CursorRendererMacTest.CursorDuringMouseMovement in "browser_side_navigation_content_unittests on Mac-10.11" failed on "Mac10.11 Tests" |
|||
Issue description[ RUN ] CursorRendererMacTest.CursorDuringMouseMovement ../../content/browser/media/capture/cursor_renderer_mac_unittest.mm:134: Failure Value of: CursorDisplayed() Actual: false Expected: true [ FAILED ] CursorRendererMacTest.CursorDuringMouseMovement (10 ms) [ RUN ] CursorRendererMacTest.CursorDuringMouseMovement ../../content/browser/media/capture/cursor_renderer_mac_unittest.mm:134: Failure Value of: CursorDisplayed() Actual: false Expected: true [ FAILED ] CursorRendererMacTest.CursorDuringMouseMovement (11 ms) [ RUN ] CursorRendererMacTest.CursorDuringMouseMovement ../../content/browser/media/capture/cursor_renderer_mac_unittest.mm:134: Failure Value of: CursorDisplayed() Actual: false Expected: true ../../content/browser/media/capture/cursor_renderer_mac_unittest.mm:139: Failure Value of: CursorDisplayed() Actual: true Expected: false [ FAILED ] CursorRendererMacTest.CursorDuringMouseMovement (16 ms) [ RUN ] CursorRendererMacTest.CursorDuringMouseMovement ../../content/browser/media/capture/cursor_renderer_mac_unittest.mm:139: Failure Value of: CursorDisplayed() Actual: true Expected: false [ FAILED ] CursorRendererMacTest.CursorDuringMouseMovement (18 ms)
,
Sep 22 2017
This failed again on 10.11: https://uberchromegw.corp.google.com/i/chromium.mac/builders/Mac10.11%20Tests/builds/18390 Bumping priority. braveyao@, can you take a peek?
,
Sep 25 2017
Sorry for the delay due to my OOO. Still didn't get the Mac10.11 machine. But most probably, this wouldn't be reproduced locally. Then I will add some debugging info and see if we can isolate the issue to OSX side or Chrome side. And take some action according to it.
,
Sep 29 2017
Questions: is there any mouse or touch-pad connected to the mac bots and is there any possibility that someone may move the mouse cursor during tests running? If the answers are YES, then I think that's the reason. In those test cases I need to actually move mouse cursor into/out_of certain range with OS API. If the cursor is moved by real device at same time, the the position will be unexpected and test case may fail.
,
Oct 11 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/1e6094789708c75e7dd0628eec61df502154cbf2 commit 1e6094789708c75e7dd0628eec61df502154cbf2 Author: braveyao <braveyao@chromium.org> Date: Wed Oct 11 21:12:35 2017 CursorRendererMacTest: dis-associate mouse and cursor in testing The CursorRendererMacTest may fail under certain occasions on the bots. So far the only possible replication I can find on local machine is to move mouse during tests running. It seems that when both test app and mouse both try to manipulate cursor at same time, we'll get unexpected cursor position and tests will fail. One way to solve this issue is to intercept and discard the kCGEventMouseMoved event and only let the test control cursor position. With this fix, moving mouse during test won't let test down locally. Bug: 761902 Change-Id: I6519110167b32292228bc5522908d8dbce1fff66 Reviewed-on: https://chromium-review.googlesource.com/710865 Reviewed-by: Yuri Wiitala <miu@chromium.org> Commit-Queue: Weiyong Yao <braveyao@chromium.org> Cr-Commit-Position: refs/heads/master@{#508115} [modify] https://crrev.com/1e6094789708c75e7dd0628eec61df502154cbf2/content/browser/media/capture/cursor_renderer_mac_unittest.mm
,
Oct 11 2017
Landed a cl to disconnect mouse and cursor during CursorRendererMacTest. Let see if it will fix the problem on bots. |
|||
►
Sign in to add a comment |
|||
Comment 1 by braveyao@chromium.org
, Sep 6 2017