Context Navigation

← Previous Period
Next Period →

Timeline

Dec 2, 2012:

11:58 PM Changeset in webkit [136369] by commit-queue@webkit.org

2 edits in trunk/Tools

[EFL][WK2] Add support for a zoom level setting to MiniBrowser
https://bugs.webkit.org/show_bug.cgi?id=99285

Patch by KyungTae Kim <ktf.kim@samsung.com> on 2012-12-02
Reviewed by Gyuyoung Kim.

Add support for a zoom level setting to MiniBrowser.
Shortcuts for setting operations are same as common browsers:
Ctrl + '+' to zoom in, Ctrl + '-' to zoom out and Ctrl + '0' to zoom to default.

MiniBrowser/efl/main.c:

(on_key_down):

11:47 PM Changeset in webkit [136368] by Christophe Dumez

2 edits in trunk/LayoutTests

Unreviewed, WebKit2 gardening.

Skip storage/domstorage/localstorage/close-idle-localstorage-databases-immediately.html
on WebKit2 since testRunner.closeIdleLocalStorageDatabases() is not implemented yet.

platform/wk2/TestExpectations:

11:27 PM Changeset in webkit [136367] by commit-queue@webkit.org

2 edits in trunk/Source/WebKit2

Coordinated Graphics: Reorder messages to LayerTreeCoordinatorProxy
https://bugs.webkit.org/show_bug.cgi?id=103843

Patch by Huang Dongsung <luxtella@company100.net> on 2012-12-02
Reviewed by Noam Rosenthal.

Send messages to the UI process by the tree order.
This is in preparation for refactoring TextureMapper to work in an actor
model (http://webkit.org/b/103854).

WebProcess/WebPage/CoordinatedGraphics/LayerTreeCoordinator.cpp:

(WebKit::LayerTreeCoordinator::flushPendingLayerChanges):

10:54 PM Changeset in webkit [136366] by commit-queue@webkit.org

3 edits in trunk/LayoutTests

[EFL] Unskip fast/js/array-sort-modifying-tostring.html in release mode
https://bugs.webkit.org/show_bug.cgi?id=103850

Unreviewed, EFL gardening.

This case has been passed in release mode, whereas crashed occasionally in debug mode.

Patch by Kangil Han <kangil.han@samsung.com> on 2012-12-02

platform/efl-wk2/TestExpectations:
platform/efl/TestExpectations:

10:24 PM Changeset in webkit [136365] by noel.gordon@gmail.com

2 edits in trunk/Source/WebCore

ImageOrientation::transformFromDefault() case labels should be in numerical order
https://bugs.webkit.org/show_bug.cgi?id=101596

Reviewed by Simon Fraser.

Place the orientation case labels in numerical order to better match the EXIF2.2 spec
definitions of orientation values and their ImageOrientation.h enumeration.

No change in behavior. Covered by existing tests:

fast/images/exif-orientation-css.html
fast/images/exif-orientation-image-document.html
fast/images/exif-orientation.html

platform/graphics/ImageOrientation.cpp:

(WebCore::ImageOrientation::transformFromDefault): Numerically order the orientation
case statement labels.

10:13 PM Changeset in webkit [136364] by tasak@google.com

1 edit
3 adds in trunk/LayoutTests

Unreviewed, webkit gardening. Rebaseline of fast/css/font-weight-1.

platform/chromium-mac-snowleopard/fast/css/font-weight-1-expected.png: Added.
platform/chromium-mac/fast/css/font-weight-1-expected.png: Added.
platform/chromium-mac/fast/css/font-weight-1-expected.txt: Added.

9:23 PM Changeset in webkit [136363] by commit-queue@webkit.org

3 edits in trunk/LayoutTests

[chromium] platform/chromium/fast/forms/calendar-picker/calendar-picker-datetime-with-step.html is failing on all platforms.
https://bugs.webkit.org/show_bug.cgi?id=103817

Patch by Kunihiko Sakamoto <ksakamoto@chromium.org> on 2012-12-02
Reviewed by Kent Tamura.

Adding initial value to the datetime input. Without it, the test fails
because calendar picker opens current month.

platform/chromium/TestExpectations:
platform/chromium/fast/forms/calendar-picker/calendar-picker-datetime-with-step.html:

9:04 PM Changeset in webkit [136362] by tasak@google.com

2 edits in trunk/LayoutTests

Unreviewed, webkit gardening.

https://bugs.webkit.org/show_bug.cgi?id=103834

platform/chromium/TestExpectations:

Mark month-multiple-fields-keyboard-events.html as FAIL for WinXP.

7:30 PM Changeset in webkit [136361] by mary.wu@torchmobile.com.cn

2 edits in trunk/Source/WebCore

should sniff it if mimetype don't contain a slash.
https://bugs.webkit.org/show_bug.cgi?id=102778

Reviewed by George Staikos.

Based on http://mimesniff.spec.whatwg.org/, a valid media type
is the string represented by concatenating type, the string "/",
and subtype. So if the mimetype don't contain slash, we should
possible get sniffed type from the content like Firefox/chrome/safari.

RIM Bug# 219590, internally reviewed by Liam Quinn.

No new tests, the patch only impact qt/blackberry porting to be consistent
with chrome, other portings have their own behavior.

platform/network/MIMESniffing.cpp:

(std::isUnknownType):

7:12 PM Changeset in webkit [136360] by fpizlo@apple.com

2 edits in trunk/Source/JavaScriptCore

DFG CSE should not keep alive things that aren't relevant to OSR
https://bugs.webkit.org/show_bug.cgi?id=103849

Reviewed by Oliver Hunt.

Most Phantom nodes are inserted by CSE, and by default have the same children as the
node that CSE had eliminated. This change makes CSE inspect all Phantom nodes (both
those it creates and those that were created by other phases) to see if they have
children that are redundant - i.e. children that are not interesting to OSR, which
is the only reason why Phantoms exist in the first place. Being relevant to OSR is
defined as one of: (1) you're a Phi, (2) you're a SetLocal, (3) somewhere between
your definition and the Phantom there was a SetLocal that referred to you.

This is a slight speed-up in a few places.

dfg/DFGCSEPhase.cpp:

(JSC::DFG::CSEPhase::CSEPhase):
(JSC::DFG::CSEPhase::run):
(JSC::DFG::CSEPhase::performSubstitution):
(CSEPhase):
(JSC::DFG::CSEPhase::eliminateIrrelevantPhantomChildren):
(JSC::DFG::CSEPhase::setReplacement):
(JSC::DFG::CSEPhase::eliminate):
(JSC::DFG::CSEPhase::performNodeCSE):
(JSC::DFG::CSEPhase::performBlockCSE):

6:30 PM Changeset in webkit [136359] by ryuan.choi@samsung.com

4 edits in trunk/Source/WebKit2

[EFL][WK2] Update comments of header files which mentions deprecated API
https://bugs.webkit.org/show_bug.cgi?id=103718

Patch by Ryuan Choi <ryuan.choi@gmail.com> on 2012-12-02
Reviewed by Gyuyoung Kim.

Updated comments which mentions XXX_unref() which was replaced to ewk_object_unref()

UIProcess/API/efl/ewk_context.h:
UIProcess/API/efl/ewk_database_manager.h:
UIProcess/API/efl/ewk_storage_manager.h:

6:08 PM Changeset in webkit [136358] by commit-queue@webkit.org

2 edits in trunk/LayoutTests

[EFL][WK1] Change AC bug url appropriately
https://bugs.webkit.org/show_bug.cgi?id=103808

Unreviewed, EFL gardening.

From https://bugs.webkit.org/show_bug.cgi?id=103488 to https://bugs.webkit.org/show_bug.cgi?id=103847

Patch by Kangil Han <kangil.han@samsung.com> on 2012-12-02

platform/efl-wk1/TestExpectations:

5:59 PM Changeset in webkit [136357] by fpizlo@apple.com

9 edits in trunk/Source

It should be possible to build and run with DFG_ENABLE(PROPAGATION_VERBOSE)
https://bugs.webkit.org/show_bug.cgi?id=103848

Reviewed by Sam Weinig.

Source/JavaScriptCore:

Fix random dataLog() and print() statements.

dfg/DFGArgumentsSimplificationPhase.cpp:

(JSC::DFG::ArgumentsSimplificationPhase::run):

dfg/DFGByteCodeParser.cpp:

(JSC::DFG::ByteCodeParser::parseCodeBlock):

dfg/DFGGraph.cpp:

(JSC::DFG::Graph::dumpBlockHeader):

dfg/DFGPredictionPropagationPhase.cpp:

(JSC::DFG::PredictionPropagationPhase::propagate):

dfg/DFGStructureCheckHoistingPhase.cpp:

(JSC::DFG::StructureCheckHoistingPhase::run):

Source/WTF:

Increase the variadicity of dataLog() and print(), and add a pointerDump()
which allows passing a pointer to something that has a printInternal(). If
the pointer is null, it'll print "(null)"; otherwise it'll invoke the
printInternal() function.

wtf/DataLog.h:

(WTF):
(WTF::dataLog):

wtf/PrintStream.h:

(PrintStream):
(WTF::PrintStream::print):
(WTF):
(PointerDump):
(WTF::PointerDump::PointerDump):
(WTF::PointerDump::dump):
(WTF::pointerDump):

5:46 PM Changeset in webkit [136356] by tasak@google.com

2 edits in trunk/LayoutTests

Unreviewed, webkit gardening, fix webkit lint.

platform/chromium/TestExpectations:

4:45 PM Changeset in webkit [136355] by commit-queue@webkit.org

3 edits in trunk/Source/WebKit2

REGRESSION(r134376): ASSERT(!m_mainBackingStore) hits in CoordinatedGraphicsLayer::syncImageBacking().
https://bugs.webkit.org/show_bug.cgi?id=103845

Patch by Huang Dongsung <luxtella@company100.net> on 2012-12-02
Reviewed by Noam Rosenthal.

flushCompositingStateForThisLayerOnly() calls syncImageBacking() before calling
updateContentBuffers(). It means there can be the moment that
CoordinatedGraphicsLayer has m_mainBackingStore although
shouldHaveBackingStore() return false.
This patch changes syncImageBacking() to check shouldHaveBackingStore() instead
of m_mainBackingStore.

WebProcess/WebPage/CoordinatedGraphics/CoordinatedGraphicsLayer.cpp:

(WebCore::CoordinatedGraphicsLayer::syncImageBacking):
(WebCore::CoordinatedGraphicsLayer::adjustContentsScale):
(WebCore::CoordinatedGraphicsLayer::updateContentBuffers):
(WebCore::CoordinatedGraphicsLayer::shouldHaveBackingStore):
(WebCore):

WebProcess/WebPage/CoordinatedGraphics/CoordinatedGraphicsLayer.h:

(CoordinatedGraphicsLayer):

4:21 PM Changeset in webkit [136354] by commit-queue@webkit.org

2 edits
1 move
1 add in trunk/Source/WebKit2

Rename WebSocketServerGtk.cpp as WebSocketServerSoup.cpp
https://bugs.webkit.org/show_bug.cgi?id=103743

Patch by Seokju Kwon <Seokju Kwon> on 2012-12-02
Reviewed by Gustavo Noronha Silva.

EFL port is using a GSocket based implementation as well.
WebSocketServerSoup can be used together.

GNUmakefile.list.am:
UIProcess/InspectorServer/soup/WebSocketServerSoup.cpp: Renamed from Source/WebKit2/UIProcess/InspectorServer/gtk/WebSocketServerGtk.cpp.

(WebKit):
(WebKit::connectionCallback):
(WebKit::WebSocketServer::platformInitialize):
(WebKit::WebSocketServer::platformListen):
(WebKit::WebSocketServer::platformClose):

3:43 PM Changeset in webkit [136353] by commit-queue@webkit.org

2 edits in trunk/Tools

[EFL][WK2] ECORE_GETOPT_VALUE_BOOL need to be used for Eina_Bool
https://bugs.webkit.org/show_bug.cgi?id=103835

Patch by KyungTae Kim <ktf.kim@samsung.com> on 2012-12-02
Reviewed by Kentaro Hara.

Because legacy_behavior_enabled is Eina_Bool, ECORE_GETOPT_VALUE_BOOL need to be used

MiniBrowser/efl/main.c:

(elm_main):

1:18 PM Changeset in webkit [136352] by mkwst@chromium.org

3 edits in trunk/Tools

Garden-o-matic shouldn't die when blank JSON results are received.
https://bugs.webkit.org/show_bug.cgi?id=103838

Reviewed by Adam Barth.

BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/base.js:

If the JSONP string doesn't contain '(' or ')', return an empty object.

BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/base_unittests.js:

Check that this behavior stays in place.

12:48 PM Changeset in webkit [136351] by eric@webkit.org

2 edits in trunk/Tools

run-perf-tests --profile should use iprofiler instead of instruments on Mac
https://bugs.webkit.org/show_bug.cgi?id=103765

Unreviewed. Updating the unittests after my previous change.

Scripts/webkitpy/common/system/profiler_unittest.py:

(ProfilerFactoryTest.test_basic):

12:29 PM Changeset in webkit [136350] by commit-queue@webkit.org

2 edits in trunk/Source/WebCore

[BlackBerry] Adding checks for null when checking cookie status in CookieJarBlackBerry
https://bugs.webkit.org/show_bug.cgi?id=103748

Patch by Otto Derek Cheung <otcheung@rim.com> on 2012-12-02
Reviewed by Rob Buis.

The logic is flipped in the previous patch.

loader/blackberry/CookieJarBlackBerry.cpp:

(WebCore::cookiesEnabled):

11:56 AM Changeset in webkit [136349] by commit-queue@webkit.org

5 edits in trunk

[chromium] Correct the location of synthetic mouse events when page is zoomed
https://bugs.webkit.org/show_bug.cgi?id=103823

Patch by Sadrul Habib Chowdhury <sadrul@chromium.org> on 2012-12-02
Reviewed by Adam Barth.

Source/WebKit/chromium:

The event-location conversion function expects absolute locations. Change the
function name to reflect that, and use the absolute location of the touch-points
when converting from touch events.

src/WebInputEventConversion.cpp:

(WebKit::convertAbsoluteLocationForRenderObject):
(WebKit::updateWebMouseEventFromWebCoreMouseEvent):
(WebKit::WebMouseEventBuilder::WebMouseEventBuilder):
(WebKit::addTouchPoints):
(WebKit::WebGestureEventBuilder::WebGestureEventBuilder):

LayoutTests:

platform/chromium/plugins/touch-events-synthesized-expected.txt:
platform/chromium/plugins/touch-events-synthesized.html:

10:49 AM Changeset in webkit [136348] by jparent@chromium.org

2 edits in trunk/Tools

Remove dead code in flakiness dashboard for obsolete buildDir option
https://bugs.webkit.org/show_bug.cgi?id=103792

Reviewed by Dirk Pranke.

TestResultServer/static-dashboards/dashboard_base.js:

(initBuilders):

9:25 AM Changeset in webkit [136347] by robert@webkit.org

3 edits
4 adds in trunk

Bad rendering of web page because of image's height is set to 100%
https://bugs.webkit.org/show_bug.cgi?id=28035

Reviewed by Eric Seidel.

Source/WebCore:

Call computeReplacedLogicalHeight() to get the correct logical height of a replaced element
for calculating the element's width. RenderBox::computeLogicalHeightUsing() is a helper function designed for finding
the min and max height values and is not the way to find the actual used height of a replaced element.

WebKit's quirks-mode behaviour for the test case is unaffected by this change.

Tests: fast/css/computed-image-width-with-percent-height-quirksmode.html

fast/css/computed-image-width-with-percent-height.html

rendering/RenderReplaced.cpp:

(WebCore::RenderReplaced::computeReplacedLogicalWidth):

LayoutTests:

fast/css/computed-image-width-with-percent-height-expected.html: Added.
fast/css/computed-image-width-with-percent-height-quirksmode-expected.html: Added.
fast/css/computed-image-width-with-percent-height-quirksmode.html: Added.
fast/css/computed-image-width-with-percent-height.html: Added.

7:35 AM Changeset in webkit [136346] by mihnea@adobe.com

5 edits
2 adds in trunk

[CSS Regions] regionlayoutupdate event fires continuously
https://bugs.webkit.org/show_bug.cgi?id=102101

Reviewed by David Hyatt.

Source/WebCore:

According to the CSSRegions spec, http://dev.w3.org/csswg/css3-regions/#region-flow-layout-events, the regionlayoutupdate event
should be dispatched when there is a possible layout change in the region chain.
In the current implementation, regionlayoutupdate is dispatched when:

a region from region chain changed its regionOverset value
a region from region chain has regionOverset = fit| overset

In the case of a named flow having an auto-height region with regionOverset = fit, the second step of layout was done
also when there was not required, therefore the regionlayoutuodate was unnecessary triggered.

We should proceed with the 2-pass layout only if the flow-thread having auto-height regions in region chain needs layout.

Test: fast/regions/autoheight-regionlayoutupdate-multipletimes.html

rendering/FlowThreadController.cpp:

(WebCore::FlowThreadController::hasRenderNamedFlowThreadsNeedingLayout):

rendering/FlowThreadController.h:
rendering/RenderView.cpp:

(WebCore::RenderView::layout):

LayoutTests:

Add a test to make sure that for a flow thread with auto-height region, the regionlayoutupdate event is not unnecessary triggered.

fast/regions/autoheight-regionlayoutupdate-multipletimes-expected.txt: Added.
fast/regions/autoheight-regionlayoutupdate-multipletimes.html: Added.

7:14 AM WebInspector edited by mousewiki@yahoo.com

(diff)

7:13 AM WebInspector edited by zandobersek@gmail.com

Remove spam content. (diff)

7:12 AM BadContent edited by zandobersek@gmail.com

Relist a tireless spam bot. (diff)

5:22 AM Changeset in webkit [136345] by dpranke@chromium.org

2 edits in trunk/Tools

nrwt: clean up logic for retrying failed tests a bit
https://bugs.webkit.org/show_bug.cgi?id=103829

Reviewed by Ojan Vafai.

The code for determining when to retry test failures and
show the results.html file was awkward and could stand some
cleaning up. This patch should have no functional changes.

The routine called _get_failures() returned a dict of test names
to failure types, but we only ever used the names, so I've converted
it to return a list (and renamed it); also, the show_results_html()
file was calling that routine to figure out if anything failed, when
the information was already in the result_summary (as either
unexpected_results or total_failures), so we were doing unnecessary work.

Scripts/webkitpy/layout_tests/controllers/manager.py:

(Manager.init):
(Manager.run):
(Manager._run_tests):
(Manager._failed_test_names):
(Manager._copy_results_html_file):

5:12 AM Changeset in webkit [136344] by dpranke@chromium.org

2 edits in trunk/Tools

nrwt: remove unused methods in manager.py
https://bugs.webkit.org/show_bug.cgi?id=103828

Reviewed by Eric Seidel.

These two methods weren't being called by anything and were no
longer needed.

Scripts/webkitpy/layout_tests/controllers/manager.py:

(Manager._get_failures):
(Manager._upload_json_files):

5:09 AM Changeset in webkit [136343] by commit-queue@webkit.org

6 edits
1 copy in trunk/Source/WebKit/chromium

[Chromium] Create an enum for the kind of date input and use that on WebDateTimeChooserParams instead of a bare WebString
https://bugs.webkit.org/show_bug.cgi?id=103746

Patch by Miguel Garcia <miguelg@chromium.org> on 2012-12-02
Reviewed by Kent Tamura.

Using an enum makes the chromium part much simpler and prevents makes sure that new items are added explicitly.

WebKit.gyp:
public/WebDateTimeChooserParams.h:

(WebDateTimeChooserParams):

public/WebDateTimeInputType.h: Copied from Source/WebKit/chromium/public/WebTextInputType.h.

(WebKit):

public/WebTextInputType.h:
public/WebViewClient.h:

(WebViewClient):

src/ExternalDateTimeChooser.cpp:

(WebKit::ExternalDateTimeChooser::openDateTimeChooser):
(WebKit::ExternalDateTimeChooser::webType):
(WebKit):

src/ExternalDateTimeChooser.h:

(ExternalDateTimeChooser):

5:00 AM Changeset in webkit [136342] by Christophe Dumez

2 edits in trunk/Source/WebKit2

[CoordinatedGraphics] LayerTreeRenderer::removeImageBacking() calls HashMap::find() twice
https://bugs.webkit.org/show_bug.cgi?id=103815

Reviewed by Noam Rosenthal.

Call HashMap::take() in LayerTreeRenderer::removeImageBacking()
to avoid calling HashMap::find() twice. find() was called once
explicitly to get an iterator and a second time to remove the
value from the value from the HashMap. We could have used
remove(iterator) instead of remove(key) but using take() results
in simpler code.

UIProcess/CoordinatedGraphics/LayerTreeRenderer.cpp:

(WebKit::LayerTreeRenderer::removeImageBacking):

4:49 AM Changeset in webkit [136341] by commit-queue@webkit.org

3 edits in trunk/Source/WebKit2

Coordinated Graphics: Images disappear randomly
https://bugs.webkit.org/show_bug.cgi?id=103522

Patch by Huang Dongsung <luxtella@company100.net> on 2012-12-02
Reviewed by Kenneth Rohde Christiansen.

Currently, we delete layers before synchronizing layer states of layers. It
causes flash. This patch deletes layers after the synchronization.

In addiation, this patch removes updateViewport() in
LayerTreeCoordinatorProxy::deleteCompositingLayer() because
LayerTreeCoordinatorProxy::didRenderFrame() is always called after
deleting layers.

UIProcess/CoordinatedGraphics/LayerTreeCoordinatorProxy.cpp:

(WebKit::LayerTreeCoordinatorProxy::deleteCompositingLayer):

WebProcess/WebPage/CoordinatedGraphics/LayerTreeCoordinator.cpp:

(WebKit::LayerTreeCoordinator::flushPendingLayerChanges):

4:43 AM Changeset in webkit [136340] by commit-queue@webkit.org

2 edits in trunk/Tools

Correct the spelling of the cmakeargs argument
https://bugs.webkit.org/show_bug.cgi?id=103814

Patch by Laszlo Gombos <l.gombos@samsung.com> on 2012-12-02
Reviewed by Darin Adler.

The script expects cmakeargs (plural), so lets get the
documentation of the script report the same.

Scripts/build-webkit:

4:40 AM Changeset in webkit [136339] by mkwst@chromium.org

7 edits in trunk

[gtk] Enable the CSP_NEXT runtime flag.
https://bugs.webkit.org/show_bug.cgi?id=103810

Reviewed by Adam Barth.

Source/WebKit/gtk:

This patch adds methods to DumpRenderTreeSupport in order to ensure that
CSP 1.1 features are enabled when running DRT tests in the GTK port.

WebCoreSupport/DumpRenderTreeSupportGtk.cpp:

(DumpRenderTreeSupportGtk::setExperimentalContentSecurityPolicyFeaturesEnabled):

WebCoreSupport/DumpRenderTreeSupportGtk.h:

(DumpRenderTreeSupportGtk):

Tools:

DumpRenderTree/gtk/DumpRenderTree.cpp:

(resetDefaultsToConsistentValues):

Turn on the CSP_NEXT runtime flag when running tests under GTK.

LayoutTests:

platform/gtk/TestExpectations:

Unskipping CSP 1.1 tests for GTK.

4:38 AM Changeset in webkit [136338] by mkwst@chromium.org

2 edits in trunk/LayoutTests

[chromium] Unreviewed gardening.

Skipping failing tests on mac.

platform/chromium/TestExpectations:

Mark fast/forms/month-multiple-fields/month-multiple-fields-keyboard-events.html
and widening fast/css/font-weight-1.html to include 'Mac'.

4:22 AM Changeset in webkit [136337] by commit-queue@webkit.org

7 edits
2 adds in trunk

Extend JavaScript support for blending in canvas
https://bugs.webkit.org/show_bug.cgi?id=101804

Patch by Rik Cabanier <cabanier@adobe.com> on 2012-12-02
Reviewed by Dirk Schulze.

Source/WebCore:

Added parsing code and variables to support additional canvas blend modes
see spec: https://dvcs.w3.org/hg/FXTF/rawfile/tip/compositing/index.html#canvascompositingandblending

Test: canvas/philip/tests/2d.composite.globalComposite.html

html/HTMLImageElement.cpp:

(WebCore::HTMLImageElement::parseAttribute):

html/canvas/CanvasRenderingContext2D.cpp:

(WebCore::CanvasRenderingContext2D::State::State):
(WebCore::CanvasRenderingContext2D::State::operator=):
(WebCore::CanvasRenderingContext2D::globalCompositeOperation):
(WebCore::CanvasRenderingContext2D::setGlobalCompositeOperation):
(WebCore::CanvasRenderingContext2D::drawImageFromRect):

html/canvas/CanvasRenderingContext2D.h:

(State):

platform/graphics/GraphicsTypes.cpp:

(WebCore):
(WebCore::parseCompositeAndBlendOperator):
(WebCore::compositeOperatorName):

platform/graphics/GraphicsTypes.h:

LayoutTests:

Added a test that checks that all blendingmodes are supported

canvas/philip/tests/2d.composite.globalComposite-expected.txt: Added.
canvas/philip/tests/2d.composite.globalComposite.html: Added.

4:11 AM Changeset in webkit [136336] by bfulgham@webkit.org

2 edits in trunk/Source/WebKit2

[Windows, WinCairo] Revise export definitions to match mac.
https://bugs.webkit.org/show_bug.cgi?id=103687

Reviewed by Tim Horton.

Update export definition files to use the same feature exclusion
macros as other generated ports.

win/WebKit2.def.in: Added additional build macro guards.

4:09 AM Changeset in webkit [136335] by commit-queue@webkit.org

2 edits in trunk/Source/WebCore

[EFL] Fix build warning in GraphicsContext3DEfl.cpp
https://bugs.webkit.org/show_bug.cgi?id=103807

Patch by KyungTae Kim <ktf.kim@samsung.com> on 2012-12-02
Reviewed by Laszlo Gombos.

Because GraphicsContext3D::ImageExtractor::extractImage is not implemented yet,
comment out the parameter names for removing -Wunused-parameter warnings.

platform/graphics/efl/GraphicsContext3DEfl.cpp:

(WebCore::GraphicsContext3D::ImageExtractor::extractImage):

4:03 AM Changeset in webkit [136334] by akling@apple.com

11 edits
2 adds in trunk/Source/WebCore

Keep ElementAttributeData sharing cache open for a while after document parsing finishes.
<http://webkit.org/b/103720>

Reviewed by Antti Koivisto.

Allow web pages that generate content dynamically to benefit more from the ElementAttributeData sharing
cache by keeping it open for 10 seconds after document parsing finishes. This enables elements constructed
from HTML fragments to share attribute data with other identical elements.

Elements created via Document.createElement are still unsupported since we don't know the list of attributes
at the time of construction.

401kB progression on Membuster3.

Document now holds on to a DocumentSharedObjectPool, accessible via Document::sharedObjectPool().
It is non-null during parsing and for some time afterwards, and can be used to implement additional
caches that are not worth holding on to permanently.

CMakeLists.txt:
GNUmakefile.list.am:
Target.pri:
WebCore.gypi:
WebCore.xcodeproj/project.pbxproj:
WebCore.vcproj/WebCore.vcproj:

Add new class DocumentSharedObjectPool.

dom/Document.cpp:

(WebCore::Document::Document):
(WebCore::Document::setParsing):
(WebCore::Document::finishedParsing):
(WebCore::Document::sharedObjectPoolClearTimerFired):

dom/Document.h:

(WebCore::Document::sharedObjectPool):

Create DocumentSharedObjectPool when parsing starts, kill it on a 10s timer after parsing finishes.

dom/DocumentSharedObjectPool.cpp: Added.

(ImmutableElementAttributeDataCacheKey):
(WebCore::ImmutableElementAttributeDataCacheKey::ImmutableElementAttributeDataCacheKey):
(WebCore::ImmutableElementAttributeDataCacheKey::operator!=):
(WebCore::ImmutableElementAttributeDataCacheKey::hash):
(ImmutableElementAttributeDataCacheEntry):
(WebCore::ImmutableElementAttributeDataCacheEntry::ImmutableElementAttributeDataCacheEntry):
(WebCore::DocumentSharedObjectPool::cachedImmutableElementAttributeData):
(WebCore::DocumentSharedObjectPool::DocumentSharedObjectPool):
(WebCore::DocumentSharedObjectPool::~DocumentSharedObjectPool):

dom/DocumentSharedObjectPool.h: Added.

(DocumentSharedObjectPool):
(WebCore::DocumentSharedObjectPool::create):

dom/Element.cpp:

(WebCore::Element::parserSetAttributes):

Moved the ElementAttributeData cache to DocumentSharedObjectPool.

3:56 AM Changeset in webkit [136333] by tkent@chromium.org

2 edits in trunk/LayoutTests

[Chromium] fast/forms/*-multiple-fields/*-multiple-fields-preserve-value-after-history-back.html is flaky
https://bugs.webkit.org/show_bug.cgi?id=96720

platform/chromium/TestExpectations: Mark them Slow

3:54 AM Changeset in webkit [136332] by commit-queue@webkit.org

2 edits in trunk/Source/WebCore

[chromium] Let Pepper Flash use integrated GPU
https://bugs.webkit.org/show_bug.cgi?id=103800

Patch by John Bauman <jbauman@chromium.org> on 2012-12-02
Reviewed by Kenneth Russell.

Pepper Flash uses the SharedGraphicsContext3D as a parent, so don't
require that to use the discrete GPU. This means that accelerated
canvas will use the integrated GPU, but that will save battery life
and seems acceptable.

platform/graphics/gpu/SharedGraphicsContext3D.cpp:

(WebCore::SharedGraphicsContext3DImpl::createContext):

3:38 AM Changeset in webkit [136331] by commit-queue@webkit.org

5 edits
2 adds in trunk

documentElement should not always get a renderer
https://bugs.webkit.org/show_bug.cgi?id=103475

Patch by Elliott Sprehn <Elliott Sprehn> on 2012-12-02
Reviewed by Ojan Vafai.

Source/WebCore:

Only SVG documents need a renderer on the documentElement when it's
display: none. This behavior is wrong, but there's assumptions in
FrameView and SVGImage that depend on a RenderSVGRoot existing for all
loaded SVG document and images.

The current code has logic that makes it appear as if we always create
a renderer for the documentElement even when it's display: none, but
because HTMLElement::createRenderer doesn't call super we end up with
HTMLElement::rendererIsNeeded returning true for display: none, but
HTMLElement::createRenderer returning null.

Instead of this generic documentElement logic only special case SVG
until someone can fix Bug 103493 so we don't need it.

Test: fast/css/display-none-documentElement.html

dom/Element.cpp:

(WebCore::Element::rendererIsNeeded):
(WebCore::Element::createRenderer):

svg/SVGSVGElement.cpp:

(WebCore::SVGSVGElement::rendererIsNeeded):

svg/SVGSVGElement.h:

(SVGSVGElement):

LayoutTests:

Test that using display: none on the documentElement hides the document.

fast/css/display-none-documentElement-expected.html: Added.
fast/css/display-none-documentElement.html: Added.

3:30 AM Changeset in webkit [136330] by commit-queue@webkit.org

4 edits in trunk/Source/WebKit/blackberry

[BlackBerry]Tap highlight
https://bugs.webkit.org/show_bug.cgi?id=103787

Patch by Tiancheng Jiang <tijiang@rim.com> on 2012-12-02
Reviewed by Rob Buis.

RIM PR 248626
Internally Reviewed by Genevieve Mak.

Draw tap highlight as soon as tap on element.

Api/WebPage.cpp:

(BlackBerry::WebKit::WebPage::touchEvent):

WebKitSupport/TouchEventHandler.cpp:

(BlackBerry::WebKit::TouchEventHandler::handleTouchPoint):

WebKitSupport/TouchEventHandler.h:

(TouchEventHandler):

3:23 AM Changeset in webkit [136329] by commit-queue@webkit.org

4 edits
5 adds in trunk

Source/WebCore: [Resource Timing] implementation of cross origin resouce timing restrictions.
https://bugs.webkit.org/show_bug.cgi?id=84886.

Patch by Pan Deng <pan.deng@intel.com> on 2012-12-02
Reviewed by Tony Gentilcore.

This patch implemented resource timing behaviors of cross origin. By default, detailed timing info is hided in cross origin resource timing, only startTime, duration, fetchStart and responseEnd can be observed. Exceptions are, server side allow its origin can be timing by another through a header with "timing-allow-origin" field.

Tests: http/tests/w3c/webperf/submission/Intel/resource-timing/test_resource_timing_cross_origin_resource_request.html

http/tests/w3c/webperf/submission/Intel/resource-timing/test_resource_timing_timing_allow_cross_origin_resource_request.html

page/Performance.cpp:

(WebCore::passesTimingAllowCheck):
(WebCore):
(WebCore::Performance::addResourceTiming):

page/Performance.h:

(WebCore):
(Performance):

page/PerformanceResourceTiming.cpp:

(WebCore):

page/PerformanceResourceTiming.h:

(WebCore::PerformanceResourceTiming::create):
(PerformanceResourceTiming):

LayoutTests: [Resource Timing]Test cases of cross origin resource timing.
https://bugs.webkit.org/show_bug.cgi?id=84886.

Patch by Pan Deng <pan.deng@intel.com> on 2012-12-02
Reviewed by Tony Gentilcore.

Test cases in this patch validate resource timing behavior when cross origin request with/without "allow-timing-origin" response header.

http/tests/w3c/webperf/resources/blank_page_green_with_allow_timing.php: Added.
http/tests/w3c/webperf/submission/Intel/resource-timing/test_resource_timing_cross_origin_resource_request-expected.txt: Added.
http/tests/w3c/webperf/submission/Intel/resource-timing/test_resource_timing_cross_origin_resource_request.html: Added.
http/tests/w3c/webperf/submission/Intel/resource-timing/test_resource_timing_timing_allow_cross_origin_resource_request-expected.txt: Added.
http/tests/w3c/webperf/submission/Intel/resource-timing/test_resource_timing_timing_allow_cross_origin_resource_request.html: Added.

3:16 AM Changeset in webkit [136328] by commit-queue@webkit.org

10 edits in trunk/Source/WebCore

Simplify treeScope and setTreeScope
https://bugs.webkit.org/show_bug.cgi?id=103708

Patch by Elliott Sprehn <Elliott Sprehn> on 2012-12-02
Reviewed by Hajime Morita.

By making the default treeScope in NodeRareData the document and then
moving the m_treeScope field to NodeRareDataBase we can make treeScope
inline and make setTreeScope much simpler.

There's also no reason to save calls to rareData() now that the map has
been eliminated by r133372 so we can eliminate the return value from
setTreeScope.

No new tests, just refactoring.

WebCore.exp.in:
dom/Document.h:

(WebCore::Node::treeScope):
(WebCore):

dom/Element.cpp:

(WebCore::Element::createRareData):

dom/ElementRareData.h:

(ElementRareData):
(WebCore::ElementRareData::ElementRareData):

dom/Node.cpp:

(WebCore::Node::setTreeScope):
(WebCore::Node::createRareData):

dom/Node.h:

(NodeRareDataBase):
(WebCore::NodeRareDataBase::treeScope):
(WebCore::NodeRareDataBase::setTreeScope):
(WebCore::NodeRareDataBase::NodeRareDataBase):
(Node):

dom/NodeRareData.cpp:

(WebCore::NodeRareData::reportMemoryUsage):

dom/NodeRareData.h:

(WebCore::NodeRareData::NodeRareData):
(NodeRareData):

dom/TreeScopeAdopter.cpp:

(WebCore::TreeScopeAdopter::moveTreeToNewScope):

3:00 AM Changeset in webkit [136327] by eric@webkit.org

2 edits in trunk/Tools

run-perf-tests --profile should use iprofiler instead of instruments on Mac
https://bugs.webkit.org/show_bug.cgi?id=103765

Reviewed by Ryosuke Niwa.

Stephanie pointed out that iprofiler is the newer command-line interface
to replace /usr/bin/instruments in Instruments 4 (which according to the
interblags is supported on everything since 10.6.6), so I've moved --profile
to use iprofiler on Mac.

I also made the IProfiler class wait for iprofiler to complete so that
iprofiler won't continue running after run-perf-tests is complete (and
spam over the top of your shell output).

There are a couple of "got-chas" to this approach:

iprofiler will ask for your password the first time you run it, which may

cause your first test to fail. /usr/bin/instruments doesn't seem to do that.

iprofiler has nice (but kinda awkward for our purposes) support for separate

directory and basename arguments, making IProfiler a somewhat awkward
baseclass of SingleFileOutputProfiler

I suspect we'll iterate from here. It does seem slightly faster than
/usr/bin/instruments was.

Scripts/webkitpy/common/system/profiler.py:

(ProfilerFactory.create_profiler):
(IProfiler):
(IProfiler.init):
(IProfiler.attach_to_pid):

2:55 AM Changeset in webkit [136326] by junov@google.com

5 edits
2 adds in trunk

Fix occlusion culling logic to handle css background layer clipping
https://bugs.webkit.org/show_bug.cgi?id=103276

Reviewed by Simon Fraser.

Source/WebCore:

Bug was introduced by r135629
This patch fixes it by verifying layer clip. A fill layer must
not be treated as if it occludes the layers under it unless it has
a larger or equal clip setting than the layers under it.

Test: fast/backgrounds/background-opaque-clipped-gradients.html

rendering/RenderBox.cpp:

(WebCore::RenderBox::paintFillLayers):

rendering/style/FillLayer.cpp:

(WebCore::clipMax):
(WebCore::FillLayer::computeClipMax):
(WebCore::FillLayer::clipOccludesNextLayers):

rendering/style/FillLayer.h:

(FillLayer):

LayoutTests:

New ref test that verifies the superposition of background layers with
the layers having smaller clip regions than the layers beneath them.
This test ensures the bottom layers are not removed by culling
optimizations.

fast/backgrounds/background-opaque-clipped-gradients-expected.html: Added.
fast/backgrounds/background-opaque-clipped-gradients.html: Added.

2:42 AM Changeset in webkit [136325] by rjkroege@chromium.org

3 edits in trunk/Source/WebKit/chromium

[chromium] make sourceDevice available to more WebGestureEvent subtypes
https://bugs.webkit.org/show_bug.cgi?id=103551

Reviewed by James Robinson.

Added an additional data payload to WebGestureEvent that can
contain the source device (touchpad or touchscreen).

public/WebInputEvent.h:
src/WebInputEvent.cpp:

(SameSizeAsWebGestureEvent):

2:39 AM Changeset in webkit [136324] by tony@chromium.org

5 edits
2 adds in trunk

Avoid a second layout of flex items in layoutAndPlaceChildren()
https://bugs.webkit.org/show_bug.cgi?id=102352

Reviewed by Ojan Vafai.

Source/WebCore:

Avoid doing a second layout if we're going to get the same size as before.
This prevents us from doing an exponential number of layouts in some
common cases.

Test: css3/flexbox/stretch-after-sibling-size-change.html

html/shadow/SliderThumbElement.cpp:

(WebCore::RenderSliderContainer::layout): Force a layout of the track, which positions the thumb.

rendering/RenderFlexibleBox.cpp:

(WebCore::RenderFlexibleBox::needToStretchChild): Determine if a child is going to stretch.
(WebCore::RenderFlexibleBox::resetAutoMarginsAndLogicalTopInCrossAxis): Makes sure we're in a consistent state before
we apply auto margins.
(WebCore::RenderFlexibleBox::layoutAndPlaceChildren):

rendering/RenderFlexibleBox.h: Add needToStretchChild.

LayoutTests:

Add a test case to make sure we relayout when a sibling is stretching.

css3/flexbox/stretch-after-sibling-size-change-expected.txt: Added.
css3/flexbox/stretch-after-sibling-size-change.html: Added.

2:31 AM Changeset in webkit [136323] by commit-queue@webkit.org

24 edits
3 adds in trunk

Need a method to close all idle localstorage databases immediately.
https://bugs.webkit.org/show_bug.cgi?id=103469

Patch by Yongjun Zhang <yongjun_zhang@apple.com> on 2012-12-02
Reviewed by David Kilzer.

Source/WebCore:

Add a new PageGroup method closeIdleLocalStorageDatabases, which walks through
all local storage areas that don't have document referencing to them and closes
their underlying databases.

Test: storage/domstorage/localstorage/close-idle-localstorage-databases-immediately.html

WebCore.exp.in:
page/PageGroup.cpp:

(WebCore::PageGroup::clearLocalStorageForOrigin):
(WebCore):
(WebCore::PageGroup::closeIdleLocalStorageDatabases): walking through all idle local storages and

closing their databases.

page/PageGroup.h:

(PageGroup):

storage/StorageArea.h:

(WebCore::StorageArea::closeDatabaseIfIdle):

storage/StorageAreaImpl.cpp:

(WebCore::StorageAreaImpl::closeDatabaseIfIdle): close the database if it is idle.
(WebCore):

storage/StorageAreaImpl.h:

(StorageAreaImpl):

storage/StorageNamespace.h:

(StorageNamespace):

storage/StorageNamespaceImpl.cpp:

(WebCore::StorageNamespaceImpl::closeIdleLocalStorageDatabases):
(WebCore):

storage/StorageNamespaceImpl.h:

(StorageNamespaceImpl):

Source/WebKit/mac:

Add a new method to WebStorageManager that we can call to close all
idle local storage databases.

Storage/WebStorageManager.mm:

(+[WebStorageManager closeIdleLocalStorageDatabases]):

Storage/WebStorageManagerPrivate.h:

Tools:

Add an JS method to TestRunner, to enable testing closing idle local storage databases in
DumpRenderTree layout tests.

DumpRenderTree/TestRunner.cpp:

(closeIdleLocalStorageDatabasesCallback):
(TestRunner::staticFunctions):

DumpRenderTree/TestRunner.h:

(TestRunner):

DumpRenderTree/blackberry/TestRunnerBlackBerry.cpp: