In bug 1439472 I disabled browser/components/extensions/test/browser/browser_ext_popup_select.js on debug build. The 1st patch there fixes this issue but introduces failure on browser/components/extensions/test/browser/test-oop-extensions/browser_ext_browserAction_popup_resize.js, bug 1442101. we should fix that one and re-enable browser/components/extensions/test/browser/browser_ext_popup_select.js on debug build.

bug 1442101 is reproducible locally with win10 debug build, with the previous patch for bug 1439472 applied.

(In reply to Tooru Fujisawa [:arai] from comment #1) > bug 1442101 is reproducible locally with win10 debug build, with the > previous patch for bug 1439472 applied. on m-c 0b997836c7cf

(sorry for taking time, the issue is rarely reproducible even on win10 debug, and hard to investigate :P

The original patch from bug 1439472. this patch itself should be fine. we should fix the failure in browser_ext_browserAction_popup_resize.js, in order to land this patch. (also, we should re-enable browser_ext_popup_select.js in browser.ini)

I have questions about browser_ext_browserAction_popup_resize.js. it opens panel and just after that it gets the size of the overflow panel, to compare later. but apparently the overflow panel is not yet visible at that point (confirmed locally). :kmag, could you tell me what size it's actually trying to get and compare? https://searchfox.org/mozilla-central/rev/b80994a43e5d92c2f79160ece176127eed85dcc9/browser/components/extensions/test/browser/browser_ext_browserAction_popup_resize.js#163 > let browser = await openPanel(extension, browserWin); > let origPanelRect = panel.getBoundingClientRect(); and what's the intention of the following assertion (that failed in bug 1442101)? https://searchfox.org/mozilla-central/rev/b80994a43e5d92c2f79160ece176127eed85dcc9/browser/components/extensions/test/browser/browser_ext_browserAction_popup_resize.js#171-172 > ok(panelRect.bottom >= origPanelRect.bottom, `Panel has not shrunk from original size (${panelRect.bottom} >= ${origPanelRect.bottom})`);

(In reply to Tooru Fujisawa [:arai] from comment #5) > it opens panel and just after that it gets the size of the overflow panel, > to compare later. > but apparently the overflow panel is not yet visible at that point > (confirmed locally). perhaps this is related to bug 1444684 and not related to the previous failure. anyway, I'd like to ask what the test is trying to do.

what I found so far is, the panel grows vertically while horizontal sliding transition (apparently between overflow menu and extension popup?), https://searchfox.org/mozilla-central/rev/b80994a43e5d92c2f79160ece176127eed85dcc9/browser/components/customizableui/PanelMultiView.jsm#946 > var PanelMultiView = class extends AssociatedToNode { > ... > async _transitionViews(previousViewNode, viewNode, reverse) { > ... > // Now set the viewContainer dimensions to that of the new view, which > // kicks of the height animation. > this._viewContainer.style.height = viewRect.height + "px"; on local Linux 64, it sets the height to "640px", from "600px", and transition starts there. after the transition, the height style is removed here: https://searchfox.org/mozilla-central/rev/b80994a43e5d92c2f79160ece176127eed85dcc9/browser/components/customizableui/PanelMultiView.jsm#1009 > var PanelMultiView = class extends AssociatedToNode { > ... > _cleanupTransitionPhase() { > ... > if (phase >= TRANSITION_PHASES.START) { > ... > this._viewContainer.style.removeProperty("height"); In successful case, the actual height of the _viewContainer box decreases here. on local Linux 64, from 640px to 600px. In failure case, the actual height grows more. on local Linux 64, from 640px to 692px. In failure case, origPanelRect.bottom becomes 768 in the test, while it's expected to be 676. https://searchfox.org/mozilla-central/rev/b80994a43e5d92c2f79160ece176127eed85dcc9/browser/components/extensions/test/browser/browser_ext_browserAction_popup_resize.js#163 > async function testPopupSize(standardsMode, browserWin = window, arrowSide = "top") { > ... > let origPanelRect = panel.getBoundingClientRect(); the height seems to become the expected value shortly after getting origPanelRect (within 100ms), and the test fails because only origPanelRect has unexpected value, while panelRect is expected value. So I think there's a race between getBoundingClientRect call above and something that decreases the height of popup. :kmag, do you have any ideas?

Comment on attachment 8964230 [details] [diff] [review] Wait for ViewShown event of extension panel before synthesizing mouse event. r=Paolo clearing review flag for tracking reason. the patch is incomplete for the whole fix. I need info from kmag anyway.