Is this fallout from bug 1244899? It seems to have spiked shortly after that commit. Should it be disabled like browser_UITour3.js, etc?

This is easy to reproduce with rr chaos mode.

(In reply to Ben Kelly [:bkelly] from comment #2) > Is this fallout from bug 1244899? It seems to have spiked shortly after > that commit. Should it be disabled like browser_UITour3.js, etc? It's because I just enabled this test for e10s in bug 1073247.

This test was doomed to be intermittent in E10S from the initial landing (which I reviewed :( ) since it's relying on an async message from the child to be handled before the test finishes.

The UITour content API call uses sendAsyncMessage so we can't assume a synchronous close. Review commit: https://reviewboard.mozilla.org/r/35515/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/35515/

Comment on attachment 8720949 [details] MozReview Request: Bug 1245107 - browser_closeTab.js: wait for the tab to close. r=jaws https://reviewboard.mozilla.org/r/35515/#review32195 ::: browser/components/uitour/test/browser_closeTab.js:14 (Diff revision 1) > + let closePromise = BrowserTestUtils.waitForEvent(gBrowser.tabContainer, "TabClose"); > yield gContentAPI.closeTab(); > + yield closePromise; It would be nice if closeTab returned a promise that was resolved either when the tab was closed or a no-op since this was the only tab. I'm not sure what we are yielding on gContentAPI.closeTab() for, as far as I can tell UITour-content.js returns undefined for this function, and UITour.jsm returns true for the page event.

(In reply to Jared Wein [:jaws] (please needinfo? me) from comment #13) > Comment on attachment 8720949 [details] > MozReview Request: Bug 1245107 - browser_closeTab.js: wait for the tab to > close. r=jaws > > https://reviewboard.mozilla.org/r/35515/#review32195 > > ::: browser/components/uitour/test/browser_closeTab.js:14 > (Diff revision 1) > > + let closePromise = BrowserTestUtils.waitForEvent(gBrowser.tabContainer, "TabClose"); > > yield gContentAPI.closeTab(); > > + yield closePromise; > > It would be nice if closeTab returned a promise that was resolved either > when the tab was closed or a no-op since this was the only tab. Ideally that's how it would work but when UITour-content.js was implemented I don't think promises were available. > I'm not sure what we are yielding on gContentAPI.closeTab() for, as far as I > can tell UITour-content.js returns undefined for this function, and > UITour.jsm returns true for the page event. gContentAPI is now a Proxy that returns a promise when the UITour-content.js API is called. The try push looks really good at the moment: https://treeherder.mozilla.org/#/jobs?repo=try&revision=2789859775d7

(In reply to Matthew N. [:MattN] (behind on bugmail) from comment #14) > gContentAPI is now a Proxy that returns a promise when the UITour-content.js > API is called. Forgot the link: https://mxr.mozilla.org/mozilla-central/source/browser/components/uitour/test/head.js?rev=4754d1254086#270

This is also passing locally on 10.10 with chaos mode enabled so I will push.

Comment on attachment 8720949 [details] MozReview Request: Bug 1245107 - browser_closeTab.js: wait for the tab to close. r=jaws Approval Request Comment Since bug 1073247 got uplifted then this should too. We can wait to see if it fixes the problem.

Comment on attachment 8720949 [details] MozReview Request: Bug 1245107 - browser_closeTab.js: wait for the tab to close. r=jaws Test only change, ok to uplift to aurora.