Closed Bug 1009643 Opened 11 years ago Closed 10 years ago

test_wheel_default_action.html fails on OSX 10.9

Categories

(Core :: DOM: Events, defect)

Product:
Core
Component:
DOM: Events
Platform:
All
macOS
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED DUPLICATE of bug 1121480

People

(Reporter: jgriffin, Unassigned)

References

Details

When run on OSX 10.9, the test test_wheel_default_action.html consistently fails, both opt and debug: 19:12:07 INFO - 1918 INFO TEST-UNEXPECTED-FAIL | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to top (page): unexpected scrollTop - got 2800, expected 0 19:12:07 INFO - 1919 INFO TEST-INFO | dumping last 1 message(s) 19:12:07 INFO - 1920 INFO TEST-INFO | if you need more context, please use SimpleTest.requestCompleteLog() in your test 19:12:07 INFO - 1921 INFO TEST-PASS | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to top (page): unexpected scrollLeft 19:12:07 INFO - 1922 INFO TEST-UNEXPECTED-FAIL | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to top when scrollTop is already top-most (page): unexpected scrollTop - got 2800, expected 0 19:12:07 INFO - 1923 INFO TEST-INFO | dumping last 1 message(s) 19:12:07 INFO - 1924 INFO TEST-INFO | if you need more context, please use SimpleTest.requestCompleteLog() in your test 19:12:07 INFO - 1925 INFO TEST-PASS | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to top when scrollTop is already top-most (page): unexpected scrollLeft 19:12:07 INFO - 1926 INFO TEST-UNEXPECTED-FAIL | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to bottom (page): unexpected scrollTop - got 0, expected 2800 19:12:07 INFO - 1927 INFO TEST-INFO | dumping last 1 message(s) 19:12:07 INFO - 1928 INFO TEST-INFO | if you need more context, please use SimpleTest.requestCompleteLog() in your test 19:12:07 INFO - 1929 INFO TEST-PASS | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to bottom (page): unexpected scrollLeft 19:12:07 INFO - 1930 INFO TEST-UNEXPECTED-FAIL | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to bottom when scrollTop is already bottom-most (page): unexpected scrollTop - got 0, expected 2800 19:12:08 INFO - 1931 INFO TEST-INFO | dumping last 2 message(s) 19:12:08 INFO - 1932 INFO TEST-INFO | if you need more context, please use SimpleTest.requestCompleteLog() in your test 19:12:08 INFO - 1933 INFO TEST-PASS | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to bottom when scrollTop is already bottom-most (page): unexpected scrollLeft 19:12:08 INFO - 1934 INFO TEST-PASS | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to left (page): unexpected scrollTop 19:12:08 INFO - 1935 INFO TEST-UNEXPECTED-FAIL | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to left (page): unexpected scrollLeft - got 2800, expected 0 19:12:08 INFO - 1936 INFO TEST-INFO | dumping last 1 message(s) 19:12:08 INFO - 1937 INFO TEST-INFO | if you need more context, please use SimpleTest.requestCompleteLog() in your test 19:12:08 INFO - 1938 INFO TEST-PASS | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to left when scrollLeft is already left-most (page): unexpected scrollTop 19:12:08 INFO - 1939 INFO TEST-UNEXPECTED-FAIL | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to left when scrollLeft is already left-most (page): unexpected scrollLeft - got 2800, expected 0 19:12:08 INFO - 1940 INFO TEST-INFO | dumping last 1 message(s) 19:12:08 INFO - 1941 INFO TEST-INFO | if you need more context, please use SimpleTest.requestCompleteLog() in your test 19:12:08 INFO - 1942 INFO TEST-PASS | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to right (page): unexpected scrollTop 19:12:08 INFO - 1943 INFO TEST-UNEXPECTED-FAIL | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to right (page): unexpected scrollLeft - got 0, expected 2800 19:12:08 INFO - 1944 INFO TEST-INFO | dumping last 1 message(s) 19:12:08 INFO - 1945 INFO TEST-INFO | if you need more context, please use SimpleTest.requestCompleteLog() in your test 19:12:08 INFO - 1946 INFO TEST-PASS | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to right when scrollLeft is already right-most (page): unexpected scrollTop 19:12:08 INFO - 1947 INFO TEST-UNEXPECTED-FAIL | /tests/dom/events/test/test_wheel_default_action.html | doTestWholeScroll, try whole-scroll to right when scrollLeft is already right-most (page): unexpected scrollLeft - got 0, expected 2800 full log: https://tbpl.mozilla.org/php/getParsedLog.php?id=39530884&tree=Cedar&full=1#error0
Hi Masayuki, Do you have any cycles to help out here? Many of the test failures occur along with this assertion: 08:32:10 INFO - [Parent 1478] ###!!! ASSERTION: The sign of overflowDeltaY is different from the scroll direction: 'aEvent->overflowDeltaY == 0 || (aEvent->overflowDeltaY > 0) == (aEvent->deltaY > 0)', file /builds/slave/ced-osx64-d-000000000000000000/build/dom/events/EventStateManager.cpp, line 2535 08:32:10 INFO - mozilla::EventStateManager::PostHandleEvent(nsPresContext*, mozilla::WidgetEvent*, nsIFrame*, nsEventStatus*) [dom/events/EventStateManager.cpp:2960] 08:32:10 INFO - PresShell::HandleEventInternal(mozilla::WidgetEvent*, nsEventStatus*) [layout/base/nsPresShell.cpp:7551] 08:32:10 INFO - PresShell::HandlePositionedEvent(nsIFrame*, mozilla::WidgetGUIEvent*, nsEventStatus*) [layout/base/nsPresShell.cpp:7268] 08:32:10 INFO - PresShell::HandleEvent(nsIFrame*, mozilla::WidgetGUIEvent*, bool, nsEventStatus*) [layout/base/nsPresShell.cpp:7065] 08:32:10 INFO - nsViewManager::DispatchEvent(mozilla::WidgetGUIEvent*, nsView*, nsEventStatus*) [obj-firefox/dist/include/nsCOMPtr.h:536] 08:32:10 INFO - nsView::HandleEvent(mozilla::WidgetGUIEvent*, bool) [obj-firefox/dist/include/nsAutoPtr.h:870] 08:32:10 INFO - nsChildView::DispatchEvent(mozilla::WidgetGUIEvent*, nsEventStatus&) [widget/cocoa/nsChildView.mm:1768] 08:32:10 INFO - nsDOMWindowUtils::SendWheelEvent(float, float, double, double, double, unsigned int, int, int, int, unsigned int) [dom/base/nsDOMWindowUtils.cpp:902] 08:32:10 INFO - NS_InvokeByIndex [xpcom/reflect/xptcall/src/md/unix/xptcinvoke_x86_64_unix.cpp:162] 08:32:10 INFO - CallMethodHelper::Call() [js/xpconnect/src/XPCWrappedNative.cpp:1730] 08:32:10 INFO - XPCWrappedNative::CallMethod(XPCCallContext&, XPCWrappedNative::CallMode) [js/xpconnect/src/XPCWrappedNative.cpp:1697] 08:32:10 INFO - XPC_WN_CallMethod(JSContext*, unsigned int, JS::Value*) [js/xpconnect/src/XPCWrappedNativeJSOps.cpp:1273] 08:32:10 INFO - js::CallJSNative(JSContext*, bool (*)(JSContext*, unsigned int, JS::Value*), JS::CallArgs const&) [js/src/jscntxtinlines.h:239] 08:32:10 INFO - js::Invoke(JSContext*, JS::CallArgs, js::MaybeConstruct) [js/src/vm/Interpreter.cpp:448] 08:32:10 INFO - js_fun_apply(JSContext*, unsigned int, JS::Value*) [js/src/jsfun.cpp:1139] 08:32:10 INFO - js::CallJSNative(JSContext*, bool (*)(JSContext*, unsigned int, JS::Value*), JS::CallArgs const&) [js/src/jscntxtinlines.h:239] 08:32:10 INFO - js::Invoke(JSContext*, JS::CallArgs, js::MaybeConstruct) [js/src/vm/Interpreter.cpp:448] 08:32:10 INFO - js_fun_call(JSContext*, unsigned int, JS::Value*) [js/src/jsfun.cpp:1068] We can disable this test on 10.9 if you don't have time to investigate.
Flags: needinfo?(masayuki)
That's very odd. The code should behave same across all platforms. Steven, do you have idea of the reason why synthesized widget wheel events handled differently only on 10.9? If you don't have any idea, I'll investigate next week.
Flags: needinfo?(masayuki) → needinfo?(smichaud)
> Steven, do you have idea of the reason why synthesized widget wheel events handled > differently only on 10.9? No, I don't. How many machines has this failure happened on? Might it have something to do with the way they're set up?
Flags: needinfo?(smichaud)
This occurs on RelEng's osx 10.9 test slaves. It could have something to do with the machine config, not sure what.
Kim, do you know what the "Scroll Direction" setting on the OSX 10.9 slaves is, and does it differ from the 10.8 slaves? If it's set to "natural direction" it causes the mouse wheel to scroll in the opposite direction; I wonder if that's what is happening here.
Flags: needinfo?(kmoir)
(In reply to Jonathan Griffin (:jgriffin) from comment #5) > If it's set to "natural direction" it causes the mouse wheel to scroll in > the opposite direction; I wonder if that's what is happening here. I think that it must not be a matter since sendWheelEvent() does NOT synthesize a wheel event with a native event. So, the given arguments are the all of the things.
I can't tell the direction of the OSX 10.9 slaves since it doesn't have a mouse attached you can't look at the preferences for that setting via the gui (I checked). From the command line it is (enabled natural direction shown below) defaults find com.apple.swipescrolldirection Found 1 keys in domain 'Apple Global Domain': { "com.apple.swipescrolldirection" = 1; from my laptop but this value doesn't appear in defaults until you toggle between the two. This value doesn't appear the slave because you have to toggle it to appear in defaults. I believe natural direction is the default setting on mavericks.
Flags: needinfo?(kmoir)
Status: NEW → RESOLVED
Closed: 10 years ago
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.