Taking screenshots in Capybara
Capybara-screenshot can automatically save screenshots and the HTML for failed Capybara tests in Cucumber, RSpec or Minitest.
Requires Capybara-Webkit, Selenium or poltergeist for making screenshots. Screenshots are saved into $APPLICATION_ROOT/tmp/capybara
.
Manually saving a page
Additionally you can trigger the same behavior manually from the test using Capybara::Session#save_and_open_page and [...
Testing for Performance: How to Ensure Your Web Vitals Stay Green
Frontend performance and user experience are orthogonal to feature development. If care is not taken, adding features usually degrades frontend performance over time.
Many years, frontend user experience has been hard to quantify. However, Google has developed some metrics to capture user experience on the web: the Web Vitals. The Core Web Vitals are about "perceived loading speed" (Largest Contentful Paint), reactivity (Interaction to Next Paint) and visual stability (Content Layout Shift).
I have recent...
CI Template for GitHub Actions
Usually our code lives on GitLab, therefore our documentation for CI testing is extensive in this environment. If you are tied to GitHub e.g. because your customer uses it, you may use the following GitHub Actions template for the CI integration. It includes jobs for rspec
(parallelized using knapsack, unit + feature specs), rubocop
, eslint
, coverage
and license_finder
.
Note that GitHub does not allow the use of YAML anchors and aliases. You can instead use [compos...
ActiveRecord: count vs size vs length on associations
TL;DR: You should generally use #size
to count associated records.
size
- Counts already loaded elements
- If the association is not loaded, falls back to a
COUNT
query
count
- If a counter cache is set up, returns the cached value
- Issues a
COUNT
query else
Caveats
- If you trigger a
COUNT
query for an association of an an unsaved record, Rails will try to load all children where the foreign keyIS NULL
. This is not what you want. To prevent this behavior, you can useunsaved_record.association.to_a.size
. - `c...
Code splitting in esbuild: Caveats and setup
Code splitting is a feature of esbuild that can keep huge libraries out of the main bundle.
How code splitting works
Like Webpack esbuild lets you use the await import()
function to load code on demand:
// application.js
const { fun } = await import('library.js')
fun()
However, esbuild's code splitting is disabled by default. The code above would simply inline (copy) `l...
Async control flow in JavaScript: Promises, Microtasks, async/await
Slides for Henning's talk on Sep 21st 2017.
Understanding sync vs. async control flow
Talking to synchronous (or "blocking") API
print('script start')
html = get('/foo')
print(html)
print('script end')
Script outputs 'script start'
, (long delay), '<html>...</html>'
, 'script end'
.
Talking to asynchronous (or "evented") API
print('script start')
get('foo', done: function(html) {
print(html)
})
print('script end')
Script outputs 'script start'
, 'script end'
, (long ...
HTML: Making browsers wrap long words
By default, browsers will not wrap text at syllable boundaries. Text is wrapped at word boundaries only.
This card explains some options to make browsers wrap inside a long word like "Donaudampfschifffahrt"
.
Option 1: hyphens CSS property (preferred)
Modern browsers can hyphenate natively. Use the hyphens CSS property:
hyphens: auto
There is also hyphens: none
(disable hyphenations even at ­
entities) and hyphens: manual
(hy...
How to list updateable dependencies with Bundler and Yarn
Bundler
bundle outdated [--filter-major|--filter-minor|--filter-patch]
Example output for bundle outdated --filter-major
Other examples
A useful flag is --strict
as it will only list versions that are allowed by your Gemfile requirements (e.g. does not show rails update to 6 if your Gemfile has the line gem 'rails', '~>5.2'
).
I also experienced that doing upgrades per group (test, development) are easier to do. Thus --groups
might also be helpful.
$ bundle...
Using ngrok for exposing your development server to the internet
Sometimes you need to access a dev server running on localhost from another machine that is not part of the same network. Maybe you want to use your phone to test a web page, but are only in a guest WiFi. In the past, we often used some port forwarding or other techniques to expose the service to the internet.
Enter ngrok, a command line tool that gives you an on-the-fly internet...
Timeouts for long-running SQL queries
While the main goal always is to prevent long-running queries in the first place, automatic timeouts can serve as a safety net to terminate problematic queries automatically if a set time limit is exceeded. This prevents single queries from taking up all of your databaseās resources and reduces the need for manual intervention that might destabilize or even crash the application.
As Rails does not set a timeout on database statements by default, the following query will run for an entire day:
ActiveRecord::Base.connection.execute("S...
Integrating ESLint
Introduction
To ensure a consistent code style for JavaScript code, we use ESLint. The workflow is similar to integrating rubocop for Ruby code.
1. Adding the gem to an existing code base
You can add the following lines to your package.json
under devDependencies
:
"devDependencies": {
"@eslint/js": "x",
"@stylistic/eslint-plugin": "x",
"eslint": "x",
"eslint-plugin-import": "x",
"globals": "x",
}
...
Common mistakes when storing file uploads with Rails
1. Saving files to a directory that is not shared between deploys or servers
If you save your uploads to a made up directory like "RAILS_ROOT/uploads"
, this directory goes away after every deploy (since every release gets a new). Also this directory is not shared between multiple application servers, so your uploads are randomly saved to one local filesystem or another. Fixing this afterwards is a lot of fun.
Only two folders are, by default, shared between our application servers and deployments: "RAILS_ROOT/storage"
and `"RAILS...
How to configure Selenium WebDriver to not automatically close alerts or other browser dialogs
tl;dr
We recommend configuring Selenium's unhandled prompt behavior to
{ default: 'ignore' }
with the monkey patch below.
When running tests in a real browser, we use Selenium. Each browser is controlled by a specific driver, e.g. Selenium::WebDriver::Chrome
for Chrome.
There is one quirk to all drivers (at least those following the W3C webdriver spec) that can be impractical:
When any user prompt (like an alert
) is encountered when trying to perform an action, they will [...
Testing Accessibility using Orca
Orca is a Linux screen reader. Since it is part of the GNOME project it should come preinstalled with Ubuntu installations.
Getting started
To turn on the screen reader you can either go to Settings > Accessibility and then activate Screen Reader in the "Seeing" section or you can simply type orca
in your terminal. Alternatively you can use the default keyboard shortcut super
+ alt
+ s
to toggle the screen reader.
Note
It may feel quite strange in the beginning to use a screen reader. It is constantly commenting on everyth...
Bash: How to grep logs for a pattern and expand it to the full request
Example
I, [2024-01-21T06:22:17.484221 #2698200] INFO -- : [4cdad7a4-8617-4bc9-84e9-c40364eea2e4] test
I, [2024-01-21T06:22:17.484221 #2698200] INFO -- : [4cdad7a4-8617-4bc9-84e9-c40364eea2e4] more
I, [2024-01-21T06:22:17.484221 #2698200] INFO -- : [6e047fb3-05df-4df7-808e-efa9fcd05f87] test
I, [2024-01-21T06:22:17.484221 #2698200] INFO -- : [6e047fb3-05df-4df7-808e-efa9fcd05f87] more
I, [2024-01-21T06:22:17.484221 #2698200] INFO -- : [53a240c1-489e-4936-bbeb-d6f77284cf38] nope
I, [2024-01-21T06:22:17.484221 #2698200] INFO -- ...
How to allow testing beforeunload confirmation dialogs with modern ChromeDrivers
Starting with ChromeDriver 127, if your application displays a beforeunload
confirmation dialog, ChromeDriver will immediately close it. In consequence, any automated tests which try to interact with unload prompts will fail.
This is because ChromeDriver now follows the W3C WebDriver spec which states that any unload prompts should be closed automatically.
However, this applies only to "HTTP" test sessions, i.e. what you're using by default. The spec also defines that bi-directional test se...
Best practices: Large data migrations from legacy systems
Migrating data from a legacy into a new system can be a surprisingly large undertaking. We have done this a few times. While there are significant differences from project to project, we do have a list of general suggestions.
Before you start, talk to someone who has done it before, and read the following hints:
Understand the old system
Before any technical considerations, you need to understand the old system as best as possible. If feasible, do not only look at its API, or database, or frontend, but let a user of the old system sho...
Don't assert exceptions in feature specs
As we are slowly switching from Cucumber scenarios to RSpec feature specs, you might be tempted to write assertions like this one:
feature 'authorization for cards management' do
let(:guest_user) { create(:user, :guest) }
scenario "rejects guest users from adding new cards", js: true do
sign_in guest_user
expect { visit new_cards_path }.to raise_error(Consul::Powerless)
end
end
While this might work under certain circumstances¹, there is a good chance you'll see two exceptions when running this single spec:
- ...
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...
How to use Parallel to speed up building the same html partial multiple times (for different data)
The parallel-gem is quite easy to use and can speed up rendering time if you want to render the same partial multiple times (e.g. for rendering long lists of things).
Parallel supports execution using forked processes (the default), threads (mind the GVL) and Ractors (some limitations for data sharing).
If your parallelized code talks to the database, you should [ensure not to leak database connections](https://makandracards.com/makandra/45360-using-activerecord-with-threads-will-leak-database-connect...
RSpec: How to write isolated specs with cookies
Background
Rails offers several methods to manage three types of different cookies along with a session storage for cookies. These are normal, signed and encrypted cookies.
By following the happy path of testing a web application, that is only the main use-case is tested as a integration test and the rest as isolated (more unit ...
How to upgrade Rails: Workflow advice
When upgrading Rails versions -- especially major versions -- you will run into a lot of unique issues, depending on the exact version, and depending on your app.
However, it is still possible to give some generic advice on how you want to tackle the update in principle.
If you are not really confident about upgrading Rails, have a look at Rails LTS.
How many update steps?
Besides the Rails upgrade itself, you might also want to upgrade your other gems and upgrade your Ruby version.
First decide in how many st...
RSpec: Ensuring a method is called on an object that will be created in the future
rspec >= 3.1 brings a method and_wrap_original
. It seems a bit complicated at first, but there are use cases where it helps to write precise tests. For example it allows to add expectations on objects that will only be created when your code is called.
If you have older rspec, you could use expect_any_instance_of
, but with the drawback, that you can't be sure if it really was the correct instance which got the message.
Example
The example model uses different validators based on a flag:
class MyModel < ApplicationRecord
...
Heads up: expect(object).to receive(:method_name) does not execute the original implementation of the method
Let's assume that we have a model Movie
that registers a callback function when a new instance of Movie
is created (Note: For the purpose of this card it is not important what that callback does or which type of callback it is).
This is how we test whether the callback function (here it is named :my_method
) is called when a new movie is created:
expect_any_instance_of(Movie).to receive(:my_method)
create(:movie) # <-- this is where the method :my_method should be called
You might expect that when calling `create(:mo...