Tests are way too slow when this preprocessor is used

Tried a simple button test

1. button.spec.ts - finishes in 4s
2. button.feature - finishes in 40s

button.spec.ts

describe('Button - vanilla', () => {
  it('should launch button page', () => {
    cy.visit('/components/ui/button');
    cy.contains('Examples').click();
    cy.contains('Secondary Buttons').click();
    cy.get('#button-secondary').should('be.visible');
  });
});

button.feature

Feature: Button

  TRQ-XXXXX

  Background:
    Given button page is open

  Scenario: Default Button
    When I click on examples tab
    And I click on secondary-buttons example
    Then I should see secondary button

button.steps.ts

import { Given, When, And, Then } from 'cypress-cucumber-preprocessor/steps';

Given(`button page is open`, () => {
  cy.visit('/components/ui/button');
});

When(`I click on examples tab`, () => {
  cy.contains('Examples').click();
});

And(`I click on secondary-buttons example`, () => {
  cy.contains('Secondary Buttons').click();
});

Then(`I should see secondary button`, () => {
  cy.get('#button-secondary').should('be.visible');
});

[rodrigo-vazquez]

Hi! Are there any news on this? We recently migrated from stock Cypress Mocha to cucumber using the preprocessor and the Suite of 10 somewhat long tests we have currently is taking double the time it used to with Mocha.

We’re using Cypress 7.5.0 and cypress-cucumber-preprocessor 4.1.2.

Things we already tried…

1. Using or not using Data Tables
2. Using global Steps or Feature Steps
3. Going back versions of both Cypress and the Cucumber Preprocessor
4. Turning the ‘numTestsKeptInMemory’ to 0

Thanks!!

[badeball]

I did some investigating myself and unfortunately found no performance overhead associated with using the preprocessor (note: using latest version, see https://github.com/badeball/cypress-cucumber-preprocessor/issues/689). I took one of the tests that are auto-generated when opening Cypress for the first time, duplicated and re-wrote it using steps. This can be found at badeball/cypress-cucumber-performance-comparison. Then I ran everything ten times. Below are the results.

	todo.feature	todo.spec.js
1. run	03:04	03:00
2. run	03:02	03:00
3. run	03:06	03:01
4. run	03:03	02:59
5. run	03:05	03:01
6. run	03:04	02:58
7. run	03:02	02:58
8. run	03:02	02:58
9. run	03:02	03:01
10. run	03:02	03:01

As you can see, there was little to no difference.

Unless anyone can show me otherwise, I’m considering this a non-issue. That’s not so say that you’re not experiencing any issues, but I can’t investigate and fix something that I can’t observe myself.