Fixed performance regression introduced MySQL 8.0

Clear query cache when insert all is used

Call `while_preventing_writes` from `connected_to`

Improve MySQL error detection

Preserve user supplied joins order as much as possible

[ci skip]

This was incorrectly added at 06401bb3.

Allow separate database env variables per-connection

Add a warning for enum elements with 'not_' prefix.

Make currency symbols optional for money column type in PostgreSQL

ref https://github.com/rails/rails/pull/36621#discussion_r301208961

When SQLite connects it will silently create a database if the database does not
exist. This behaviour causes different issues because of inconsistent behaviour
between adapters: #36383, #32914. This commit adds a `database_exists?` method
as a way to check the database without creating it. This is a stepping stone to
fully resolving the above issues.

PostgreSQL: Fix GROUP BY with ORDER BY virtual count attribute

Add support for multiple databases to `rails db:abort_if_pending_migrations`

* Make ActiveRecord `ConnectionPool.connections` thread-safe.

ConnectionPool documentation is clear on the need to synchronize
access to @connections but also states that public methods do not
require synchronization. Existing code exposed @connections
directly via attr_reader. The fix uses synchronize() to lock
@connections then returns a copy to the caller using Array.dup().

Includes comments on the connections method that thread-safe access
to the connections array does not imply thread-safety of accessing
methods on the actual connections.

Adds a test-case that modifies the pool using a supported method
in one thread  while a second thread accesses pool.connections.
The test fails without this patch.

Fixes #36465.

* Update activerecord/test/cases/connection_pool_test.rb

[jeffdoering + Rafael Mendonça França]

Fix sqlite3 collation parsing when using decimal columns.

Fix invalid schema dump when primary key column has a comment

Fix table comment also being applied to the primary key column

Fix merging left_joins to maintain its own `join_type` context

`where.not` with polymorphic association is partly fixed incidentally at
213796fb (refer #33493, #26207, #17010, #16983, #14161), and I've added
test case e9ba12f7 to avoid lose that fix accidentally in the future.

In Rails 5.2, `where.not(polymorphic: object)` works as expected as
NAND, but `where.not(polymorphic_type: object.class.polymorphic_name,
polymorphic_id: object.id)` still unexpectedly works as NOR.

To will make `where.not` working desiredly as NAND in Rails 6.1, this
deprecates `where.not` working as NOR. If people want to continue NOR
conditions, we'd encourage to them to `where.not` each conditions
manually.

```ruby
all = [treasures(:diamond), treasures(:sapphire), cars(:honda), treasures(:sapphire)]
assert_equal all, PriceEstimate.all.map(&:estimate_of)
```

In Rails 6.0:

```ruby
sapphire = treasures(:sapphire)

nor = all.reject { |e|
  e.estimate_of_type == sapphire.class.polymorphic_name
}.reject { |e|
  e.estimate_of_id == sapphire.id
}
assert_equal [cars(:honda)], nor

without_sapphire = PriceEstimate.where.not(
  estimate_of_type: sapphire.class.polymorphic_name, estimate_of_id: sapphire.id
)
assert_equal nor, without_sapphire.map(&:estimate_of)
```

In Rails 6.1:

```ruby
sapphire = treasures(:sapphire)

nand = all - [sapphire]
assert_equal [treasures(:diamond), cars(:honda)], nand

without_sapphire = PriceEstimate.where.not(
  estimate_of_type: sapphire.class.polymorphic_name, estimate_of_id: sapphire.id
)
assert_equal nand, without_sapphire.map(&:estimate_of)
```

Resolves #31209.

Cache versioning enables the same cache key to be reused when the object
being cached changes by moving the volatile part of the cache key out of
the cache key and into a version that is embedded in the cache entry.

This is already occurring when the object being cached is an
`ActiveRecord::Base`, but when caching an `ActiveRecord::Relation`
we are currently still putting the volatile information (max updated at
and count) as part of the cache key.

This PR moves the volatile part of the relations `cache_key` into the
`cache_version` to support recycling cache keys for
`ActiveRecord::Relation`s.

Currently the rollback only restores primary key value, `new_record?`,
`destroyed?`, and `frozen?`. Since the `save` clears current dirty
attribute states, retrying save after rollback will causes no change
saved if partial writes is enabled (by default).

This makes `remember_transaction_record_state` remembers original values
then restores dirty attribute states after rollback.

Fixes #15018.
Fixes #30167.
Fixes #33868.
Fixes #33443.
Closes #33444.
Closes #34504.

[ci skip]

We can revert migrations using `change_column_comment` or
`change_table_comment` at current master.
However, results are not what we expect: comments are remained in new
status.
This change tells previous comment to these methods in a way like
`change_column_default`.

Regardless of a record isn't saved (e.g. validation is failed),
`after_commit` / `after_rollback` callbacks are invoked for now.

To fix the issue, this adds a record to the current transaction only
when a record is actually saved.

Fixes #29747.
Closes #29833.

When I've added new `:size` option in #35071, I've found that invalid
`:limit` and `:precision` raises `ActiveRecordError` unlike other
invalid options.

I think that is hard to distinguish argument errors and statement
invalid errors since the `StatementInvalid` is a subclass of the
`ActiveRecordError`.

https://github.com/rails/rails/blob/c9e4c848eeeb8999b778fa1ae52185ca5537fffe/activerecord/lib/active_record/errors.rb#L103

```ruby
begin
  # execute any migration
rescue ActiveRecord::StatementInvalid
  # statement invalid
rescue ActiveRecord::ActiveRecordError, ArgumentError
  # `ActiveRecordError` except `StatementInvalid` is maybe an argument error
end
```

I'd say this is the inconsistency worth fixing.

Before:

```ruby
add_column :items, :attr1, :binary,   size: 10      # => ArgumentError
add_column :items, :attr2, :decimal,  scale: 10     # => ArgumentError
add_column :items, :attr3, :integer,  limit: 10     # => ActiveRecordError
add_column :items, :attr4, :datetime, precision: 10 # => ActiveRecordError
```

After:

```ruby
add_column :items, :attr1, :binary,   size: 10      # => ArgumentError
add_column :items, :attr2, :decimal,  scale: 10     # => ArgumentError
add_column :items, :attr3, :integer,  limit: 10     # => ArgumentError
add_column :items, :attr4, :datetime, precision: 10 # => ArgumentError
```

Follow-up of 5c71000d, #29834, and #30271.

Currently, preloading and eager loading are not to be affected by
scoping, with the exception of `unscoped`.

But non eager loaded association access is still affected by scoping.

Although this is a breaking change, the association loading will work
consistently whether preloaded / eager loaded or not.

Before:

```ruby
Post.where("1=0").scoping do
  Comment.find(1).post                   # => nil
  Comment.preload(:post).find(1).post    # => #<Post id: 1, ...>
  Comment.eager_load(:post).find(1).post # => #<Post id: 1, ...>
end
```

After:

```ruby
Post.where("1=0").scoping do
  Comment.find(1).post                   # => #<Post id: 1, ...>
  Comment.preload(:post).find(1).post    # => #<Post id: 1, ...>
  Comment.eager_load(:post).find(1).post # => #<Post id: 1, ...>
end
```

Fixes #34638.
Fixes #35398.

* s/Postgres/PostgreSQL/
* s/MYSQL/MySQL/, s/Mysql/MySQL/
* s/Sqlite/SQLite/

Replaced all newly added them after 6089b314.

Add `after_save_commit` callback as shortcut for `after_commit :hook, on: [ :create, :update ]`. (#35804)