browser/base/content/test/general/browser_trackingUI_5.js starts testing on new tab just after adding the new tab, that results in triggering page load in unstable state, and the finally loaded page can differ depending on the timing.

this test deterministically fails with non-e10s locally, after bug 1193394. so I think there's some more issue there.

On non-e10s, the tracking protection is disabled (just like unblock button is clicked) at the beginning of testExceptionPersistence. and commenting out testExceptionAddition except `await UrlClassifierTestUtils.addTestTrackers(); await pushPrefs([PB_PREF, true]);` solves the issue. So I think there's some unexpected interaction between those 2 tests.

turns out that it should also wait for the close of private window.

promiseWindowWillBeClosed needs executeSoon just like other cases. also, the test should wait for the tab's load event to run in stable state, by using BrowserTestUtils.openNewForegroundTab instead of browser.addTab.

Pushed by arai_a@mac.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/04761f378126 Wait for new tab load and window close in browser/base/content/test/general/browser_trackingUI_5.js to make sure state transition does not overlap. r=felipe

https://hg.mozilla.org/integration/mozilla-inbound/rev/04761f378126ee866a1a25807782e9d504bade4b Bug 1421653 - Wait for new tab load and window close in browser/base/content/test/general/browser_trackingUI_5.js to make sure state transition does not overlap. r=felipe