Title: [121703] trunk/LayoutTests
- Revision
- 121703
- Author
- alexis.men...@openbossa.org
- Date
- 2012-07-02 12:56:36 -0700 (Mon, 02 Jul 2012)
Log Message
fast/box-decoration-break/box-decoration-break-rendering.html failing on mac bots
https://bugs.webkit.org/show_bug.cgi?id=89620
Reviewed by Simon Fraser.
Remove the border-radius from the test case as it can produce differences between
the test case and the reference html. At first we thought the 3 pixels difference
was only happening on Mac but after investigation it also happens in Qt (at least).
It was not triggered because ImageDiff was called with some tolerance. Turning the
tolerance to 0 on Qt also trigger some pixel differences.
The differences on how the radius is drawn can be explained by the fact that
the reference html and the test case uses two different paths for drawing the border.
The reference on the slice case only draw three borders which leads into the contruction
of three different drawing paths that we use to clip and clip out. We then draw the border with
antialiasing off as we have constructed the outer and inner paths pretty accurately.
The test case in the other hand uses the fast path for drawing the borders. The slice effect
is in fact is happening because the graphics context is clipped to the current InlineFlowBox rect but still
the border has 4 edges and its drawing will happen using the following : create a path with
an outer rounded rectangle and an inner rounded rectangle. The path is then filled with RULE_EVENODD
and anti-aliasing turn on to achieve a nice looking effect.
To conclude the pixel differences are the fact of the two different drawing techniques used here. Fast
drawing and anti-aliasing to get a decent result or more accurate algorithm but slower. The original test
was not about testing the radiuses of the borders but rather if the borders are cloned or sliced. Even
removing the radius still cover the feature.
* fast/box-decoration-break/box-decoration-break-rendering-expected.html:
* fast/box-decoration-break/box-decoration-break-rendering.html:
* platform/mac/Skipped:
Modified Paths
Diff
Modified: trunk/LayoutTests/ChangeLog (121702 => 121703)
--- trunk/LayoutTests/ChangeLog 2012-07-02 19:46:16 UTC (rev 121702)
+++ trunk/LayoutTests/ChangeLog 2012-07-02 19:56:36 UTC (rev 121703)
@@ -1,3 +1,36 @@
+2012-07-02 Alexis Menard <alexis.men...@openbossa.org>
+
+ fast/box-decoration-break/box-decoration-break-rendering.html failing on mac bots
+ https://bugs.webkit.org/show_bug.cgi?id=89620
+
+ Reviewed by Simon Fraser.
+
+ Remove the border-radius from the test case as it can produce differences between
+ the test case and the reference html. At first we thought the 3 pixels difference
+ was only happening on Mac but after investigation it also happens in Qt (at least).
+ It was not triggered because ImageDiff was called with some tolerance. Turning the
+ tolerance to 0 on Qt also trigger some pixel differences.
+
+ The differences on how the radius is drawn can be explained by the fact that
+ the reference html and the test case uses two different paths for drawing the border.
+ The reference on the slice case only draw three borders which leads into the contruction
+ of three different drawing paths that we use to clip and clip out. We then draw the border with
+ antialiasing off as we have constructed the outer and inner paths pretty accurately.
+ The test case in the other hand uses the fast path for drawing the borders. The slice effect
+ is in fact is happening because the graphics context is clipped to the current InlineFlowBox rect but still
+ the border has 4 edges and its drawing will happen using the following : create a path with
+ an outer rounded rectangle and an inner rounded rectangle. The path is then filled with RULE_EVENODD
+ and anti-aliasing turn on to achieve a nice looking effect.
+
+ To conclude the pixel differences are the fact of the two different drawing techniques used here. Fast
+ drawing and anti-aliasing to get a decent result or more accurate algorithm but slower. The original test
+ was not about testing the radiuses of the borders but rather if the borders are cloned or sliced. Even
+ removing the radius still cover the feature.
+
+ * fast/box-decoration-break/box-decoration-break-rendering-expected.html:
+ * fast/box-decoration-break/box-decoration-break-rendering.html:
+ * platform/mac/Skipped:
+
2012-07-02 Dan Bernstein <m...@apple.com>
<rdar://problem/11787030> In vertical writing modes, child following float-clearing block has incorrect logical top
Modified: trunk/LayoutTests/fast/box-decoration-break/box-decoration-break-rendering-expected.html (121702 => 121703)
--- trunk/LayoutTests/fast/box-decoration-break/box-decoration-break-rendering-expected.html 2012-07-02 19:46:16 UTC (rev 121702)
+++ trunk/LayoutTests/fast/box-decoration-break/box-decoration-break-rendering-expected.html 2012-07-02 19:56:36 UTC (rev 121703)
@@ -17,10 +17,6 @@
border-right: 0px solid red;
border-left: 2px solid red;
border-bottom: 2px solid red;
- border-top-left-radius: 4px;
- border-top-right-radius: 0px;
- border-bottom-right-radius: 0px;
- border-bottom-left-radius: 4px;
background: yellow;
box-shadow: 0px 1px 3px dimgrey;
}
@@ -32,10 +28,6 @@
border-right: 2px solid red;
border-left: 0px solid red;
border-bottom: 2px solid red;
- border-top-left-radius: 0px;
- border-top-right-radius: 4px;
- border-bottom-right-radius: 4px;
- border-bottom-left-radius: 0px;
background: yellow;
box-shadow: 0px 1px 3px dimgrey;
}
@@ -44,7 +36,6 @@
margin: 0px;
padding: 0px;
border: 2px solid red;
- border-radius: 4px;
background: yellow;
box-shadow: 0px 1px 3px dimgrey;
}
Modified: trunk/LayoutTests/fast/box-decoration-break/box-decoration-break-rendering.html (121702 => 121703)
--- trunk/LayoutTests/fast/box-decoration-break/box-decoration-break-rendering.html 2012-07-02 19:46:16 UTC (rev 121702)
+++ trunk/LayoutTests/fast/box-decoration-break/box-decoration-break-rendering.html 2012-07-02 19:56:36 UTC (rev 121703)
@@ -13,7 +13,6 @@
}
span highlight {
border: 2px solid red;
- border-radius: 4px;
background: yellow;
box-shadow: 0px 1px 3px dimgrey;
}
Modified: trunk/LayoutTests/platform/mac/Skipped (121702 => 121703)
--- trunk/LayoutTests/platform/mac/Skipped 2012-07-02 19:46:16 UTC (rev 121702)
+++ trunk/LayoutTests/platform/mac/Skipped 2012-07-02 19:56:36 UTC (rev 121703)
@@ -894,9 +894,6 @@
# https://bugs.webkit.org/show_bug.cgi?id=88736
http/tests/security/mixedContent/blob-url-in-iframe.html
-# https://bugs.webkit.org/show_bug.cgi?id=89620
-fast/box-decoration-break/box-decoration-break-rendering.html
-
# https://bugs.webkit.org/show_bug.cgi?id=89680
fast/canvas/canvas-imageSmoothingEnabled-repaint.html
_______________________________________________
webkit-changes mailing list
webkit-changes@lists.webkit.org
http://lists.webkit.org/mailman/listinfo.cgi/webkit-changes
