Using unstub

unstub removes a method stub, essentially cleaning up the method stub early, rather than waiting for the cleanup that runs at the end of the example. The newer non-monkey-patching syntax does not have a direct equivalent but in most situations you can achieve the same behavior using and_call_original. The difference is that obj.unstub(:foo) completely cleans up the foo method stub, whereas allow(obj).to receive(:foo).and_call_original continues to observe calls to the method (important when you are using spies), which could affect the method’s behavior if it does anything with caller as it will include additional rspec stack frames.


Given a file named “spec/spec_helper.rb” with:

RSpec.configure do |config|
  config.mock_with :rspec do |mocks|
    mocks.syntax = :should

And a file named “.rspec” with:

--require spec_helper

Unstub a method

Given a file named “spec/unstub_spec.rb” with:

RSpec.describe "Unstubbing a method" do
  it "restores the original behavior" do
    string = "hello world"
    string.stub(:reverse) { "hello dlrow" }

    expect {
    }.to change { string.reverse }.from("hello dlrow").to("dlrow olleh")

When I run rspec spec/unstub_spec.rb

Then the examples should all pass.