# Preserve & Rerun (Compare)

When a test fails, the usual debugging loop is: rerun it, then compare two walls of logs to work out what changed. Preserve & Rerun collapses that into a single click. It **snapshots the failing run and re-executes the test in one action**, then shows both runs side-by-side in a **Compare** view aligned command-by-command - so you can see exactly where the two diverged without re-reading anything.

This is the fastest way to diagnose a flaky test: the command that behaved differently between the pass and the fail is highlighted for you, along with the assertion that broke.

Available across all three adapters - **WebdriverIO**, **[Selenium WebDriver](/docs/devtools/selenium.md)**, and **[Nightwatch.js](/docs/devtools/nightwatch.md)**.

## Demo[​](#demo "Direct link to Demo")

![Preserve \& Rerun Demo](/assets/images/preserve-rerun-52348ac5e421da1ec2d83f71a9848396.gif)

## How it works[​](#how-it-works "Direct link to How it works")

1. Run your tests as normal. When a test finishes in a **failed** state, hover its row in the sidebar.
2. A bug-play icon (🐞▶) appears next to the regular ▶ rerun button. It only shows on failed test/suite rows, wherever a plain rerun is already supported (e.g. Cucumber scenarios at the scenario row, Mocha/Jasmine tests at the test or suite row).
3. Click it. DevTools captures a snapshot of the failing run, then re-launches just that test.
4. The **Compare** tab opens with the two runs aligned by command. The point of divergence and the assertion error (**Expected vs Received**) are called out.

## Key Features[​](#key-features "Direct link to Key Features")

* **One-click snapshot + rerun** - Preserve the failing run and re-execute it in a single action, no code changes or full-suite restart.
* **Command-by-command alignment** - Both runs are laid out side-by-side and aligned by command, so differences stand out instantly.
* **Failure point highlighted** - Jumps you straight to the command where the two runs diverged.
* **Assertion diff** - Shows the assertion that broke with Expected vs Received side-by-side.
* **Pop-out window** - Open the comparison in a separate, themed window for a roomier view.
* **Flaky-test triage** - See which command differed between a pass and a fail without re-reading logs.

## Limitations[​](#limitations "Direct link to Limitations")

* **Cucumber**: per-step rerun is disabled because Cucumber's `--name` filter targets scenarios, not individual Gherkin steps. Scenario-level Preserve & Rerun still works.