Remove unused line

Add more documents for AR connection_adapters abstract schema_definitions. [ci skip]

- Add example to column_exists?
- Add example to index_exists?
- Add document for foreign_key
- Add document for foreign_key_exists?

Work around for upstream Ruby bug #10685

Partition routes during setup.

[ci skip] Add documentation for Helpers#all_helpers_from_path

This reverts commit a38732c8.

Since the mutation logic was reverted in
07278519, we must bring the reader
method back as well, since the implementation relies on it.

Added testcase for #18742 

This reverts commit 393e65b4 and
ec51c3fe

We don't want the records to hold hard references to transactions
because they point at records that have callbacks.

this patch makes errors slightly more expensive when someone is missing
a route key, but in exchange it drops 4 allocations per `url_for` call.
Since missing a route key is an error, optimizing for the non-error path
seems like a good trade off

we already know the length of the args, so we can use that length for
parallel iteration and cut down on allocations for `url_for` calls.

this centralizes the logic for determining the script name key and drops
object allocations when calling `engine_script_name` (which is called on
each `url_for`).

Remove unneeded comment

this lets us keep singleton instances of "state" values and precalculate
return values of things like `finalized?` and `completed?`.

this way we don't have to mutate a state object, we can just change the
state of the txn

Improve the Rubinius build

We've replaced most querues using DROP TABLE in our tests already.
This patch replaces the last couple calls.

Preserve Array#take(n) behaviour of HasManyAssociation

Move `UnknownAttributeError` to a more sane namespace

[ci skip] Fix a typo for PostgreSQL text limit, GB instead of Gb.

At the moment, `rescue_from` doesn't work with strings on Rubinius
because of rubinius/rubinius#3272.

Rubinius is still targeting compliance with Ruby 2.1.0 but we are pretty
close to a green build so let's avoid this check for the moment.

We only set the state on the record if that condition is `false` in the
first place, so we dont need to call that again. Also that call is
expensive, follow benchmark with before and after this change:

```
Calculating -------------------------------------
          persisted?    15.272k i/100ms
	  -------------------------------------------------
          persisted?    350.119k (± 4.6%) i/s -      1.756M
```

```
Calculating -------------------------------------
          persisted?    25.988k i/100ms
	  -------------------------------------------------
          persisted?      1.294M (± 5.3%) i/s -      6.445M
```

(benchmark borrowed from 57d35b2b)

The edge version ships with a patch that uses composition over
inheritance for the Mail::PartsList object (see mikel/mail#782).
Let's test Action Mailer against it to prevent eventual regressions
and experience it.

Moreover, this branch makes the Action Mailer suite green against
Rubinius.

Merge pull request #19170 from JuanitoFatas/doc/active_record/connection_adapters/abstract/schema_definitions.rb

Clarify that t.references and t.belongs_to are interchangeable. [ci skip]

directly

calling `sync_with_transaction_state` is not fast, so if we call it
once, we can improve the performance of the `persisted?` method.  This
is important because every call to `url_for(model)` will call
`persisted?`, so we want that to be fast.

Here is the benchmark:

```ruby
require 'active_record'

ActiveRecord::Base.establish_connection adapter: "sqlite3", database: ":memory:"

ActiveRecord::Base.connection.instance_eval do
  create_table(:articles)
end

class Article < ActiveRecord::Base; end
article = Article.new.tap(&:save!)

Benchmark.ips do |x|
  x.report("persisted?") do
    article.persisted?
  end
end
```

Before this patch:

```
$ bundle exec ruby -rbenchmark/ips persisted.rb
Calculating -------------------------------------
          persisted?     3.333k i/100ms
-------------------------------------------------
          persisted?     51.037k (± 8.2%) i/s -    253.308k
```

After:

```
$ bundle exec ruby -rbenchmark/ips persisted.rb
Calculating -------------------------------------
          persisted?     7.172k i/100ms
-------------------------------------------------
          persisted?    120.730k (± 5.1%) i/s -    602.448k
```

depth is always 0, so the index will always be false.  No reason to
create the instance variable if it isn't used

Rake restart task no longer loads entire Rails environment when run

The restart task does not need access to models or other classes and
helpers from the application environment.