GitLab

https://bugs.webkit.org/show_bug.cgi?id=66345

Reviewed by James Robinson.

bulk-optimize-baselines is now a superset of optimize-baselines, so
they can be combined.

* Scripts/webkitpy/tool/commands/rebaseline.py:


git-svn-id: svn://svn.chromium.org/blink/trunk@93173 bbb929c8-8fbe-4397-9dbb-9b2b20218538

* platform/chromium-mac: Added.
* platform/chromium-mac/fast: Added.
* platform/chromium-mac/fast/events: Added.
* platform/chromium-mac/http: Added.
* platform/chromium-mac/http/tests: Added.
* platform/chromium-mac/http/tests/workers: Added.
* platform/chromium-mac/storage: Added.
* platform/chromium-mac/storage/domstorage: Added.


git-svn-id: svn://svn.chromium.org/blink/trunk@93172 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Reviewed by Dan Bernstein.

In r93140 the temporary font created for a remote CSS font face
was created without being retained. We need to protect the
font from being purged while we hold onto it, so add a font cache
purge preventer.

* css/CSSFontFaceSource.cpp:
(WebCore::CSSFontFaceSource::getFontData):

* platform/chromium/test_expectations.txt:
Unskipped.

git-svn-id: svn://svn.chromium.org/blink/trunk@93171 bbb929c8-8fbe-4397-9dbb-9b2b20218538

which WebKitTestRunner does not support yet.

* platform/wk2/Skipped:


git-svn-id: svn://svn.chromium.org/blink/trunk@93170 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=66177

Reviewed by James Robinson.

* DEPS:
* WebKit.gyp:
* public/WebScrollbar.h:


git-svn-id: svn://svn.chromium.org/blink/trunk@93169 bbb929c8-8fbe-4397-9dbb-9b2b20218538

git-svn-id: svn://svn.chromium.org/blink/trunk@93168 bbb929c8-8fbe-4397-9dbb-9b2b20218538

git-svn-id: svn://svn.chromium.org/blink/trunk@93167 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=66334

Reviewed by Tony Chang.

This patch moves us closer to switching the Chromium port from
CoreGraphics to Skia.  This patch teaches our tools that the current
Chromium Mac configuration is actually chromium-cg-mac, which makes
room for a future chromium-mac configuration that uses Skia.

Shortly after this patch lands, I'll do a server-side move of the
chromium-mac and chromium-mac-leopard directories to chromium-cg-mac
and chromium-cg-mac-leopard.

* BuildSlaveSupport/build.webkit.org-config/config.json:
* Scripts/webkitpy/common/checkout/baselineoptimizer_unittest.py:
* Scripts/webkitpy/common/config/build.py:
* Scripts/webkitpy/common/config/build_unittest.py:
* Scripts/webkitpy/layout_tests/port/base.py:
* Scripts/webkitpy/layout_tests/port/builders.py:
* Scripts/webkitpy/layout_tests/port/chromium.py:
* Scripts/webkitpy/layout_tests/port/chromium_mac.py:
* Scripts/webkitpy/layout_tests/port/chromium_mac_unittest.py:
* Scripts/webkitpy/layout_tests/port/dryrun.py:
* Scripts/webkitpy/layout_tests/port/factory.py:
* Scripts/webkitpy/layout_tests/port/factory_unittest.py:
* Scripts/webkitpy/layout_tests/port/google_chrome_unittest.py:
* Scripts/webkitpy/layout_tests/port/mock_drt_unittest.py:
* Scripts/webkitpy/to_be_moved/rebaseline_chromium_webkit_tests.py:
* Scripts/webkitpy/to_be_moved/rebaseline_chromium_webkit_tests_unittest.py:
* Scripts/webkitpy/tool/commands/rebaseline_unittest.py:
* TestResultServer/static-dashboards/flakiness_dashboard.html:


git-svn-id: svn://svn.chromium.org/blink/trunk@93166 bbb929c8-8fbe-4397-9dbb-9b2b20218538

chromium builds.

Also mark 2 repaint tests as failing on mac.

* platform/chromium/test_expectations.txt:

git-svn-id: svn://svn.chromium.org/blink/trunk@93165 bbb929c8-8fbe-4397-9dbb-9b2b20218538

because of the win baselines added in r93139.

* platform/chromium/test_expectations.txt:

git-svn-id: svn://svn.chromium.org/blink/trunk@93164 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=66324

Reviewed by Adam Barth.

* BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/base.js: Added base.relativizeTime method.
* BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/base_unittests.js: Added unit tests.
* BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/ui/notifications.js: Wired up Time widget to use it.

git-svn-id: svn://svn.chromium.org/blink/trunk@93163 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The Windows results match the cross-platform results, not the Mac results.

* platform/win/inspector/debugger/scripts-panel-expected.txt: Added.

git-svn-id: svn://svn.chromium.org/blink/trunk@93162 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=66318

Reviewed by Adam Barth.

* BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/summary.js: Added a simplistic updating button.
* BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/ui/notifications.js: Added ways to check for
    equality of ui.notifications.FailingTest, and whether ui.notifications.TestFailures already contains a given failure analysis.
* BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/ui/notifications_unittests.js: Added tests.

git-svn-id: svn://svn.chromium.org/blink/trunk@93161 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Namespace the now public enum as FontCache::ShouldRetain.

* platform/graphics/chromium/FontCacheChromiumWin.cpp:
(WebCore::GetLastResortFallbackFontProcData::GetLastResortFallbackFontProcData):

git-svn-id: svn://svn.chromium.org/blink/trunk@93160 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The Windows results match the cross-platform results, not the Mac results.

* platform/win/compositing/iframes/overlapped-iframe-iframe-expected.txt: Copied from compositing/iframes/overlapped-iframe-iframe-expected.txt.

git-svn-id: svn://svn.chromium.org/blink/trunk@93159 bbb929c8-8fbe-4397-9dbb-9b2b20218538

* platform/win/Skipped: Added fast/repaint/background-scaling.html.

git-svn-id: svn://svn.chromium.org/blink/trunk@93158 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=66251

Canvas resizing was slow due to re-allocation of the ImageBuffer on
each size change (width or height).  This was introduced inadvertently
by calls to isAccelerated() during canvas reset().  Since we won't 
know if we have successfully accelerated until ImageBuffer creation,
move the compositor invalidation to createImageBuffer() as well.
This patch also attempts to unify the Skia and CG accelerated canvas
paths.  The DrawingBuffer used by the Skia path is now owned by 
ImageBuffer[Skia], similar to how the IOSurface is owned by
ImageBuffer[CG].  Some of the logic for when to accelerate was moved
into HTMLCanvasElement and unified with the CG path.  Acceleration is
also now enabled by the same "Accelerated" ImageBuffer create flag
used by the CG path.  DrawingBuffer is now re-created even for a 
same-size change (same as the memory buffer), but we speed it up
by calling GraphicsContext3D::texImage2D() with a NULL pixels ptr
instead of GraphicsContext3D::teximage2DResourceSafe() (no need to
clear it, since it's done with a glClear anyway).

Reviewed by Kenneth Russell.

Covered by existing tests in fast/canvas and canvas/philip.

* html/HTMLCanvasElement.cpp:
(WebCore::HTMLCanvasElement::reset):
Don't call isAccelerated() from reset(), since we don't want to
inadvertently create the ImageBuffer.  Also, since we won't know if
we have successfully accelerated until createImageBuffer() is
called, defer the compositor invalidation to createImageBuffer() as
well.
(WebCore::HTMLCanvasElement::shouldAccelerate):
Unify the CG and Skia ports' logic for when to accelerate (Skia
logic moved in from CanvasRenderingContext2D).
(WebCore::HTMLCanvasElement::createImageBuffer):
Use the unified shouldAccelerate() logic, and pass it as a flag
to ImageBuffer.  Do compositor invalidation as well.
* html/HTMLCanvasElement.h:
(WebCore::HTMLCanvasElement::hasCreatedImageBuffer):
Expose hasCreatedImageBuffer() publically, so
CanvasRenderingContext2D::isAccelerated() doesn't inadvertently create
it.
* html/canvas/CanvasRenderingContext2D.cpp:
(WebCore::CanvasRenderingContext2D::CanvasRenderingContext2D):
(WebCore::CanvasRenderingContext2D::~CanvasRenderingContext2D):
Remove all acceleration setting and resetting, since it's now done
during ImageBuffer creation.
(WebCore::CanvasRenderingContext2D::isAccelerated):
Check if the image buffer was created, so we don't inadvertently
create it here.
(WebCore::CanvasRenderingContext2D::paintsIntoCanvasBuffer):
Don't call GraphicsContext3D::paintsIntoCanvasBuffer(), since its
WebViewImpl may be null.  Ask the render tree instead.
(WebCore::CanvasRenderingContext2D::reset):
Don't reset acceleration here, since we don't own it anymore.
(WebCore::CanvasRenderingContext2D::platformLayer):
Call into the ImageBuffer to get our PlatformLayer.
* html/canvas/CanvasRenderingContext2D.h:
Remove m_drawingBuffer, and all acceleration-related calls.
* platform/graphics/GraphicsContext.cpp:
* platform/graphics/GraphicsContext.h:
Remove setGraphicsContext3D(), and paintsIntoImageBuffer().
* platform/graphics/ImageBuffer.cpp:
(WebCore::ImageBuffer::platformLayer):
Implement a dummy platformLayer() call for non-skia ports.
* platform/graphics/ImageBuffer.h:
Declare a platformLayer() call (USE(ACCELERATED_COMPOSITING) only).
* platform/graphics/chromium/ImageBufferDataSkia.h:
Add a DrawingBuffer data member.
* platform/graphics/gpu/DrawingBuffer.cpp:
(WebCore::DrawingBuffer::reset):
Use texImage2D() with a NULL ptr, not texImage2DResourceSafe().
Since we immediately clear the framebuffer via glClear(), this whole
resource safe business is overkill.
* platform/graphics/skia/GraphicsContextSkia.cpp:
Remove setGraphicsContext3D() and paintsIntoImageBuffer().  This
functionality is handled by ImageBuffer now.
* platform/graphics/skia/ImageBufferSkia.cpp:
(WebCore::ImageBuffer::ImageBuffer):
Give the ImageBufferData ownership of the DrawingBuffer.
(WebCore::ImageBuffer::platformLayer):
Implement an accessor for the DrawingBuffer's PlatformLayer.
* platform/graphics/skia/PlatformContextSkia.cpp:
(WebCore::PlatformContextSkia::setGraphicsContext3D):
* platform/graphics/skia/PlatformContextSkia.h:
Remove isPathSkiaSafe() extern (unused).
Remove paintsIntoImageBuffer() (now unused).  Remove IntSize param
from setGraphicsContext3D() (unused).


git-svn-id: svn://svn.chromium.org/blink/trunk@93157 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Include the proper includes!

* platform/graphics/FontCache.h:

git-svn-id: svn://svn.chromium.org/blink/trunk@93156 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The basic strategy here is that once a test has failed or passed many times in a row we
never again consider it as a possibly flaky test. It's a simple strategy but seems to result
in many fewer false positives than our current behavior.

Fixes <http://webkit.org/b/66327> TestFailures page considers far too many tests to be
flaky, including tests that failed for a while but then were fixed

Reviewed by Dan Bates.

* BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/FlakyLayoutTestDetector.js:
(FlakyLayoutTestDetector): Initialize new _buildCount property, which is used to track the
number of non-too-many-failure builds we've seen.
(FlakyLayoutTestDetector.prototype.incorporateTestResults): Don't track tests which haven't
failed in the _maximumFailOrPassCount most recent builds. For other not-yet-considered-flaky
tests, keep track of how many times they have passed or failed in a row. If they pass or
fail more than _maximumFailOrPassCount times, consider them to be non-flaky. (Once a test is
considered flaky it doesn't matter how many times it passes or fails.)

* BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/FlakyLayoutTestDetector_unittests.js:
Added tests for the above.

git-svn-id: svn://svn.chromium.org/blink/trunk@93155 bbb929c8-8fbe-4397-9dbb-9b2b20218538

in debug on Linux.

* platform/chromium/test_expectations.txt:

git-svn-id: svn://svn.chromium.org/blink/trunk@93154 bbb929c8-8fbe-4397-9dbb-9b2b20218538

bots when they cycle.

* platform/chromium/test_expectations.txt:

git-svn-id: svn://svn.chromium.org/blink/trunk@93153 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=66144

Reviewed by Adam Barth.

* BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/run-unittests.html: Added notifications tests.
* BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/summary.js: Added.
* BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/ui/notifications.js: Added.
* BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/ui/notifications_unittests.js: Added.
* BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/styles/summary.css: Added.
* BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/summary.html: Added.

git-svn-id: svn://svn.chromium.org/blink/trunk@93152 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This just makes the troublesome enum public. If this
fixes the build I will file a follow-up bug to clean-up
these build issues and either keep the enum public and
cleanup other functions or restore the enum to private.
There is no harm in making the enum public right now.

* platform/graphics/FontCache.h:

git-svn-id: svn://svn.chromium.org/blink/trunk@93151 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=66080

Patch by Mihnea Ovidenie <mihnea@adobe.com> on 2011-08-16
Reviewed by Dean Jackson.

Source/WebCore:

Current version of the algorithm negates only the scaleX while
it should negate also scaleY and scaleZ when appropriate.

Test: animations/animation-matrix-negative-scale-unmatrix.html

* platform/graphics/transforms/TransformationMatrix.cpp:
(WebCore::decompose):

LayoutTests:

* animations/animation-matrix-negative-scale-unmatrix-expected.txt: Added.
* animations/animation-matrix-negative-scale-unmatrix.html: Added.

git-svn-id: svn://svn.chromium.org/blink/trunk@93150 bbb929c8-8fbe-4397-9dbb-9b2b20218538

* platform/graphics/FontCache.h:

git-svn-id: svn://svn.chromium.org/blink/trunk@93149 bbb929c8-8fbe-4397-9dbb-9b2b20218538

* media/video-controls-no-scripting-expected.txt:
* media/video-controls-no-scripting.html:
* platform/chromium/media/video-controls-no-scripting-expected.txt: Removed.


git-svn-id: svn://svn.chromium.org/blink/trunk@93148 bbb929c8-8fbe-4397-9dbb-9b2b20218538

* platform/graphics/FontCache.h:
* platform/graphics/chromium/FontCacheChromiumWin.cpp:
(WebCore::FontCache::fontDataFromDescriptionAndLogFont):
(WebCore::getLastResortFallbackFontProc):
(WebCore::FontCache::getLastResortFallbackFont):
* platform/graphics/win/FontCacheWin.cpp:
(WebCore::FontCache::fontDataFromDescriptionAndLogFont):
(WebCore::FontCache::getLastResortFallbackFont):

git-svn-id: svn://svn.chromium.org/blink/trunk@93147 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Fixes <http://webkit.org/b/66319> Placeholder text in text fields disappears when the field
is focused on Windows, which doesn't match Lion or Windows native text fields

Covered by existing tests (which are currently failing on the bots until this patch lands).

Reviewed by Dave Hyatt.

* rendering/RenderThemeSafari.h:
(WebCore::RenderThemeSafari::shouldShowPlaceholderWhenFocused):
* rendering/RenderThemeWin.h:
(WebCore::RenderThemeWin::shouldShowPlaceholderWhenFocused):
Added overrides to return true.

git-svn-id: svn://svn.chromium.org/blink/trunk@93146 bbb929c8-8fbe-4397-9dbb-9b2b20218538

* Scripts/webkitpy/common/config/committers.py:

git-svn-id: svn://svn.chromium.org/blink/trunk@93145 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Source/WebCore: 

Make region invalidation and repainting work properly. We do this by making RenderFlowThreads into
repaint containers and then issuing new repaints in the regions that contain the invalidated flow thread
content.

Reviewed by Simon Fraser.

Added new tests in fast/repaint.

* rendering/RenderFlowThread.cpp:
(WebCore::RenderFlowThread::repaintRectangleInRegions):
* rendering/RenderFlowThread.h:
* rendering/RenderObject.cpp:
(WebCore::RenderObject::enclosingRenderFlowThread):
(WebCore::RenderObject::containerForRepaint):
(WebCore::RenderObject::repaintUsingContainer):
* rendering/RenderObject.h:
* rendering/RenderView.cpp:
(WebCore::RenderView::RenderView):
(WebCore::RenderView::renderFlowThreadWithName):
* rendering/RenderView.h:
(WebCore::RenderView::hasRenderFlowThreads):

LayoutTests: 

Make regions repaint properly. Added new repaint tests.

Reviewed by Simon Fraser.

* fast/repaint/japanese-rl-selection-repaint-in-regions.html: Added.
* fast/repaint/line-flow-with-floats-in-regions.html: Added.
* fast/repaint/overflow-flipped-writing-mode-block-in-regions.html: Added.
* platform/mac/fast/repaint/japanese-rl-selection-repaint-in-regions-expected.png: Added.
* platform/mac/fast/repaint/japanese-rl-selection-repaint-in-regions-expected.txt: Added.
* platform/mac/fast/repaint/line-flow-with-floats-in-regions-expected.png: Added.
* platform/mac/fast/repaint/line-flow-with-floats-in-regions-expected.txt: Added.
* platform/mac/fast/repaint/overflow-flipped-writing-mode-block-in-regions-expected.png: Added.
* platform/mac/fast/repaint/overflow-flipped-writing-mode-block-in-regions-expected.txt: Added.



git-svn-id: svn://svn.chromium.org/blink/trunk@93144 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=66153

Reviewed by Dan Bernstein.

While a remote CSS font face is loading we fallback to a
temporary font. We don't want to retain the fallback font
because noone takes ownership of the temporary font. This
patch adds a way to get an uncached fallback font, which
plumbs the ShouldRetain enum through the different platform
implementations of getLastResortFallbackFont.

No new tests, no functional change.

* css/CSSFontFaceSource.cpp:
(WebCore::CSSFontFaceSource::getFontData):
* platform/graphics/FontCache.cpp:
(WebCore::FontCache::getNonRetainedLastResortFallbackFont):
* platform/graphics/FontCache.h:
* platform/graphics/chromium/FontCacheChromiumWin.cpp:
(WebCore::fontDataFromDescriptionAndLogFont):
(WebCore::GetLastResortFallbackFontProcData::GetLastResortFallbackFontProcData):
(WebCore::getLastResortFallbackFontProc):
(WebCore::FontCache::getLastResortFallbackFont):
* platform/graphics/chromium/FontCacheLinux.cpp:
(WebCore::FontCache::getLastResortFallbackFont):
* platform/graphics/freetype/FontCacheFreeType.cpp:
(WebCore::FontCache::getLastResortFallbackFont):
* platform/graphics/haiku/FontCacheHaiku.cpp:
(WebCore::FontCache::getLastResortFallbackFont):
* platform/graphics/mac/FontCacheMac.mm:
(WebCore::FontCache::getLastResortFallbackFont):
* platform/graphics/pango/FontCachePango.cpp:
(WebCore::FontCache::getLastResortFallbackFont):
* platform/graphics/qt/FontCacheQt.cpp:
(WebCore::FontCache::getLastResortFallbackFont):
* platform/graphics/win/FontCacheWin.cpp:
(WebCore::fontDataFromDescriptionAndLogFont):
(WebCore::FontCache::getLastResortFallbackFont):
* platform/graphics/wince/FontCacheWinCE.cpp:
* platform/graphics/wx/FontCacheWx.cpp:
(WebCore::FontCache::getLastResortFallbackFont):


git-svn-id: svn://svn.chromium.org/blink/trunk@93140 bbb929c8-8fbe-4397-9dbb-9b2b20218538

<canvas> doesn't use accelerated compositing on Windows like it does on Lion.

* platform/win/compositing/layer-creation/spanOverlapsCanvas-expected.txt: Copied from platform/mac-snowleopard/compositing/layer-creation/spanOverlapsCanvas-expected.txt.
* platform/win/fast/repaint/canvas-putImageData-expected.txt: Copied from platform/mac-snowleopard/fast/repaint/canvas-putImageData-expected.txt.

git-svn-id: svn://svn.chromium.org/blink/trunk@93139 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Add myself to the list of individual members in the WebKit Security Group.

        Rubber-stamped by Eric Seidel.

        * security/security-group-members.html:


git-svn-id: svn://svn.chromium.org/blink/trunk@93138 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=66226

Make ScrollView::paintOverhangAreas() use the ScrollbarTheme::nativeTheme() to
draw the overhang areas.
Move default implementation to ScrollbarThemeComposite::paintOverhangAreas().
Add a different implementation for Chromium Mac.

Patch by Alexei Svitkine <asvitkine@chromium.org> on 2011-08-16
Reviewed by Dimitri Glazkov.

No new tests since this is just refactoring code and adding a Chromium-specific path for overhang drawing.

* platform/ScrollView.cpp:
(WebCore::ScrollView::wheelEvent):
* platform/ScrollbarTheme.h:
(WebCore::ScrollbarTheme::paintOverhangAreas):
* platform/ScrollbarThemeComposite.cpp:
(WebCore::ScrollbarThemeComposite::paintOverhangAreas):
* platform/ScrollbarThemeComposite.h:
* platform/chromium/ScrollbarThemeChromiumMac.h:
* platform/chromium/ScrollbarThemeChromiumMac.mm:
(WebCore::ScrollbarThemeChromiumMac::ScrollbarThemeChromiumMac):
(WebCore::scrollbarStateToThemeState):
(WebCore::ScrollbarThemeChromiumMac::paintTickmarks):
(WebCore::ScrollbarThemeChromiumMac::paintOverhangAreas):

git-svn-id: svn://svn.chromium.org/blink/trunk@93136 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=60529

Patch by Wyatt Carss <wcarss@chromium.org> on 2011-08-12
Reviewed by Ryosuke Niwa.

Source/WebCore: 

Modified default value of VisibleSelection->m_isDirectional from 'true'
to 'false' by adding a defaulted argument to several VisibleSelection 
constructors to make programmatic selection be directionless by default 
on Mac. Also modified several calls to VisibleSelection's constructors
to correctly preserve directionality.
        
Near the end of FrameSelection::modify (presently FrameSelection.cpp:869)
a call to setExtent triggers an editing delegate that uses the
FrameSelection. The direction here should have been preserved from before,
but setIsDirectional was not called until after FrameSelection.cpp:869. It
has been moved up to make things behave.

* WebCore.exp.in:
* editing/ApplyBlockElementCommand.cpp:
(WebCore::ApplyBlockElementCommand::doApply):
(WebCore::ApplyBlockElementCommand::formatSelection):
* editing/ApplyStyleCommand.cpp:
(WebCore::ApplyStyleCommand::updateStartEnd):
* editing/BreakBlockquoteCommand.cpp:
(WebCore::BreakBlockquoteCommand::doApply):
* editing/CompositeEditCommand.cpp:
(WebCore::CompositeEditCommand::moveParagraphWithClones):
(WebCore::CompositeEditCommand::moveParagraphs):
(WebCore::CompositeEditCommand::breakOutOfEmptyListItem):
(WebCore::CompositeEditCommand::breakOutOfEmptyMailBlockquotedParagraph):
* editing/CreateLinkCommand.cpp:
(WebCore::CreateLinkCommand::doApply):
* editing/DeleteSelectionCommand.cpp:
(WebCore::DeleteSelectionCommand::setStartingSelectionOnSmartDelete):
(WebCore::DeleteSelectionCommand::doApply):
* editing/Editor.cpp:
(WebCore::Editor::selectionForCommand):
* editing/FrameSelection.cpp:
(WebCore::shouldAlwaysUseDirectionalSelection):
(WebCore::FrameSelection::FrameSelection):
(WebCore::FrameSelection::moveTo):
(WebCore::FrameSelection::setNonDirectionalSelectionIfNeeded):
(WebCore::FrameSelection::modify):
(WebCore::FrameSelection::setBase):
(WebCore::FrameSelection::setExtent):
* editing/FrameSelection.h:
* editing/InsertLineBreakCommand.cpp:
(WebCore::InsertLineBreakCommand::doApply):
* editing/InsertListCommand.cpp:
(WebCore::InsertListCommand::doApply):
* editing/InsertParagraphSeparatorCommand.cpp:
(WebCore::InsertParagraphSeparatorCommand::doApply):
* editing/InsertTextCommand.cpp:
(WebCore::InsertTextCommand::performTrivialReplace):
(WebCore::InsertTextCommand::doApply):
* editing/MoveSelectionCommand.cpp:
(WebCore::MoveSelectionCommand::doApply):
* editing/ReplaceSelectionCommand.cpp:
(WebCore::ReplaceSelectionCommand::completeHTMLReplacement):
* editing/TypingCommand.cpp:
(WebCore::TypingCommand::makeEditableRootEmpty):
(WebCore::TypingCommand::deleteKeyPressed):
(WebCore::TypingCommand::forwardDeleteKeyPressed):
* editing/VisibleSelection.cpp:
(WebCore::VisibleSelection::VisibleSelection):
* editing/VisibleSelection.h:
* page/EventHandler.cpp:
(WebCore::EventHandler::updateSelectionForMouseDownDispatchingSelectStart):
(WebCore::EventHandler::updateSelectionForMouseDrag):

LayoutTests: 

Added editing/selection/programmatic-selection-on-mac-is-directionless
to test selection direction in programmatically set selections on
windows, unix, and mac, in plain text, editable divs, text-input
fields, and textareas. Modified text baselines for many editing tests,
which had a delegate change as a result of the cpp patch.

* editing/deleting/delete-ligature-003-expected.txt:
* editing/deleting/paragraph-in-preserveNewline-expected.txt:
* editing/deleting/whitespace-pre-1-expected.txt:
* editing/selection/after-line-break-expected.txt:
* platform/mac/editing/deleting/delete-all-text-in-text-field-assertion-expected.txt:
* platform/mac/editing/deleting/delete-and-undo-expected.txt:
* platform/mac/editing/deleting/delete-at-paragraph-boundaries-008-expected.txt:
* platform/mac/editing/deleting/delete-br-002-expected.txt:
* platform/mac/editing/deleting/delete-br-004-expected.txt:
* platform/mac/editing/deleting/delete-br-005-expected.txt:
* platform/mac/editing/deleting/delete-br-006-expected.txt:
* platform/mac/editing/deleting/delete-br-009-expected.txt:
* platform/mac/editing/deleting/delete-br-010-expected.txt:
* platform/mac/editing/deleting/delete-br-011-expected.txt:
* platform/mac/editing/deleting/delete-ws-fixup-003-expected.txt:
* platform/mac/editing/deleting/delete-ws-fixup-004-expected.txt:
* platform/mac/editing/deleting/smart-delete-002-expected.txt:
* platform/mac/editing/inserting/insert-3800346-fix-expected.txt:
* platform/mac/editing/inserting/insert-at-end-01-expected.txt:
* platform/mac/editing/inserting/insert-at-end-02-expected.txt:
* platform/mac/editing/inserting/insert-div-010-expected.txt:
* platform/mac/editing/inserting/insert-div-013-expected.txt:
* platform/mac/editing/inserting/insert-div-015-expected.txt:
* platform/mac/editing/inserting/insert-div-019-expected.txt:
* platform/mac/editing/inserting/insert-paragraph-03-expected.txt:
* platform/mac/editing/pasteboard/3976872-expected.txt:
* platform/mac/editing/pasteboard/4076267-2-expected.txt:
* platform/mac/editing/pasteboard/4076267-expected.txt:
* platform/mac/editing/pasteboard/drag-selected-image-to-contenteditable-expected.txt:
* platform/mac/editing/pasteboard/emacs-ctrl-a-k-y-expected.txt:
* platform/mac/editing/pasteboard/emacs-ctrl-k-y-001-expected.txt:
* platform/mac/editing/pasteboard/paste-text-014-expected.txt:
* platform/mac/editing/pasteboard/paste-text-015-expected.txt:
* platform/mac/editing/pasteboard/smart-paste-008-expected.txt:
* platform/mac/editing/selection/4932260-1-expected.txt:
* platform/mac/editing/selection/anchor-focus2-expected.txt:
* platform/mac/editing/selection/anchor-focus3-expected.txt:
* platform/mac/editing/selection/expanding-selections-expected.txt:
* platform/mac/editing/selection/expanding-selections2-expected.txt:
* platform/mac/editing/selection/fake-drag-expected.txt:
* platform/mac/editing/selection/move-by-word-001-expected.txt:
* platform/mac/editing/selection/transformed-selection-rects-expected.txt:
* platform/mac/editing/selection/triple-click-in-pre-expected.txt:
* platform/mac/editing/style/create-block-for-style-008-expected.txt:
* platform/mac/editing/undo/4063751-expected.txt:
* platform/mac/editing/undo/undo-forward-delete-boundary-expected.txt:
* platform/mac/editing/undo/undo-forward-delete-expected.txt:


git-svn-id: svn://svn.chromium.org/blink/trunk@93134 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=66317

Reviewed by Adam Barth.

* dom/DocumentFragment.cpp:
(WebCore::DocumentFragment::parseXML):
* xml/parser/NewXMLDocumentParser.cpp:
(WebCore::NewXMLDocumentParser::NewXMLDocumentParser):
(WebCore::NewXMLDocumentParser::parseDocumentFragment):
* xml/parser/NewXMLDocumentParser.h:
(WebCore::NewXMLDocumentParser::create):
* xml/parser/XMLTokenizer.h:
* xml/parser/XMLTreeBuilder.cpp:
(WebCore::XMLTreeBuilder::XMLTreeBuilder):
(WebCore::XMLTreeBuilder::pushCurrentNode):
* xml/parser/XMLTreeBuilder.h:
(WebCore::XMLTreeBuilder::create):
(WebCore::XMLTreeBuilder::NodeStackItem::node):


git-svn-id: svn://svn.chromium.org/blink/trunk@93133 bbb929c8-8fbe-4397-9dbb-9b2b20218538

* platform/chromium/test_expectations.txt:

git-svn-id: svn://svn.chromium.org/blink/trunk@93132 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=66305

Allow the hostWindow param (and m_webViewImpl) to be NULL.  This
makes it much easier to enable GPU acceleration at a lower level
in WebKit's platform/graphics layer, without needing access to the
Page* or HostWindow*.  Also bump the chromium DEPS to chrome r96940,
which includes a similar change for the in-process GraphicsContext3D
implementations.

Reviewed by James Robinson.

* DEPS:
* src/GraphicsContext3DChromium.cpp:
(WebCore::GraphicsContext3DInternal::initialize):
(WebCore::GraphicsContext3DInternal::platformTexture):
(WebCore::GraphicsContext3DInternal::paintsIntoCanvasBuffer):


git-svn-id: svn://svn.chromium.org/blink/trunk@93131 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=66269

Reviewed by Adam Barth.

This patch checks for character data in the prolog (which is illegal in XML) and discards it, instead of failing an assertion.

* xml/parser/XMLTreeBuilder.cpp:
(WebCore::XMLTreeBuilder::processProcessingInstruction):
(WebCore::XMLTreeBuilder::processXMLDeclaration):
(WebCore::XMLTreeBuilder::processDOCTYPE):
(WebCore::XMLTreeBuilder::enterText):
(WebCore::XMLTreeBuilder::failOnText):
* xml/parser/XMLTreeBuilder.h:


git-svn-id: svn://svn.chromium.org/blink/trunk@93130 bbb929c8-8fbe-4397-9dbb-9b2b20218538

debug (not just snowleopard).

* platform/chromium/test_expectations.txt:

git-svn-id: svn://svn.chromium.org/blink/trunk@93129 bbb929c8-8fbe-4397-9dbb-9b2b20218538