RspecProfiling

Collects profiles of RSpec test suites, enabling you to identify specs with interesting attributes. For example, find the slowest specs, or the spec which issues the most queries.

Collected attributes include:

• git commit SHA (or SVN revision) and date
• example file, line number and description
• example status (i.e. passed or failed)
• example exception (i.e. nil if passed, reason for failure otherwise)
• example time
• query count and time
• request count and time

RspecProfiling should work with Rails >= 3.2 and RSpec >= 2.14.

Add this line to your application's Gemfile:

gem 'rspec_profiling'

And then execute:

bundle

Require the gem to your spec_helper.rb.

require "rspec_profiling/rspec"

Lastly, run the installation rake tasks to initialize an empty database in which results will be collected.

bundle exec rake rspec_profiling:install

If you are planning on using sqlite or pg ensure to add the dependency to your gemfile

  gem 'sqlite', require: false
  gem 'pg', require: false

Results are collected based on the version control system employed e.g. revision or commit SHA for svn and git respectively.

By default, RspecProfiling expects Git as the version control system.

RspecProfiling can be configured to use svn in config/initializers/rspec_profiling.rb:

RspecProfiling.configure do |config|
  config.vcs = RspecProfiling::VCS::Svn
end

For those with a mixed project, with some developers using git svn and others regular svn, use this configuration to detect which is being used locally and behave accordingly.

RspecProfiling.configure do |config|
  config.vcs = RspecProfiling::VCS::GitSvn
end

If the repo you are running the profiler on has many teams working on it, you can use the magic_comment option to specify a comment at the top of files to scan for ownership tracking. In the example below, the profiler will look for #team: <owner> comments at the top of each file and add to the results. The default is team but can be configured to any comment you want.

RspecProfiling.configure do |config|
  config.magic_comment = 'team'
end

RspecProfiling.configure do |config|
  config.events = %w[event1 event2]
end

Note that custom events are only currently reported by the CSV collector.

It is possible to record the event metadata for a spec.

  describe 'Records all active record queries', record_events: %w[sql.active_record] do
    it 'Records Rails deprecations', record_events: %w[deprecation.rails] do
      ...
    end
    it 'Records nothing' do
      ...
    end
  end

Results are collected just by running the specs.

Make sure you've run the installation rake task before attempting.

You can configure RspecProfiling to collect results in a SQL database in config/initializers/rspec_profiling.rb:

RspecProfiling.configure do |config|
  config.collector = RspecProfiling::Collectors::SQL
end

You can review results by running the RspecProfiling console. The console has a preloaded results variable.

bundle exec rake rspec_profiling:console

> results.count
=> 1970

You can find the spec that runs the most queries:

> results.order(:query_count).last.to_s
=> "Updating my account - ./spec/features/account_spec.rb:15"

Or find the spec that takes the most time:

> results.order(:time).last.to_s
=> "Updating my account - ./spec/features/account_spec.rb:15"

There are additional attributes available on the Result instances to enable debugging, such as exception and status.

By default, profiles are collected in an a CSV file. You can configure RspecProfiling to collect results in a CSV in config/initializers/rspec_profiling.rb:

RspecProfiling.configure do |config|
  config.collector = RspecProfiling::Collectors::CSV
end

By default, the CSV is output to cat tmp/spec_benchmarks.csv. Rerunning spec will overwrite the file. You can customize the CSV path to, for example, include the sample time.

RspecProfiling.configure do |config|
  config.collector = RspecProfiling::Collectors::CSV
  config.csv_path = ->{ "tmp/spec_benchmark_#{Time.now.to_i}" }
end

You can configure RspecProfiling to collect results in a Postgres database in your spec_helper.rb file.

RspecProfiling.configure do |config|
  config.collector = RspecProfiling::Collectors::PSQL
  config.db_path   = 'profiling'
end

Configuration is performed like this:

RspecProfiling.configure do |config|
  config.<option> = <something>
end

• db_path - the location of the SQLite database file
• table_name - the database table name in which results are stored
• csv_path - the directory in which CSV files are dumped
• collector - collector to use
• magic_comment - comment to scan top of files to enable ownership tracking (EX: #team: tooling)

If you want to access the results from a Ruby script instead of the rake rspec_profiling:console shell command:

require 'rspec_profiling'
require 'rspec_profiling/console'

Then results will be available as a variable to the script.

To remove the results database, run bundle exec rake rspec_profiling:uninstall.

Please read CONTRIBUTING.md for details on our code of conduct, and the process for submitting pull requests to us.

Local tools needed:

• docker
• docker-compose
• ruby

To run the specs:

make spec