`DISTINCT` clause is applied inside aggregate function by
`operation_over_aggregate_column` if needed. Unneeded outside aggregate
function.

```ruby
  # Before
  author.unique_categorized_posts.count
  # => SELECT DISTINCT COUNT(DISTINCT "posts"."id") FROM "posts" INNER JOIN "categorizations" ON "posts"."id" = "categorizations"."post_id" WHERE "categorizations"."author_id" = ?  [["author_id", 2]]

  # After
  author.unique_categorized_posts.count
  # => SELECT COUNT(DISTINCT "posts"."id") FROM "posts" INNER JOIN "categorizations" ON "posts"."id" = "categorizations"."post_id" WHERE "categorizations"."author_id" = ?  [["author_id", 2]]
```

Closes #27615

In ca324a0d the flash middleware was effectively removed by its
constructor returning the app it was passed and the `commit_flash`
call was moved to the `ActionController::Metal#dispatch` method.
This broke any redirect routes that modified the flash because the
redirect happens before `dispatch` gets called.

To fix it, this commit adds a `commit_flash` call in the `serve`
method of `ActionDispatch::Routing::Redirect`.

Fixes #27992.

Improve docs for Rails::AppGenerator. [ci skip]

Add example so its easier to understand how one can overide
an app generator.

[close #24435] Send user_supplied_options to server

Closes #27429

In Ruby 2.4, BigDecimal(), as used by the Decimal cast, was changed so
that it will raise ArgumentError when passed an invalid string, in order
to be more consistent with Integer(), Float(), etc. The other numeric
types use ex. to_i and to_f.

Unfortunately, we can't simply change BigDecimal() to to_d. String#to_d
raises errors like BigDecimal(), unlike all the other to_* methods (this
should probably be filed as a ruby bug).

Instead, this simulates the existing behaviour and the behaviour of the
other to_* methods by finding a numeric string at the start of the
passed in value, and parsing that using BigDecimal().

See also
https://bugs.ruby-lang.org/issues/10286
https://github.com/ruby/bigdecimal/commit/3081a627cebdc1fc119425c7a9f009dbb6bec8e8

Soft-deprecate the `HashWithIndifferentAccess` constant

Allow ActiveSupport::MarshalWithAutoloading#load to take a Proc

Add missing gzip footer check in ActiveSupport::Gzip.decompress

This ensures that if we try to hard-deprecate it again in the future,
we won't break these behaviors.

Since using a `ActiveSupport::Deprecation::DeprecatedConstantProxy`
would prevent people from inheriting this class and extending it
from the `ActiveSupport::HashWithIndifferentAccess` one would break
the ancestors chain, that's the best option we have here.

A gzip file has a checksum and length for the decompressed data in its
footer which isn't checked by just calling Zlib::GzipReader#read.
Calling Zlib::GzipReader#close must be called after reading to the end
of the file causes this check to be done, which is done by
Zlib::GzipReader.wrap after its block is called.

Change `SystemTestCase.driven_by` to use `setup`/`teardown` hooks

Previously, `driven_by` would change the Capybara configuration when
the test case is loaded, and having multiple test classes with different
`driven_by` configs would fail as the last loaded would be effective.

Currently when Puma gets a `:Port` it doesn't know if it is Rails' default port or if it is one that is specified by a user. Because of this it assumes that the port passed in is always a user defined port and therefor 3000 always "wins" even if you specify `port` inside of the `config/puma.rb` file when booting your server with `rails s`.

The fix is to record the options that are explicitly passed in from the user and pass those to the Puma server (or all servers really). Puma then has enough information to know when `:Port` is the default and when it is user defined. I went ahead and did this for all values rails server exposes as server side options for completeness.

The hardest thing was converting the input say `-p` or `--port` into the appropriate "name", in this case `Port`. There may be a more straightforward way to do this with Thor, but I'm not an expert here.

Move logic for parsing user options to method

Better variable name for iteration

Explicitly test `--port` user input

✂️

Update array if environment variables are used

does not show hidden namespaces in generator's help

Merge pull request #28062 from y-yagi/make_adding_gemfile_entry_work_even_if_specify_only_the_plugin_name

Make adding gemfile entry work even if specify only the plugin name

Improve docs for ActiveRecord::Result. [ci skip]

Remove styling from `true` and `false`.

Whether the command was executed within the rails application is checked
by whether or not the application's path matches `app_path`.
https://github.com/rails/rails/blob/5-0-stable/railties/lib/rails/generators/rails/plugin/plugin_generator.rb#L439..L441

Therefore, if only plugin name is specified in `app_path`, addition to
Gemfile is not done.
However, in the rails guide an example of specifying only plugin name
is given, and it is considered that there are many cases where only
plugin name is specified.
For that reason, made it work even if only plugin name was specified.

Make version and help short-cut alias to work

Enable Active Model testing for JRuby

Make HWIA#compact not return nil when no nils

[ci skip]

Merge pull request #25285 from kamipo/fix_remove_reference_to_multiple_foreign_keys_in_the_same_table

Fix `remove_reference` to multiple foreign keys in the same table

Deprecate using `#quoted_id` in quoting / type casting

Fix w3c_validator.rb validation script