Cucumber: Wait for any requests to finish before moving on to the next scenario
Background
Generally, Selenium tests use the browser to interact with the page. If it's unavailable, a timeout error is thrown.
Now, consider a scenario like this:
@javascript
Scenario: Receive an e-mail after clicking the fancy link
When I follow "fancy link"
Then I should have an e-mail with the subject "Hello"
When the last step in the scenario passes, you are done. Right? Wrong.
Why it's not enough
What if clicking our "fancy link" above sends the e-mail that we expect, but it also does stuff on the server...
Force SSH client to use password authentication instead of public key
To test if you can connect to a host using password authentication and explicitly deny public key authentication:
ssh -o PreferredAuthentications=password -o PubkeyAuthentication=no user@host
This also works for Secure Copy:
scp -o PreferredAuthentications=password -o PubkeyAuthentication=no local.file user@host:/path/to/remote.file
VNC browser disappears while typing
We often use the Then console step from spreewald in combination with geordi vnc from geordi to debug tests within a real browser. Sometimes when you type in the browser it suddenly disappears. You will only see a grey screen then.
This will always happen if you press the d key. Press the d key again and the browser will appear again.
ActionView::Template::Error (dump format error for symbol(0x6d))
I recently encountered the error above when I was running selenium tests.
Thanks to a post on stackoverflow I found out that clearing all files in tmp/cache in my project folder made the issue go away.
Testing state_machine callbacks without touching the database
You should test the callback methods and its correct invocation in two separate tests. Understand the ActiveRecord note before you move on with this note.
Say this is your Spaceship class with a transition launch and a release_docking_clamps callback:
class Spaceship
state_machine :state, :initial => :docked do
event :launch do
transition :docked => :en_route
end
before_transition :on => :launch, :do => :release_doc...
How to silence UTF-8 warnings on Rails 2.3 with Ruby 1.9
Rails 2.3.16+ on Ruby 1.9 causes warnings like this:
.../gems/activesupport-2.3.17/lib/active_support/core_ext/string/output_safety.rb:22: warning: regexp match /.../n against to UTF-8 string
Many thanks to grosser for supplying a monkey-patch for Rails 2.3 (Commit f93e3f0ec3 fixed it for Rails 3). Just put it into config/initializers/ to make those warnings go away.
Since we're using RSpec on mos...
How to set the default monospace font on Xfce (Xubuntu)
While you can set your own font in your terminal or other tools, it will not change the default "Monospace" font that some applications use.
To change that, edit ~/.fonts.conf and add settings for the "monospace" family. Here is how it looks on my machine now:
<fontconfig>
<match target="pattern">
<test qual="any" name="family">
<string>monospace</string>
</test>
<edit name="family" mode="assign">
<string>DejaVu Sans Mono</string>
</edit>
</match>
</fontconfig>
...
Check whether a Paperclip attachment exists
Don't simply test for the presence of the magic Paperclip attribute, it will return a paperclip Attachment object and thus always be true:
- if user.photo.present? # always true
= image_tag(user.photo.url)
Use #exists? instead:
- if user.photo.exists?
= image_tag(user.photo.url)
Use byebug on Ruby 2+
The debugger gem does not seem to be properly working on Ruby 2. Use byebug instead!
Byebug is a simple to use, feature rich debugger for Ruby 2. It uses the new TracePoint API for execution control and the new Debug Inspector API for call stack navigation, so it doesn't depend on internal core sources. It's developed as a C extension, so it's fast. And it has a full test suite so it's reliable. Note that byebug works only for ruby 2.0.0 or newer. For...
How to fix: Bundler 1.13 breaks parallel_tests
When running tests via parallel_tests, you may encounter an error:
cannot load such file -- parallel_tests/gherkin/runtime_logger
Error creating formatter: ParallelTests::Gherkin::RuntimeLogger (LoadError)
This will happen when you upgrade Bundler to version 1.13.x and appears to be "by design" since there is a Bundler config option to restore previous behavior.
You can fix it by setting that flag. You should commit the resulting config file into the repository!
bundle config --local disable_exec_load true
There is a Git...
RCov: The difference between "code coverage" and "total coverage"
Code coverage is the ratio of code lines kissed by a test vs. the total number of lines in your source files. This sounds meaningful, but isn't.
Total coverage additionally ignores whitespace and comments when computing the coverage ratio, which is what you want.
Beware: Nested Spreewald patiently blocks are not patient
Note: The behaviour of Spreewald's within step is as described below for version < 1.9.0; For Spreewald >= 1.9.0 it is as described in Solution 1.
When doing integration testing with cucumber and selenium you will often encounter problems with timing - For example if your test runs faster than your application, html elements may not yet be visible when the test looks for them. That's why Spreewald (a collection of cucumber steps) has a concept of doing things patiently, which means a given b...
Change how Capybara sees or ignores hidden elements
Short version
- Capybara has a global option (
Capybara.ignore_hidden_elements) that determines whether Capybara sees or ignores hidden elements. - Prefer not to change this global option, and use the
:visibleoption when callingpage.find(...). This way the behavior is only changed for this onefindand your step doesn't have confusing side effects. - Every Capybara driver has its own notion of "visibility".
Long version
Capybara has an option (Capybara.ignore_hidden_elements) to configure the default...
How to install the `xelatex` binary on Ubuntu 14.04
Just install the texlive-xetex package:
sudo apt-get install texlive-xetex
Running integration tests without texlive-xetex will produce an error during xelatex execution:
RTeX::Document::ExecutableNotFoundError
Geordi hints
Reminder of what you can do with Geordi.
Note: If you alias Geordi to something short like g, running commands gets much faster!
Note: You only need to type the first letters of a command to run it, e.g. geordi dep will run the deploy command.
geordi deploy
Guided deployment, including push, merge, switch branches. Does nothing without confirmation.
geordi capistrano
Run something for all Capistrano environments, e.g. geordi cap deploy
geordi setup -t -d staging
When you just clon...
A nicer way to run RSpec and/or Cucumber
geordi, our collection of awesome shell scripts, has been extended by three scripts to help you call RSpec or Cucumber:
cuc
This script runs Cucumber the way you want it:
- Prints some line feeds to easily find your test results when you come back to the console later
- Configures Cucumber to use cucumber_spinner if it is available in your
Gemfile - Runs Cucumber under
bundle exec - Uses an old version of Firefox for Selenium (Javascript) features...
Error "execution expired (Timeout::Error)" when using Selenium and Webmock
If you get the above error when running tests in bulk (but not individually), it's actually the fault of Webmock.
Updating Webmock to version 1.7+ fixes this.
Bash script to list commits by Pivotal Tracker ID
The main benefit of our convention to prefix commits by their corresponding Pivotal Tracker ID is that we can easily detect commits that belong to the same story. You can either do that manually or use the bash script below by copying it somewhere to your .bashrc.
# Usage: ptcommits 123456
function ptcommits {
if test "$1"
then
local PTID=$(echo "$1" | grep "[0-9]*" -o) # Allow URLs
git log --onel...
Fix a spec that only runs when called directly
When a spec only runs when it is called directly, but not as part of the whole test suite, make sure the filename is foo_spec.rb instead of just foo.rb.
Testing Cookie Limits
TL;DR If you want to support most browsers, then don't exceed 50 cookies per domain, and don't exceed 4093 bytes per domain (i.e. total size of all cookies <= 4093 bytes)
Behind the link, you'll find a simple HTML page that offers some cookie tests (how large, how many etc) and an overview of this data for various browsers.
Fun fact: You cannot delete cookies with a key that hits the size limit and has a small value.
CoffeeScript
Imagine all the syntactical delights of Ruby and Haml for your JavaScript. You write in a nice language, but get normal JavaScript at runtime. All whilst having full access to 3rd-party JavaScript libraries (jQuery, PrototypeJS), debugging support (it becomes pure, readable JavaScript), existing support from test suites (it’s normal JavaScript) and growing support from various text editors (TextMate, Vim, Emacs).
Error "undefined method last_comment"
This error message may occur when rspec gets loaded by rake, e.g. when you migrate the test database.
NoMethodError: undefined method 'last_comment' for #<Rake::Application:0x0055a617d37ad0>
Rake 11 removes a method that rspec-core < 3.4.4 depends on. To fix, lock Rake to < 11 in your Gemfile:
gem 'rake', '< 11', # Removes a method that rspec-core < 3.4 depends on
Waiting for page loads and AJAX requests to finish with Capybara
If you're using the Capybara webdriver, steps sometimes fail because the browser hasn't finished loading the next page yet, or it still has a pending AJAX request. You'll often see workarounds like
When I wait for the page to load
Then ...
Workarounds like this do not work reliably, will result in flickering tests and should be avoided. There is no known reliable way to detect if the browser has finished loading the page.
Solution
Instead you should wait until you can observe the result of a page load. E.g. if y...
Web Operations 101 For Developers
This post is not about devops, it's not about lean startups, it's not about web scale, it's not about the cloud, and it's not about continuous deployment. This post is about you, the developer who's main purpose in life has always been to build great web applications. In a pretty traditional world you write code, you write tests for it, you deploy, and you go home. Until now.