HTTP 302 redirects for PATCH or DELETE will not redirect with GET

A HTTP 302 Found redirect to PATCH and DELETE requests will be followed with PATCH or DELETE. Redirect responses to GET and POST will be followed with a GET. The Rails form_for helper will use a workaround to send POST requests with a _method param to avoid this issue for PATCH/DELETE.

If you make requests yourself, watch out for the following behavior.

When you make an AJAX request PATCH /foo and the /foo action redirects to /bar, browsers will request PATCH /bar. You probably expected the second requ...

High-level data types with "composed_of"

I recently stumbled upon the Rails feature composed_of. One of our applications dealt with a lot of addresses and they were implemented as 7 separate columns in the DB and Rails models. This seemed like a perfect use case to try out this feature.

TLDR

The feature is still a VERY leaky abstraction. I ran into a lot of ugly edge cases.

It also doesn't solve the question of UI. We like to use simple_form. It's currently not possible to simply write `f...

Preventing users from uploading malicious content

When you allow file uploads in your app, a user might upload content that hurts other users.

Our primary concern here is users uploading .html or .svg files that can run JavaScript and possibly hijack another user's session.

A secondary concern is that malicious users can upload executables (like an .exe or .scr file) and use your server to distribute it. However, modern operating systems usually warn before executing files that were downloaded from t...

How to update a single gem conservatively

The problem

Calling bundle update GEMNAME will update a lot more gems than you think. E.g. when you do this:

bundle update cucumber-rails

... you might think this will only update cucumber-rails. But it actually updates cucumber-rails and all of its dependencies. This will explode in your face when one of these dependencies release a new version with breaking API changes. Which is all the time.

In the example above updating cucumber-rails will give you Capybara 2.0 (because capybara is a dependency of `cucumber-rail...

Be very careful with 301 and 308 redirects

Browsers support different types of redirects.

Be very careful with these status codes:

  • 301 Moved Permanently
  • 308 Permanent Redirect

Most browsers seem to cache these redirects forever, unless you set different Cache-Control headers. If you don't have any cache control headers, you can never change them without forcing users to empty their cache.

Note

By default Rails sends a ...

How to disable logging for ActiveStorage's Disk Service routes

In development, we store files using ActiveStorage's disk service. This means that stored files are served by your Rails application, and every request to a file results in (at least!) one non-trivial log entry which can be annoying. Here is how to disable those log entries.

Example

Here is an example of what loading a single <img> in an example application writes to the Rails log.

Started GET "/rails/active_storage/blobs/redirect/..." for ::1 at ...
Processing by ActiveStorage::Blobs::RedirectController#show as SVG
  Parameter...

Popular mistakes when using nested forms

Here are some popular mistakes when using nested forms:

  • You are using fields_for instead of form.fields_for.
  • You forgot to use accepts_nested_attributes in the containing model. Rails won't complain, but nothing will work. In particular, nested_form.object will be nil.
  • The :reject_if option lambda in your accepts_nested_attributes call is defined incorrectly. Raise the attributes hash given to your :reject_if lambda to see if it looks like you expect.
  • If you are nesting forms into nested forms, each model involved ne...

Compare library versions as "Gem::Version" instances, not as strings

Sometimes we have to write code that behaves differently based on the version of a specific gem or the Ruby Version itself. The version comparison can often be seen with simple string comparison like so.

# ❌ Not recommended
if Rails.version > '6.1.7.8' || RUBY_VERSION > '3.1.4'
  raise Error, 'please check if the monkey patch below is still needed'
end

If you are lucky, the version comparison above works by coincidence. But chances are that you are not: For example, Rails version 6.1.10.8 would not raise an error in the code ...

Testing ActiveRecord validations with RSpec

Validations should be covered by a model's spec.

This card shows how to test an individual validation. This is preferrable to save an entire record and see whether it is invalid.

Recipe for testing any validation

In general any validation test for an attribute :attribute_being_tested looks like this:

  1. Make a model instance (named record below)
  2. Run validations by saying record.validate
  3. Check if record.errors[:attribute_being_tested] contains the expected validation error
  4. Put the attribute into a valid state
  5. Run...

ActiveRecord: String and text fields should always validate their length

If you have a :string or :text field, you should pair it with a model validation that restricts its length.

There are two motivations for this:

  • In modern Rails, database types :string and :text no longer have a relevant size limit. Without a validation a malicious user can quickly exhaust the hard drive of your database server.
  • In legacy Rails (or database schemas migrated from legacy Rails), database types :string and :text had a database-side length constraint. When the user enters a longer string, the ActiveRecord valida...

Migrating from Elasticsearch to Opensearch: searchkick instructions (without downtime!)

General

A general overview about why and how we migrate can be found under Migrating from Elasticsearch to Opensearch

This card deals with specifics concerning the use of searchkick.

Step 1: Make Opensearch available for Searchkick

In your Gemfile

# Search
gem 'searchkick'                   # needs to be > 5, to use Opensearch 2
gem 'elasticsearch'
gem 'opensearch-ruby'

in config/initializers/searchkick.rb (or wherever you have configured your Searchkick settings) add:

SEARCHKICK_CLIENT_T...

CarrierWave: Default Configuration and Suggested Changes

CarrierWave comes with a set of default configuration options which make sense in most cases. However, you should review these defaults and adjust for your project wherever necessary.

You will also find suggestions on what to change below.

Understanding the default configuration

Here is the current default config for version 2:

config.permissions = 0644
config.directory_permissions = 0755
config.storage_engines = {
  :f...

Different ways to set attributes in ActiveRecord

Rails 5 / 6 / 7

Method Uses Default Accessor Saves to Database Runs Validations Runs Callbacks Updates updated_at/updated_on Respects Readonly
attribute= Yes No n/a n/a n/a n/a
attributes= Yes No ...

What we know about PDFKit

What PDFKit is

  • PDFKit converts a web page to a PDF document. It uses a Webkit engine under the hood.
  • For you as a web developer this means you can keep using the technology you are familar with and don't need to learn LaTeX. All you need is a pretty print-stylesheet.

How to use it from your Rails application

  • You can have PDFKit render a website by simply calling PDFKit.new('http://google.com').to_file('google.pdf'). You can then send the...

Beware of params with non-string values (nil, array, hash)

Recent rails security updates have shown that people make incorrect assumptions about the possible contents of the params hash.

Just don't make any! Treat it as what it is: potentially unsafe user input. For example:

/pages/edit?foo=   --> params == {:foo => ""}
/pages/edit?foo    --> params == {:foo => nil}
/pages/edit?foo[]  --> params == {:foo => [nil]} # at least in older rails 3 and in rails 2.x

Be especially wary about stuff like

User.find_by_password_reset_token(params[:password_reset_token])

I...

`simple_format` does not escape HTML tags

simple_format ignores Rails' XSS protection. Even when called with an unsafe string, HTML characters will not be escaped or stripped!

Instead simple_format calls sanitize on each of the generated paragraphs.

ActionView::Base.sanitized_allowed_tags
# => #<Set: {"small", "dfn", "sup", "sub", "pre", "blockquote", "ins", "ul", "var", "samp", "del", "h6", "h5", "h4", "h3", "h2", "h1", "span", "br", "hr", "em", "address", "img", "kbd", "tt", "a", "acrony...

Fixing flaky E2E tests

An end-to-end test (E2E test) is a script that remote-controls a web browser with tools like Selenium WebDriver. This card shows basic techniques for fixing a flaky E2E test suite that sometimes passes and sometimes fails.

Although many examples in this card use Ruby, Cucumber and Selenium, the techniques are applicable to all languages and testing tools.

Why tests are flaky

Your tests probably look like this:

When I click on A
And I click on B
And I click on C
Then I should see effects of C

A test like this works fine...

The many gotchas of Ruby class variables

TLDR: Ruby class variables (@@foo) are dangerous in many ways. You should avoid them at all cost. See bottom of this card for alternatives.

Class variables are shared between a class hierarchy

When you declare a class variable, it is shared between this and all descending (inheriting) classes. This is rarely what you want.

Class variables are bound at compile-time

Like unqualified constants, class variables are bound to your current scope *whe...

RSpec: Applying stubs only within a block

When you mocked method calls in RSpec, they are mocked until the end of a spec, or until you explicitly release them.

You can use RSpec::Mocks.with_temporary_scope to have all mocks applied inside a block to be released when the block ends.
Example:

RSpec::Mocks.with_temporary_scope do
  allow(Rails).to receive(:env).and_return('production'.inquiry)
  puts Rails.env # prints "production"
end
puts Rails.env # prints "test"

Note that, when overriding pre-existing mocks inside the block, they are not reverted to the previously ...

Careful: `fresh_when last_modified: ...` without an object does not generate an E-Tag

To allow HTTP 304 responses, Rails offers the fresh_when method for controllers.

The most common way is to pass an ActiveRecord instance or scope, and fresh_when will set fitting E-Tag and Last-Modified headers for you. For scopes, an extra query is sent to the database.

fresh_when @users

If you do not want that magic to happen, e.g. because your scope is expens...

Preloaded associations are filtered by conditions on the same table

When you eagerly load an association list using the .include option, and at the same time have a .where on an included table, two things happen:

  1. Rails tries to load all involved records in a huge single query spanning multiple database tables.
  2. The preloaded association list is filtered by the where condition, even though you only wanted to use the where condition to filter the containing model.

The second case's behavior is mostly unexpected, because pre-loaded associations usually don't care about the circumstances under whi...

Project management best practices: Technical debt summary

Maintaining larger projects makes it more difficult to balance refactoring and upgrade tasks according to its actual value. Consider to create and periodically maintain a summary, which helps you and your team in the decision which refactoring task should be taken next.

Template

Here is an template on how you might categorize your tasks:

| Technical debt | Estimated Efforts | Visible customer value| Customer value explained| Developer value|Developer value explained|
|-----------------------------|----------------|----------...

Eager-loading polymorphic associations

To avoid n+1 queries, you want to eager-load associated records if you know you need to access them later on.

The Rails docs say:

Eager loading is supported with polymorphic associations.

This is true, but has some caveats.

Example

Consider the following models:

class Image < ActiveRecord::Base; end
class Video < ActiveRecord::Base; end
class PageVersion < ActiveRecord::Base
  belongs_to :primary_medium, polymorphic: true # may be Image or Video
end
class Page < ActiveRecord::Base
  belongs_to ...

Writing a README for a project

Rails applications and ruby gems should have a README that gives the reader a quick overview of the project. Its size will vary as projects differ in complexity, but there should always be some introductory prose for a developer to read when starting on it.

Purpose

That's already the main purpose of a project README: Give a new developer a quick overview of the project. In sketching this outline, the README should notify the reader of any peculiarity he needs to know of.

Remember that in a few months, you'll be a kind of "new ...