- When relations return no result or 0 result then cache_key should
  handle it gracefully instead of blowing up trying to access
  `result[:size]` and `result[:timestamp]`.
- Fixes #23063.

Fix ActiveRecord::Relation#cache_key for loaded empty collection

Fix `ActionController::Parameters#==` bug

See bug #21032.

add application_mailer.rb to template of mountable engine

[ci skip] fix #23157

This is a similar case to wanting ot use bind params for limit and
offset. Right now passing a range grows the amount of prepared
statements in an unbounded fashion. We could avoid using prepared
statements in that case, similar to what we do with arrays, but there's
a known number of variants for ranges.

This ends up duplicating some of the logic from Arel for how to handle
potentially infinite ranges, and that behavior may be removed from Arel
in the future.

Fixes #23074

Fix warning introduced in API response fix

- Fix warning introduced in https://github.com/vipulnsward/rails/commit/77acc004efad07dfd4d4f83be14ef897968a3fd9 when fixing API  responses.
- render :text is deprecated, so added an assertion for it.

updating secrets.yml template to use rails command instead of rake

Improve Docs for ActionController::MimeResponds#respond_to

Re-add ActionController::ApiRendering

insert newlines into code samples in Upgrading section of Asset Pipeline doc

[ci skip]

- Fixes bug #23142.
- Bug was occurring only with ActionController::API, because `_process_options` wasn't being run for API requests, even though it was being run for normal app requests.

Fix marking of custom routes for Journey

Response etags to always be weak: Prefixed 'W/' to value returned by Act...

Small doc update per RF [ci skip]

Add CHANGELOG.md entry for #22950

[ci skip]

The Mapper build_path method marks routes where path parameters are part
of a path segment as custom routes by altering the regular expression, e.g:

    get '/foo-:bar', to: 'foo#bar'

There were some edge cases where certain constructs weren't being picked
up and this commit fixes those.

Fixes #23069.

Fix the API documentation layout of after_*_commit

Just noticed this on the [edge API].

[ci skip]

[edge API]: http://edgeapi.rubyonrails.org/classes/ActiveRecord/Transactions/ClassMethods.html

This reverts commit 5d1b7c3b.

The change here didn't actually fix the issue it was trying to fix, and
this isn't the correct way to fix either issue. The problem is switching
from the builder to grouping with find_all/regex is now very dependent
on how you structure your path pattern.

Changed html_safe to raw in AV

html_safe is not supposed to be public API for AV. This change removes usage of html_safe in favour of raw() in AV helpers.
Also changed usage of html_safe to make use of raw() instead so that the intended behaviour is verified with raw()

Pluralize counter_cache column name in example [ci skip]

Adapterize storage for ActionCable

Update CHANEGLOG for https://github.com/rails/rails/pull/20005 [ci skip]

Response etags to always be weak: Prefixed W/ to value returned by ActionDispatch::Http::Cache::Response#etag= such that etags set in fresh_when and stale? are weak. For #17556.

Revert "Its ideal to set Vary: Accept-Encoding, irrespective of whether gzipped version exists or not. This is helpful for CDN's to later distinguish assets, based on previous, current copies and introduced gzip version if any."

This reverts commit 067c52f6.

Conversation: https://github.com/rails/rails/pull/23120#issuecomment-173007011

We want this method to be the single canonical source of information
about type metadata related to a model. This is the method I've been
continuously recommending people use if they need this sort of access,
as I have no plans to remove or change it at any point in the future.

We can do ourselves a favor and get people to use this instead of
relying on some other part of the internals that they shouldn't be by
making this method public.

Its ideal to set Vary: Accept-Encoding, irrespective of whether gzipped or not