Bug 77628 – REGRESSION (r106488): fast/events/remove-target-with-shadow-in-drag.html fails

• Status: RESOLVED
• Resolution: FIXED
• Component: Tools / Tests

[János Badics]

fast/events/remove-target-with-shadow-in-drag.html started failing on Qt Linux Release in r106488 <http://trac.webkit.org/changeset/106488>.

http://build.webkit.org/results/Qt%20Linux%20Release/r106487%20(42771)/results.html passed
http://build.webkit.org/results/Qt%20Linux%20Release/r106488%20(42772)/results.html failed

[Csaba Osztrogonác]

We can't comment the original bug, becasue it is security bug. Could you check if we should add platform specific expected results or if it is a real bug.

fast/events/remove-target-with-shadow-in-drag.html introduced in r106488, but fails almost on all platform with the following diffs:

Chromium:
--- E:\b\chromium-win-release-tests\build\layout-test-results\fast/events/remove-target-with-shadow-in-drag-expected.txt 
+++ E:\b\chromium-win-release-tests\build\layout-test-results\fast/events/remove-target-with-shadow-in-drag-actual.txt 
@@ -1,3 +1,3 @@
-CONSOLE MESSAGE: line 34: NOT_FOUND_ERR: DOM Exception 8: An attempt was made to reference a Node in a context where it does not exist.
-CONSOLE MESSAGE: line 34: NOT_FOUND_ERR: DOM Exception 8: An attempt was made to reference a Node in a context where it does not exist.
+CONSOLE MESSAGE: line 34: Uncaught Error: NOT_FOUND_ERR: DOM Exception 8
+CONSOLE MESSAGE: line 34: Uncaught Error: NOT_FOUND_ERR: DOM Exception 8
 PASS. DRT didn't crash.


GTK:
--- /home/slave/webkitgtk/gtk-linux-64-release/build/layout-test-results/fast/events/remove-target-with-shadow-in-drag-expected.txt 
+++ /home/slave/webkitgtk/gtk-linux-64-release/build/layout-test-results/fast/events/remove-target-with-shadow-in-drag-actual.txt 
@@ -1,3 +1,2 @@
-CONSOLE MESSAGE: line 34: NOT_FOUND_ERR: DOM Exception 8: An attempt was made to reference a Node in a context where it does not exist.
 CONSOLE MESSAGE: line 34: NOT_FOUND_ERR: DOM Exception 8: An attempt was made to reference a Node in a context where it does not exist.
 PASS. DRT didn't crash.


Qt:
--- /ramdisk/qt-linux-64-release/build/layout-test-results/fast/events/remove-target-with-shadow-in-drag-expected.txt 
+++ /ramdisk/qt-linux-64-release/build/layout-test-results/fast/events/remove-target-with-shadow-in-drag-actual.txt 
@@ -1,3 +1,1 @@
-CONSOLE MESSAGE: line 34: NOT_FOUND_ERR: DOM Exception 8: An attempt was made to reference a Node in a context where it does not exist.
-CONSOLE MESSAGE: line 34: NOT_FOUND_ERR: DOM Exception 8: An attempt was made to reference a Node in a context where it does not exist.
 PASS. DRT didn't crash.

[Csaba Osztrogonác]

Additionally the new test made fast/events/resize-subframe.html timeout on Qt-WK2. Maybe on Mac-WK2 too, but Mac-WK2 bot is dead long long time ago, so I don't know.

You can easily reproduce it with:
$ run-tests-in-xvfb.sh -2 fast/events/remove-target-with-shadow-in-drag.html fast/events/resize-subframe.html

I skipped the new fast/events/remove-target-with-shadow-in-drag.html locally,
and then fast/events/resize-subframe.html works fine. So this is a new regression or there is a serious sideeffect bug in WebKitTestRunner somewhere.

[János Badics]

Created attachment 125124 [details]
REGRESSION: fast/events/remove-target-with-shadow-in-drag.html failing on Qt Linux Release

[Csaba Osztrogonác]

Comment on attachment 125124 [details]
REGRESSION: fast/events/remove-target-with-shadow-in-drag.html failing on Qt Linux Release

r-, I bet you uploaded it accidentally.

[János Badics]

Created attachment 125129 [details]
proposed patch

Added expectations to fast/events/remove-target-with-shadow-in-drag.html

[Csaba Osztrogonác]

Comment on attachment 125129 [details]
proposed patch

r- again. Did you add only changelog intentionally again?

[János Badics]

Created attachment 125131 [details]
proposed patch

put fast/events/remove-target-with-shadow-in-drag.html to the qt-wk2 skipped list;
added expectations to fast/events/remove-target-with-shadow-in-drag.html

[Csaba Osztrogonác]

Comment on attachment 125131 [details]
proposed patch

We skipped the new test on Qt-WK2 not to make any test timeout and added Qt specific expected file:  http://trac.webkit.org/changeset/106553

Have you got any idea how can this new test make the following test timeout? And is this committed expected file correct or not?

[Csaba Osztrogonác]

The timeout isn't WK2 specific, but Qt5 specific, so I moved the skipped test from qt-wk2 to qt-5.0

[Ryosuke Niwa]

Exceptions don't really matter. So you can check in platform-specific results. Sorry about the failures. I didn't realize this.

[Ryosuke Niwa]

I think we've rebaselined on all platforms.