fix: prevent retry config accumulation

[smalia2001]

Make each plugin/helper clean its own retry configuration

• It solves a leak in recorder.retries and make retryFailedStep plugin keep working also in the n > 1 test.
• Resolves Bug: retryOnFailedStep works only for the first test in the same run #5598 (if applicable).

Applicable helpers:

• Playwright
• Puppeteer
• WebDriver
• REST
• FileHelper
• Appium

Applicable plugins:

• aiTrace
• autoDelay
• autoLogin
• customLocator
• pause
• coverage
• heal
• retryFailedStep
• screenshot
• selenoid
• stepTimeout
• subtitles

Type of change

• 🔥 Breaking changes
• 🚀 New functionality
• 🐛 Bug fix
• 🧹 Chore
• 📋 Documentation changes/updates
• ♨️ Hot fix
• 🔨 Markdown files fix - not related to source code
• 💅 Polish code

Checklist:

• Tests have been added
• Documentation has been added (Run npm run docs)
• Lint checking (Run npm run lint)
• Local tests are passed (Run npm test)

[DavertMik]

@smalia2001 could be resolve the real issue
Please prove it with writing a unit test for it
Also descrbe the issue in PR descriptoin

[smalia2001]

@DavertMik done.

[DenysKuchma]

@smalia2001 Can we fix the new exec tests so they fail with the real error instead of timing out, and double-check that the new retry accumulation cases pass? Also worth covering retryFailedStep together with Playwright/Puppeteer retries, since this PR touches both

[smalia2001]

@DenysKuchma I understand your point about playwright and puppeteer, but I don't get you about failing instead of timing out... This tests are failing if the failing step doesn't get retried.

[mirao]

@smalia2001 Your proposed fix works for me.

A patch I'm having in my project now is a bit differrent (and works for me too):

diff --git a/lib/plugin/retryFailedStep.js b/lib/plugin/retryFailedStep.js
index 394fa4e..bd889f9 100644
--- a/lib/plugin/retryFailedStep.js
+++ b/lib/plugin/retryFailedStep.js
@@ -141,12 +141,6 @@ export default function (config) {
       return
     }
 
-    const hasManualRetries = recorder.retries.some(retry => retry !== config)
-    if (hasManualRetries) {
-      store.autoRetries = false
-      return
-    }
-
     store.autoRetries = true
     test.opts.conditionalRetries = config.retries
     test.opts.stepRetryPriority = stepRetryPriority
@@ -158,9 +152,6 @@ export default function (config) {
   event.dispatcher.on(event.test.started, test => {
     if (test.opts?.disableRetryFailedStep || test.disableRetryFailedStep) return
 
-    const hasManualRetries = recorder.retries.some(retry => retry !== config)
-    if (hasManualRetries) return
-
     const scenarioRetries = typeof test.retries === 'function' ? test.retries() : -1
     if (scenarioRetries > 0 && config.deferToScenarioRetries !== false) {
       return

[smalia2001]

@DenysKuchma I understand your point about playwright and puppeteer, but I don't get you about failing instead of timing out... This tests are failing if the failing step doesn't get retried.

@DenysKuchma ??

[DenysKuchma]

@smalia2001 Sorry for the delay. I meant that the new exec tests should pass child process errors to done(err) instead of potentially timing out. Also, please update the branch with the latest 4.x and rerun CI