rspec_n is a Ruby gem that makes it easy to run a project's RSpec test suite N times. You can customize the command that is used to start RSpec, or let rspec_n guess the best command (based on the files in your project). rspec_n is useful for finding repeatability issues in RSpec test suites.
Releases are versioned using SemVer 2.0.0 with the following caveats:
- Support for Ruby versions, that reach EOL, can be removed in a major or minor release.
Supported Ruby Versions
Ruby 2.7, 3.0 and 3.1.
Install by executing:
$ gem install rspec_n
The gem will install an executable called
Add the following to your project's
.gitignore to exclude the output generated by rspec_n from your project's repo:
Usage in a Gemfile
If you add rspec_n to your Gemfile, use the
require: false option so rspec_n isn't loaded into your app. rspec_n doesn't provide any runtime benefit to apps and requiring it will add unnecessary code to your project. Also, rspec_n is designed as a standalone commandline tool and isn't tested for compatibility inside other apps.
gem 'rspec_n', require: false
The simplest way to run rspec_n is to give it a positive integer which tells it how many times to run RSpec:
$ rspec_n 5
As iterations complete, summary output is sent to the screen and detailed output is written to files (in the project's root).
To target specific specs, provide one or more paths. You can do anything you would normally do when giving paths to RSpec:
$ rspec_n 5 spec/path/to/something_spec.rb $ rspec_n 5 spec/path/to/folder spec/path/to/some/other/file_spec.rb $ rspec_n 5 spec/path/to/folder $ rspec_n 5 spec/path/to/something_spec.rb:5
--order rand is always sent to RSpec to force it to run specs in random order. You can use a
defined order if you don't want randomness:
$ rspec_n 5 --order defined
Or, let the RSpec configuration in the project determine the order:
$ rspec_n 5 --order project
You can create a
.rspec_n file in your project's root and add command line options to it. They'll be used if
rspec_n is run without options. Options are any arguments that start with
--. For example,
rspec_n 10 or
rspec_n 10 spec/some_spec.rb will make rspec_n consider
rspec_n 10 -c "rm -rf /tmp/* && bundle exec rspec" won't.
Example file format:
--no-file -s --order defined -c "rm -rf tmp && bundle exec rspec"
The config file can be multi line or single line.
Automatic Command Selection
rspec_n inspects files in your project so it can pick the best way to start RSpec. If it can't make an educated guess, it will use
bundle exec rspec as the base command and add any extra information you've entered on the command line (like the order or paths). The following is a list of project types that rspec_n can identify and the associated commands it will try to execute:
- Ruby on Rails Applications:
DISABLE_DATABASE_ENVIRONMENT_CHECK=1 RAILS_ENV=test bundle exec rake db:drop db:create db:migrate && bundle exec rspec.
- Everything else:
bundle exec rspec.
Use a Custom Command to Start RSpec
-c option if you want to specify your own command. The following example deletes the
tmp folder before starting RSpec:
$ rspec_n 5 -c 'rm -rf tmp && bundle exec rspec'
There are a couple points to consider:
- Wrap your entire command in a single or double quoted string.
&&to create compound commands.
- Avoid trying to change test order within a custom command. rspec_n was created to help discover flaky test suites so it adds
--order randto your custom command. Use the
--order projectoptions to the control the order.
Control File Output
rspec_n writes output for each iteration to files with the iteration number (
rspec_n_iteration.2, etc...). If you want to disable this, add the
--no-file option to the command.
$ rspec_n 5 --no-file
Note: rspec_n deletes all files matching
rspec_n_iteration.* when it starts, so be sure to move those files to another location if you want to save them.
Stop on First Failure
You can tell rspec_n to abort the first time an iteration fails by using the
-s flag. All remaining iterations are skipped.
Understanding the Results
rspec_n uses the
EXIT STATUS of the
rspec command to decide if the run passed or failed. The run is considered successful if RSpec returns an
EXIT STATUS of
0 regardless of any content in the STDERR stream. rspec_n considers the run to be a failure if RSpec's
There are times when
STDERR has content, even though RSpec returns an
EXIT STATUS of
0. This frequently happens with deprecation notices and RSpec itself will pass the test suite in this situation. Also, it's not uncommon for code to write messages to
STDERR that look like errors, but not actually cause RSpec to fail a spec example. If rspec_n is able to detect this situation, it adds
(Warning) to the results to indicate there's something extra in the
STDERR that you might want to investigate.
The results of each run (number of tests, number of failures, etc...) are determined by parsing
STDOUT and extracting information from the line that says
xyz examples, xyz failures, xyz pending.
Run Automated Tests
Run the automated test suite with:
Start a Console
Choose one of the following to start a console with the gem loaded:
$ bin/console # for Pry $ rake console # for IRB
Contributions are welcome. Please us the following process when submitting work:
- Create an issue on the issue page that targets a single problem/enhancement. All PRs should be tied to an issue.
- Fork the project.
- Create a branch. The name of the branch should start with the issue number that the branch will address.
- Submit the PR. In the PR comment (in the UI), add
xyzis the issue number that the PR addresses. The title of the PR (in the UI) should start with
[PR for #xyz].