveRecord::QueryMethods::VALID_UNSCOPING_VALUES
+    for the full list of supported unscopable scopes.
 
-    production:
-      sslmode: verify-full
-      sslcert: client.crt
-      sslkey: client.key
-      sslrootcert: ca.crt
+    ```ruby
+    query.unscope(:eager_load, :preload).group(:id).select(:id)
     ```
 
-    Environment variables
+    *David Morehouse*
 
-    ```
-    PGSSLMODE=verify-full
-    PGSSLCERT=client.crt
-    PGSSLKEY=client.key
-    PGSSLROOTCERT=ca.crt
-    ```
+*   Add automatic filtering of encrypted attributes on inspect
 
-    Fixes #42994
+    This feature is enabled by default but can be disabled with
 
-    *Michael Bayucot*
+    ```ruby
+    config.active_record.encryption.add_to_filter_parameters = false
+    ```
+
+    *Hartley McGuire*
 
-*   Avoid scoping update callbacks in `ActiveRecord::Relation#update!`.
+*   Clear locking column on #dup
 
-    Making it consistent with how scoping is applied only to the query in `ActiveRecord::Relation#update`
-    and not also to the callbacks from the update itself.
+    This change fixes not to duplicate locking_column like id and timestamps.
 
-    *Dylan Thacker-Smith*
+    ```
+    car = Car.create!
+    car.touch
+    car.lock_version #=> 1
+    car.dup.lock_version #=> 0
+    ```
 
-*   Fix 2 cases that inferred polymorphic class from the association's `foreign_type`
-    using `String#constantize` instead of the model's `polymorphic_class_for`.
+    *Shouichi Kamiya*, *Seonggi Yang*, *Ryohei UEDA*
 
-    When updating a polymorphic association, the old `foreign_type` was not inferred correctly when:
-    1. `touch`ing the previously associated record
-    2. updating the previously associated record's `counter_cache`
+*   Invalidate transaction as early as possible
 
-    *Jimmy Bourassa*
+    After rescuing a `TransactionRollbackError` exception Rails invalidates transactions earlier in the flow
+    allowing the framework to skip issuing the `ROLLBACK` statement in more cases.
+    Only affects adapters that have `savepoint_errors_invalidate_transactions?` configured as `true`,
+    which at this point is only applicable to the `mysql2` adapter.
 
-*   Add config option for ignoring tables when dumping the schema cache.
+    *Nikita Vasilevsky*
 
-    Applications can now be configured to ignore certain tables when dumping the schema cache.
+*   Allow configuring columns list to be used in SQL queries issued by an `ActiveRecord::Base` object
 
-    The configuration option can table an array of tables:
+    It is now possible to configure columns list that will be used to build an SQL query clauses when
+    updating, deleting or reloading an `ActiveRecord::Base` object
 
     ```ruby
-    config.active_record.schema_cache_ignored_tables = ["ignored_table", "another_ignored_table"]
+    class Developer < ActiveRecord::Base
+      query_constraints :company_id, :id
+    end
+    developer = Developer.first.update(name: "Bob")
+    # => UPDATE "developers" SET "name" = 'Bob' WHERE "developers"."company_id" = 1 AND "developers"."id" = 1
     ```
 
-    Or a regex:
+    *Nikita Vasilevsky*
 
-    ```ruby
-    config.active_record.schema_cache_ignored_tables = [/^_/]
-    ```
+*   Adds `validate` to foreign keys and check constraints in schema.rb
 
-    *Eileen M. Uchitelle*
+    Previously, `schema.rb` would not record if `validate: false` had been used when adding a foreign key or check
+    constraint, so restoring a database from the schema could result in foreign keys or check constraints being
+    incorrectly validated.
 
-*   Make schema cache methods return consistent results.
+    *Tommy Graves*
 
-    Previously the schema cache methods `primary_keys`, `columns`, `columns_hash`, and `indexes`
-    would behave differently than one another when a table didn't exist and differently across
-    database adapters. This change unifies the behavior so each method behaves the same regardless
-    of adapter.
+*   Adapter `#execute` methods now accept an `allow_retry` option. When set to `true`, the SQL statement will be
+    retried, up to the database's configured `connection_retries` value, upon encountering connection-related errors.
 
-    The behavior now is:
+    *Adrianna Chang*
 
-    `columns`: (unchanged) raises a db error if the table does not exist.
-    `columns_hash`: (unchanged) raises a db error if the table does not exist.
-    `primary_keys`: (unchanged) returns `nil` if the table does not exist.
-    `indexes`: (changed for mysql2) returns `[]` if the table does not exist.
+*   Only trigger `after_commit :destroy` callbacks when a database row is deleted.
 
-    *Eileen M. Uchitelle*
+    This prevents `after_commit :destroy` callbacks from being triggered again
+    when `destroy` is called multiple times on the same record.
 
-*   Reestablish connection to previous database after after running `db:schema:load:name`
+    *Ben Sheldon*
 
-    After running `db:schema:load:name` the previous connection is restored.
+*   Fix `ciphertext_for` for yet-to-be-encrypted values.
 
-    *Jacopo Beschi*
+    Previously, `ciphertext_for` returned the cleartext of values that had not
+    yet been encrypted, such as with an unpersisted record:
 
-*   Add database config option `database_tasks`
+      ```ruby
+      Post.encrypts :body
 
-    If you would like to connect to an external database without any database
-    management tasks such as schema management, migrations, seeds, etc. you can set
-    the per database config option `database_tasks: false`
+      post = Post.create!(body: "Hello")
+      post.ciphertext_for(:body)
+      # => "{\"p\":\"abc..."
 
-    ```yaml
-    # config/database.yml
+      post.body = "World"
+      post.ciphertext_for(:body)
+      # => "World"
+      ```
 
-    production:
-      primary:
-        database: my_database
-        adapter: mysql2
-      animals:
-        database: my_animals_database
-        adapter: mysql2
-        database_tasks: false
-    ```
+    Now, `ciphertext_for` will always return the ciphertext of encrypted
+    attributes:
 
-    *Weston Ganger*
+      ```ruby
+      Post.encrypts :body
 
-*   Fix `ActiveRecord::InternalMetadata` to not be broken by `config.active_record.record_timestamps = false`
+      post = Post.create!(body: "Hello")
+      post.ciphertext_for(:body)
+      # => "{\"p\":\"abc..."
 
-    Since the model always create the timestamp columns, it has to set them, otherwise it breaks
-    various DB management tasks.
+      post.body = "World"
+      post.ciphertext_for(:body)
+      # => "{\"p\":\"xyz..."
+      ```
 
-    Fixes #42983
+    *Jonathan Hefner*
 
-*   Add `ActiveRecord::QueryLogs`.
+*   Fix a bug where using groups and counts with long table names would return incorrect results.
 
-    Configurable tags can be automatically added to all SQL queries generated by Active Record.
+    *Shota Toguchi*, *Yusaku Ono*
 
-    ```ruby
-    # config/application.rb
-    module MyApp
-      class Application < Rails::Application
-        config.active_record.query_log_tags_enabled = true
-      end
-    end
-    ```
+*   Fix encryption of column default values.
 
-    By default the application, controller and action details are added to the query tags:
+    Previously, encrypted attributes that used column default values appeared to
+    be encrypted on create, but were not:
 
-    ```ruby
-    class BooksController < ApplicationController
-      def index
-        @books = Book.all
-      end
-    end
-    ```
+      ```ruby
+      Book.encrypts :name
 
-    ```ruby
-    GET /books
-    # SELECT * FROM books /*application:MyApp;controller:books;action:index*/
-    ```
+      book = Book.create!
+      book.name
+      # => "<untitled>"
+      book.name_before_type_cast
+      # => "{\"p\":\"abc..."
+      book.reload.name_before_type_cast
+      # => "<untitled>"
+      ```
 
-    Custom tags containing static values and Procs can be defined in the application configuration:
+    Now, attributes with column default values are encrypted:
 
-    ```ruby
-    config.active_record.query_log_tags = [
-      :application,
-      :controller,
-      :action,
-      {
-        custom_static: "foo",
-        custom_dynamic: -> { Time.now }
-      }
-    ]
-    ```
+      ```ruby
+      Book.encrypts :name
 
-    *Keeran Raj Hawoldar*, *Eileen M. Uchitelle*, *Kasper Timm Hansen*
+      book = Book.create!
+      book.name
+      # => "<untitled>"
+      book.name_before_type_cast
+      # => "{\"p\":\"abc..."
+      book.reload.name_before_type_cast
+      # => "{\"p\":\"abc..."
+      ```
 
-*   Added support for multiple databases to `rails db:setup` and `rails db:reset`.
+    *Jonathan Hefner*
 
-    *Ryan Hall*
+*   Deprecate delegation from `Base` to `connection_handler`.
 
-*   Add `ActiveRecord::Relation#structurally_compatible?`.
+    Calling `Base.clear_all_connections!`, `Base.clear_active_connections!`, `Base.clear_reloadable_connections!` and `Base.flush_idle_connections!` is deprecated. Please call these methods on the connection handler directly. In future Rails versions, the delegation from `Base` to the `connection_handler` will be removed.
 
-    Adds a query method by which a user can tell if the relation that they're
-    about to use for `#or` or `#and` is structurally compatible with the
-    receiver.
+    *Eileen M. Uchitelle*
 
-    *Kevin Newton*
+*   Allow ActiveRecord::QueryMethods#reselect to receive hash values, similar to ActiveRecord::QueryMethods#select
 
-*   Add `ActiveRecord::QueryMethods#in_order_of`.
+    *Sampat Badhe*
 
-    This allows you to specify an explicit order that you'd like records
-    returned in based on a SQL expression. By default, this will be accomplished
-    using a case statement, as in:
+*   Validate options when managing columns and tables in migrations.
 
-    ```ruby
-    Post.in_order_of(:id, [3, 5, 1])
-    ```
+    If an invalid option is passed to a migration method like `create_table` and `add_column`, an error will be raised
+    instead of the option being silently ignored. Validation of the options will only be applied for new migrations
+    that are created.
 
-    will generate the SQL:
+    *Guo Xiang Tan*, *George Wambold*
 
-    ```sql
-    SELECT "posts".* FROM "posts" ORDER BY CASE "posts"."id" WHEN 3 THEN 1 WHEN 5 THEN 2 WHEN 1 THEN 3 ELSE 4 END ASC
-    ```
+*   Update query log tags to use the [SQLCommenter](https://open-telemetry.github.io/opentelemetry-sqlcommenter/) format by default. See [#46179](https://github.com/rails/rails/issues/46179)
 
-    However, because this functionality is built into MySQL in the form of the
-    `FIELD` function, that connection adapter will generate the following SQL
-    instead:
+    To opt out of SQLCommenter-formatted query log tags, set `config.active_record.query_log_tags_format = :legacy`. By default, this is set to `:sqlcommenter`.
 
-    ```sql
-    SELECT "posts".* FROM "posts" ORDER BY FIELD("posts"."id", 1, 5, 3) DESC
-    ```
+    *Modulitos* and *Iheanyi*
 
-    *Kevin Newton*
+*   Allow any ERB in the database.yml when creating rake tasks.
 
-*   Fix `eager_loading?` when ordering with `Symbol`.
+    Any ERB can be used in `database.yml` even if it accesses environment
+    configurations.
 
-    `eager_loading?` is triggered correctly when using `order` with symbols.
+    Deprecates `config.active_record.suppress_multiple_database_warning`.
 
-    ```ruby
-    scope = Post.includes(:comments).order(:"comments.label")
-    => true
-    ```
+    *Eike Send*
 
-    *Jacopo Beschi*
+*   Add table to error for duplicate column definitions.
 
-*   Two change tracking methods are added for `belongs_to` associations.
+    If a migration defines duplicate columns for a table, the error message
+    shows which table it concerns.
 
-    The `association_changed?` method (assuming an association named `:association`) returns true
-    if a different associated object has been assigned and the foreign key will be updated in the
-    next save.
+    *Petrik de Heus*
 
-    The `association_previously_changed?` method returns true if the previous save updated the
-    association to reference a different associated object.
+*   Fix erroneous nil default precision on virtual datetime columns.
 
-    *George Claghorn*
+    Prior to this change, virtual datetime columns did not have the same
+    default precision as regular datetime columns, resulting in the following
+    being erroneously equivalent:
 
-*   Add option to disable schema dump per-database.
+        t.virtual :name, type: datetime,                 as: "expression"
+        t.virtual :name, type: datetime, precision: nil, as: "expression"
 
-    Dumping the schema is on by default for all databases in an application. To turn it off for a
-    specific database, use the `schema_dump` option:
+    This change fixes the default precision lookup, so virtual and regular
+    datetime column default precisions match.
 
-    ```yaml
-    # config/database.yml
+    *Sam Bostock*
 
-    production:
-      schema_dump: false
-    ```
+*   Use connection from `#with_raw_connection` in `#quote_string`.
 
-    *Luis Vasconcellos*, *Eileen M. Uchitelle*
+    This ensures that the string quoting is wrapped in the reconnect and retry logic
+    that `#with_raw_connection` offers.
 
-*   Fix `eager_loading?` when ordering with `Hash` syntax.
+    *Adrianna Chang*
 
-    `eager_loading?` is triggered correctly when using `order` with hash syntax
-    on an outer table.
+*   Add `expires_in` option to `signed_id`.
 
-    ```ruby
-    Post.includes(:comments).order({ "comments.label": :ASC }).eager_loading?
-    # => true
-    ```
+    *Shouichi Kamiya*
 
-    *Jacopo Beschi*
+*   Allow applications to set retry deadline for query retries.
 
-*   Move the forcing of clear text encoding to the `ActiveRecord::Encryption::Encryptor`.
+    Building on the work done in #44576 and #44591, we extend the logic that automatically
+    reconnects database connections to take into account a timeout limit. We won't retry
+    a query if a given amount of time has elapsed since the query was first attempted. This
+    value defaults to nil, meaning that all retryable queries are retried regardless of time elapsed,
+    but this can be changed via the `retry_deadline` option in the database config.
 
-    Fixes #42699.
+    *Adrianna Chang*
 
-    *J Smith*
+*   Fix a case where the query cache can return wrong values. See #46044
 
-*   `partial_inserts` is now disabled by default in new apps.
+    *Aaron Patterson*
 
-    This will be the default for new apps in Rails 7. To opt in:
+*   Support MySQL's ssl-mode option for MySQLDatabaseTasks.
 
-    ```ruby
-    config.active_record.partial_inserts = true
-    ```
+    Verifying the identity of the database server requires setting the ssl-mode
+    option to VERIFY_CA or VERIFY_IDENTITY. This option was previously ignored
+    for MySQL database tasks like creating a database and dumping the structure.
 
-    If a migration removes the default value of a column, this option
-    would cause old processes to no longer be able to create new records.
+    *Petrik de Heus*
 
-    If you need to remove a column, you should first use `ignored_columns`
-    to stop using it.
+*   Move `ActiveRecord::InternalMetadata` to an independent object.
 
-    *Jean Boussier*
+    `ActiveRecord::InternalMetadata` no longer inherits from `ActiveRecord::Base` and is now an independent object that should be instantiated with a `connection`. This class is private and should not be used by applications directly. If you want to interact with the schema migrations table, please access it on the connection directly, for example: `ActiveRecord::Base.connection.schema_migration`.
 
-*   Rails can now verify foreign keys after loading fixtures in tests.
+    *Eileen M. Uchitelle*
 
-    This will be the default for new apps in Rails 7. To opt in:
+*   Deprecate quoting `ActiveSupport::Duration` as an integer
 
-    ```ruby
-    config.active_record.verify_foreign_keys_for_fixtures = true
+    Using ActiveSupport::Duration as an interpolated bind parameter in a SQL
+    string template is deprecated. To avoid this warning, you should explicitly
+    convert the duration to a more specific database type. For example, if you
+    want to use a duration as an integer number of seconds:
+    ```
+    Record.where("duration = ?", 1.hour.to_i)
+    ```
+    If you want to use a duration as an ISO 8601 string:
+    ```
+    Record.where("duration = ?", 1.hour.iso8601)
     ```
 
-    Tests will not run if there is a foreign key constraint violation in your fixture data.
-
-    The feature is supported by SQLite and PostgreSQL, other adapters can also add support for it.
+    *Aram Greenman*
 
-    *Alex Ghiculescu*
+*   Allow `QueryMethods#in_order_of` to order by a string column name.
 
-*   Clear cached `has_one` association after setting `belongs_to` association to `nil`.
+    ```ruby
+    Post.in_order_of("id", [4,2,3,1]).to_a
+    Post.joins(:author).in_order_of("authors.name", ["Bob", "Anna", "John"]).to_a
+    ```
 
-    After setting a `belongs_to` relation to `nil` and updating an unrelated attribute on the owner,
-    the owner should still return `nil` on the `has_one` relation.
+    *Igor Kasyanchuk*
 
-    Fixes #42597.
+*   Move `ActiveRecord::SchemaMigration` to an independent object.
 
-    *Michiel de Mare*
+    `ActiveRecord::SchemaMigration` no longer inherits from `ActiveRecord::Base` and is now an independent object that should be instantiated with a `connection`. This class is private and should not be used by applications directly. If you want to interact with the schema migrations table, please access it on the connection directly, for example: `ActiveRecord::Base.connection.schema_migration`.
 
-*   OpenSSL constants are now used for Digest computations.
+    *Eileen M. Uchitelle*
 
-    *Dirkjan Bussink*
+*   Deprecate `all_connection_pools` and make `connection_pool_list` more explicit.
 
-*   Adds support for `if_not_exists` to `add_foreign_key` and `if_exists` to `remove_foreign_key`.
+    Following on #45924 `all_connection_pools` is now deprecated. `connection_pool_list` will either take an explicit role or applications can opt into the new behavior by passing `:all`.
 
-    Applications can set their migrations to ignore exceptions raised when adding a foreign key
-    that already exists or when removing a foreign key that does not exist.
+    *Eileen M. Uchitelle*
 
-    Example Usage:
+*   Fix connection handler methods to operate on all pools.
 
-    ```ruby
-    class AddAuthorsForeignKeyToArticles < ActiveRecord::Migration[7.0]
-      def change
-        add_foreign_key :articles, :authors, if_not_exists: true
-      end
-    end
-    ```
+    `active_connections?`, `clear_active_connections!`, `clear_reloadable_connections!`, `clear_all_connections!`, and `flush_idle_connections!` now operate on all pools by default. Previously they would default to using the `current_role` or `:writing` role unless specified.
 
-    ```ruby
-    class RemoveAuthorsForeignKeyFromArticles < ActiveRecord::Migration[7.0]
-      def change
-        remove_foreign_key :articles, :authors, if_exists: true
-      end
-    end
-    ```
+    *Eileen M. Uchitelle*
 
-    *Roberto Miranda*
 
-*   Prevent polluting ENV during postgresql structure dump/load.
+*   Allow ActiveRecord::QueryMethods#select to receive hash values.
 
-    Some configuration parameters were provided to pg_dump / psql via
-    environment variables which persisted beyond the command being run, and may
-    have caused subsequent commands and connections to fail. Tasks running
-    across multiple postgresql databases like `rails db:test:prepare` may have
-    been affected.
+    Currently, `select` might receive only raw sql and symbols to define columns and aliases to select.
 
-    *Samuel Cochran*
+    With this change we can provide `hash` as argument, for example:
 
-*   Set precision 6 by default for `datetime` columns.
+    ```ruby
+    Post.joins(:comments).select(posts: [:id, :title, :created_at], comments: [:id, :body, :author_id])
+    #=> "SELECT \"posts\".\"id\", \"posts\".\"title\", \"posts\".\"created_at\", \"comments\".\"id\", \"comments\".\"body\", \"comments\".\"author_id\"
+    #   FROM \"posts\" INNER JOIN \"comments\" ON \"comments\".\"post_id\" = \"posts\".\"id\""
 
-    By default, datetime columns will have microseconds precision instead of seconds precision.
+    Post.joins(:comments).select(posts: { id: :post_id, title: :post_title }, comments: { id: :comment_id, body: :comment_body })
+    #=> "SELECT posts.id as post_id, posts.title as post_title, comments.id as comment_id, comments.body as comment_body
+    #    FROM \"posts\" INNER JOIN \"comments\" ON \"comments\".\"post_id\" = \"posts\".\"id\""
+    ```
+    *Oleksandr Holubenko*, *Josef Šimánek*, *Jean Boussier*
 
-    *Roberto Miranda*
+*   Adapts virtual attributes on `ActiveRecord::Persistence#becomes`.
 
-*   Allow preloading of associations with instance dependent scopes.
+    When source and target classes have a different set of attributes adapts
+    attributes such that the extra attributes from target are added.
 
-    *John Hawthorn*, *John Crepezzi*, *Adam Hess*, *Eileen M. Uchitelle*, *Dinah Shi*
+    ```ruby
+    class Person < ApplicationRecord
+    end
 
-*   Do not try to rollback transactions that failed due to a `ActiveRecord::TransactionRollbackError`.
+    class WebUser < Person
+      attribute :is_admin, :boolean
+      after_initialize :set_admin
 
-    *Jamie McCarthy*
+      def set_admin
+        write_attribute(:is_admin, email =~ /@ourcompany\.com$/)
+      end
+    end
 
-*   Active Record Encryption will now encode values as UTF-8 when using deterministic
-    encryption. The encoding is part of the encrypted payload, so different encodings for
-    different values result in different ciphertexts. This can break unique constraints and
-    queries.
+    person = Person.find_by(email: "email@ourcompany.com")
+    person.respond_to? :is_admin
+    # => false
+    person.becomes(WebUser).is_admin?
+    # => true
+    ```
 
-    The new behavior is configurable via `active_record.encryption.forced_encoding_for_deterministic_encryption`
-    that is `Encoding::UTF_8` by default. It can be disabled by setting it to `nil`.
+    *Jacopo Beschi*, *Sampson Crowley*
 
-    *Jorge Manrubia*
+*   Fix `ActiveRecord::QueryMethods#in_order_of` to include `nil`s, to match the
+    behavior of `Enumerable#in_order_of`.
 
-*   The MySQL adapter now cast numbers and booleans bind parameters to string for safety reasons.
+    For example, `Post.in_order_of(:title, [nil, "foo"])` will now include posts
+    with `nil` titles, the same as `Post.all.to_a.in_order_of(:title, [nil, "foo"])`.
 
-    When comparing a string and a number in a query, MySQL converts the string to a number. So for
-    instance `"foo" = 0`, will implicitly cast `"foo"` to `0` and will evaluate to `TRUE` which can
-    lead to security vulnerabilities.
+    *fatkodima*
 
-    Active Record already protect against that vulnerability when it knows the type of the column
-    being compared, however until now it was still vulnerable when using bind parameters:
+*   Optimize `add_timestamps` to use a single SQL statement.
 
     ```ruby
-    User.where("login_token = ?", 0).first
+    add_timestamps :my_table
     ```
 
-    Would perform:
+    Now results in the following SQL:
 
     ```sql
-    SELECT * FROM `users` WHERE `login_token` = 0 LIMIT 1;
+    ALTER TABLE "my_table" ADD COLUMN "created_at" datetime(6) NOT NULL, ADD COLUMN "updated_at" datetime(6) NOT NULL
     ```
 
-    Now it will perform:
-
-    ```sql
-    SELECT * FROM `users` WHERE `login_token` = '0' LIMIT 1;
-    ```
+    *Iliana Hadzhiatanasova*
 
-    *Jean Boussier*
+*   Add `drop_enum` migration command for PostgreSQL
 
-*   Fixture configurations (`_fixture`) are now strictly validated.
+    This does the inverse of `create_enum`. Before dropping an enum, ensure you have
+    dropped columns that depend on it.
 
-    If an error will be raised if that entry contains unknown keys while previously it
-    would silently have no effects.
+    *Alex Ghiculescu*
 
-    *Jean Boussier*
+*   Adds support for `if_exists` option when removing a check constraint.
 
-*   Add `ActiveRecord::Base.update!` that works like `ActiveRecord::Base.update` but raises exceptions.
+    The `remove_check_constraint` method now accepts an `if_exists` option. If set
+    to true an error won't be raised if the check constraint doesn't exist.
 
-    This allows for the same behavior as the instance method `#update!` at a class level.
+    *Margaret Parsa* and *Aditya Bhutani*
 
-    ```ruby
-    Person.update!(:all, state: "confirmed")
-    ```
+*   `find_or_create_by` now try to find a second time if it hits a unicity constraint.
 
-    *Dorian Marié*
+    `find_or_create_by` always has been inherently racy, either creating multiple
+    duplicate records or failing with `ActiveRecord::RecordNotUnique` depending on
+    whether a proper unicity constraint was set.
 
-*   Add `ActiveRecord::Base#attributes_for_database`.
+    `create_or_find_by` was introduced for this use case, however it's quite wasteful
+    when the record is expected to exist most of the time, as INSERT require to send
+    more data than SELECT and require more work from the database. Also on some
+    databases it can actually consume a primary key increment which is undesirable.
 
-    Returns attributes with values for assignment to the database.
+    So for case where most of the time the record is expected to exist, `find_or_create_by`
+    can be made race-condition free by re-trying the `find` if the `create` failed
+    with `ActiveRecord::RecordNotUnique`. This assumes that the table has the proper
+    unicity constraints, if not, `find_or_create_by` will still lead to duplicated records.
 
-    *Chris Salzberg*
+    *Jean Boussier*, *Alex Kitchens*
 
-*   Use an empty query to check if the PostgreSQL connection is still active.
+*   Introduce a simpler constructor API for ActiveRecord database adapters.
 
-    An empty query is faster than `SELECT 1`.
+    Previously the adapter had to know how to build a new raw connection to
+    support reconnect, but also expected to be passed an initial already-
+    established connection.
 
-    *Heinrich Lee Yu*
+    When manually creating an adapter instance, it will now accept a single
+    config hash, and only establish the real connection on demand.
 
-*   Add `ActiveRecord::Base#previously_persisted?`.
+    *Matthew Draper*
 
-    Returns `true` if the object has been previously persisted but now it has been deleted.
+*   Avoid redundant `SELECT 1` connection-validation query during DB pool
+    checkout when possible.
 
-*   Deprecate `partial_writes` in favor of `partial_inserts` and `partial_updates`.
+    If the first query run during a request is known to be idempotent, it can be
+    used directly to validate the connection, saving a network round-trip.
 
-    This allows to have a different behavior on update and create.
+    *Matthew Draper*
 
-    *Jean Boussier*
+*   Automatically reconnect broken database connections when safe, even
+    mid-request.
 
-*   Fix compatibility with `psych >= 4`.
+    When an error occurs while attempting to run a known-idempotent query, and
+    not inside a transaction, it is safe to immediately reconnect to the
+    database server and try again, so this is now the default behavior.
 
-    Starting in Psych 4.0.0 `YAML.load` behaves like `YAML.safe_load`. To preserve compatibility,
-    Active Record's schema cache loader and `YAMLColumn` now uses `YAML.unsafe_load` if available.
+    This new default should always be safe -- to support that, it's consciously
+    conservative about which queries are considered idempotent -- but if
+    necessary it can be disabled by setting the `connection_retries` connection
+    option to `0`.
 
-    *Jean Boussier*
+    *Matthew Draper*
 
-*   `ActiveRecord::Base.logger` is now a `class_attribute`.
+*   Avoid removing a PostgreSQL extension when there are dependent objects.
 
-    This means it can no longer be accessed directly through `@@logger`, and that setting `logger =`
-    on a subclass won't change the parent's logger.
+    Previously, removing an extension also implicitly removed dependent objects. Now, this will raise an error.
 
-    *Jean Boussier*
+    You can force removing the extension:
 
-*   Add `.asc.nulls_first` for all databases. Unfortunately MySQL still doesn't like `nulls_last`.
+    ```ruby
+    disable_extension :citext, force: :cascade
+    ```
 
-    *Keenan Brock*
+    Fixes #29091.
 
-*   Improve performance of `one?` and `many?` by limiting the generated count query to 2 results.
+    *fatkodima*
 
-    *Gonzalo Riestra*
+*   Allow nested functions as safe SQL string
 
-*   Don't check type when using `if_not_exists` on `add_column`.
+    *Michael Siegfried*
 
-    Previously, if a migration called `add_column` with the `if_not_exists` option set to true
-    the `column_exists?` check would look for a column with the same name and type as the migration.
+*   Allow `destroy_association_async_job=` to be configured with a class string instead of a constant.
 
-    Recently it was discovered that the type passed to the migration is not always the same type
-    as the column after migration. For example a column set to `:mediumblob` in the migration will
-    be casted to `binary` when calling `column.type`. Since there is no straightforward way to cast
-    the type to the database type without running the migration, we opted to drop the type check from
-    `add_column`. This means that migrations adding a duplicate column with a different type will no
-    longer raise an error.
+    Defers an autoloading dependency between `ActiveRecord::Base` and `ActiveJob::Base`
+    and moves the configuration of `ActiveRecord::DestroyAssociationAsyncJob`
+    from ActiveJob to ActiveRecord.
 
-    *Eileen M. Uchitelle*
+    Deprecates `ActiveRecord::ActiveJobRequiredError` and now raises a `NameError`
+    if the job class is unloadable or an `ActiveRecord::ConfigurationError` if
+    `dependent: :destroy_async` is declared on an association but there is no job
+    class configured.
 
-*   Log a warning message when running SQLite in production.
+    *Ben Sheldon*
 
-    Using SQLite in production ENV is generally discouraged. SQLite is also the default adapter
-    in a new Rails application.
-    For the above reasons log a warning message when running SQLite in production.
+*   Fix `ActiveRecord::Store` to serialize as a regular Hash
 
-    The warning can be disabled by setting `config.active_record.sqlite3_production_warning=false`.
+    Previously it would serialize as an `ActiveSupport::HashWithIndifferentAccess`
+    which is wasteful and cause problem with YAML safe_load.
 
-    *Jacopo Beschi*
+    *Jean Boussier*
 
-*   Add option to disable joins for `has_one` associations.
+*   Add `timestamptz` as a time zone aware type for PostgreSQL
 
-    In a multiple database application, associations can't join across
-    databases. When set, this option instructs Rails to generate 2 or
-    more queries rather than generating joins for `has_one` associations.
+    This is required for correctly parsing `timestamp with time zone` values in your database.
 
-    Set the option on a has one through association:
+    If you don't want this, you can opt out by adding this initializer:
 
     ```ruby
-    class Person
-      has_one :dog
-      has_one :veterinarian, through: :dog, disable_joins: true
-    end
-    ```
-
-    Then instead of generating join SQL, two queries are used for `@person.veterinarian`:
-
-    ```
-    SELECT "dogs"."id" FROM "dogs" WHERE "dogs"."person_id" = ?  [["person_id", 1]]
-    SELECT "veterinarians".* FROM "veterinarians" WHERE "veterinarians"."dog_id" = ?  [["dog_id", 1]]
+    ActiveRecord::Base.time_zone_aware_types -= [:timestamptz]
     ```
 
-    *Sarah Vessels*, *Eileen M. Uchitelle*
-
-*   `Arel::Visitors::Dot` now renders a complete set of properties when visiting
-    `Arel::Nodes::SelectCore`, `SelectStatement`, `InsertStatement`, `UpdateStatement`, and
-    `DeleteStatement`, which fixes #42026. Previously, some properties were omitted.
-
-    *Mike Dalessio*
+    *Alex Ghiculescu*
 
-*   `Arel::Visitors::Dot` now supports `Arel::Nodes::Bin`, `Case`, `CurrentRow`, `Distinct`,
-    `DistinctOn`, `Else`, `Except`, `InfixOperation`, `Intersect`, `Lock`, `NotRegexp`, `Quoted`,
-    `Regexp`, `UnaryOperation`, `Union`, `UnionAll`, `When`, and `With`. Previously, these node
-    types caused an exception to be raised by `Arel::Visitors::Dot#accept`.
+*   Add new `ActiveRecord::Base.generates_token_for` API.
 
-    *Mike Dalessio*
+    Currently, `signed_id` fulfills the role of generating tokens for e.g.
+    resetting a password.  However, signed IDs cannot reflect record state, so
+    if a token is intended to be single-use, it must be tracked in a database at
+    least until it expires.
 
-*   Optimize `remove_columns` to use a single SQL statement.
+    With `generates_token_for`, a token can embed data from a record.  When
+    using the token to fetch the record, the data from the token and the current
+    data from the record will be compared.  If the two do not match, the token
+    will be treated as invalid, the same as if it had expired.  For example:
 
     ```ruby
-    remove_columns :my_table, :col_one, :col_two
-    ```
+    class User < ActiveRecord::Base
+      has_secure_password
 
-    Now results in the following SQL:
+      generates_token_for :password_reset, expires_in: 15.minutes do
+        # A password's BCrypt salt changes when the password is updated.
+        # By embedding (part of) the salt in a token, the token will
+        # expire when the password is updated.
+        BCrypt::Password.new(password_digest).salt[-10..]
+      end
+    end
 
-    ```sql
-    ALTER TABLE "my_table" DROP COLUMN "col_one", DROP COLUMN "col_two"
-    ```
+    user = User.first
+    token = user.generate_token_for(:password_reset)
 
-    *Jon Dufresne*
+    User.find_by_token_for(:password_reset, token) # => user
 
-*   Ensure `has_one` autosave association callbacks get called once.
+    user.update!(password: "new password")
+    User.find_by_token_for(:password_reset, token) # => nil
+    ```
 
-    Change the `has_one` autosave callback to be non cyclic as well.
-    By doing this the autosave callback are made more consistent for
-    all 3 cases: `has_many`, `has_one`, and `belongs_to`.
+    *Jonathan Hefner*
 
-    *Petrik de Heus*
+*   Optimize Active Record batching for whole table iterations.
 
-*   Add option to disable joins for associations.
+    Previously, `in_batches` got all the ids and constructed an `IN`-based query for each batch.
+    When iterating over the whole tables, this approach is not optimal as it loads unneeded ids and
+    `IN` queries with lots of items are slow.
 
-    In a multiple database application, associations can't join across
-    databases. When set, this option instructs Rails to generate 2 or
-    more queries rather than generating joins for associations.
+    Now, whole table iterations use range iteration (`id >= x AND id <= y`) by default which can make iteration
+    several times faster. E.g., tested on a PostgreSQL table with 10 million records: querying (`253s` vs `30s`),
+    updating (`288s` vs `124s`), deleting (`268s` vs `83s`).
 
-    Set the option on a has many through association:
+    Only whole table iterations use this style of iteration by default. You can disable this behavior by passing `use_ranges: false`.
+    If you iterate over the table and the only condition is, e.g., `archived_at: nil` (and only a tiny fraction
+    of the records are archived), it makes sense to opt in to this approach:
 
     ```ruby
-    class Dog
-      has_many :treats, through: :humans, disable_joins: true
-      has_many :humans
+    Project.where(archived_at: nil).in_batches(use_ranges: true) do |relation|
+      # do something
     end
     ```
 
-    Then instead of generating join SQL, two queries are used for `@dog.treats`:
-
-    ```
-    SELECT "humans"."id" FROM "humans" WHERE "humans"."dog_id" = ?  [["dog_id", 1]]
-    SELECT "treats".* FROM "treats" WHERE "treats"."human_id" IN (?, ?, ?)  [["human_id", 1], ["human_id", 2], ["human_id", 3]]
-    ```
-
-    *Eileen M. Uchitelle*, *Aaron Patterson*, *Lee Quarella*
-
-*   Add setting for enumerating column names in SELECT statements.
+    See #45414 for more details.
 
-    Adding a column to a PostgreSQL database, for example, while the application is running can
-    change the result of wildcard `SELECT *` queries, which invalidates the result
-    of cached prepared statements and raises a `PreparedStatementCacheExpired` error.
-
-    When enabled, Active Record will avoid wildcards and always include column names
-    in `SELECT` queries, which will return consistent results and avoid prepared
-    statement errors.
+    *fatkodima*
 
-    Before:
+*   `.with` query method added. Construct common table expressions with ease and get `ActiveRecord::Relation` back.
 
     ```ruby
-    Book.limit(5)
-    # SELECT * FROM books LIMIT 5
+    Post.with(posts_with_comments: Post.where("comments_count > ?", 0))
+    # => ActiveRecord::Relation
+    # WITH posts_with_comments AS (SELECT * FROM posts WHERE (comments_count > 0)) SELECT * FROM posts
     ```
 
-    After:
-
-    ```ruby
-    # config/application.rb
-    module MyApp
-      class Application < Rails::Application
-        config.active_record.enumerate_columns_in_select_statements = true
-      end
-    end
+    *Vlado Cingel*
 
-    # or, configure per-model
-    class Book < ApplicationRecord
-      self.enumerate_columns_in_select_statements = true
-    end
-    ```
+*   Don't establish a new connection if an identical pool exists already.
 
-    ```ruby
-    Book.limit(5)
-    # SELECT id, author_id, name, format, status, language, etc FROM books LIMIT 5
-    ```
+    Previously, if `establish_connection` was called on a class that already had an established connection, the existing connection would be removed regardless of whether it was the same config. Now if a pool is found with the same values as the new connection, the existing connection will be returned instead of creating a new one.
 
-    *Matt Duszynski*
+    This has a slight change in behavior if application code is depending on a new connection being established regardless of whether it's identical to an existing connection. If the old behavior is desirable, applications should call `ActiveRecord::Base#remove_connection` before establishing a new one. Calling `establish_connection` with a different config works the same way as it did previously.
 
-*   Allow passing SQL as `on_duplicate` value to `#upsert_all` to make it possible to use raw SQL to update columns on conflict:
+    *Eileen M. Uchitelle*
 
-    ```ruby
-    Book.upsert_all(
-      [{ id: 1, status: 1 }, { id: 2, status: 1 }],
-      on_duplicate: Arel.sql("status = GREATEST(books.status, EXCLUDED.status)")
-    )
-    ```
+*   Update `db:prepare` task to load schema when an uninitialized database exists, and dump schema after migrations.
 
-    *Vladimir Dementyev*
+    *Ben Sheldon*
 
-*   Allow passing SQL as `returning` statement to `#upsert_all`:
+*   Fix supporting timezone awareness for `tsrange` and `tstzrange` array columns.
 
     ```ruby
-    Article.insert_all(
-      [
-        { title: "Article 1", slug: "article-1", published: false },
-        { title: "Article 2", slug: "article-2", published: false }
-      ],
-      returning: Arel.sql("id, (xmax = '0') as inserted, name as new_name")
-    )
+    # In database migrations
+    add_column :shops, :open_hours, :tsrange, array: true
+    # In app config
+    ActiveRecord::Base.time_zone_aware_types += [:tsrange]
+    # In the code times are properly converted to app time zone
+    Shop.create!(open_hours: [Time.current..8.hour.from_now])
     ```
 
-    *Vladimir Dementyev*
+    *Wojciech Wnętrzak*
 
-*   Deprecate `legacy_connection_handling`.
+*   Introduce strategy pattern for executing migrations.
 
-    *Eileen M. Uchitelle*
+    By default, migrations will use a strategy object that delegates the method
+    to the connection adapter. Consumers can implement custom strategy objects
+    to change how their migrations run.
 
-*   Add attribute encryption support.
+    *Adrianna Chang*
 
-    Encrypted attributes are declared at the model level. These
-    are regular Active Record attributes backed by a column with
-    the same name. The system will transparently encrypt these
-    attributes before saving them into the database and will
-    decrypt them when retrieving their values.
+*   Add adapter option disallowing foreign keys
 
+    This adds a new option to be added to `database.yml` which enables skipping
+    foreign key constraints usage even if the underlying database supports them.
 
-    ```ruby
-    class Person < ApplicationRecord
-      encrypts :name
-      encrypts :email_address, deterministic: true
-    end
+    Usage:
+    ```yaml
+    development:
+        <<: *default
+        database: storage/development.sqlite3
+        foreign_keys: false
     ```
 
-    You can learn more in the [Active Record Encryption
-    guide](https://edgeguides.rubyonrails.org/active_record_encryption.html).
+    *Paulo Barros*
 
-    *Jorge Manrubia*
+*   Add configurable deprecation warning for singular associations
 
-*   Changed Arel predications `contains` and `overlaps` to use
-    `quoted_node` so that PostgreSQL arrays are quoted properly.
+    This adds a deprecation warning when using the plural name of a singular associations in `where`.
+    It is possible to opt into the new more performant behavior with `config.active_record.allow_deprecated_singular_associations_name = false`
 
-    *Bradley Priest*
+    *Adam Hess*
 
-*   Add mode argument to record level `strict_loading!`.
+*   Run transactional callbacks on the freshest instance to save a given
+    record within a transaction.
 
-    This argument can be used when enabling strict loading for a single record
-    to specify that we only want to raise on n plus one queries.
+    When multiple Active Record instances change the same record within a
+    transaction, Rails runs `after_commit` or `after_rollback` callbacks for
+    only one of them. `config.active_record.run_commit_callbacks_on_first_saved_instances_in_transaction`
+    was added to specify how Rails chooses which instance receives the
+    callbacks. The framework defaults were changed to use the new logic.
 
-    ```ruby
-    developer.strict_loading!(mode: :n_plus_one_only)
+    When `config.active_record.run_commit_callbacks_on_first_saved_instances_in_transaction`
+    is `true`, transactional callbacks are run on the first instance to save,
+    even though its instance state may be stale.
 
-    developer.projects.to_a # Does not raise
-    developer.projects.first.client # Raises StrictLoadingViolationError
-    ```
+    When it is `false`, which is the new framework default starting with version
+    7.1, transactional callbacks are run on the instances with the freshest
+    instance state. Those instances are chosen as follows:
 
-    Previously, enabling strict loading would cause any lazily loaded
-    association to raise an error. Using `n_plus_one_only` mode allows us to
-    lazily load belongs_to, has_many, and other associations that are fetched
-    through a single query.
+    - In general, run transactional callbacks on the last instance to save a
+      given record within the transaction.
+    - There are two exceptions:
+        - If the record is created within the transaction, then updated by
+          another instance, `after_create_commit` callbacks will be run on the
+          second instance. This is instead of the `after_update_commit`
+          callbacks that would naively be run based on that instance’s state.
+        - If the record is destroyed within the transaction, then
+          `after_destroy_commit` callbacks will be fired on the last destroyed
+          instance, even if a stale instance subsequently performed an update
+          (which will have affected 0 rows).
 
-    *Dinah Shi*
+    *Cameron Bothner and Mitch Vollebregt*
 
-*   Fix Float::INFINITY assignment to datetime column with postgresql adapter.
+*   Enable strict strings mode for `SQLite3Adapter`.
 
-    Before:
+    Configures SQLite with a strict strings mode, which disables double-quoted string literals.
 
-    ```ruby
-    # With this config
-    ActiveRecord::Base.time_zone_aware_attributes = true
+    SQLite has some quirks around double-quoted string literals.
+    It first tries to consider double-quoted strings as identifier names, but if they don't exist
+    it then considers them as string literals. Because of this, typos can silently go unnoticed.
+    For example, it is possible to create an index for a non existing column.
+    See [SQLite documentation](https://www.sqlite.org/quirks.html#double_quoted_string_literals_are_accepted) for more details.
 
-    # and the following schema:
-    create_table "postgresql_infinities" do |t|
-      t.datetime "datetime"
-    end
+    If you don't want this behavior, you can disable it via:
 
-    # This test fails
-    record = PostgresqlInfinity.create!(datetime: Float::INFINITY)
-    assert_equal Float::INFINITY, record.datetime # record.datetime gets nil
+    ```ruby
+    # config/application.rb
+    config.active_record.sqlite3_adapter_strict_strings_by_default = false
     ```
 
-    After this commit, `record.datetime` gets `Float::INFINITY` as expected.
+    Fixes #27782.
+
+    *fatkodima*, *Jean Boussier*
 
-    *Shunichi Ikegami*
+*   Resolve issue where a relation cache_version could be left stale.
 
-*   Type cast enum values by the original attribute type.
+    Previously, when `reset` was called on a relation object it did not reset the cache_versions
+    ivar. This led to a confusing situation where despite having the correct data the relation
+    still reported a stale cache_version.
 
-    The notable thing about this change is that unknown labels will no longer match 0 on MySQL.
+    Usage:
 
     ```ruby
-    class Book < ActiveRecord::Base
-      enum :status, { proposed: 0, written: 1, published: 2 }
-    end
-    ```
+    developers = Developer.all
+    developers.cache_version
 
-    Before:
+    Developer.update_all(updated_at: Time.now.utc + 1.second)
 
-    ```ruby
-    # SELECT `books`.* FROM `books` WHERE `books`.`status` = 'prohibited' LIMIT 1
-    Book.find_by(status: :prohibited)
-    # => #<Book id: 1, status: "proposed", ...> (for mysql2 adapter)
-    # => ActiveRecord::StatementInvalid: PG::InvalidTextRepresentation: ERROR:  invalid input syntax for type integer: "prohibited" (for postgresql adapter)
-    # => nil (for sqlite3 adapter)
+    developers.cache_version # Stale cache_version
+    developers.reset
+    developers.cache_version # Returns the current correct cache_version
     ```
 
-    After:
+    Fixes #45341.
+
+    *Austen Madden*
+
+*   Add support for exclusion constraints (PostgreSQL-only).
 
     ```ruby
-    # SELECT `books`.* FROM `books` WHERE `books`.`status` IS NULL LIMIT 1
-    Book.find_by(status: :prohibited)
-    # => nil (for all adapters)
+    add_exclusion_constraint :invoices, "daterange(start_date, end_date) WITH &&", using: :gist, name: "invoices_date_overlap"
+    remove_exclusion_constraint :invoices, name: "invoices_date_overlap"
     ```
 
-    *Ryuta Kamizono*
+    See PostgreSQL's [`CREATE TABLE ... EXCLUDE ...`](https://www.postgresql.org/docs/12/sql-createtable.html#SQL-CREATETABLE-EXCLUDE) documentation for more on exclusion constraints.
 
-*   Fixtures for `has_many :through` associations now load timestamps on join tables.
+    *Alex Robbin*
 
-    Given this fixture:
+*   `change_column_null` raises if a non-boolean argument is provided
 
-    ```yml
-    ### monkeys.yml
-    george:
-      name: George the Monkey
-      fruits: apple
+    Previously if you provided a non-boolean argument, `change_column_null` would
+    treat it as truthy and make your column nullable. This could be surprising, so now
+    the input must be either `true` or `false`.
 
-    ### fruits.yml
-    apple:
-      name: apple
+    ```ruby
+    change_column_null :table, :column, true # good
+    change_column_null :table, :column, false # good
+    change_column_null :table, :column, from: true, to: false # raises (previously this made the column nullable)
     ```
 
-    If the join table (`fruit_monkeys`) contains `created_at` or `updated_at` columns,
-    these will now be populated when loading the fixture. Previously, fixture loading
-    would crash if these columns were required, and leave them as null otherwise.
-
     *Alex Ghiculescu*
 
-*   Allow applications to configure the thread pool for async queries.
+*   Enforce limit on table names length.
 
-    Some applications may want one thread pool per database whereas others want to use
-    a single global thread pool for all queries. By default, Rails will set `async_query_executor`
-    to `nil` which will not initialize any executor. If `load_async` is called and no executor
-    has been configured, the query will be executed in the foreground.
+    Fixes #45130.
 
-    To create one thread pool for all database connections to use applications can set
-    `config.active_record.async_query_executor` to `:global_thread_pool` and optionally define
-    `config.active_record.global_executor_concurrency`. This defaults to 4. For applications that want
-    to have a thread pool for each database connection, `config.active_record.async_query_executor` can
-    be set to `:multi_thread_pool`. The configuration for each thread pool is set in the database
-    configuration.
+    *fatkodima*
 
-    *Eileen M. Uchitelle*
+*   Adjust the minimum MariaDB version for check constraints support.
 
-*   Allow new syntax for `enum` to avoid leading `_` from reserved options.
+    *Eddie Lebow*
 
-    Before:
+*   Fix Hstore deserialize regression.
 
-    ```ruby
-    class Book < ActiveRecord::Base
-      enum status: [ :proposed, :written ], _prefix: true, _scopes: false
-      enum cover: [ :hard, :soft ], _suffix: true, _default: :hard
-    end
-    ```
+    *edsharp*
 
-    After:
+*   Add validity for PostgreSQL indexes.
 
     ```ruby
-    class Book < ActiveRecord::Base
-      enum :status, [ :proposed, :written ], prefix: true, scopes: false
-      enum :cover, [ :hard, :soft ], suffix: true, default: :hard
-    end
+    connection.index_exists?(:users, :email, valid: true)
+    connection.indexes(:users).select(&:valid?)
     ```
 
-    *Ryuta Kamizono*
+    *fatkodima*
 
-*   Add `ActiveRecord::Relation#load_async`.
+*   Fix eager loading for models without primary keys.
 
-    This method schedules the query to be performed asynchronously from a thread pool.
+    *Anmol Chopra*, *Matt Lawrence*, and *Jonathan Hefner*
 
-    If the result is accessed before a background thread had the opportunity to perform
-    the query, it will be performed in the foreground.
+*   Avoid validating a unique field if it has not changed and is backed by a unique index.
 
-    This is useful for queries that can be performed long enough before their result will be
-    needed, or for controllers which need to perform several independent queries.
+    Previously, when saving a record, Active Record will perform an extra query to check for the
+    uniqueness of each attribute having a `uniqueness` validation, even if that attribute hasn't changed.
+    If the database has the corresponding unique index, then this validation can never fail for persisted
+    records, and we could safely skip it.
 
-    ```ruby
-    def index
-      @categories = Category.some_complex_scope.load_async
-      @posts = Post.some_complex_scope.load_async
-    end
-    ```
+    *fatkodima*
 
-    Active Record logs will also include timing info for the duration of how long
-    the main thread had to wait to access the result. This timing is useful to know
-    whether or not it's worth to load the query asynchronously.
+*   Stop setting `sql_auto_is_null`
 
-    ```
-    DEBUG -- :   Category Load (62.1ms)  SELECT * FROM `categories` LIMIT 50
-    DEBUG -- :   ASYNC Post Load (64ms) (db time 126.1ms)  SELECT * FROM `posts` LIMIT 100
-    ```
+    Since version 5.5 the default has been off, we no longer have to manually turn it off.
 
-    The duration in the first set of parens is how long the main thread was blocked
-    waiting for the results, and the second set of parens with "db time" is how long
-    the entire query took to execute.
+    *Adam Hess*
 
-    *Jean Boussier*
+*   Fix `touch` to raise an error for readonly columns.
 
-*   Implemented `ActiveRecord::Relation#excluding` method.
+    *fatkodima*
 
-    This method excludes the specified record (or collection of records) from
-    the resulting relation:
+*   Add ability to ignore tables by regexp for SQL schema dumps.
 
     ```ruby
-    Post.excluding(post)
-    Post.excluding(post_one, post_two)
+    ActiveRecord::SchemaDumper.ignore_tables = [/^_/]
     ```
 
-    Also works on associations:
-
-    ```ruby
-    post.comments.excluding(comment)
-    post.comments.excluding(comment_one, comment_two)
-    ```
+    *fatkodima*
 
-    This is short-hand for `Post.where.not(id: post.id)` (for a single record)
-    and `Post.where.not(id: [post_one.id, post_two.id])` (for a collection).
+*   Avoid queries when performing calculations on contradictory relations.
 
-    *Glen Crawford*
+    Previously calculations would make a query even when passed a
+    contradiction, such as `User.where(id: []).count`. We no longer perform a
+    query in that scenario.
 
-*   Skip optimised #exist? query when #include? is called on a relation
-    with a having clause.
+    This applies to the following calculations: `count`, `sum`, `average`,
+    `minimum` and `maximum`
 
-    Relations that have aliased select values AND a having clause that
-    references an aliased select value would generate an error when
-    #include? was called, due to an optimisation that would generate
-    call #exists? on the relation instead, which effectively alters
-    the select values of the query (and thus removes the aliased select
-    values), but leaves the having clause intact. Because the having
-    clause is then referencing an aliased column that is no longer
-    present in the simplified query, an ActiveRecord::InvalidStatement
-    error was raised.
+    *Luan Vieira, John Hawthorn and Daniel Colson*
 
-    A sample query affected by this problem:
+*   Allow using aliased attributes with `insert_all`/`upsert_all`.
 
     ```ruby
-    Author.select('COUNT(*) as total_posts', 'authors.*')
-          .joins(:posts)
-          .group(:id)
-          .having('total_posts > 2')
-          .include?(Author.first)
+    class Book < ApplicationRecord
+      alias_attribute :title, :name
+    end
+
+    Book.insert_all [{ title: "Remote", author_id: 1 }], returning: :title
     ```
 
-    This change adds an addition check to the condition that skips the
-    simplified #exists? query, which simply checks for the presence of
-    a having clause.
+    *fatkodima*
 
-    Fixes #41417.
+*   Support encrypted attributes on columns with default db values.
 
-    *Michael Smart*
+    This adds support for encrypted attributes defined on columns with default values.
+    It will encrypt those values at creation time. Before, it would raise an
+    error unless `config.active_record.encryption.support_unencrypted_data` was true.
 
-*   Increment postgres prepared statement counter before making a prepared statement, so if the statement is aborted
-    without Rails knowledge (e.g., if app gets killed during long-running query or due to Rack::Timeout), app won't end
-    up in perpetual crash state for being inconsistent with PostgreSQL.
+    *Jorge Manrubia* and *Dima Fatko*
 
-    *wbharding*, *Martin Tepper*
+*   Allow overriding `reading_request?` in `DatabaseSelector::Resolver`
 
-*   Add ability to apply `scoping` to `all_queries`.
+    The default implementation checks if a request is a `get?` or `head?`,
+    but you can now change it to anything you like. If the method returns true,
+    `Resolver#read` gets called meaning the request could be served by the
+    replica database.
 
-    Some applications may want to use the `scoping` method but previously it only
-    worked on certain types of queries. This change allows the `scoping` method to apply
-    to all queries for a model in a block.
+    *Alex Ghiculescu*
 
-    ```ruby
-    Post.where(blog_id: post.blog_id).scoping(all_queries: true) do
-      post.update(title: "a post title") # adds `posts.blog_id = 1` to the query
-    end
-    ```
+*   Remove `ActiveRecord.legacy_connection_handling`.
 
     *Eileen M. Uchitelle*
 
-*   `ActiveRecord::Calculations.calculate` called with `:average`
-    (aliased as `ActiveRecord::Calculations.average`) will now use column-based
-    type casting. This means that floating-point number columns will now be
-    aggregated as `Float` and decimal columns will be aggregated as `BigDecimal`.
+*   `rails db:schema:{dump,load}` now checks `ENV["SCHEMA_FORMAT"]` before config
 
-    Integers are handled as a special case returning `BigDecimal` always
-    (this was the case before already).
+    Since `rails db:structure:{dump,load}` was deprecated there wasn't a simple
+    way to dump a schema to both SQL and Ruby formats. You can now do this with
+    an environment variable. For example:
 
-    ```ruby
-    # With the following schema:
-    create_table "measurements" do |t|
-      t.float "temperature"
-    end
+    ```
+    SCHEMA_FORMAT=sql rake db:schema:dump
+    ```
 
-    # Before:
-    Measurement.average(:temperature).class
-    # => BigDecimal
+    *Alex Ghiculescu*
 
-    # After:
-    Measurement.average(:temperature).class
-    # => Float
-    ```
+*   Fixed MariaDB default function support.
+
+    Defaults would be written wrong in "db/schema.rb" and not work correctly
+    if using `db:schema:load`. Further more the function name would be
+    added as string content when saving new records.
 
-    Before this change, Rails just called `to_d` on average aggregates from the
-    database adapter. This is not the case anymore. If you relied on that kind
-    of magic, you now need to register your own `ActiveRecord::Type`
-    (see `ActiveRecord::Attributes::ClassMethods` for documentation).
+    *kaspernj*
 
-    *Josua Schmid*
+*   Add `active_record.destroy_association_async_batch_size` configuration
 
-*   PostgreSQL: introduce `ActiveRecord::ConnectionAdapters::PostgreSQLAdapter.datetime_type`.
+    This allows applications to specify the maximum number of records that will
+    be destroyed in a single background job by the `dependent: :destroy_async`
+    association option. By default, the current behavior will remain the same:
+    when a parent record is destroyed, all dependent records will be destroyed
+    in a single background job. If the number of dependent records is greater
+    than this configuration, the records will be destroyed in multiple
+    background jobs.
 
-    This setting controls what native type Active Record should use when you call `datetime` in
-    a migration or schema. It takes a symbol which must correspond to one of the configured
-    `NATIVE_DATABASE_TYPES`. The default is `:timestamp`, meaning `t.datetime` in a migration
-    will create a "timestamp without time zone" column. To use "timestamp with time zone",
-    change this to `:timestamptz` in an initializer.
+    *Nick Holden*
 
-    You should run `bin/rails db:migrate` to rebuild your schema.rb if you change this.
+*   Fix `remove_foreign_key` with `:if_exists` option when foreign key actually exists.
 
-    *Alex Ghiculescu*
+    *fatkodima*
+
+*   Remove `--no-comments` flag in structure dumps for PostgreSQL
 
-*   PostgreSQL: handle `timestamp with time zone` columns correctly in `schema.rb`.
+    This broke some apps that used custom schema comments. If you don't want
+    comments in your structure dump, you can use:
 
-    Previously they dumped as `t.datetime :column_name`, now they dump as `t.timestamptz :column_name`,
-    and are created as `timestamptz` columns when the schema is loaded.
+    ```ruby
+    ActiveRecord::Tasks::DatabaseTasks.structure_dump_flags = ['--no-comments']
+    ```
 
     *Alex Ghiculescu*
 
-*   Removing trailing whitespace when matching columns in
-    `ActiveRecord::Sanitization.disallow_raw_sql!`.
+*   Reduce the memory footprint of fixtures accessors.
 
-    *Gannon McGibbon*, *Adrian Hirt*
+    Until now fixtures accessors were eagerly defined using `define_method`.
+    So the memory usage was directly dependent of the number of fixtures and
+    test suites.
 
-*   Expose a way for applications to set a `primary_abstract_class`.
+    Instead fixtures accessors are now implemented with `method_missing`,
+    so they incur much less memory and CPU overhead.
 
-    Multiple database applications that use a primary abstract class that is not
-    named `ApplicationRecord` can now set a specific class to be the `primary_abstract_class`.
+    *Jean Boussier*
 
-    ```ruby
-    class PrimaryApplicationRecord
-      self.primary_abstract_class
-    end
-    ```
+*   Fix `config.active_record.destroy_association_async_job` configuration
 
-    When an application boots it automatically connects to the primary or first database in the
-    database configuration file. In a multiple database application that then call `connects_to`
-    needs to know that the default connection is the same as the `ApplicationRecord` connection.
-    However, some applications have a differently named `ApplicationRecord`. This prevents Active
-    Record from opening duplicate connections to the same database.
+    `config.active_record.destroy_association_async_job` should allow
+    applications to specify the job that will be used to destroy associated
+    records in the background for `has_many` associations with the
+    `dependent: :destroy_async` option. Previously, that was ignored, which
+    meant the default `ActiveRecord::DestroyAssociationAsyncJob` always
+    destroyed records in the background.
 
-    *Eileen M. Uchitelle*, *John Crepezzi*
+    *Nick Holden*
 
-*   Support hash config for `structure_dump_flags` and `structure_load_flags` flags.
-    Now that Active Record supports multiple databases configuration,
-    we need a way to pass specific flags for dump/load databases since
-    the options are not the same for different adapters.
-    We can use in the original way:
+*   Fix `change_column_comment` to preserve column's AUTO_INCREMENT in the MySQL adapter
 
-    ```ruby
-    ActiveRecord::Tasks::DatabaseTasks.structure_dump_flags = ['--no-defaults', '--skip-add-drop-table']
-    # or
-    ActiveRecord::Tasks::DatabaseTasks.structure_dump_flags = '--no-defaults --skip-add-drop-table'
-    ```
+    *fatkodima*
 
-    And also use it passing a hash, with one or more keys, where the key
-    is the adapter
+*   Fix quoting of `ActiveSupport::Duration` and `Rational` numbers in the MySQL adapter.
 
-    ```ruby
-    ActiveRecord::Tasks::DatabaseTasks.structure_dump_flags = {
-      mysql2: ['--no-defaults', '--skip-add-drop-table'],
-      postgres: '--no-tablespaces'
-    }
-    ```
+    *Kevin McPhillips*
 
-    *Gustavo Gonzalez*
+*   Allow column name with COLLATE (e.g., title COLLATE "C") as safe SQL string
 
-*   Connection specification now passes the "url" key as a configuration for the
-    adapter if the "url" protocol is "jdbc", "http", or "https". Previously only
-    urls with the "jdbc" prefix were passed to the Active Record Adapter, others
-    are assumed to be adapter specification urls.
+    *Shugo Maeda*
 
-    Fixes #41137.
+*   Permit underscores in the VERSION argument to database rake tasks.
 
-    *Jonathan Bracy*
+    *Eddie Lebow*
 
-*   Allow to opt-out of `strict_loading` mode on a per-record base.
+*   Reversed the order of `INSERT` statements in `structure.sql` dumps
 
-    This is useful when strict loading is enabled application wide or on a
-    model level.
+    This should decrease the likelihood of merge conflicts. New migrations
+    will now be added at the top of the list.
 
-    ```ruby
-    class User < ApplicationRecord
-      has_many :bookmarks
-      has_many :articles, strict_loading: true
-    end
+    For existing apps, there will be a large diff the next time `structure.sql`
+    is generated.
 
-    user = User.first
-    user.articles                        # => ActiveRecord::StrictLoadingViolationError
-    user.bookmarks                       # => #<ActiveRecord::Associations::CollectionProxy>
+    *Alex Ghiculescu*, *Matt Larraz*
 
-    user.strict_loading!(true)           # => true
-    user.bookmarks                       # => ActiveRecord::StrictLoadingViolationError
+*   Fix PG.connect keyword arguments deprecation warning on ruby 2.7
 
-    user.strict_loading!(false)          # => false
-    user.bookmarks                       # => #<ActiveRecord::Associations::CollectionProxy>
-    user.articles.strict_loading!(false) # => #<ActiveRecord::Associations::CollectionProxy>
-    ```
+    Fixes #44307.
 
-    *Ayrton De Craene*
+    *Nikita Vasilevsky*
 
-*   Add `FinderMethods#sole` and `#find_sole_by` to find and assert the
-    presence of exactly one record.
+*   Fix dropping DB connections after serialization failures and deadlocks.
 
-    Used when you need a single row, but also want to assert that there aren't
-    multiple rows matching the condition; especially for when database
-    constraints aren't enough or are impractical.
+    Prior to 6.1.4, serialization failures and deadlocks caused rollbacks to be
+    issued for both real transactions and savepoints. This breaks MySQL which
+    disallows rollbacks of savepoints following a deadlock.
 
-    ```ruby
-    Product.where(["price = %?", price]).sole
-    # => ActiveRecord::RecordNotFound      (if no Product with given price)
-    # => #<Product ...>                    (if one Product with given price)
-    # => ActiveRecord::SoleRecordExceeded  (if more than one Product with given price)
+    6.1.4 removed these rollbacks, for both transactions and savepoints, causing
+    the DB connection to be left in an unknown state and thus discarded.
 
-    user.api_keys.find_sole_by(key: key)
-    # as above
-    ```
+    These rollbacks are now restored, except for savepoints on MySQL.
 
-    *Asherah Connor*
+    *Thomas Morgan*
 
-*   Makes `ActiveRecord::AttributeMethods::Query` respect the getter overrides defined in the model.
+*   Make `ActiveRecord::ConnectionPool` Fiber-safe
 
-    Before:
+    When `ActiveSupport::IsolatedExecutionState.isolation_level` is set to `:fiber`,
+    the connection pool now supports multiple Fibers from the same Thread checking
+    out connections from the pool.
+
+    *Alex Matchneer*
+
+*   Add `update_attribute!` to `ActiveRecord::Persistence`
+
+    Similar to `update_attribute`, but raises `ActiveRecord::RecordNotSaved` when a `before_*` callback throws `:abort`.
 
     ```ruby
-    class User
-      def admin
-        false # Overriding the getter to always return false
+    class Topic < ActiveRecord::Base
+      before_save :check_title
+
+      def check_title
+        throw(:abort) if title == "abort"
       end
     end
 
-    user = User.first
-    user.update(admin: true)
-
-    user.admin # false (as expected, due to the getter overwrite)
-    user.admin? # true (not expected, returned the DB column value)
+    topic = Topic.create(title: "Test Title")
+    # #=> #<Topic title: "Test Title">
+    topic.update_attribute!(:title, "Another Title")
+    # #=> #<Topic title: "Another Title">
+    topic.update_attribute!(:title, "abort")
+    # raises ActiveRecord::RecordNotSaved
     ```
 
-    After this commit, `user.admin?` above returns false, as expected.
+    *Drew Tempelmeyer*
 
-    Fixes #40771.
+*   Avoid loading every record in `ActiveRecord::Relation#pretty_print`
 
-    *Felipe*
+    ```ruby
+    # Before
+    pp Foo.all # Loads the whole table.
+
+    # After
+    pp Foo.all # Shows 10 items and an ellipsis.
+    ```
 
-*   Allow delegated_type to be specified primary_key and foreign_key.
+    *Ulysse Buonomo*
 
-    Since delegated_type assumes that the foreign_key ends with `_id`,
-    `singular_id` defined by it does not work when the foreign_key does
-    not end with `id`. This change fixes it by taking into account
-    `primary_key` and `foreign_key` in the options.
+*   Change `QueryMethods#in_order_of` to drop records not listed in values.
 
-    *Ryota Egusa*
+    `in_order_of` now filters down to the values provided, to match the behavior of the `Enumerable` version.
 
-*   Expose an `invert_where` method that will invert all scope conditions.
+    *Kevin Newton*
 
-    ```ruby
-    class User
-      scope :active, -> { where(accepted: true, locked: false) }
-    end
+*   Allow named expression indexes to be revertible.
 
-    User.active
-    # ... WHERE `accepted` = 1 AND `locked` = 0
+    Previously, the following code would raise an error in a reversible migration executed while rolling back, due to the index name not being used in the index removal.
 
-    User.active.invert_where
-    # ... WHERE NOT (`accepted` = 1 AND `locked` = 0)
+    ```ruby
+    add_index(:settings, "(data->'property')", using: :gin, name: :index_settings_data_property)
     ```
 
-    *Kevin Deisz*
+    Fixes #43331.
+
+    *Oliver Günther*
 
-*   Restore possibility of passing `false` to :polymorphic option of `belongs_to`.
+*   Fix incorrect argument in PostgreSQL structure dump tasks.
 
-    Previously, passing `false` would trigger the option validation logic
-    to throw an error saying :polymorphic would not be a valid option.
+    Updating the `--no-comment` argument added in Rails 7 to the correct `--no-comments` argument.
 
-    *glaszig*
+    *Alex Dent*
 
-*   Remove deprecated `database` kwarg from `connected_to`.
+*   Fix migration compatibility to create SQLite references/belongs_to column as integer when migration version is 6.0.
 
-    *Eileen M. Uchitelle*, *John Crepezzi*
+    Reference/belongs_to in migrations with version 6.0 were creating columns as
+    bigint instead of integer for the SQLite Adapter.
 
-*   Allow adding nonnamed expression indexes to be revertible.
+    *Marcelo Lauxen*
 
-    Previously, the following code would raise an error, when executed while rolling back,
-    and the index name should be specified explicitly. Now, the index name is inferred
-    automatically.
+*   Fix `QueryMethods#in_order_of` to handle empty order list.
 
     ```ruby
-    add_index(:items, "to_tsvector('english', description)")
+    Post.in_order_of(:id, []).to_a
     ```
 
-    Fixes #40732.
-
-    *fatkodima*
-
-*   Only warn about negative enums if a positive form that would cause conflicts exists.
-
-    Fixes #39065.
-
-    *Alex Ghiculescu*
+    Also more explicitly set the column as secondary order, so that any other
+    value is still ordered.
 
-*   Add option to run `default_scope` on all queries.
+    *Jean Boussier*
 
-    Previously, a `default_scope` would only run on select or insert queries. In some cases, like non-Rails tenant sharding solutions, it may be desirable to run `default_scope` on all queries in order to ensure queries are including a foreign key for the shard (i.e. `blog_id`).
+*   Fix quoting of column aliases generated by calculation methods.
 
-    Now applications can add an option to run on all queries including select, insert, delete, and update by adding an `all_queries` option to the default scope definition.
+    Since the alias is derived from the table name, we can't assume the result
+    is a valid identifier.
 
     ```ruby
-    class Article < ApplicationRecord
-      default_scope -> { where(blog_id: Current.blog.id) }, all_queries: true
+    class Test < ActiveRecord::Base
+      self.table_name = '1abc'
     end
+    Test.group(:id).count
+    # syntax error at or near "1" (ActiveRecord::StatementInvalid)
+    # LINE 1: SELECT COUNT(*) AS count_all, "1abc"."id" AS 1abc_id FROM "1...
     ```
 
-    *Eileen M. Uchitelle*
+    *Jean Boussier*
 
-*   Add `where.associated` to check for the presence of an association.
+*   Add `authenticate_by` when using `has_secure_password`.
 
-    ```ruby
-    # Before:
-    account.users.joins(:contact).where.not(contact_id: nil)
+    `authenticate_by` is intended to replace code like the following, which
+    returns early when a user with a matching email is not found:
 
-    # After:
-    account.users.where.associated(:contact)
+    ```ruby
+    User.find_by(email: "...")&.authenticate("...")
     ```
 
-    Also mirrors `where.missing`.
+    Such code is vulnerable to timing-based enumeration attacks, wherein an
+    attacker can determine if a user account with a given email exists. After
+    confirming that an account exists, the attacker can try passwords associated
+    with that email address from other leaked databases, in case the user
+    re-used a password across multiple sites (a common practice). Additionally,
+    knowing an account email address allows the attacker to attempt a targeted
+    phishing ("spear phishing") attack.
 
-    *Kasper Timm Hansen*
+    `authenticate_by` addresses the vulnerability by taking the same amount of
+    time regardless of whether a user with a matching email is found:
 
-*   Allow constructors (`build_association` and `create_association`) on
-    `has_one :through` associations.
+    ```ruby
+    User.authenticate_by(email: "...", password: "...")
+    ```
 
-    *Santiago Perez Perret*
+    *Jonathan Hefner*
 
 
-Please check [6-1-stable](https://github.com/rails/rails/blob/6-1-stable/activerecord/CHANGELOG.md) for previous changes.
+Please check [7-0-stable](https://github.com/rails/rails/blob/7-0-stable/activerecord/CHANGELOG.md) for previous changes.