This stems from [a comment](rails#17227 (comment)) by @dhh.
In summary:

* New Rails 5.0 apps will not accept `return false` as a way to halt callback chains, and will not display a deprecation warning.
* Existing apps ported to Rails 5.0 will still accept `return false` as a way to halt callback chains, albeit with a deprecation warning.

For this purpose, this commit introduces a Rails configuration option:

```ruby
config.active_support.halt_callback_chains_on_return_false
```

For new Rails 5.0 apps, this option will be set to `false` by a new initializer
`config/initializers/callback_terminator.rb`:

```ruby
Rails.application.config.active_support.halt_callback_chains_on_return_false = false
```

For existing apps ported to Rails 5.0, the initializers above will not exist.
Even running `rake rails:update` will not create this initializer.

Since the default value of `halt_callback_chains_on_return_false` is set to
`true`, these apps will still accept `return true` as a way to halt callback
chains, displaying a deprecation warning.

Developers will be able to switch to the new behavior (and stop the warning)
by manually adding the line above to their `config/application.rb`.

A gist with the suggested release notes to add to Rails 5.0 after this
commit is available at https://gist.github.com/claudiob/614c59409fb7d11f2931

After this commit, returning `false` in a callback will display a deprecation
warning to make developers aware of the fact that they need to explicitly
`throw(:abort)` if their intention is to halt a callback chain.

This commit also patches two internal uses of AS::Callbacks (inside
ActiveRecord and ActionDispatch) which sometimes return `false` but whose
returned value is not meaningful for the purpose of execution.

In both cases, the returned value is set to `true`, which does not affect the
execution of the callbacks but prevents unrequested deprecation warnings from
showing up.

This commit changes arguments and default value of CallbackChain's :terminator
option.

After this commit, Chains of callbacks defined **without** an explicit
`:terminator` option will be halted as soon as a `before_` callback throws
`:abort`.

Chains of callbacks defined **with** a `:terminator` option will maintain their
existing behavior of halting as soon as a `before_` callback matches the
terminator's expectation. For instance, ActiveModel's callbacks will still
halt the chain when a `before_` callback returns `false`.

`MissingSourceFile` was just an alias to `LoadError` and was not
being raised inside the framework.

[ci skip]

As confirmed by @lleger (the author of `verified`) [in this comment](https://github.com/rails/rails/pull/17727#issuecomment-65488743):

> Actually, it no longer returns false explicitly (bc8cc56a), so I guess the CHANGELOG isn't totally accurate. It returns nil instead (but the functionality isn't practically different).

This commit adds a `#verified` method to
`ActiveSupport::MessageVerifier` which will return either `false` when
it encounters an error or the message. `#verify` continues to raise an
`InvalidSignature` exception on error.

This commit also adds a convenience boolean method on `MessageVerifier`
as a way to check if a message is valid without performing the
decoding.

We will support only Ruby >= 2.1.

But right now we don't accept pull requests with syntax changes to drop
support to Ruby 1.9.

/cc @chancancode

References f92ac245.

Fixes #16956.

1. spacing issues
2. spelling correction
3. grammar correction
4. Add missing docs

Closes #16392.

Otherwise this will lead to another error later on 
when generating a signature:
TypeError (no implicit conversion of nil into String).

Goals:

1. Default to :random for newly generated applications
2. Default to :sorted for existing applications with a warning
3. Only show the warning once
4. Only show the warning if the app actually uses AS::TestCase

Fixes #16769

Fixes #8015, #9756.

[Fred Wu & Matthew Draper]

https://github.com/rails/rails/commit/98b46bf5e201307cae56ee14bf41363a539779c5
did not properly handled out-of-range `:usec`s.

Passing a `:usec` that's out of range now throws an `ArgumentError` as it
should.

Fixes #16759.

We tenderized the wrong method! Object#try already had the yield option, just needed some tenderloving instance_eval to fit the bill

For the sake of backward-compatibility, we need to make #instance_of?
return true for Fixnum. On the other hand, the method should still
give true for ActiveSupport::Duration itself which was not the case
before.

Since Duration is extending from ProxyObject which extends itself from
BasicObject, the Duration object doesn't respond to the #instance_of?
method. Thus, the #method_missing hook get triggered, delegating the
method to its `value` attribute.

However, Rubinius' #eql? definition relies on #instance_of?, thus this
will equal to true with a Fixnum (since its `value` attribute is a
Fixnum) while it should not.

The previous behavior was wrong anyway, no matter the implementation.

Fix rounding errors with #travel_to by resetting the usec on any passed time to zero, so we only travel with per-second precision, not anything deeper than that.

Fixes #16406