Test Isolation

info

What you'll learn

• What is test isolation
• How it impacts E2E Testing vs Component Testing
• Test isolation trade-offs

What is Test Isolation?

tip

Best Practice: Tests should always be able to be run independently from one another and still pass.

As stated in our mission, we hold ourselves accountable to champion a testing process that actually works, and have built Cypress to guide developers towards writing independent tests from the start.

We do this by cleaning up state before each test to ensure that the operation of one test does not affect another test later on. The goal for each test should be to reliably pass whether run in isolation or consecutively with other tests. Having tests that depend on the state of an earlier test can potentially cause nondeterministic test failures which make debugging challenging.

Cypress will start each test with a clean test slate by restoring and clearing all:

• aliases
• clock mocks
• intercepts
• spies
• stubs
• viewport changes

In additional to a clean test slate, Cypress also believes in running tests in a clean browser context such that the application or component under test behaves consistently when ran. This behavior is described as testIsolation.

The test isolation is a global configuration and can be overridden for end-to-end testing at the describe level with the testIsolation option.

Test Isolation in End-to-End Testing

Cypress supports enabling or disabling test isolation in end-to-end testing to describe if a suite of tests should run in a clean browser context or not.

Test Isolation Enabled

When test isolation is enabled, Cypress resets the browser context before each test by:

• clearing the dom state by visiting about:blank
• clearing cookies in all domains
• clearing localStorage in all domains
• clearing sessionStorage in all domains

Because the test starts in a fresh browser context, you must re-visit your application and perform the series of interactions needed to build the dom and browser state for each test.

Additionally, the cy.session() command will inherit this configuration and will clear the page and current browser context when establishing a browser session. This is so tests can reliably pass when run standalone or in a randomized order.

Test Isolation Disabled

When test isolation is disabled, Cypress will not alter the browser context before the test starts. The page does not clear between tests and cookies, local storage and session storage will be available across tests in that suite. Additionally, the cy.session() command will only clear the current browser context when establishing the browser session - the current page is not cleared.

Quick Comparison

testIsolation	beforeEach test	cy.session()
true	- clears page by visiting about:blank - clears cookies in all domains - local storage in all domains - session storage in all domains	- clears page by visiting about:blank - clears cookies in all domains - local storage in all domains - session storage in all domains
false	does not alter the current browser context	- clears cookies in all domains - local storage in all domains - session storage in all domains

Test Isolation in Component Testing

Cypress does not support configuring the test isolation behavior in component testing.

When running component tests, Cypress always resets the browser context before each test by:

• unmounting the rendered component under test
• clearing cookies in all domains
• clearing localStorage in all domains
• clearing sessionStorage in all domains

Test Isolation Trade-offs

It is important to note that disabling test isolation may improve the overall performance of end-to-end tests, however, it can also cause state to "leak" between tests. This can make later tests dependent on the results of earlier tests, and potentially cause misleading test failures. It is important to be extremely mindful of how tests are written when using this mode, and ensure that tests continue to run independently of one another.

The best way to ensure your tests are independent is to add a .only() to your test and verify it can run successfully without the test before it.