Some improve your rspec speed

Do you ever run rspec in your project take time 20 mn or much that this or not ? It maybe become normal for you, if you want to run it all for making sure that your editing code is not effected to any function expectation. Unfortunately, if you meet some case fails or you need to update current rspec to correct with current logic, Clearly that you need to run fails case few times and worse than this sometime you need to run all case in project. It seem so borde when it waste much time to run slow rspec. It have a lot of cases that make your rspec become slow. In this article, i want to show all of you some cases effected to your rspec performace.

Slow because of configure in spec_helper

Plz take a look example below:

# spec/models/user_spec.rb

describe "associations" do
  subject(:user) { User.new }

  it { should have_many(:orders) }
  # 12 additional association specs
end

Result:

It takes more than 3 seconds to test the model relationships, even without call any database,it seem take too long. And it is cause from large logic in file spec_helper so it make our rspec become slow also. Now let's command all logic in file spec_helper and run it again:

# spec/spec_helper.rb`

RSpec.configure do |config|
  # > 300 lines of around, before and after blocks
end

As you can see speed is faster nearly 9 times that before. Now we will start toinvestigate about the causes and how to solve the problems that lead to our rspec running so long in spec_helper.

Case 1

# spec/spec_helper.rb

config.before(:each) do
  stub_something
end

The before block above is only needed for the feature spec but if we write it like this, it will run before all spec. So we need write it as below to make it faster:

# spec/spec_helper.rb

config.before(:each, type: :feature) do
  stub_something
end

But sometime doesn't mean all feature spec need to run block code above. So we can make better as below:

# spec/spec_helper.rb

config.before(:each, :stub_something) do
  stub_something
end

Case 2

# spec/spec_helper.rb

config.before(:each) do |example|
  DatabaseCleaner.strategy = :truncation
end

Sometime alot of data can make our rspec become slow also, So we need database cleaning strategies to reset the db, We can create config.before block to clean db as below:

# spec/spec_helper.rb

config.before(:each, :clean_database_with_truncation) do |example|
  DatabaseCleaner.strategy = :truncation
end

Slow in Spec Setup

Using the RSpec flag profile, we can determine the slowest parameters when running rspec for the file.

rspec spec/models/user_spec.rb --profile 5

Now let's check slow logic in our rspec:

# spec/models/user_spec.rb

it "does something" do
  user = FactoryGirl.create(:user)

  # expectation on user
end

Even comment all logic except FactoryGirl and can see speed is not improved at all. So the cause of the slowdown is most likely due FactoryGirlwe looked at whether it was needed FactoryGirl or could be used build_stubbed. In addition, we still consider whether we can optimize the internal code FactoryGirl or not.

# spec/factories/user_factory.rb

FactoryGirl.define do
  factory :user do

  # more code

  after(:create) do |user|
    create(:billing_profile, user: user)
  end
end

In FactoryGirl, has created an additional active record for the relationship without using in some spec. This part can be further optimized.

# spec/factories/user_factory.rb

FactoryGirl.define do
  factory :user do

  # more code

  trait :with_billing_profile do
    after(:create) do |user|
      create(:billing_profile, user: user)
    end
  end
end

So now it is possible to create users with or without creating relationships through the use of the option :with_billing_profile Now run rspec again to see if the speed has increased.