rspec_n

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.

rspec_n inspects files in your project and determines 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:

1. Ruby on Rails Applications: DISABLE_DATABASE_ENVIRONMENT_CHECK=1 RAILS_ENV=test bundle exec rake db:drop db:create db:schema:load && bundle exec rspec.
2. Everything else: bundle exec rspec.

Note: You can override this behavior by using the -c option, which lets you specify your own command. See the Use a Custom Command to Start RSpec section for more info.

Releases are versioned using SemVer 2.0.0 with the following caveats:

1. Support for Ruby versions, that reach EOL, can be removed in a major or minor release.

Ruby 3.1+ is supported.

Install by executing:

$ gem install rspec_n

The gem will install an executable called rspec_n.

Add the following to your project's .gitignore to exclude the output generated by rspec_n from your project's repo:

rspec_n_iteration.*

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 command line tool and isn't tested for compatibility inside other apps.

gem 'rspec_n', require: false

1. By default, rspec_n will attempt to prepare the test database by executing the following code before each run:

   DISABLE_DATABASE_ENVIRONMENT_CHECK=1 RAILS_ENV=test bundle exec rake db:drop db:create db:schema:load rake db:drop db:create db:schema:load
   

   1. DISABLE_DATABASE_ENVIRONMENT_CHECK=1 disables a Rails check that can prevent the test database from being dropped.
   2. Since db:schema:load is used to prepare the test database, projects must have a correct schema.rb or structure.sql.

2. By default, rspec_n tries to ensure an empty database before each run so it does not load seed files. If your test suite depends on seeded data, or other special setup, you will need to specify the exact command that rspec_n uses to invoke each run. See the Use a Custom Command to Start RSpec section for more info.

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 passing 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

By default, --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

Use the -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:

1. Wrap your entire command in a single or double quoted string.

2. Use && to create compound commands.

3. rspec_n was created to help discover flaky test suites so it will add --order rand to your custom command (which gets passed to RSpec and causes RSpec to run tests randomly). If you need to change the test order, you must add the --order defined or --order project options to the control the order like this:

   $ rspec_n 5 -c 'rm -rf tmp && bundle exec rspec' --order defined
   

rspec_n writes output to the project's root folder, for each iteration, to files with the iteration number in the name (rspec_n_iteration.1, 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.

You can tell rspec_n to abort the first time an iteration fails by using the -s flag. All remaining iterations are skipped.

rspec_n doesn't directly support running specs in parallel, however, you should be able to use the -c option to pass a rspec_n the command you use to run parallel specs.

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 - or --. For example, rspec_n 10 or rspec_n 10 spec/some_spec.rb will make rspec_n consider .rspec_n, but 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.

rspec_n uses the STDOUT, STDERR and 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 EXIT STATUS > 0.

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.

Install dependencies:

$ bin/setup

That's it!

Run the automated test suite with:

$ rspec

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:

1. Create an issue on the issue page that targets a single problem/enhancement. All PRs should be tied to an issue.
2. Fork the project.
3. Create a branch. The name of the branch should start with the issue number that the branch will address.
4. Submit the PR. In the PR comment (in the UI), add Closes #xyz or Supports #xyz where xyz is the issue number that the PR addresses. The title of the PR (in the UI) should start with [PR for #xyz] .