Bug 36642 – fast/replaced/border-radius-clip.html failed on Snow Leopard Debug bot

• Status: REOPENED
• Component: Tools / Tests

[Eric Seidel (no email)]

fast/replaced/border-radius-clip.html failed on Snow Leopard Debug bot

I believe this is failing intermittently.  Here is one failure:
http://build.webkit.org/results/SnowLeopard%20Intel%20Leaks/r56605%20(5249)/fast/replaced/border-radius-clip-diffs.txt
--- /Volumes/Data/WebKit-BuildSlave/snowleopard-intel-leaks/build/layout-test-results/fast/replaced/border-radius-clip-expected.txt	2010-03-25 23:43:57.000000000 -0700
+++ /Volumes/Data/WebKit-BuildSlave/snowleopard-intel-leaks/build/layout-test-results/fast/replaced/border-radius-clip-actual.txt	2010-03-25 23:43:57.000000000 -0700
@@ -1,8 +1,8 @@
-layer at (0,0) size 785x952
+layer at (0,0) size 785x902
   RenderView at (0,0) size 785x600
-layer at (0,0) size 785x952
-  RenderBlock {HTML} at (0,0) size 785x928
-    RenderBody {BODY} at (8,8) size 769x912
+layer at (0,0) size 785x902
+  RenderBlock {HTML} at (0,0) size 785x878
+    RenderBody {BODY} at (8,8) size 769x862
       RenderText {#text} at (0,0) size 457x18
         text run at (0,0) width 457: "All of the replaced elements below should be clipped to a border-radius. "
       RenderBR {BR} at (0,0) size 0x0
@@ -24,11 +24,11 @@
       RenderText {#text} at (304,366) size 4x18
         text run at (304,366) width 4: " "
       RenderBR {BR} at (0,0) size 0x0
-      RenderText {#text} at (300,520) size 4x18
-        text run at (300,520) width 4: " "
+      RenderText {#text} at (100,470) size 4x18
+        text run at (100,470) width 4: " "
       RenderBR {BR} at (0,0) size 0x0
-      RenderEmbeddedObject {EMBED} at (0,538) size 556x406 [bgcolor=#00FF00] [border: (3px solid #000000)]
+      RenderEmbeddedObject {EMBED} at (0,488) size 556x406 [bgcolor=#00FF00] [border: (3px solid #000000)]
       RenderText {#text} at (0,0) size 0x0
       RenderText {#text} at (0,0) size 0x0
-layer at (8,392) size 300x150
-  RenderVideo {VIDEO} at (0,384) size 300x150
+layer at (8,392) size 100x100
+  RenderVideo {VIDEO} at (0,384) size 100x100

This may be related to bug 29824.

[Eric Seidel (no email)]

http://trac.webkit.org/browser/trunk/LayoutTests/fast/replaced/border-radius-clip.html was added 12 months ago by Hyatt.

[Eric Seidel (no email)]

Failed again:
http://build.webkit.org/results/SnowLeopard%20Intel%20Release%20(Tests)/r56619%20(7256)/fast/replaced/border-radius-clip-diffs.txt

[Eric Seidel (no email)]

Failed again:
http://build.webkit.org/results/SnowLeopard%20Intel%20Release%20(Tests)/r56629%20(7265)/fast/replaced/border-radius-clip-diffs.txt

This seems to be a pretty common flaky test.

Is the video dynamically sized and failing to load?

[Eric Seidel (no email)]

@eric:  Looks like the <video> tag has completed the load.  It looks like commonly it does not complete the load on time and thus does not resize itself.

The resource in question:
http://trac.webkit.org/browser/trunk/LayoutTests/fast/replaced/resources/sample.gif

Does that sound right to anyone else?

[Eric Seidel (no email)]

Seems to fail every 3rd build or so:
http://build.webkit.org/results/SnowLeopard%20Intel%20Release%20(Tests)/r56631%20(7267)/fast/replaced/border-radius-clip-diffs.txt

[Eric Seidel (no email)]

Failed again:
http://build.webkit.org/results/SnowLeopard%20Intel%20Leaks/r56735%20(5356)/fast/replaced/border-radius-clip-pretty-diff.html

[Eric Seidel (no email)]

Failed again on the commit bot just now.  I recommend skipping this test.

[Eric Seidel (no email)]

Created attachment 52053 [details]
Patch

[Adam Barth]

Comment on attachment 52053 [details]
Patch

ok.  We need a better way to track these tests.

[Eric Seidel (no email)]

Agreed.  When the new-run-webkit-tests messiah comes at least we'll have a way to mark tests as flaky in the expectations instead of having to just skip them.

[WebKit Commit Bot]

Comment on attachment 52053 [details]
Patch

Clearing flags on attachment: 52053

Committed r56898: <http://trac.webkit.org/changeset/56898>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Simon Fraser (smfr)]

Test uses Flash. Bad test.

[Ryosuke Niwa]

Turns out that this test is still flaky :(