Bug 10171 – REGRESSION: failing layout test: fast/events/objc-event-api.html

• Status: RESOLVED
• Resolution: FIXED
• Component: DOM

[Geoffrey Garen]

This check-in caused the regression:

Index: WebKitTools/ChangeLog
===================================================================
--- WebKitTools/ChangeLog       (revision 15045)
+++ WebKitTools/ChangeLog       (revision 15046)
@@ -1,3 +1,14 @@
[phone]  Jonas Witt  <[e-mail]>
+
+        Reviewed by Darin.
+
+        Bug 9579: LayoutTests/fast/events/objc-event-api.html failing due to screenY output
+        http://bugzilla.opendarwin.org/show_bug.cgi?id=9579
+
+        Report screenY values as (height of zero screen - screenY)
+
+        * DumpRenderTree/EventSendingController.m: (-[EventSendingController handleEvent:]):
+
 2006-06-25  Timothy Hatcher  <[e-mail]>
 
         Reviewed by Geoff.

Here's the diff:

--- /Users/buildbot/Desktop/BuildData/WebKit-BuildSlave/post-commit-powerpc-mac-os-x/build/LayoutTests/fast/events/objc-event-api-expected.txt	2006-07-31 01:46:45.000000000 -0700
+++ layout-test-results/fast/events/objc-event-api-actual.txt	2006-07-31 02:43:06.000000000 -0700
@@ -274,8 +274,8 @@
   detail:        0
   view:          OK (document: OK)
   button:        0
-  screenX:       0
-  screenY:       0 (flipped)
+  screenX:       -1
+  screenY:       -1 (flipped)
   clientX:       -1
   clientY:       601
   modifier keys: c:0 s:0 a:0 m:0

[Darin Adler]

Strangely, on my local machine the results I get are -1 for screenX but 0 for screenY. The expected results have 0 for both, and the buildbot results have -1 for both.

[Darin Adler]

Reading the test code more closely, I don't see how flipping the screen Y coordinate could possibly be correct.

[Darin Adler]

Created attachment 9789 [details]
patch to fix the issue, which was a flaw in DumpRenderTree

[Maciej Stachowiak]

Comment on attachment 9789 [details]
patch to fix the issue, which was a flaw in DumpRenderTree

r=me

[Darin Adler]

Committed revision 15731.