Jul 7, 2014:

10:56 PM Changeset in webkit [170880] by psolanki@apple.com
  • 2 edits in trunk/Source/WebCore

Unreviewed. iOS build fix after r170871.

  • rendering/RenderThemeIOS.mm:


10:51 PM Changeset in webkit [170879] by achristensen@apple.com
  • 2 edits in trunk/Tools

Unreviewed build fix.

  • WinLauncher/WinLauncher.vcxproj/WinLauncherLib.vcxproj:

Make Common.cpp excluded from both 32-bit and 64-bit builds.

9:35 PM Changeset in webkit [170878] by benjamin@webkit.org
  • 2 edits in trunk/Source/JavaScriptCore

Fix the build after r170876

Patch by Benjamin Poulain <bpoulain@apple.com> on 2014-07-07

  • assembler/LinkBuffer.cpp:


9:26 PM Changeset in webkit [170877] by Alan Bujtas
  • 3 edits
    2 adds in trunk

Subpixel rendering: icloud.com password arrow has clipped circle at some window sizes.

Reviewed by Simon Fraser.

Transparent layer clipping needs device pixel snapping.

Test: fast/layers/wrong-clipping-semi-transparent-layer-on-subpixel-position.html

  • rendering/RenderLayer.cpp:



  • fast/layers/wrong-clipping-semi-transparent-layer-on-subpixel-position-expected.html: Added.
  • fast/layers/wrong-clipping-semi-transparent-layer-on-subpixel-position.html: Added.
9:23 PM Changeset in webkit [170876] by benjamin@webkit.org
  • 35 edits in trunk/Source

LinkBuffer should not keep a reference to the MacroAssembler

Reviewed by Geoffrey Garen.

In FTL, the LinkBuffer can outlive the MacroAssembler that was used for code generation.
When that happens, the pointer m_assembler points to released memory. That was not causing
issues because the attribute is not used after linking, but that was not particularily
future proof.

This patch refactors LinkBuffer to avoid any lifetime risk. The MacroAssembler is now passed
as a reference, it is used for linking but no reference is ever stored with the LinkBuffer.

While fixing the call sites to use a reference, I also discovered LinkBuffer.h was included
everywhere. I refactored some #include to avoid that.

  • assembler/LinkBuffer.cpp:


  • assembler/LinkBuffer.h:


  • bytecode/Watchpoint.cpp:
  • dfg/DFGDisassembler.cpp:
  • dfg/DFGDisassembler.h:
  • dfg/DFGJITCompiler.cpp:


  • dfg/DFGOSRExitCompiler.cpp:
  • dfg/DFGPlan.cpp:
  • dfg/DFGThunks.cpp:


  • ftl/FTLCompile.cpp:


  • ftl/FTLJSCall.cpp:
  • ftl/FTLJSCall.h:
  • ftl/FTLLink.cpp:


  • ftl/FTLLowerDFGToLLVM.cpp:
  • ftl/FTLOSRExitCompiler.cpp:


  • ftl/FTLThunks.cpp:


  • jit/ArityCheckFailReturnThunks.cpp:


  • jit/JIT.cpp:


  • jit/JITCall.cpp:


  • jit/JITCall32_64.cpp:


  • jit/JITDisassembler.cpp:
  • jit/JITDisassembler.h:
  • jit/JITOpcodes.cpp:
  • jit/JITPropertyAccess.cpp:


  • jit/JITPropertyAccess32_64.cpp:


  • jit/RegisterPreservationWrapperGenerator.cpp:


  • jit/Repatch.cpp:


  • jit/SpecializedThunkJIT.h:


  • jit/ThunkGenerators.cpp:


  • llint/LLIntThunks.cpp:


  • yarr/YarrJIT.cpp:



  • cssjit/SelectorCompiler.cpp:


9:13 PM Changeset in webkit [170875] by Alan Bujtas
  • 6 edits
    2 adds in trunk

Subpixel rendering: Inline box decoration rounds to integral.

Reviewed by Darin Adler.

This patch removes 2 integral roundings from InlineFlowBox:

  1. Border and padding sizes are implicitly integral truncated by the 'int' return type of borderLogicalLeft/Right()/paddingLogicalLeft/Right(). It results in losing fractional border/padding values.
  2. Painting rectangle is explicitly rounded which pushes border and other decoration elements to odd device pixel positions on retina displays.

These values get pixel snapped right before calling in to GraphicsContext::*.

Test: fast/inline/hidpi-inline-text-decoration-with-subpixel-value.html

  • rendering/InlineBox.h:


  • rendering/InlineFlowBox.cpp:

(WebCore::InlineFlowBox::roundedFrameRect): Deleted.

  • rendering/InlineFlowBox.h:



  • fast/inline/hidpi-inline-text-decoration-with-subpixel-value-expected.html: Added.
  • fast/inline/hidpi-inline-text-decoration-with-subpixel-value.html: Added.
  • platform/mac/css1/formatting_model/inline_elements-expected.txt:
9:00 PM Changeset in webkit [170874] by commit-queue@webkit.org
  • 2 edits in trunk/LayoutTests

[EFL] gardening after r170864

Unreviewed, EFL gardening.

Patch by Hunseop Jeong <Hunseop Jeong> on 2014-07-07

  • platform/efl/TestExpectations: Report and mark several unexpected image-only failures.
8:22 PM Changeset in webkit [170873] by Alan Bujtas
  • 3 edits in trunk/Source/WebCore

Pass RenderLayer reference instead of pointer to RenderLayer::paintingExtent().

Reviewed by Simon Fraser.

No change in functionality.

  • rendering/RenderLayer.cpp: * -> &

(WebCore::RenderLayer::paintingExtent): Deleted.

  • rendering/RenderLayer.h:
7:40 PM Changeset in webkit [170872] by timothy_horton@apple.com
  • 2 edits in trunk/Source/WebKit2

Turn on accelerated drawing for WebKit2 by default

Reviewed by Simon Fraser.

  • Shared/WebPreferencesDefinitions.h:

Make Mac match iOS, in that accelerated drawing is on by default.

7:27 PM Changeset in webkit [170871] by mmaxfield@apple.com
  • 8 edits in trunk/Source/WebCore

Remove misleadingly-named Font::isSVGFont()

Reviewed by Dean Jackson.

Being a SVG font is a property of the FontData object, not
the Font object. Providing an isSVGFont() function on the
Font class will likely lead to incorrect code and non-obvious

No new tests because there is no behavior change.

  • platform/graphics/Font.h:

(WebCore::Font::isSVGFont): Deleted.

  • platform/graphics/mac/FontMac.mm:


  • rendering/InlineTextBox.cpp:


  • rendering/RenderBlock.cpp:


  • rendering/SimpleLineLayout.cpp:


  • rendering/svg/SVGInlineTextBox.cpp:


  • rendering/svg/SVGTextMetrics.cpp:


6:56 PM Changeset in webkit [170870] by timothy_horton@apple.com
  • 2 edits in trunk/Source/WebKit2

Don't leak _WKRemoteObjectRegistry in WKBrowsingContextController

Reviewed by Simon Fraser.

  • UIProcess/API/Cocoa/WKBrowsingContextController.mm:

(-[WKBrowsingContextController _remoteObjectRegistry]):
Adoption is important!!

6:52 PM Changeset in webkit [170869] by beidson@apple.com
  • 13 edits in trunk/Source/WebCore

Move Gamepad index down into PlatformGamepad.

Reviewed by Darin Adler.

PlatformGamepads always have an inherent index anyways.
This will make things easier in all upcoming work.

  • Modules/gamepad/Gamepad.cpp:


  • Modules/gamepad/Gamepad.h:
  • Modules/gamepad/GamepadManager.cpp:


  • Modules/gamepad/GamepadManager.h:
  • Modules/gamepad/NavigatorGamepad.cpp:


  • Modules/gamepad/NavigatorGamepad.h:
  • platform/GamepadProviderClient.h:
  • platform/PlatformGamepad.h:


  • platform/mac/HIDGamepad.cpp:


  • platform/mac/HIDGamepad.h:
  • platform/mac/HIDGamepadProvider.cpp:


  • platform/mac/HIDGamepadProvider.h:
6:38 PM Changeset in webkit [170868] by jonowells@apple.com
  • 2 edits in trunk/Source/WebInspectorUI

Web Inspector: CSS style declaration editor should ignore empty lines when determining prefix whitespace

Reviewed by Joseph Pecoraro.

When checking for zero-length prefix whitespace, completely empty lines are now ignored
as these lines are most often used for organizing rules into groups and often have their
whitespace removed due to trailing whitespace removal conventions.

  • UserInterface/Views/CSSStyleDeclarationTextEditor.js:
5:14 PM Changeset in webkit [170867] by Lucas Forschler
  • 5 edits in branches/safari-537.75-branch/Source


5:10 PM Changeset in webkit [170866] by Lucas Forschler
  • 1 copy in tags/Safari-537.75.18

New Tag.

5:07 PM Changeset in webkit [170865] by Lucas Forschler
  • 4 edits in branches/safari-537.75-branch/Source

Merged r169927. <rdar://problem/17563093>

4:44 PM Changeset in webkit [170864] by Simon Fraser
  • 15 edits in trunk/Source

[UI-side compositing] Support reflections on custom layers like video

Reviewed by Tim Horton.


Add a LayerTypeWebGLLayer layer type, and support casting to PlatformCALayerRemoteCustom.

  • WebCore.exp.in: Export PlatformCALayerMac::layerTypeForPlatformLayer(CALayer*), and sort.
  • platform/graphics/ca/PlatformCALayer.h:


  • platform/graphics/ca/mac/PlatformCALayerMac.h:
  • platform/graphics/ca/mac/PlatformCALayerMac.mm:



For video reflections, we have to support cloning of PlatformCALayerRemoteCustom
in the web process. Do so by implementing PlatformCALayerRemoteCustom::clone(),
which does the right gyrations to get AVPlayerLayers cloned, then makes a
new PlatformCALayerRemoteCustom to wrap the new layer. This ends up getting
its own context hosting ID, allowing the clone to show in the UI process.

Attempt to do the same for WebGL, but turn it off because it breaks.

  • Shared/mac/RemoteLayerBackingStore.mm:


  • Shared/mac/RemoteLayerTreeTransaction.mm:

(WebKit::RemoteLayerTreeTransaction::LayerCreationProperties::encode): Unconditionally encode/decode
the hostingContextID. It will be 0 for most layers.

  • UIProcess/ios/RemoteLayerTreeHostIOS.mm:


  • UIProcess/mac/RemoteLayerTreeHost.mm:


  • WebProcess/WebPage/mac/PlatformCALayerRemote.cpp:

(WebKit::PlatformCALayerRemote::create): Creation with a custom PlatformLayer* always
creates a PlatformCALayerRemoteCustom.
(WebKit::PlatformCALayerRemote::clone): Factor some code.

  • WebProcess/WebPage/mac/PlatformCALayerRemote.h:
  • WebProcess/WebPage/mac/PlatformCALayerRemoteCustom.h:
  • WebProcess/WebPage/mac/PlatformCALayerRemoteCustom.mm:

(WebKit::PlatformCALayerRemoteCustom::clone): Clone by making an instance of the
correct type of platform layer when possible, then wrapping a PlatformCALayerRemoteCustom
around it.

  • WebProcess/WebPage/mac/RemoteLayerTreeContext.mm:


4:44 PM Changeset in webkit [170863] by commit-queue@webkit.org
  • 3 edits in trunk/Source/WebInspectorUI

Web Inspector: Uncaught Exception debugging iOS 7 NetworkAgent.loadResource is unavailable

Patch by Joseph Pecoraro <Joseph Pecoraro> on 2014-07-07
Reviewed by Darin Adler.

Gracefully handle if NetworkAgent.loadResource is unavailable.

  • UserInterface/Controllers/SourceMapManager.js:


  • UserInterface/Models/SourceMapResource.js:
4:44 PM Changeset in webkit [170862] by Simon Fraser
  • 5 edits in trunk/Source

[UI-side compositing] Crash when starting a filter transition on a reflected layer

Reviewed by Tim Horton.


Don't call the owner if we failed to find the animation key (which actually
isn't used by PlatformCALayerMac anyway).

  • platform/graphics/ca/mac/PlatformCALayerMac.mm:

(-[WebAnimationDelegate animationDidStart:]):


When cloned layers had animations, we would fire two animationDidStart callbacks,
but the second would pass an empty animationKey string to the web process, resulting
in a crash.

Fix by not blindly copying all layer properties when cloning PlatformCALayerRemotes,
since the clone would include addedAnimations, and then get the same animations
added on top by the caller.

Also protect against an empty animation key in the animationDidStart callback.

  • UIProcess/mac/RemoteLayerTreeHost.mm:


  • WebProcess/WebPage/mac/PlatformCALayerRemote.cpp:

(WebKit::PlatformCALayerRemote::clone): Don't copy all the properties; copy
them manually as PlatformCALayerMac does. Only copy the big things if they don't
have their default values.
(WebKit::PlatformCALayerRemote::copyFiltersFrom): Need an implementation of this
for clone() to call.

4:44 PM Changeset in webkit [170861] by Simon Fraser
  • 2 edits in trunk/Tools

Have MiniBrowser log and reload if the Web Content Process crashes

Reviewed by Tim Horton.

Implement _webViewWebProcessDidCrash, log and reload when the web process

  • MiniBrowser/mac/WK2BrowserWindowController.m:

(-[WK2BrowserWindowController _webViewWebProcessDidCrash:]):

4:33 PM Changeset in webkit [170860] by mhahnenberg@apple.com
  • 2 edits in branches/ftlopt/Source/JavaScriptCore

Merge r170011 from trunk.

2014-06-16 Michael Saboff <msaboff@apple.com>

Convert ASSERT in inlineFunctionForCapabilityLevel to early return

Reviewed by Mark Hahnenberg.

Hardened code by Converting ASSERT to return CannotCompile.

  • dfg/DFGCapabilities.h: (JSC::DFG::inlineFunctionForCapabilityLevel):
4:18 PM Changeset in webkit [170859] by timothy_horton@apple.com
  • 2 edits in trunk/Source/WebKit2

Nearly everything in the UIProcess "leaks" when WKWebView is torn down

Reviewed by Simon Fraser.

  • UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView initWithFrame:configuration:]):
Adoption is important!

2:42 PM Changeset in webkit [170858] by enrica@apple.com
  • 9 edits in trunk/Source/WebKit2

REGRESSION(iOS WK2): arrow keys movements don't work.

Reviewed by Benjamin Poulain.

This change add the implementations for cursor movement selectors.
It also provides a mechanism to ensure that every selection
change in WebKit is reflected in UIKit so that it can update the
autocorrection data. This way we ensure that the autocorraction state
is consistent even when selection changes are originated by JavaScript.
WebPageProxy::editorStateChanged now notifies the page client when the
selection is about to change as well as when it actually changed.

  • UIProcess/PageClient.h:
  • UIProcess/WebPageProxy.cpp:


  • UIProcess/WebPageProxy.h:
  • UIProcess/ios/PageClientImplIOS.h:
  • UIProcess/ios/PageClientImplIOS.mm:


  • UIProcess/ios/WKContentViewInteraction.h:
  • UIProcess/ios/WKContentViewInteraction.mm:

(-[WKContentView cut:]): Removed incorrect calls to textWillChange and
(-[WKContentView paste:]):
(-[WKContentView _moveUp:withHistory:]):
(-[WKContentView _moveDown:withHistory:]):
(-[WKContentView _moveLeft:withHistory:]):
(-[WKContentView _moveRight:withHistory:]):
(-[WKContentView _moveToStartOfWord:withHistory:]):
(-[WKContentView _moveToStartOfParagraph:withHistory:]):
(-[WKContentView _moveToStartOfLine:withHistory:]):
(-[WKContentView _moveToStartOfDocument:withHistory:]):
(-[WKContentView _moveToEndOfWord:withHistory:]):
(-[WKContentView _moveToEndOfParagraph:withHistory:]):
(-[WKContentView _moveToEndOfLine:withHistory:]):
(-[WKContentView _moveToEndOfDocument:withHistory:]):
(-[WKContentView _selectionWillChange]):
(-[WKContentView _selectionChanged]):

  • UIProcess/ios/WebPageProxyIOS.mm:


2:06 PM Changeset in webkit [170857] by akling@apple.com
  • 2 edits in trunk/Source/JavaScriptCore

Fast path for jsStringWithCache() when asked for the same string repeatedly.

Reviewed by Darin Adler.

Follow-up to r170818 addressing a review comment by Geoff Garen.

  • runtime/JSString.cpp:


1:52 PM Changeset in webkit [170856] by andersca@apple.com
  • 8 edits in trunk/Source/WebKit2

Add SPI for saving and restoring a WKWebView's _WKSessionState

Reviewed by Dan Bernstein.

  • UIProcess/API/C/WKPage.cpp:


  • UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView _sessionState]):
(-[WKWebView _restoreFromSessionStateData:]):
(-[WKWebView _restoreFromSessionState:]):
(-[WKWebView _restoreSessionState:andNavigate:]):

  • UIProcess/API/Cocoa/WKWebViewPrivate.h:
  • UIProcess/API/Cocoa/_WKSessionState.mm:

(-[_WKSessionState _initWithSessionState:]):

  • UIProcess/API/Cocoa/_WKSessionStateInternal.h:
  • UIProcess/WebPageProxy.cpp:


  • UIProcess/WebPageProxy.h:
1:39 PM Changeset in webkit [170855] by fpizlo@apple.com
  • 36 edits
    14 adds in branches/ftlopt

[ftlopt] Infer immutable object properties

Reviewed by Mark Hahnenberg.


This introduces a new way of inferring immutable object properties. A property is said to
be immutable if after its creation (i.e. the transition that creates it), we never
overwrite it (i.e. replace it) or delete it. Immutability is a property of an "own
property" - so if we say that "f" is immutable at "o" then we are implying that "o" has "f"
directly and not on a prototype. More specifically, the immutability inference will prove
that a property on some structure is immutable. This means that, for example, we may have a
structure S1 with property "f" where we claim that "f" at S1 is immutable, but S1 has a
transition to S2 that adds a new property "g" and we may claim that "f" at S2 is actually
mutable. This is mainly for convenience; it allows us to decouple immutability logic from
transition logic. Immutability can be used to constant-fold accesses to objects at
DFG-time. The DFG needs to prove the following to constant-fold the access:

  • The base of the access must be a constant object pointer. We prove that a property at a structure is immutable, but that says nothing of its value; each actual instance of that property may have a different value. So, a constant object pointer is needed to get an actual constant instance of the immutable value.

  • A check (or watchpoint) must have been emitted proving that the object has a structure that allows loading the property in question.

  • The replacement watchpoint set of the property in the structure that we've proven the object to have is still valid and we add a watchpoint to it lazily. The replacement watchpoint set is the key new mechanism that this change adds. It's possible that we have proven that the object has one of many structures, in which case each of those structures needs a valid replacement watchpoint set.

The replacement watchpoint set is created the first time that any access to the property is
cached. A put replace cache will create, and immediately invalidate, the watchpoint set. A
get cache will create the watchpoint set and make it start watching. Any non-cached put
access will invalidate the watchpoint set if one had been created; the underlying algorithm
ensures that checking for the existence of a replacement watchpoint set is very fast in the
common case. This algorithm ensures that no cached access needs to ever do any work to
invalidate, or check the validity of, any replacement watchpoint sets. It also has some
other nice properties:

  • It's very robust in its definition of immutability. The strictest that it will ever be is that for any instance of the object, the property must be written to only once, specifically at the time that the property is created. But it's looser than this in practice. For example, the property may be written to any number of times before we add the final property that the object will have before anyone reads the property; this works since for optimization purposes we only care if we detect immutability on the structure that the object will have when it is most frequently read from, not any previous structure that the object had. Also, we may write to the property any number of times before anyone caches accesses to it.

  • It is mostly orthogonal to structure transitions. No new structures need to be created to track the immutability of a property. Hence, there is no risk from this feature causing more polymorphism. This is different from the previous "specificValue" constant inference, which did cause additional structures to be created and sometimes those structures led to fake polymorphism. This feature does leverage existing transitions to do some of the watchpointing: property deletions don't fire the replacement watchpoint set because that would cause a new structure and so the mandatory structure check would fail. Also, this feature is guaranteed to never kick in for uncacheable dictionaries because those wouldn't allow for cacheable accesses - and it takes a cacheable access for this feature to be enabled.

  • No memory overhead is incurred except when accesses to the property are cached. Dictionary properties will typically have no meta-data for immutability. The number of replacement watchpoint sets we allocate is proportional to the number of inline caches in the program, which is typically must smaller than the number of structures or even the number of objects.

This inference is far more powerful than the previous "specificValue" inference, so this
change also removes all of that code. It's interesting that the amount of code that is
changed to remove that feature is almost as big as the amount of code added to support the
new inference - and that's if you include the new tests in the tally. Without new tests,
it appears that the new feature actually touches less code!

There is one corner case where the previous "specificValue" inference was more powerful.
You can imagine someone creating objects with functions as self properties on those
objects, such that each object instance had the same function pointers - essentially,
someone might be trying to create a vtable but failing at the whole "one vtable for many
instances" concept. The "specificValue" inference would do very well for such programs,
because a structure check would be sufficient to prove a constant value for all of the
function properties. This new inference will fail because it doesn't track the constant
values of constant properties; instead it detects the immutability of otherwise variable
properties (in the sense that each instance of the property may have a different value).
So, the new inference requires having a particular object instance to actually get the
constant value. I think it's OK to lose this antifeature. It took a lot of code to support
and was a constant source of grief in our transition logic, and there doesn't appear to be
any real evidence that programs benefited from that particular kind of inference since
usually it's the singleton prototype instance that has all of the functions.

This change is a speed-up on everything. date-format-xparb and both SunSpider/raytrace and
V8/raytrace seem to be the biggest winners among the macrobenchmarks; they see >5%
speed-ups. Many of our microbenchmarks see very large performance improvements, even 80% in
one case.

  • bytecode/ComplexGetStatus.cpp:


  • bytecode/GetByIdStatus.cpp:


  • bytecode/GetByIdVariant.cpp:


  • bytecode/GetByIdVariant.h:

(JSC::GetByIdVariant::specificValue): Deleted.

  • bytecode/PutByIdStatus.cpp:


  • bytecode/PutByIdVariant.cpp:


  • bytecode/PutByIdVariant.h:

(JSC::PutByIdVariant::specificValue): Deleted.

  • bytecode/Watchpoint.cpp:

(JSC::WatchpointSet::fireAll): Deleted.

  • bytecode/Watchpoint.h:


  • dfg/DFGAbstractInterpreterInlines.h:


  • dfg/DFGByteCodeParser.cpp:


  • dfg/DFGConstantFoldingPhase.cpp:


  • dfg/DFGFixupPhase.cpp:


  • dfg/DFGGraph.cpp:


  • dfg/DFGGraph.h:
  • dfg/DFGWatchableStructureWatchingPhase.cpp:


  • ftl/FTLLowerDFGToLLVM.cpp:


  • jit/JITOperations.cpp:
  • jit/Repatch.cpp:


  • llint/LLIntSlowPaths.cpp:


  • runtime/CommonSlowPaths.h:


  • runtime/IntendedStructureChain.cpp:


  • runtime/JSCJSValue.cpp:


  • runtime/JSGlobalObject.cpp:


  • runtime/JSObject.cpp:

(JSC::getCallableObjectSlow): Deleted.
(JSC::JSObject::getPropertySpecificValue): Deleted.

  • runtime/JSObject.h:

(JSC::getCallableObject): Deleted.

  • runtime/JSScope.cpp:


  • runtime/PropertyMapHashTable.h:


  • runtime/PropertyTable.cpp:

(JSC::PropertyTable::visitChildren): Deleted.

  • runtime/Structure.cpp:

(JSC::Structure::despecifyDictionaryFunction): Deleted.
(JSC::Structure::despecifyFunctionTransition): Deleted.
(JSC::Structure::despecifyFunction): Deleted.
(JSC::Structure::despecifyAllFunctions): Deleted.
(JSC::Structure::putSpecificValue): Deleted.

  • runtime/Structure.h:

(JSC::Structure::transitionDidInvolveSpecificValue): Deleted.
(JSC::Structure::disableSpecificFunctionTracking): Deleted.

  • runtime/StructureInlines.h:


  • runtime/StructureRareData.cpp:


  • runtime/StructureRareData.h:
  • tests/stress/infer-constant-global-property.js: Added.


  • tests/stress/infer-constant-property.js: Added.


  • tests/stress/jit-cache-poly-replace-then-cache-get-and-fold-then-invalidate.js: Added.


  • tests/stress/jit-cache-replace-then-cache-get-and-fold-then-invalidate.js: Added.


  • tests/stress/jit-put-to-scope-global-cache-watchpoint-invalidate.js: Added.


  • tests/stress/llint-cache-replace-then-cache-get-and-fold-then-invalidate.js: Added.


  • tests/stress/llint-put-to-scope-global-cache-watchpoint-invalidate.js: Added.


  • tests/stress/repeat-put-to-scope-global-with-same-value-watchpoint-invalidate.js: Added.



  • js/regress/infer-constant-global-property-expected.txt: Added.
  • js/regress/infer-constant-global-property.html: Added.
  • js/regress/infer-constant-property-expected.txt: Added.
  • js/regress/infer-constant-property.html: Added.
  • js/regress/script-tests/infer-constant-global-property.js: Added.
  • js/regress/script-tests/infer-constant-property.js: Added.
1:28 PM Changeset in webkit [170854] by andersca@apple.com
  • 3 edits
    5 copies in trunk/Source/WebKit2

Add a stubbed out _WKSessionState class

Reviewed by Geoffrey Garen.

  • UIProcess/API/Cocoa/_WKSessionState.h: Copied from Source/WebKit2/UIProcess/LegacySessionStateCoding.h.
  • UIProcess/API/Cocoa/_WKSessionState.mm: Copied from Source/WebKit2/UIProcess/LegacySessionStateCoding.h.

(-[_WKSessionState initWithData:]):
(-[_WKSessionState data]):

  • UIProcess/API/Cocoa/_WKSessionStateInternal.h: Copied from Source/WebKit2/UIProcess/LegacySessionStateCoding.h.
  • UIProcess/Cocoa/SessionStateCoding.h: Copied from Source/WebKit2/UIProcess/LegacySessionStateCoding.h.
  • UIProcess/Cocoa/SessionStateCoding.mm: Copied from Source/WebKit2/UIProcess/LegacySessionStateCoding.h.


  • UIProcess/LegacySessionStateCoding.h:
  • WebKit2.xcodeproj/project.pbxproj:
12:52 PM Changeset in webkit [170853] by andersca@apple.com
  • 3 edits in trunk/Source/WebKit2

Some of the WKPreferences (API) property getters read like questions

Reviewed by Dan Bernstein.

  • UIProcess/API/Cocoa/WKPreferences.h:
  • UIProcess/API/Cocoa/WKPreferences.mm:

(-[WKPreferences javaScriptIsEnabled]):
(-[WKPreferences javaIsEnabled]):
(-[WKPreferences plugInsAreEnabled]):
(-[WKPreferences isJavaScriptEnabled]): Deleted.
(-[WKPreferences isJavaEnabled]): Deleted.
(-[WKPreferences arePlugInsEnabled]): Deleted.

11:50 AM Changeset in webkit [170852] by beidson@apple.com
  • 2 edits in trunk/Source/WebKit2

ServicesOverlayController menus show up in the wrong place.
<rdar://problem/17130576> and https://bugs.webkit.org/show_bug.cgi?id=134684

Reviewed by Tim Horton.

  • WebProcess/WebPage/mac/ServicesOverlayController.mm:

(WebKit::ServicesOverlayController::selectedTelephoneNumberRangesChanged): Handle cases where

the telephone number is in a subframe.

(WebKit::ServicesOverlayController::handleClick): The click point is always in main frame document

coordinates, so convert it to window coordinates using the main frame’s FrameView.

11:31 AM Changeset in webkit [170851] by commit-queue@webkit.org
  • 4 edits in trunk/Source/JavaScriptCore

Add missing ENABLE(FTL_JIT) guards

Patch by Tibor Meszaros <tmeszaros.u-szeged@partner.samsung.com> on 2014-07-07
Reviewed by Darin Adler.

  • ftl/FTLDWARFDebugLineInfo.cpp:
  • ftl/FTLDWARFDebugLineInfo.h:
  • ftl/FTLGeneratedFunction.h:
11:30 AM Changeset in webkit [170850] by bshafiei@apple.com
  • 5 edits in branches/safari-537.75-branch/Source


11:21 AM Changeset in webkit [170849] by commit-queue@webkit.org
  • 2 edits in trunk/Source/WebCore

[iOS WebGL] Fix crash with too many nested glsl functions.

Patch by Alex Christensen <achristensen@webkit.org> on 2014-07-07
Reviewed by Geoffrey Garen.

This fixes the Khronos WebGL 1.0.3 test nested-functions-should-not-crash.html.

  • platform/graphics/opengl/Extensions3DOpenGLCommon.cpp:

Add SH_LIMIT_CALL_STACK_DEPTH to extraCompileOptions.

11:18 AM Changeset in webkit [170848] by berto@igalia.com
  • 2 edits in trunk/Tools

update-webkitgtk-libs script fails by unsatisfied dependencies

Reviewed by Martin Robinson.

Install libxcb-xfixes0-dev, needed to build Mesa.

  • gtk/install-dependencies:
10:52 AM Changeset in webkit [170847] by dbates@webkit.org
  • 1 edit
    2 adds in trunk/LayoutTests

Add test to ensure that SVGDocumentExtensions::rebuildElements() doesn't rebuild invalid elements

Reviewed by David Kilzer.

Add a test that implicitly checks that the list of SVG elements to be rebuilt doesn't contain
removed elements. Run this test in DRT with Guard Malloc/MallocScribble (or the platform-
specific equivalent) enabled.

  • svg/custom/overwrite-page-that-has-use-elements-expected.txt: Added.
  • svg/custom/overwrite-page-that-has-use-elements.html: Added.
10:20 AM Changeset in webkit [170846] by Alan Bujtas
  • 2 edits in trunk/LayoutTests

Unreviewed Mac gardening after r170817.

  • platform/mac-wk2/TestExpectations:
9:21 AM Changeset in webkit [170845] by mitz@apple.com
  • 2 edits in trunk/Source/WebKit2

[Cocoa] Assertion failure in NavigationState::LoaderClient::didStartProvisionalLoadForFrame (navigationID) when navigating through the page cache

Reviewed by Tim Horton.

  • UIProcess/Cocoa/NavigationState.mm:

(WebKit::NavigationState::LoaderClient::didStartProvisionalLoadForFrame): Replaced the
assertion with a FIXME.
(WebKit::NavigationState::LoaderClient::didFailProvisionalLoadWithErrorForFrame): Ditto.
(WebKit::NavigationState::LoaderClient::didCommitLoadForFrame): Ditto.
(WebKit::NavigationState::LoaderClient::didFinishDocumentLoadForFrame): Ditto.
(WebKit::NavigationState::LoaderClient::didFinishLoadForFrame): Ditto.
(WebKit::NavigationState::LoaderClient::didFailLoadWithErrorForFrame): Ditto.

4:48 AM Changeset in webkit [170844] by commit-queue@webkit.org
  • 2 edits in trunk/Source/WebKit2

[EFL] Fix cursor artifacts on naver map site.

Patch by KwangHyuk Kim <hyuki.kim@samsung.com> on 2014-07-07
Reviewed by Gyuyoung Kim.

Old evas cursor and ecore x cursor are reset before new cursor is applied.

  • UIProcess/API/efl/EwkView.cpp:


2:04 AM Changeset in webkit [170843] by zandobersek@gmail.com
  • 5 edits in trunk/Source

Enable ARMv7 disassembler for the GTK port

Reviewed by Benjamin Poulain.


  • CMakeLists.txt: Add ARMv7DOpcode.cpp file to the build.
  • disassembler/ARMv7/ARMv7DOpcode.cpp: Include the string.h header for strlen().


  • wtf/Platform.h: Enable the ARMv7 disassembler for the GTK port when

targeting the ARMv7 architecture.

12:38 AM Changeset in webkit [170842] by commit-queue@webkit.org
  • 2 edits in trunk/LayoutTests

[GTK] Unreviewed GTK gardening.

Remove failure expectations for tests that have been passing after r168046, r168350, r168400,
r168479, r169309, r169620, r169689 and r169799.

Patch by Lorenzo Tilve <ltilve@igalia.com> on 2014-07-07

  • platform/gtk/TestExpectations:
12:16 AM Changeset in webkit [170841] by Ion Rosca
  • 6 edits
    6 adds in trunk

[CSS Blending]The background images set on the root element will blend on an initial white backdrop.

Reviewed by Dean Jackson.

Tests: css3/blending/background-blend-mode-body-image.html


  • rendering/RenderBox.cpp:

If the background layers have blend modes, we paint the base background color first,
before starting a new transparency layer. After creating the transparency layer,
the base background will be skipped, and the background layers will blend on top of
a transparent backdrop.

  • rendering/RenderBox.h:
  • rendering/RenderBoxModelObject.cpp:

This method is able to paint only the base background color (BaseBackgroundColorOnly),
skipping the background-color and background-images. This option is available only
on bottom layer of the root renderer and only if the background color is not opaque,
otherwise the method will return early.
Another option is to paint the root background without using the base background color

  • rendering/RenderBoxModelObject.h:

Adding the BaseBackgroundColorUsage enum.


  • css3/blending/background-blend-mode-body-image-expected.html: Added.
  • css3/blending/background-blend-mode-body-image.html: Added.
  • css3/blending/background-blend-mode-body-transparent-color-and-image-expected.html: Added.
  • css3/blending/background-blend-mode-body-transparent-color-and-image.html: Added.
  • css3/blending/background-blend-mode-body-transparent-image-expected.html: Added.
  • css3/blending/background-blend-mode-body-transparent-image.html: Added.
12:14 AM WebKitGTK/KeepingTheTreeGreen edited by ltilve@igalia.com

Jul 6, 2014:

11:26 PM Changeset in webkit [170840] by benjamin@webkit.org
  • 2 edits in trunk/Source/WebKit2

[iOS][WK2] The tap highlight ID is not invalidated when a long press ends

Patch by Benjamin Poulain <bpoulain@apple.com> on 2014-07-06
Reviewed by Darin Adler.

  • UIProcess/ios/WKContentViewInteraction.mm:

(-[WKContentView _cancelInteraction]):
(-[WKContentView _finishInteraction]):
I messed that up in r170600. I only accounted for the web process being too fast.

If the web process is too slow, _cancelInteraction or _finishInteraction happen before
_didGetTapHighlightForRequest:, and we also need to nuke the tap highlight.

11:24 PM Changeset in webkit [170839] by benjamin@webkit.org
  • 5 edits in trunk/Source/JavaScriptCore

[ARMv7] Use 16 bits instructions for push/pop when possible

Reviewed by Andreas Kling.

  • assembler/ARMv7Assembler.h:

Add the 16 bits version of push and pop.

  • assembler/MacroAssemblerARMv7.h:

Use the new push/pop instead of a regular load/store.

  • disassembler/ARMv7/ARMv7DOpcode.cpp:


  • disassembler/ARMv7/ARMv7DOpcode.h:

Fix the disassembler for push/pop:
-The register mask was on 7 bits for some reason.
-The code printing the registers was comparing a register ID with a register


11:04 PM Changeset in webkit [170838] by timothy_horton@apple.com
  • 5 edits in trunk/Source/WebKit2

[WK2] Don't support dynamically enabling the RemoteLayerTree debug indicator

Reviewed by Darin Adler.

  • UIProcess/DrawingAreaProxy.h:

(WebKit::DrawingAreaProxy::setShouldShowDebugIndicator): Deleted.

  • UIProcess/WebPageProxy.cpp:


  • UIProcess/mac/RemoteLayerTreeDrawingAreaProxy.h:
  • UIProcess/mac/RemoteLayerTreeDrawingAreaProxy.mm:

(WebKit::RemoteLayerTreeDrawingAreaProxy::setShouldShowDebugIndicator): Deleted.
The indicator depends on the live layer tree commits coming in, including layer creation,
so it can't be turned on after the root layer is created. Even a refresh is insufficient
to get it working; the debug indicator really needs to exist from the first commit.
So, don't even attempt to dynamically enable the indicator; create it at
RemoteLayerTreeDrawingAreaProxy construction time if the pref is enabled.

10:02 PM Changeset in webkit [170837] by yoav@yoav.ws
  • 18 edits in trunk

Turn on img@sizes compile flag

Reviewed by Benjamin Poulain.


  • Configurations/FeatureDefines.xcconfig: Moved compile flag to alphabetical order.


  • Configurations/FeatureDefines.xcconfig: Moved compile flag to alphabetical order.


  • mac/Configurations/FeatureDefines.xcconfig: Moved compile flag to alphabetical order.


  • Configurations/FeatureDefines.xcconfig: Moved compile flag to alphabetical order.


  • wtf/FeatureDefines.h: Added compile flag definition.


  • Scripts/webkitperl/FeatureList.pm: Turn on the img@sizes feature.


  • win/tools/vsprops/FeatureDefines.props: Added compile flag.


9:49 PM Changeset in webkit [170836] by fpizlo@apple.com
  • 27 edits in branches/ftlopt/Source/JavaScriptCore

Add more coverage for the profile_types_with_high_fidelity op code.

Patch by Saam Barati <sbarati@apple.com> on 2014-07-03
Reviewed by Filip Pizlo.

More operations are now being recorded by the profile_types_with_high_fidelity
opcode. Specifically: function parameters, function return values,
function 'this' value, get_by_id, get_by_value, resolve nodes, function return
values at the call site. Added more flags to the profile_types_with_high_fidelity
opcode so more focused tasks can take place when the instruction is
being linked in CodeBlock. Re-worked the type profiler to search
through character offset ranges when asked for the type of an expression
at a given offset. Removed redundant calls to Structure::toStructureShape
in HighFidelityLog and TypeSet by caching calls based on StructureID.

  • bytecode/BytecodeList.json:
  • bytecode/BytecodeUseDef.h:


  • bytecode/CodeBlock.cpp:


  • bytecode/CodeBlock.h:


  • bytecode/TypeLocation.h:
  • bytecode/UnlinkedCodeBlock.cpp:


  • bytecode/UnlinkedCodeBlock.h:
  • bytecompiler/BytecodeGenerator.cpp:


  • bytecompiler/BytecodeGenerator.h:


  • bytecompiler/NodesCodegen.cpp:


  • inspector/agents/InspectorRuntimeAgent.cpp:

(Inspector::InspectorRuntimeAgent::getRuntimeTypeForVariableInTextRange): Deleted.

  • inspector/agents/InspectorRuntimeAgent.h:
  • inspector/protocol/Runtime.json:
  • llint/LLIntSlowPaths.cpp:


  • llint/LLIntSlowPaths.h:
  • llint/LowLevelInterpreter.asm:
  • runtime/HighFidelityLog.cpp:

(JSC::HighFidelityLog::recordTypeInformationForLocation): Deleted.

  • runtime/HighFidelityLog.h:


  • runtime/HighFidelityTypeProfiler.cpp:

(JSC::HighFidelityTypeProfiler::getTypesForVariableInRange): Deleted.
(JSC::HighFidelityTypeProfiler::getGlobalTypesForVariableInRange): Deleted.
(JSC::HighFidelityTypeProfiler::getLocalTypesForVariableInRange): Deleted.
(JSC::HighFidelityTypeProfiler::getLocationBasedHash): Deleted.

  • runtime/HighFidelityTypeProfiler.h:

(JSC::LocationKey::LocationKey): Deleted.
(JSC::LocationKey::hash): Deleted.
(JSC::LocationKey::operator==): Deleted.

  • runtime/Structure.cpp:


  • runtime/Structure.h:
  • runtime/TypeSet.cpp:

(JSC::TypeSet::removeDuplicatesInStructureHistory): Deleted.

  • runtime/TypeSet.h:


  • runtime/VM.cpp:

(JSC::VM::getTypesForVariableInRange): Deleted.

  • runtime/VM.h:
8:42 PM Changeset in webkit [170835] by hw1008.kim@samsung.com
  • 2 edits in trunk/Tools

Unreviewed. Adding myself to committers.

  • Scripts/webkitpy/common/config/contributors.json:
7:06 PM Changeset in webkit [170834] by commit-queue@webkit.org
  • 2 edits in trunk/Source/JavaScriptCore

Flags value of SourceCodeKey should be unique for each case.

Patch by Daewoong Jang <daewoong.jang@navercorp.com> on 2014-07-06
Reviewed by Darin Adler.

Different combinations of CodeType and JSParserStrictness could generate same m_flags value because
the value of CodeType and the value of JSParserStrictness shares a bit inside m_flags member variable.
Shift the value of CodeType one bit farther to the left so those values don't overlap.

  • runtime/CodeCache.h:


6:22 PM Changeset in webkit [170833] by gyuyoung.kim@samsung.com
  • 4 edits in trunk/Source/WebCore

Generate isMathMLFoo() toMathMLFoo() by using NODE_TYPE_CASTS macro and generateTypeHelpers

Reviewed by Andreas Kling.

Use type cast macro and generation tag for isFoo() and toFoo() in some MathML classes.

No new tests, no behavior change.

  • mathml/MathMLElement.h: Place MathMLElementTypeHelpers.h to the bottom of the

file so that functions from that file can use the Element|Node classes and other things defined in this file.

  • mathml/MathMLMencloseElement.h:

(WebCore::toMathMLMencloseElement): Deleted.

  • mathml/mathtags.in:
6:18 PM Changeset in webkit [170832] by Yusuke Suzuki
  • 10 edits in trunk/Source/WebCore

CSS JIT: Ensure resolvingMode size is 1 byte

Reviewed by Benjamin Poulain.

Since selector compiler uses MacroAssembler::branch8 for SelectorChecker::Mode,
we should ensure sizeof(SelectorChecker::Mode) == 1.
Old implementation works only in little endianness environments.

  • css/ElementRuleCollector.cpp:


  • css/ElementRuleCollector.h:


  • css/SelectorChecker.cpp:


  • css/SelectorChecker.h:
  • css/StyleInvalidationAnalysis.cpp:


  • css/StyleResolver.cpp:


  • css/StyleResolver.h:


  • cssjit/SelectorCompiler.cpp:


  • dom/SelectorQuery.cpp:


5:01 PM Changeset in webkit [170831] by beidson@apple.com
  • 3 edits in trunk/Source/WebKit2

Speculative fix for: NetworkProcess sometimes hangs under copyDefaultCredentialForProtectionSpace

Reviewed by Tim Horton.

  • Shared/mac/SecItemRequestData.cpp:

(WebKit::SecItemRequestData::encode): Encode whether or not the query dictionary exists.
(WebKit::SecItemRequestData::decode): Only fail when query dictionary fails to decode if we expect one.

  • UIProcess/mac/SecItemShimProxy.cpp:

(WebKit::SecItemShimProxy::secItemRequest): If the request is of type Invalid, log an error message asking

for a bug, and respond with an "invalid parameter" error.

1:05 PM Changeset in webkit [170830] by Antti Koivisto
  • 5 edits in trunk/Source

Don't throttle layer flushes when the main resource is a GIF


Reviewed by Simon Fraser.

Avoid throttling big image animations.

  • page/FrameView.cpp:


Disable for image documents.


Refactor a bit.

  • page/LayerFlushThrottleState.h:

Rename the flag.


Reviewed by Simon Fraser.

  • WebProcess/WebPage/mac/RemoteLayerTreeDrawingArea.mm:


11:03 AM Changeset in webkit [170829] by akling@apple.com
  • 2 edits in trunk/Source/WebCore

Inline NodeListsNodeData helpers more aggressively.

Get rid of call overhead costs when retrieving NodeLists and
HTMLCollections from Document. These functions have very few clients,
so there's no real reason to keep them out of line.

Reviewed by Antti Koivisto.

  • dom/NodeRareData.h:


11:01 AM Changeset in webkit [170828] by akling@apple.com
  • 4 edits in trunk/Source/WebCore

Inline Node's rare data getters.

There was a fair bit of overhead when calling Node::ensureRareData()
from Document's collection getters. This patch inlines the case where
we've already materialized the rare data.

Reviewed by Antti Koivisto.

  • dom/Node.cpp:

(WebCore::Node::rareData): Deleted.
(WebCore::Node::ensureRareData): Deleted.

  • dom/Node.h:
  • dom/NodeRareData.h:


Jul 5, 2014:

5:03 PM Changeset in webkit [170827] by dino@apple.com
  • 2 edits in trunk/Source/WebCore

[iOS] MediaDocument should set a viewport

Reviewed by Zalan Bujtas.

A MediaDocument should fill the viewport, so set its
initial-scale to 1 and make it non-zoomable.

  • html/MediaDocument.cpp:

(WebCore::MediaDocumentParser::createDocumentStructure): Create
a <head> element and a <meta> element with a 1:1 viewport.

10:56 AM Changeset in webkit [170826] by commit-queue@webkit.org
  • 4 edits in trunk/Source/WebCore

Unreviewed, rolling out r170825.

broke debug builds (Requested by kling on #webkit).

Reverted changeset:

"Generate isMathMLFoo() toMathMLFoo() by using NODE_TYPE_CASTS
macro and generateTypeHelpers"

10:36 AM Changeset in webkit [170825] by gyuyoung.kim@samsung.com
  • 4 edits in trunk/Source/WebCore

Generate isMathMLFoo() toMathMLFoo() by using NODE_TYPE_CASTS macro and generateTypeHelpers

Reviewed by Andreas Kling.

Use type cast macro and generation tag for isFoo() and toFoo() in some MathML classes.

No new tests, no behavior change.

  • mathml/MathMLElement.h: Place MathMLElementTypeHelpers.h to the bottom of the

file so that functions from that file can use the Element|Node classes and other things defined in this file.
(WebCore::isMathMLElement): Deleted.

  • mathml/MathMLMencloseElement.h:

(WebCore::toMathMLMencloseElement): Deleted.

  • mathml/mathtags.in:
3:29 AM Changeset in webkit [170824] by commit-queue@webkit.org
  • 2 edits in trunk/Tools

[EFL] Add libgeoclue-dev to Tools/efl/install-dependencies.

Patch by KwangHyuk Kim <hyuki.kim@samsung.com> on 2014-07-05
Reviewed by Gyuyoung Kim.

As Geoclue is enabled for EFL, libgeoclue-dev is also added to Tools/efl/install-dependencies.

  • efl/install-dependencies:

Jul 4, 2014:

11:33 PM Changeset in webkit [170823] by gyuyoung.kim@samsung.com
  • 8 edits in trunk

[EFL] Support Geolocation

Reviewed by Antonio Gomes.


  • Source/cmake/OptionsEfl.cmake: Turn on ENABLE_GEOLOCATION.

Add geoclue files in order to enable geolocation for EFL port.

  • PlatformEfl.cmake:


  • Scripts/webkitperl/FeatureList.pm: Enable geolocation for EFL port.


  • platform/efl/TestExpectations: Unskip geolocation tests.
10:25 PM Changeset in webkit [170822] by benjamin@webkit.org
  • 4 edits in trunk/Source/WebCore

CSS JIT: compile :in-range and :out-of-range

Reviewed by Andreas Kling.

The pseudo class :in-range and :out-of-range are trivial non-inlined selectors.

  • css/SelectorChecker.cpp:


  • css/SelectorCheckerTestFunctions.h:


  • cssjit/SelectorCompiler.cpp:


9:43 PM Changeset in webkit [170821] by fpizlo@apple.com
  • 2 edits in branches/ftlopt/Source/JavaScriptCore

Merge trunk r169595.

2014-06-04 Filip Pizlo <fpizlo@apple.com>

DFG::Safepoint::begin() should set m_didCallBegin before releasing the rightToRun lock, because otherwise, Safepoint::checkLivenessAndVisitChildren() may assert due to a race

Reviewed by Oliver Hunt.

  • dfg/DFGSafepoint.cpp: (JSC::DFG::Safepoint::begin):
9:28 PM Changeset in webkit [170820] by commit-queue@webkit.org
  • 2 edits in trunk/Tools

[EFL][WK2] Enhance the Minibrowser to show selected/highlighted text in search bar, if any.

Patch by Shivakumar JM <shiva.jm@samsung.com> on 2014-07-04
Reviewed by Gyuyoung Kim.

Use script_execute_callback to get selected/highlighted text and show in search bar.

  • MiniBrowser/efl/main.c:


9:13 PM Changeset in webkit [170819] by fpizlo@apple.com
  • 2 edits in branches/ftlopt/Source/JavaScriptCore

[ftlopt][REGRESSION] debug tests fail because PutByIdDirect is now implemented in terms of In

Rubber stamped by Andreas Kling.

  • ftl/FTLLowerDFGToLLVM.cpp:


8:36 PM Changeset in webkit [170818] by akling@apple.com
  • 10 edits in trunk/Source

Fast path for jsStringWithCache() when asked for the same string repeatedly.

Also moved the whole thing from WebCore to JavaScriptCore since it
makes more sense here, and inline the lightweight checks, leaving only
the hashmap stuff out of line.

Reviewed by Darin Adler.

  • runtime/JSString.cpp:


  • runtime/JSString.h:


  • runtime/VM.h:

Reviewed by Darin Adler.

  • WebCore.exp.in:
  • bindings/js/JSDOMBinding.cpp:

(WebCore::jsStringWithCache): Deleted.

  • bindings/js/JSDOMBinding.h:

(WebCore::jsStringWithCache): Deleted.

  • bridge/c/c_utility.cpp:


  • loader/cache/CachedResourceHandle.h:
8:03 PM Changeset in webkit [170817] by Alan Bujtas
  • 3 edits
    2 adds in trunk

Subpixel rendering: ebay.com rotating billboard on the main page has cut off buttons.

Reviewed by Darin Adler.

Missing pixel snapping when clipping the transparency layer for BackgroundBleedUseTransparencyLayer.
Clipping needs to take the same snapping strategy as the rest of the painting code.

Test: fast/borders/border-radius-on-subpixel-position-non-hidpi.html

  • rendering/RenderBox.cpp:



  • fast/borders/border-radius-on-subpixel-position-non-hidpi-expected.html: Added.
  • fast/borders/border-radius-on-subpixel-position-non-hidpi.html: Added.
6:29 PM Changeset in webkit [170816] by akling@apple.com
  • 4 edits in trunk/Source/WebCore

QualifiedName should use RefPtr<QualifiedNameImpl> internally.

We were not inlining ~QualifiedName() and it was showing up in some
profiles (over 1% on dom-attr.html.)

This patch modernizes QualifiedName a bit by using RefPtr to store
the QualifiedNameImpl instead of doing manual ref() and deref()
in out-of-line functions.

Reviewed by Darin Adler.

  • WebCore.exp.in:
  • dom/QualifiedName.cpp:

(WebCore::QualifiedName::~QualifiedName): Deleted.
(WebCore::QualifiedName::deref): Deleted.

  • dom/QualifiedName.h:

(WebCore::QualifiedName::ref): Deleted.

4:03 PM Changeset in webkit [170815] by dino@apple.com
  • 2 edits in trunk/Source/WebCore

[iOS] [Media] Prevent page zoom if it started in the video element

Reviewed by Sam Weinig.

In r170572 I attempted to prevent the page zoom from happening
when the user was attempting to fullscreen the video. That fix was
not complete, because it relied on the touchstart event being
sent before a gesture event. It turns out this doesn't always happen,
so we need to apply the same check in the gesture event - if the
gesture began with two fingers on the media, don't do a page zoom.
The exception is zooming out, if you've managed to make the video
cover the entire screen (also, the pinch-out-of-fullscreen gesture
does not get handled by WebKit).

  • Modules/mediacontrols/mediaControlsiOS.js:

(ControllerIOS.prototype.handleBaseGestureChange): Add a check to
make sure our pinch didn't start in the video element.

3:27 PM Changeset in webkit [170814] by ryuan.choi@samsung.com
  • 3 edits in trunk/Source/WebKit2

[EFL][WK2]Fix build break in EFL WK2

Unreviewed build fix.

Patch by Rohit Kumar <kumar.rohit@samsung.com> on 2014-07-04

  • WebProcess/WebPage/CoordinatedGraphics/CoordinatedDrawingArea.cpp:


  • WebProcess/WebPage/CoordinatedGraphics/CoordinatedDrawingArea.h:
3:26 PM Changeset in webkit [170813] by commit-queue@webkit.org
  • 5 edits in trunk/Source/WebCore

[CoordinatedGraphics] Use the C++11 syntax to clean-up for loops

Patch by Hunseop Jeong <Hunseop Jeong> on 2014-07-04
Reviewed by Darin Adler.

Convert a couple of loops to range-based for

No change in functionality, no new tests.

  • platform/graphics/TiledBackingStore.cpp:


  • platform/graphics/texmap/coordinated/CompositingCoordinator.cpp:


  • platform/graphics/texmap/coordinated/CoordinatedBackingStore.cpp:


  • platform/graphics/texmap/coordinated/CoordinatedGraphicsScene.cpp:


2:54 PM Changeset in webkit [170812] by mitz@apple.com
  • 2 edits in trunk/Tools

Build fix after r170811.

  • MiniBrowser/mac/WK2BrowserWindowController.m:

(-[WK2BrowserWindowController reload:]):

2:00 PM Changeset in webkit [170811] by mitz@apple.com
  • 3 edits in trunk/Source/WebKit2

[Cocoa] -[WKWebView _reload] is unused

Reviewed by Sam Weinig.

  • UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView _reload]): Deleted.

  • UIProcess/API/Cocoa/WKWebViewPrivate.h:
1:49 PM Changeset in webkit [170810] by timothy_horton@apple.com
  • 2 edits in trunk/Source/WebKit2

[WK2] Take TopContentInset into account when sending dictionary lookup point to PluginView

Reviewed by Sam Weinig.

  • WebProcess/Plugins/PDF/PDFPlugin.mm:

Perform the same coordinate conversion we use for mouse events, which happens
to take the top content inset into account.

12:54 PM Changeset in webkit [170809] by akling@apple.com
  • 6 edits in trunk/Source/WebCore

CTTE: ScriptElement always has a HTTPScriptElement or SVGScriptElement.

Have ScriptElement store the pointer to the more specific subclass
element as a reference. Also made the constructor protected since
nobody should instantiate ScriptElement directly.

Reviewed by Antti Koivisto.

  • dom/ScriptElement.cpp:


  • dom/ScriptElement.h:


  • dom/ScriptRunner.cpp:


  • html/HTMLScriptElement.cpp:


  • svg/SVGScriptElement.cpp:


12:12 PM Changeset in webkit [170808] by commit-queue@webkit.org
  • 10 edits in trunk

input type=range element should only fire change events after committing a value

Patch by Julien Quint <pom@graougraou.com> on 2014-07-04
Reviewed by Dean Jackson.

A "change" event was fired every time the slider thumb element was dragged
by the user. The "change" event is now fired only after the thumb
element has stopped moving; previously, both "input" and "change" events
where dispatched while changes were being made. This new behavior is
consistent with the specification (cf.
as well as other implementations such as Firefox and Chrome.

  • Modules/mediacontrols/mediaControlsApple.js:

(Controller.prototype.createControls): Listen to the "input" event
rather than the "change" event for the timeline control in order to
keep track of value changes when the user is dragging the thumb.

  • accessibility/AccessibilitySlider.cpp:

(WebCore::AccessibilitySlider::setValue): Dispatch "change" event while
setting the new value rather than dispatching later, since setting the
value now clears the change flag.

  • html/RangeInputType.cpp:

(WebCore::RangeInputType::setValue): Update the text value of the
control in the case when no event is to be dispatched, so that this
value can be checked the next time a "change" event dispatch is

  • html/shadow/SliderThumbElement.cpp:

(WebCore::SliderThumbElement::setPositionFromPoint): Removed the
dispatch of the "change" event, and no longer track the text value of
the element as a result of dispatching a "change" event.
(WebCore::SliderThumbElement::stopDragging): Dispatch the "change" event
on completing the drag.

Two existing tests are updated to count "input" events as well as
"change" events. The tests now verify that "change" is only fired once
after every slider drag completes, whereas "input" may be fired more
than once.

  • fast/forms/range/range-drag-expected.txt:
  • fast/forms/range/range-drag-when-toggled-disabled-expected.txt:
  • fast/forms/range/range-drag-when-toggled-disabled.html:
  • fast/forms/range/range-drag.html:
11:07 AM Changeset in webkit [170807] by akling@apple.com
  • 4 edits in trunk/Source/WebCore

CTTE: ApplicationCacheHost always has a DocumentLoader.

Have ApplicationCacheHost store the backpointer to its owning
DocumentLoader as a reference.

Reviewed by Sam Weinig.

  • loader/DocumentLoader.cpp:


  • loader/appcache/ApplicationCacheHost.cpp:


  • loader/appcache/ApplicationCacheHost.h:

(WebCore::ApplicationCacheHost::documentLoader): Deleted.

8:31 AM Changeset in webkit [170806] by mario.prada@samsung.com
  • 17 edits
    1 add in trunk/LayoutTests

buildAccessibilityTree() needs to be removed from individual layout tests

Reviewed by Chris Fleizach.

Extracted two functions to 'dump' and 'touch' the accessibility
hierarchies out of different layout tests and into a single file,
living in LayoutTests/resources/accessibility-helper.js

  • resources/accessibility-helper.js: Added.

(dumpAccessibilityTree): New, based on buildAccessibilityTree()
for the cases where it was used to actually 'dump' it into the output.
(touchAccessibilityTree): Same, but for the cases where we don't
want to dump the accessibility tree, just to make sure is generated.

Updated affected layout tests to use the newly added helper functions.

  • accessibility/adjacent-continuations-cause-assertion-failure.html: Updated.
  • accessibility/children-changed-sends-notification.html: Ditto.
  • accessibility/deleting-iframe-destroys-axcache.html: Ditto.
  • accessibility/div-within-anchors-causes-crash.html: Ditto.
  • accessibility/first-letter-text-transform-causes-crash.html: Ditto.
  • accessibility/image-map-update-parent-crash.html: Ditto.
  • accessibility/table-with-empty-thead-causes-crash.html: Ditto.
  • platform/efl/accessibility/media-emits-object-replacement.html: Ditto.
  • platform/gtk/accessibility/entry-and-password.html: Ditto.
  • platform/gtk/accessibility/media-emits-object-replacement.html: Ditto.
  • platform/gtk/accessibility/replaced-objects-in-anonymous-blocks.html: Ditto.

Updated Mac specific expectations not to print the "End of test" element.

  • platform/mac-wk2/accessibility/deleting-iframe-destroys-axcache-expected.txt: Updated.
  • platform/mac/accessibility/adjacent-continuations-cause-assertion-failure-expected.txt: Ditto.
  • platform/mac/accessibility/deleting-iframe-destroys-axcache-expected.txt: Ditto.
  • platform/mac/accessibility/div-within-anchors-causes-crash-expected.txt: Ditto.
6:55 AM Changeset in webkit [170805] by Antti Koivisto
  • 3 edits
    2 adds in trunk

REGRESSION(r159345): Lines are clipped between pages when printing web content from Safari


Reviewed by Zalan Bujtas.

Test: printing/simple-lines-break.html

  • rendering/SimpleLineLayout.cpp:


Disable simple lines if global pagination is enabled.
Printing does not currently use flow threads which we already test for.


Reviewed by Zalan Bujtas.

  • printing/simple-lines-break-expected.html: Added.
  • printing/simple-lines-break.html: Added.
4:19 AM Changeset in webkit [170804] by Carlos Garcia Campos
  • 3 edits in releases/WebKitGTK/webkit-2.4/Source/WebKit2

Merge r170707 - [GTK] WebKitVersion.h should be shared between UI and Web Process APIs

Reviewed by Martin Robinson.

To be able to check current version from web extensions too.

  • UIProcess/API/gtk/WebKitVersion.h.in: Allow to include WebKitVersion.h also from webkit-web-extension.h.
  • WebProcess/InjectedBundle/API/gtk/webkit-web-extension.h: Include WebKitVersion.h.
3:46 AM Changeset in webkit [170803] by Carlos Garcia Campos
  • 7 edits in releases/WebKitGTK/webkit-2.4


Reviewed by Andreas Kling.


Remove the WTF_COMPILER_SUPPORTS_CXX_GENERALIZED_INITIALIZERS macro that was manually defined for compilers
that support the C++11 initializer lists. The feature is well supported in Clang, MSVC and also GCC. It
was actually already used unguarded in HashSet.

  • wtf/Compiler.h:
  • wtf/HashMap.h:


  • wtf/Vector.h:



Remove the COMPILER_SUPPORTS(CXX_GENERALIZED_INITIALIZERS) as the underlying macro is also
being removed since the feature is supported by all the compilers the project approves.

  • TestWebKitAPI/Tests/WTF/HashMap.cpp:


  • TestWebKitAPI/Tests/WTF/Vector.cpp:


3:33 AM Changeset in webkit [170802] by timothy_horton@apple.com
  • 2 edits in trunk/Tools

Typo in watchlist config means four specific people get Cc'd on everything

Reviewed by Gyuyoung Kim.

  • Scripts/webkitpy/common/config/watchlist:

Remove leading pipe, which meant that this rule matched everything!

2:58 AM Changeset in webkit [170801] by Philippe Normand
  • 3 edits in trunk/Source/WebKit2

2014-07-04 Philippe Normand <pnormand@igalia.com>

Unreviewed, GTK WK2 build fix after r170787.

  • WebProcess/WebPage/DrawingAreaImpl.cpp: (WebKit::DrawingAreaImpl::scheduleCompositingLayerFlushImmediately):
  • WebProcess/WebPage/DrawingAreaImpl.h:
1:55 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
1:55 AM Changeset in webkit [170800] by Carlos Garcia Campos
  • 3 edits in releases/WebKitGTK/webkit-2.4/Source/WebCore

Merge r170526 - [GTK] 8tracks.com triggers annoying pop-up window/installation of "About protocol source plugin" (GStreamer?)

Reviewed by Carlos Garcia Campos.

Don't attempt to load blank URLs with the GStreamer media
player. Those URLs trigger the codec installer which is useless in
this scenario. This patch also renames some of the variables of
the ::load method, as suggested by Carlos.

  • platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.cpp:

Check video-sink validity before disconnecting its signal handlers.
(WebCore::MediaPlayerPrivateGStreamer::load): Don't load blank
URLs (about:blank).

  • platform/graphics/gstreamer/MediaPlayerPrivateGStreamerBase.cpp:

Disconnect repaint handler only if it's valid.

1:50 AM Changeset in webkit [170799] by zandobersek@gmail.com
  • 7 edits in trunk/Source

Source/WebCore: Unreviewed. Fixing the build after r170795.

Mac didn't like that change as it seems to properly typedef long long to int64_t.
That's not the case on 64-bit Linux (for some reason), so I'm changing the type of
the two long long variables in FormDataElement to int64_t to avoid any further trouble.

  • platform/network/FormData.h:

Source/WebKit2: Unreviewed. Reverting the bad changes introduced in r170795.

  • Platform/IPC/ArgumentDecoder.cpp:

(IPC::ArgumentDecoder::decode): Deleted.

  • Platform/IPC/ArgumentDecoder.h:
  • Platform/IPC/ArgumentEncoder.cpp:

(IPC::ArgumentEncoder::encode): Deleted.

  • Platform/IPC/ArgumentEncoder.h:
1:48 AM Changeset in webkit [170798] by Carlos Garcia Campos
  • 2 edits in releases/WebKitGTK/webkit-2.4/Source/WebCore

Merge r169896 - [GTK] fails to build with --disable-webgl --disable-accelerated-compositing

Reviewed by Philippe Normand.

Add missing headers needed for Float32Array::create().

  • Modules/webaudio/AudioBuffer.cpp:
1:48 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
1:38 AM Changeset in webkit [170797] by Carlos Garcia Campos
  • 2 edits in releases/WebKitGTK/webkit-2.4/Source/WebCore

Merge r169935 - [GTK] GLContextEGL.cpp: conflicting declaration 'typedef ptrdiff_t GLintptr'

Reviewed by Gustavo Noronha Silva.

Cairo includes some definitions from GLX that conflict with the
ones provided by our GLES2 headers (GLintptr, GLsizeiptr).

Since GLContextEGL doesn't use any GLX functions we can safely
disable the inclusion of those functions in that source file.

  • platform/graphics/egl/GLContextEGL.cpp:
1:36 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
1:28 AM Changeset in webkit [170796] by Carlos Garcia Campos
  • 2 edits in releases/WebKitGTK/webkit-2.4/Tools

Merge r169768 - [GTK][gtkdoc] remove -Wcast-align

Apparently gcc warns that GParamSpec is not castable to
GParamSpecInt64/GParamSpecUInt64/GParamSpecDouble due they are 64bit, even
though ARM hackers claim that those only need 4byte alignment. As long as gcc
behaves that way, this warning is not very useful, also they break the Debian

This patch appends the compiler flag -Wno-align-cast for the gtkdoc
scanner compilation, thus the compiler ignores the previous
-Waling-cast flag.

Reviewed by Martin Robinson.

  • gtk/generate-gtkdoc:


1:27 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
1:24 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
1:23 AM Changeset in webkit [170795] by zandobersek@gmail.com
  • 5 edits in trunk/Source/WebKit2

Unreviewed. Fixing the build for ports on 64-bit Linux.

Add IPC encoding and decoding overloads for the long long type, needed
after r170755 started encoding and decoding long long variables in

  • Platform/IPC/ArgumentDecoder.cpp:


  • Platform/IPC/ArgumentDecoder.h:
  • Platform/IPC/ArgumentEncoder.cpp:


  • Platform/IPC/ArgumentEncoder.h:
1:09 AM Changeset in webkit [170794] by Carlos Garcia Campos
  • 3 edits in releases/WebKitGTK/webkit-2.4/Source/JavaScriptCore

[GTK] [Stable] Crash while playing a video

Unreviewed. Rolling out r160688.

  • jit/CCallHelpers.h:
  • jit/Repatch.cpp:


12:55 AM Changeset in webkit [170793] by Carlos Garcia Campos
  • 5 edits in releases/WebKitGTK/webkit-2.4

Merge r170359 - Source/WebCore: [ATK] Don't ignore links rendered as blocks.

Patch by Jarek Czekalski <Jarek Czekalski> on 2014-06-24
Reviewed by Mario Sanchez Prada.

  • /m/usr/src/webkit-trunk/Source/WebCore/accessibility/atk/WebKitAccessibleWrapperAtk.cpp:

(WebCore::AccessibilityObject::objectFocusedAndCaretOffsetUnignored): Don't ignore links
when they are renderered with display:block.

LayoutTests: [ATK] Extended the test for accessibility of text inside "span" block in "a" block
with the tests for caret movements in these cases.

Patch by Jarek Czekalski <Jarek Czekalski> on 2014-06-24
Reviewed by Mario Sanchez Prada.

  • platform/gtk/accessibility/text-in-span-block-in-a-block-expected.txt:
  • platform/gtk/accessibility/text-in-span-block-in-a-block.html:

Added testing for text-caret-moved signal. Added one more "a" element to the set.

12:55 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
12:50 AM Changeset in webkit [170792] by Carlos Garcia Campos
  • 3 edits
    2 adds in releases/WebKitGTK/webkit-2.4

Merge r169936 - Source/WebCore: [ATK] Fix text inside "span" block in "a" block was not accessible.

Patch by Jarek Czekalski <Jarek Czekalski> on 2014-06-13
Reviewed by Mario Sanchez Prada.

  • accessibility/atk/WebKitAccessibleWrapperAtk.cpp: Add link role to the list of roles whose objects always expose atk_text interface.

LayoutTests: [ATK] Add test for accessibility of text inside "span" block in "a" block.

Patch by Jarek Czekalski <Jarek Czekalski> on 2014-06-13
Reviewed by Mario Sanchez Prada.

  • platform/gtk/accessibility/text-in-span-block-in-a-block-expected.txt: Added.
  • platform/gtk/accessibility/text-in-span-block-in-a-block.html: Added.
12:48 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
12:43 AM Changeset in webkit [170791] by Carlos Garcia Campos
  • 10 edits
    6 adds in releases/WebKitGTK/webkit-2.4

Merge r169483 - Tools: [ATK] Added new accessibility signal: text-caret-moved.
Make AccessibilityUIElement.setSelectedTextRange work
as atk_text_set_caret_offset when given offset == 0.

Patch by Jarek Czekalski <Jarek Czekalski> on 2014-05-30
Reviewed by Mario Sanchez Prada.

text-caret-moved signal in accessibilityController is needed to
implement tests from Tools/TestWebKitAPI/Tests/WebKitGtk/testatk.c
as LayoutTests. Add the signal to the controller.

  • WebKitTestRunner/InjectedBundle/Bindings/AccessibilityUIElement.idl:
  • WebKitTestRunner/InjectedBundle/AccessibilityUIElement.cpp:
  • WebKitTestRunner/InjectedBundle/AccessibilityUIElement.h:
  • WebKitTestRunner/InjectedBundle/ios/AccessibilityUIElementIOS.mm:
  • WebKitTestRunner/InjectedBundle/mac/AccessibilityUIElementMac.mm:

(setSelectedTextRange): Change return type to bool.

  • WebKitTestRunner/InjectedBundle/atk/AccessibilityUIElementAtk.cpp:

(setSelectedTextRange): Call atk_text_set_caret_offset and return bool.

  • WebKitTestRunner/InjectedBundle/atk/AccessibilityNotificationHandlerAtk.cpp:

(axObjectEventListener): Add notification name AXTextCaretMoved, allow
for extra arguments passed to handler and use it for text-caret-moved offset.
Connect callback to new listener for ATK:AtkText:text-caret-moved.
Refactor so that failures be reported.
Initialize atk interfaces before trying to connect to them, see
Simplify code by using a vector in place of dozen variables.

  • DumpRenderTree/atk/AccessibilityCallbacksAtk.cpp:

Comment about not synchronizing with accessibilityController.

LayoutTests: [ATK] Convert caret offset tests from testatk.c.

Patch by Jarek Czekalski <Jarek Czekalski> on 2014-05-30
Reviewed by Mario Sanchez Prada.

The tests are replicated from testatk.c file, functions


The file was present in webkit1. Removed from trunk in r166977, so
the last version was
What could not be verified in the same way:

  1. Whether element implements ATK_TEXT. If setCaretOffset

succeeds, it confirms that element is text as well. But in cases
when setCaretOffset must fail I see no way to confirm it is text

  1. Whether position is correct through a call to

atk_text_get_caret_offset. text-caret-moved signal value is
tested instead.

Anyway these cases don't look crucial, so they are skipped.

  • platform/gtk/resources/atk-helpers.js: Added.

(setCaretOffset): Added.

  • platform/gtk/accessibility/caret-offsets-and-extraneous-white-spaces-expected.txt: Added.
  • platform/gtk/accessibility/caret-offsets-and-extraneous-white-spaces.html: Added.
  • platform/gtk/accessibility/caret-offsets-expected.txt: Added.
  • platform/gtk/accessibility/caret-offsets.html: Added.
12:43 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
12:26 AM Changeset in webkit [170790] by Carlos Garcia Campos
  • 19 edits in releases/WebKitGTK/webkit-2.4/Source/WebKit2

Merge r170423 - [GTK] Windowed plugins visibility doesn't work

Reviewed by Anders Carlsson.

Implement plugins visibility changes and add a new message to
notify the UI process when a windowed plugin is shown/hidden to
show/hide the plugin widget.

  • PluginProcess/PluginControllerProxy.cpp:

(WebKit::PluginControllerProxy::visibilityDidChange): Add
implementation to notify the plugin about visibility change.
Send WindowedPluginVisibilityDidChange to the plugin proxy.

  • PluginProcess/PluginControllerProxy.h:
  • PluginProcess/PluginControllerProxy.messages.in: Add VisibilityDidChange message.
  • UIProcess/WebPageProxy.h: Add windowedPluginVisibilityDidChange

to handle WindowedPluginVisibilityDidChange message.

  • UIProcess/WebPageProxy.messages.in: Add WindowedPluginVisibilityDidChange message.
  • UIProcess/efl/WebPageProxyEfl.cpp:


  • UIProcess/gtk/WebPageProxyGtk.cpp:

(WebKit::WebPageProxy::createPluginContainer): Do not show the
plugins by default.
(WebKit::WebPageProxy::windowedPluginVisibilityDidChange): Show or hide the plugin widget.

  • WebProcess/Plugins/Netscape/NetscapePlugin.cpp:

(WebKit::NetscapePlugin::NetscapePlugin): Initialize m_isVisible.
(WebKit::NetscapePlugin::visibilityDidChange): Add visible parameter and save it in m_isVisible
member, calling platformVisibilityDidChange() only when it has actually changed.

  • WebProcess/Plugins/Netscape/NetscapePlugin.h:
  • WebProcess/Plugins/Netscape/x11/NetscapePluginX11.cpp:

(WebKit::NetscapePlugin::platformVisibilityDidChange): Notify the controller about visibility change.

  • WebProcess/Plugins/PDF/PDFPlugin.h:
  • WebProcess/Plugins/Plugin.h:
  • WebProcess/Plugins/PluginController.h:
  • WebProcess/Plugins/PluginProxy.cpp:

(WebKit::PluginProxy::visibilityDidChange): Send VisibilityDidChange message to the plugin controller proxy.
(WebKit::PluginProxy::windowedPluginVisibilityDidChange): Notify the controller about visibility change.

  • WebProcess/Plugins/PluginProxy.h:
  • WebProcess/Plugins/PluginProxy.messages.in: Add WindowedPluginVisibilityDidChange message.
  • WebProcess/Plugins/PluginView.cpp:

(WebKit::PluginView::didInitializePlugin): Also call viewVisibilityDidChange() when the plugin is initialized.
(WebKit::PluginView::setParentVisible): Override this Widget method to update the plugin visibility when parent
widget is shown/hidden.
(WebKit::PluginView::viewVisibilityDidChange): Pass visible parameter to visibilityDidChange().
(WebKit::PluginView::windowedPluginVisibilityDidChange): Send WindowedPluginVisibilityDidChange message to the UI process.

  • WebProcess/Plugins/PluginView.h:
12:26 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
12:16 AM Changeset in webkit [170789] by Carlos Garcia Campos
  • 2 edits in releases/WebKitGTK/webkit-2.4/Source/WebKit2

Merge r170633 - [GTK] Annotations for WebResource.get_data produce wrong signature in Vala binding

Patch by Evan Nemerson <evan@nemerson.com> on 2014-07-01
Reviewed by Carlos Garcia Campos.

  • UIProcess/API/gtk/WebKitWebResource.cpp: Add array length and

element-type annotations to return value of webkit_web_resource_get_data_finish().

12:14 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
12:13 AM Changeset in webkit [170788] by Carlos Garcia Campos
  • 2 edits in releases/WebKitGTK/webkit-2.4/Source/ThirdParty/ANGLE

[GTK] [Stable] Crash in TSymbolTableLevel::~TSymbolTableLevel with GCC 4.9

Unreviewed. Backported from Mozilla by Mike Hommey <mh+mozilla@glandium.org>

  • src/compiler/SymbolTable.cpp:


12:10 AM Changeset in webkit [170787] by timothy_horton@apple.com
  • 9 edits in trunk/Source/WebKit2

[iOS][WK2] Black web view after un-suspending process

Reviewed by Simon Fraser.

  • UIProcess/WebPageProxy.cpp:

Add an argument to viewStateDidChange that allows callers (-[WKContentView _applicationWillEnterForeground:])
to force us to wait for a synchronous reply from the Web process after performing a view state change.

Move the has-been-in-window-and-now-is-newly-in-window check into dispatchViewStateChange.
Adjust the logic surrounding going into/out of window by factoring out the IsInWindow-did-change check, for clarity.

  • UIProcess/WebPageProxy.h:
  • UIProcess/ios/WKContentView.mm:

(-[WKContentView _applicationWillEnterForeground:]):
As previously mentioned, wait for a reply when foregrounding.

  • WebProcess/WebPage/DrawingArea.h:
  • WebProcess/WebPage/mac/RemoteLayerTreeDrawingArea.h:
  • WebProcess/WebPage/mac/RemoteLayerTreeDrawingArea.mm:

Make sure to schedule a commit immediately if the UI process is waiting for a reply.
Previously we assumed that a commit would be scheduled anyway because we would have to reparent the
layer tree, but that doesn't happen in the suspension-without-unparenting case. Also, we want to skip
all throttling in this case.

  • WebProcess/WebPage/mac/TiledCoreAnimationDrawingArea.h:
  • WebProcess/WebPage/mac/TiledCoreAnimationDrawingArea.mm:


Jul 3, 2014:

10:38 PM Changeset in webkit [170786] by commit-queue@webkit.org
  • 2 edits in trunk/Tools

[EFL][WK2] Enhance the Minibrowser to show text search count and close search bar.

Patch by Shivakumar JM <shiva.jm@samsung.com> on 2014-07-03
Reviewed by Gyuyoung Kim.

Use on_search_close_button_clicked() callback to close search bar and on_search_text_found() callback to show search count.

  • MiniBrowser/efl/main.c:


6:12 PM Changeset in webkit [170785] by barraclough@apple.com
  • 7 edits in trunk/Source/WebKit2

Should not take background task assertion for NetworkProcess

Reviewed by Tim Horton.

When the WebContent wants to complete a task in the background we take a process assertion on the child process, and also need to prevent the UI process from suspending.
However for the NetworkProcess we always just hold a process assertion, and don't want this to interfere with UIApp suspension.

  • Platform/IPC/mac/ConnectionMac.mm:


  • ProcessAssertion -> ProcessAndUIAssertion
  • UIProcess/ProcessAssertion.cpp:


  • assignment -> initializer list.


  • added no-op implementation.

(WebKit::ProcessAssertion::~ProcessAssertion): Deleted.

  • UIProcess/ProcessAssertion.h:
    • removed ~ProcessAssertion, added ProcessAndUIAssertion class.
  • UIProcess/ProcessThrottler.cpp:


  • ProcessAssertion -> ProcessAndUIAssertion
  • UIProcess/ProcessThrottler.h:
    • ProcessAssertion -> ProcessAndUIAssertion
  • UIProcess/ios/ProcessAssertionIOS.mm:


  • assignment -> initializer list.
  • moved code to manage WKProcessAssertionBackgroundTaskManager


  • moved code to manage WKProcessAssertionBackgroundTaskManager


  • moved code to manage WKProcessAssertionBackgroundTaskManager from ProcessAssertion.

(WebKit::ProcessAssertion::~ProcessAssertion): Deleted.

  • delete - moved code to manage WKProcessAssertionBackgroundTaskManager
6:09 PM Changeset in webkit [170784] by Simon Fraser
  • 5 edits in trunk/Source/WebKit2

Clean up Brady's mess on iOS.

  • Platform/mac/StringUtilities.h:
  • Platform/mac/StringUtilities.mm:
5:48 PM Changeset in webkit [170783] by Simon Fraser
  • 3 edits in trunk/Source/WebKit2

[iOS WK2] Compositing layers draw outside page bounds

Reviewed by Benjamin Poulain.

With the new rubber-banding behavior, we no longer have the root content layer clipping all
the web layers, but this resulted in web layers which were moved outside the viewport by CSS
transforms not being clipped.

Fix by adding a clipping layer inside the scroll view, above _rootContentView. This layer
normally has the same bounds as the content, but when rubber-banding or pinching,
it takes the union of the content bounds and the rect used for fixed positioning.
To make scrolling work as expected in this scenario, when it has non-zero offset, it
has to compensate by setting its bounds origin.

The bounds of the clipping layer are updated on scrolling/zooming, and when the
layer commit tells us that the content size changed.

  • UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView _updateScrollViewBackground]):

  • UIProcess/ios/WKContentView.mm:

(-[WKContentView initWithFrame:context:configuration:webView:]):
(-[WKContentView updateFixedClippingView:]):
(-[WKContentView didUpdateVisibleRect:unobscuredRect:unobscuredRectInScrollViewCoordinates:scale:minimumScale:inStableState:isChangingObscuredInsetsInteractively:]):
(-[WKContentView _didCommitLayerTree:]):

5:39 PM Changeset in webkit [170782] by beidson@apple.com
  • 15 edits
    1 copy
    1 add in trunk/Source

When showing the selection menu, include menu options for all selected phone numbers.
<rdar://problem/16983434>, <rdar://problem/16874568>, and https://bugs.webkit.org/show_bug.cgi?id=134613

Reviewed by Tim Horton.


  • English.lproj/Localizable.strings:
  • dom/Range.cpp:

(WebCore::rangesOverlap): Fix reversed arguments that caused some types of overlaps to not be detected.

Move some telephone number menu logic from WebKitSystemInterface here.
Also provide utility to get just the one menu item relevant to a telephone number.

  • Platform/mac/MenuUtilities.h:
  • Platform/mac/MenuUtilities.mm: Added.


  • Platform/mac/StringUtilities.h:
  • Platform/mac/StringUtilities.mm:

(WebKit::formattedPhoneNumberString): Return a localized formatted phone number.

Add a member to track the selected phone numbers:

  • Shared/ContextMenuContextData.h:


  • UIProcess/WebPageProxy.h:
  • UIProcess/WebPageProxy.messages.in:
  • UIProcess/mac/WebContextMenuProxyMac.mm:

(WebKit::WebContextMenuProxyMac::setupServicesMenu): Also add menu items for each phone number that is

in the selection.

Switch away from a WKSI method, and remove a lot of unneeded forward declarations and soft linking:

  • UIProcess/mac/WebPageProxyMac.mm:


  • WebKit2.xcodeproj/project.pbxproj:
  • WebProcess/WebPage/mac/ServicesOverlayController.mm:

(WebKit::ServicesOverlayController::handleClick): Pass the phone numbers along.

  • WebProcess/WebPage/WebPage.h:
  • WebProcess/WebPage/mac/WebPageMac.mm:

(WebKit::WebPage::handleSelectionServiceClick): Pass the phone numbers along.

5:36 PM Changeset in webkit [170781] by barraclough@apple.com
  • 2 edits in trunk/Source/WebKit2

WKProcessAssertionBackgroundTaskManager should clear member if task expires

Reviewed by Anders Carlson.

Failure to do so results in a warning, and could result in us erroneously completing a background task too soon.

  • UIProcess/ios/ProcessAssertionIOS.mm:

(-[WKProcessAssertionBackgroundTaskManager _updateBackgroundTask]):

  • remove local variable, _backgroundTask = UIBackgroundTaskInvalid;
5:15 PM Changeset in webkit [170780] by Joseph Pecoraro
  • 2 edits in trunk/Source/WebKit2

Unreviewed iOS build fix for r170774.

Remove some unnecessary WTF::moves on iOS:
error: static_assert failed "T is not an lvalue reference; move() is unnecessary."

  • WebProcess/WebPage/EventDispatcher.cpp:


5:08 PM Changeset in webkit [170779] by andersca@apple.com
  • 3 edits
    1 add in trunk/Source/WebKit2

Try to fix the iOS build.

Add WebPageProxyCocoa.mm for code shared between iOS and OS X. Put the recent searches save/restore code there.

  • UIProcess/Cocoa/WebPageProxyCocoa.mm: Added.


  • UIProcess/mac/WebPageProxyMac.mm:
  • WebKit2.xcodeproj/project.pbxproj:
4:47 PM Changeset in webkit [170778] by gyuyoung.kim@samsung.com
  • 2 edits in trunk/Source/WTF

Unreivewed, fix build break on EFL and GTK since r170767.

  • wtf/CMakeLists.txt: Exclude Decoder.h and Encoder.h because those files were removed.
4:46 PM Changeset in webkit [170777] by Lucas Forschler
  • 5 edits in trunk/Source


4:31 PM Changeset in webkit [170776] by mitz@apple.com
  • 14 edits in trunk/Source/WebKit2

<rdar://problem/16337741> The UI process needs to track the number of HTTP subresource loads in progress (or at least whether there are any)

Reviewed by Anders Carlsson.

Added a _networkRequestsInProgress boolean property to WKWebView.

  • UIProcess/API/Cocoa/WKBrowsingContextController.mm: Added no-op overrides of new

PageLoadStateObserver member functions.

  • UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView _networkRequestsInProgress]): New getter that gets this state from the

  • UIProcess/API/Cocoa/WKWebViewPrivate.h: Declared new property.
  • UIProcess/Cocoa/NavigationState.h:
  • UIProcess/Cocoa/NavigationState.mm:

(WebKit::NavigationState::willChangeNetworkRequestsInProgress):: Override this new
PageLoadState::Observer member function by sending the appropriate KVO change message to the
(WebKit::NavigationState::didChangeNetworkRequestsInProgress): Ditto.

  • UIProcess/PageLoadState.cpp:

(WebKit::PageLoadState::commitChanges): Check for changes to networkRequestsInProgress
and call the observers if needed.
(WebKit::PageLoadState::reset): Reset networkRequestsInProgress in the uncommitted state.
(WebKit::PageLoadState::setNetworkRequestsInProgress): Set networkRequestsInProgress in the
uncommitted state.

  • UIProcess/PageLoadState.h:

(WebKit::PageLoadState::networkRequestsInProgress): Added this getter.
(WebKit::PageLoadState::Data::Data): Initialize new networkRequestsInProgress member.

  • UIProcess/WebPageProxy.cpp:

(WebKit::WebPageProxy::setNetworkRequestsInProgress): Added. Updates the PageLoadState.

  • UIProcess/WebPageProxy.h:
  • UIProcess/WebPageProxy.messages.in: Added SetNetworkRequestsInProgress message.
  • WebProcess/WebCoreSupport/WebFrameLoaderClient.cpp:

(WebKit::WebFrameLoaderClient::assignIdentifierToInitialRequest): Call
(WebKit::WebFrameLoaderClient::dispatchDidFinishLoading): Call
(WebKit::WebFrameLoaderClient::dispatchDidFailLoading): Ditto.

  • WebProcess/WebPage/WebPage.cpp:

(WebKit::WebPage::addResourceRequest): Added. If the new request is for an HTTP-family URL,
add its identifier to the set of network resourece request identifiers. If the set was
previously empty, send the WebPageProxy a message.
(WebKit::WebPage::removeResourceRequest): Added. Remove the identifier from the set. If it
becomes empty, send the WebPageProxy a message.

  • WebProcess/WebPage/WebPage.h:
3:31 PM Changeset in webkit [170775] by andersca@apple.com
  • 3 edits in trunk/Source/WebKit2

Remove two unused functions

Reviewed by Andreas Kling.

  • UIProcess/LegacySessionStateCoding.h:
  • UIProcess/mac/LegacySessionStateCoding.cpp:

(WebKit::encodeLegacySessionHistoryEntryData): Deleted.
(WebKit::decodeLegacySessionHistoryEntryData): Deleted.

3:13 PM Changeset in webkit [170774] by dbates@webkit.org
  • 689 edits in trunk

Add WTF::move()

Rubber-stamped by Anders Carlsson.

Substitute WTF::move() for std::move().

  • bytecode/CodeBlock.h:
  • bytecode/UnlinkedCodeBlock.cpp:
  • bytecompiler/BytecodeGenerator.cpp:
  • dfg/DFGGraph.cpp:
  • dfg/DFGJITCompiler.cpp:
  • dfg/DFGStackLayoutPhase.cpp:
  • dfg/DFGWorklist.cpp:
  • heap/DelayedReleaseScope.h:
  • heap/HeapInlines.h:


Rubber-stamped by Anders Carlsson.

Substitute WTF::move() for std::move().

  • Modules/battery/NavigatorBattery.cpp:
  • Modules/encryptedmedia/MediaKeys.cpp:
  • Modules/gamepad/NavigatorGamepad.cpp:
  • Modules/gamepad/deprecated/NavigatorGamepad.cpp:
  • Modules/geolocation/NavigatorGeolocation.cpp:
  • Modules/indexeddb/DOMWindowIndexedDatabase.cpp:
  • Modules/indexeddb/IDBDatabaseBackend.cpp:
  • Modules/indexeddb/IDBDatabaseBackend.h:


Rubber-stamped by Anders Carlsson.

Substitute WTF::move() for std::move().

  • WebCoreSupport/WebFixedPositionContent.mm:

(-[WebFixedPositionContent setViewportConstrainedLayers:stickyContainerMap:]):

  • WebCoreSupport/WebInspectorClientIOS.mm:


Rubber-stamped by Anders Carlsson.

Substitute WTF::move() for std::move().

  • History/WebHistoryItem.mm:

(-[WebHistoryItem initFromDictionaryRepresentation:]):

  • Plugins/Hosted/NetscapePluginInstanceProxy.h:


  • Plugins/WebNetscapePluginView.mm:

(-[WebNetscapePluginView scheduleTimerWithInterval:repeat:timerFunc:]):

  • WebCoreSupport/WebFrameLoaderClient.mm:


  • WebCoreSupport/WebInspectorClient.mm:


  • WebView/WebArchive.mm:

(-[WebArchive initWithMainResource:subresources:subframeArchives:]):

Rubber-stamped by Anders Carlsson.

Substitute WTF::move() for std::move().

  • WebCoreSupport/WebInspectorClient.cpp:


  • WebHistory.cpp:


  • WebHistoryItem.cpp:


  • WebView.cpp:


Rubber-stamped by Anders Carlsson.

Substitute WTF::move() for std::move().

  • DatabaseProcess/DatabaseProcess.cpp:
  • DatabaseProcess/IndexedDB/UniqueIDBDatabase.cpp:
  • NetworkProcess/EntryPoint/mac/XPCService/NetworkServiceEntryPoint.mm:
  • NetworkProcess/FileAPI/NetworkBlobRegistry.cpp:
  • NetworkProcess/NetworkConnectionToWebProcess.cpp:
  • Platform/IPC/ArgumentCoders.h:
  • Platform/IPC/ArgumentDecoder.cpp:
  • Platform/IPC/Connection.cpp:
  • Platform/IPC/Connection.h:


Reviewed by Darin Adler.

Substitution of WTF::move() for std::move() rubber-stamped by Anders Carlsson.

Implement WTF::move(), a replacement for std::move(). WTF::move() is less error
prone than std::move() as it compile asserts that its argument is a non-const
lvalue reference before ultimately casting it to a rvalue reference (via std::move).

  • wtf/Deque.h: Substitute WTF::move() for std::move().
  • wtf/HashMap.h: Ditto.
  • wtf/HashSet.h: Ditto.
  • wtf/HashTable.h: Ditto.
  • wtf/IteratorAdaptors.h: Ditto.
  • wtf/IteratorRange.h: Ditto.
  • wtf/ListHashSet.h: Ditto.
  • wtf/MainThread.cpp: Ditto.


  • wtf/StdLibExtras.h:

(WTF::move): Added.

Rubber-stamped by Anders Carlsson.

Substitute WTF::move() for std::move().

  • DumpRenderTree/win/DRTDataObject.cpp:
  • TestWebKitAPI/Tests/WTF/HashMap.cpp:
  • TestWebKitAPI/Tests/WTF/HashSet.cpp:
  • TestWebKitAPI/Tests/WTF/RefPtr.cpp:
  • TestWebKitAPI/Tests/WTF/Vector.cpp:
  • TestWebKitAPI/Tests/WTF/WTFString.cpp:
  • TestWebKitAPI/Tests/WTF/ns/RetainPtr.mm: Added FIXME comments that explain that some of the sub-tests

aren't actually testing move semantics in tests: MoveAssignmentFromSameType, MoveAssignmentFromSimilarType,
ConstructionFromSameType, and ConstructionFromSimilarType.

  • TestWebKitAPI/Tests/WebCore/CalculationValue.cpp:
3:11 PM Changeset in webkit [170773] by benjamin@webkit.org
  • 6 edits in trunk/Source

[iOS][WK2] Fix small bugs of dynamicViewportSizeUpdate that were causing inaccuracies in the computed target

Patch by Benjamin Poulain <bpoulain@apple.com> on 2014-07-03
Reviewed by Tim Horton.


  • WebCore.exp.in:
  • platform/ScrollView.h:


  • platform/ios/ScrollViewIOS.mm:


This patch fixes two small bugs that were causing "jumpiness" on animated resize:
1) The new scale was less accurate than the target scale, making the WebProcess force a worse scale on the content.
2) The code putting back the rects in view was ignoring the obscured insets, constraining the scroll position more

than it should be.

The first problem was cause by a series of rounding issues accumulating to the fact that "scale != targetScale"
was almost never true.

The first issue is that the unobscured content size was stored in integer coordinates. Because of that, it was
impossible to determine accurately how much content is in the view.
The outcome was that visibleHorizontalFraction was very inaccurate compared to what the UIProcess computed.

Another issue affecting the scale is that scaleAfterViewportWidthChange was computing the widthToKeepInView
on floating point. Since ARM64 does all those computations on doubles in the UIProcess, the value would be quite
different from the received targetScale.

Finally, the code uses withinEpsilon() instead of a strict equality to reduce the effect of rounding errors.

For the second issue, the problem was that I was bounding the exposed rect to the page bounds. It is completely
valid to have an exposed rect outside the page bounds, that area is just under the obscured insets.

On the other hand, the unobscured rect needs to be within the bounds as we do not want to rotate to a rubberbanding
position. The fix is simply to put the right rect into bounds, and the horizontal/vertical adjustement applies on
both rects.

  • WebProcess/WebPage/ios/WebPageIOS.mm:


2:52 PM Changeset in webkit [170772] by andersca@apple.com
  • 2 edits in trunk/Source/WebKit2

Fix build.

  • UIProcess/mac/WebPageProxyMac.mm:
2:34 PM Changeset in webkit [170771] by bshafiei@apple.com
  • 4 edits in tags/Safari-538.44.1/Source/WebCore

Merged r170762.

2:29 PM Changeset in webkit [170770] by bshafiei@apple.com
  • 4 edits in tags/Safari-538.44.1/Source/WebCore

Relanding r170578, incorrect roll out was done in r170742.

2:04 PM Changeset in webkit [170769] by fpizlo@apple.com
  • 3 edits
    1 add in trunk/Source/JavaScriptCore

SSA DCE should process blocks in forward order

Reviewed by Andreas Kling.

  • dfg/DFGDCEPhase.cpp:


  • ftl/FTLLowerDFGToLLVM.cpp:


  • tests/stress/dead-value-with-mov-hint-in-another-block.js: Added.


2:04 PM Changeset in webkit [170768] by andersca@apple.com
  • 3 edits
    1 delete in trunk/Source/WebKit2

Get rid of WebPageProxyCF.cpp

Reviewed by Dan Bernstein.

  • UIProcess/cf/WebPageProxyCF.cpp: Removed.
  • UIProcess/mac/WebPageProxyMac.mm:


  • WebKit2.xcodeproj/project.pbxproj:
1:46 PM Changeset in webkit [170767] by andersca@apple.com
  • 6 edits
    2 deletes in trunk/Source

Get rid of Encoder.h and Decoder.h from WTF

Reviewed by Darin Adler.

  • WTF.vcxproj/WTF.vcxproj:
  • WTF.vcxproj/WTF.vcxproj.filters:
  • WTF.xcodeproj/project.pbxproj:
  • wtf/Decoder.h: Removed.
  • wtf/Encoder.h: Removed.
1:22 PM Changeset in webkit [170766] by fpizlo@apple.com
  • 3 edits
    1 add in trunk/Source/JavaScriptCore

JSActivation::symbolTablePut() should invalidate variable watchpoints

Reviewed by Oliver Hunt.

Usually stores to captured variables cause us to invalidate the variable watchpoint because CodeBlock does so
during linking - we essentially assume that if it's at all possible for an inner function to store to a
variable we declare then this variable cannot be a constant. But this misses the dynamic store case, i.e.
JSActivation::symbolTablePut(). Part of the problem here is that JSActivation duplicates
JSSymbolTableObject's symbolTablePut() logic, which did have the invalidation. This patch keeps that code
duplicated, but fixes JSActivation::symbolTablePut() to do the right thing.

  • runtime/JSActivation.cpp:


  • runtime/JSSymbolTableObject.h:


  • tests/stress/constant-closure-var-with-dynamic-invalidation.js: Added.


1:10 PM Changeset in webkit [170765] by Brent Fulgham
  • 5 edits
    2 adds in trunk

[Mac] WebKit1 WebView iframe not responding to scroll gestures

Reviewed by Simon Fraser.

New test: platform/mac/fast/scrolling/scroll-iframe-fragment.html

WK1 scrollable areas are implemented on top of PlatformWidget (i.e., NSView) objects. They only
perform a scroll operation if the EventHandler::handleWheelEvent method returns 'false', indicating
that the Cocoa layer should do scrolling work. This is in contrast to WK2, where the scrolling is
handled separately and we return 'true' from EventHandler::handleWheelEvent to indicate that the
scroll event has been handled and that no other work is needed. This mismatch in the meaning of
EventHandler::handleWheelEvent means that WK1 views do not properly handle scroll wheel events.

To correct this, we need to see if we are acting on an Element backed by a PlatformWidget. When we
are (the WK1 case) we want to return 'false' to instruct the Cocoa object to do the scroll operation.
To implement proper latching behavior in WK1, we must return 'true' in the case where we are latched
to a PlatformWidget, and that PlatformWidget did not start the scroll gesture at the edge of
the region in the dominant scroll direction.

Finally, additional logic was needed to determine whether the PlatformWidget is scrolled to the
edge in the dominant scroll direction.

  • page/EventHandler.cpp:

(WebCore::EventHandler::platformCompleteWheelEvent): Added stub.
(WebCore::EventHandler::platformCompletePlatformWidgetWheelEvent): Added stub.
(WebCore::EventHandler::handleWheelEvent): Call new platformCompletePlatformWidgetWheelEvent if the
event was handled for a PlatformWidget.

  • page/EventHandler.h:
  • page/mac/EventHandlerMac.mm:

(WebCore::widgetForEventTarget): Added helper function.
(WebCore::scrollViewForEventTarget): Added helper function.
(WebCore::eventTargetIsPlatformWidget): Added helper function.
(WebCore::EventHandler::platformPrepareForWheelEvents): Check whether we are at a scrolled edge
for the NSView (PlatformWidget) case.
(WebCore::EventHandler::platformCompleteWheelEvent): Return 'false' when handling WK1-style scrolls.
(WebCore::EventHandler::platformCompletePlatformWidgetWheelEvent): Added. We generally want to
return false so the PlatformWidget handles scrolling, except in the case where we are latched to
a PlatformWidget, which has already performed its scroll, and we want to avoid propegating a
scroll event to an enclosing region.


  • platform/mac/fast/scrolling/scroll-iframe-fragment-expected.txt: Added.
  • platform/mac/fast/scrolling/scroll-iframe-fragment.html: Added.
1:06 PM Changeset in webkit [170764] by ryuan.choi@samsung.com
  • 4 edits in trunk

Broken build with build-webkit --no-video

Reviewed by Darin Adler.



  • Source/cmake/WebKitFeatures.cmake: Added the dependency for ENABLE_MEDIA_CONTROLS_SCRIPT.


  • wtf/FeatureDefines.h: Added error message for dependency failure.
1:01 PM Changeset in webkit [170763] by timothy_horton@apple.com
  • 2 edits in trunk/Source/WebKit2

[WK2] Revise the flat find indicator secondary highlight shadows

Reviewed by Brady Eidson.

  • WebProcess/WebPage/FindController.cpp:

Bring back the secondary highlight shadow, with new constants.

12:48 PM Changeset in webkit [170762] by psolanki@apple.com
  • 4 edits in trunk/Source/WebCore

Preserve old behavior of creating an NSURLRequest of the exact same type as passed to us

Reviewed by Andreas Kling.

We have client code that passes us a subclass of NSURLRequest to load resource. Later when
we call willSendRequest, they test to make sure they get an object of the same type. My
optimization in r170642 broke this path when I cleared out the NSURLRequest object. We
already had code in updateNSURLRequest() that was taking this quirk into account but I broke
that. Fix it by reverting to old behavior for such clients.

No new tests though we need one. I will add it later.

  • platform/network/cf/ResourceRequest.h:
  • platform/network/cf/ResourceRequestCFNet.cpp:


  • platform/network/ios/ResourceRequestIOS.mm:


12:30 PM Changeset in webkit [170761] by timothy_horton@apple.com
  • 12 edits in trunk/Source/WebKit2

[iOS][WK2] Sometimes the swipe snapshot stays up too long

Reviewed by Simon Fraser.

Implement a transaction callback mechanism.
The UI process can register a callback at any point, generally immediately after sending a message to the Web process.
It will then send another message (in-order) with the callback ID to the Web process, which will put it into the next transaction
(scheduling a new one if needed). When the transaction comes back to the UI process, the callbacks are performed.
This ensures that the callback fires alongside a commit that includes the results of whatever messages were sent before it was registered.
For now, all callbacks are fired just before committing layer changes, but it's possible future patches will want post-commit callbacks.

Make use of this to remove the ViewGestureControllerIOS snapshots at the right time.

  • Shared/mac/RemoteLayerTreeTransaction.h:


  • Shared/mac/RemoteLayerTreeTransaction.mm:

Add a vector of callback IDs to the transaction; encode and decode as appropriate.

  • UIProcess/DrawingAreaProxy.h:

(WebKit::DrawingAreaProxy::lastVisibleTransactionID): Deleted.

  • UIProcess/mac/RemoteLayerTreeDrawingAreaProxy.h:
  • UIProcess/mac/RemoteLayerTreeDrawingAreaProxy.mm:

Add dispatchAfterEnsuringDrawing, which takes a function.
When a callback is added, we send the callback ID to the Web process via the AddTransactionCallbackID message.
Perform callbacks listed in the incoming transaction's vector of callback IDs.

  • WebProcess/WebPage/DrawingArea.h:


  • WebProcess/WebPage/DrawingArea.messages.in:
  • WebProcess/WebPage/mac/RemoteLayerTreeDrawingArea.h:
  • WebProcess/WebPage/mac/RemoteLayerTreeDrawingArea.mm:

Keep track of the pending callback IDs on the drawing area, and move them into the transaction.
We schedule a flush when installing a transaction callback on the premise that
sometimes the action (goToBackForwardListItem in the swipe case) might have already occurred
and been committed by the time the Web process receives AddTransactionCallbackID, so we need
to cause another commit to send the callbacks across. If said commit is still pending, this is a no-op.

  • UIProcess/ios/ViewGestureControllerIOS.mm:


  • UIProcess/mac/ViewGestureController.h:

Keep a side-map of page IDs to ViewGestureControllers, so that we can safely get back to
our ViewGestureController from the callback.

When the callback fires, if it succeeded, set m_shouldRemoveSnapshotWhenTargetRenderTreeSizeHit,
so that the commit (which is about to occur immediately after the callback returns) which calls
setRenderTreeSize will (perhaps) remove the snapshot.

If it failed, we remove the snapshot immediately, as this usually happens if the Web process crashed.

12:16 PM Changeset in webkit [170760] by beidson@apple.com
  • 2 edits in trunk/Source/WebKit2

Followup to "rects sent to ServicesOverlayController are wrong"

Rubberstamped by Tim Horton.

  • WebProcess/WebPage/mac/ServicesOverlayController.mm:

(WebKit::ServicesOverlayController::drawSelectionHighlight): Don’t check to see if the

proposed rect intersects the dirty rect. We always need to include all rects that form
the selection when creating the DDHighlight.

12:04 PM Changeset in webkit [170759] by timothy_horton@apple.com
  • 5 edits in trunk/Source/WebKit2

[WK2] RemoteLayerTreeDrawingAreaProxy re-checks a preference every commit for no reason

Reviewed by Benjamin Poulain.

  • UIProcess/DrawingAreaProxy.h:

(WebKit::DrawingAreaProxy::showDebugIndicator): Deleted.
(WebKit::DrawingAreaProxy::isShowingDebugIndicator): Remove this, nobody is using it.

  • UIProcess/WebPageProxy.cpp:


  • UIProcess/mac/RemoteLayerTreeDrawingAreaProxy.h:
  • UIProcess/mac/RemoteLayerTreeDrawingAreaProxy.mm:

(WebKit::RemoteLayerTreeDrawingAreaProxy::showDebugIndicator): Deleted.
Check the debug indicator pref at construction time, and update the state when it changes.
Checking it every commit was surprisingly expensive.

10:56 AM Changeset in webkit [170758] by beidson@apple.com
  • 16 edits in trunk/Source

Selection rects sent to ServicesOverlayController are wrong.
<rdar://problem/16727796> and https://bugs.webkit.org/show_bug.cgi?id=134568

Reviewed by Darin Adler (and Tim Horton and Ryosuke Niwa).


  • WebCore.exp.in:

Update the gatherer to keep GapRects separate from LayoutRects:

  • editing/SelectionRectGatherer.cpp:


  • editing/SelectionRectGatherer.h:
  • page/EditorClient.h:

(WebCore::EditorClient::selectionRectsDidChange): Updated to take LayoutRects and GapRects separately.

Change RenderSelectionInfo to also hang on to the individual rects that formed the final bounding rect:

  • rendering/RenderSelectionInfo.h:

(WebCore::RenderSelectionInfo::RenderSelectionInfo): If the RenderObject is a RenderText, then call

collectSelectionRectsForLineBoxes instead of selectionRectForRepaint.


  • rendering/RenderText.cpp:

(WebCore::RenderText::collectSelectionRectsForLineBoxes): Added

  • rendering/RenderText.h:
  • rendering/RenderTextLineBoxes.cpp:

(WebCore::RenderTextLineBoxes::collectSelectionRectsForRange): Added

  • rendering/RenderTextLineBoxes.h:
  • rendering/RenderView.cpp:

(WebCore::RenderView::setSubtreeSelection): Add the list of rects to the gatherer instead of just

the bounding rect.


  • WebProcess/WebCoreSupport/WebEditorClient.cpp:

(WebKit::WebEditorClient::selectionRectsDidChange): Also pass the GapRects to the ServicesOverlayController.

  • WebProcess/WebCoreSupport/WebEditorClient.h:
  • WebProcess/WebPage/ServicesOverlayController.h:
  • WebProcess/WebPage/mac/ServicesOverlayController.mm:

(WebKit::compactRectsWithGapRects): Combine 3+ rects down to exactly 3 rects, then expand them based on GapRects.
(WebKit::ServicesOverlayController::selectionRectsDidChange): Call compactRectsWithGapRects, then reverse the list.
(WebKit::ServicesOverlayController::drawSelectionHighlight): Tell data detectors to flip this.
(WebKit::ServicesOverlayController::drawTelephoneNumberHighlight): Tell data detectors to flip this.
(WebKit::ServicesOverlayController::drawCurrentHighlight): No need to flip this anymore.

10:13 AM Changeset in webkit [170757] by Antti Koivisto
  • 2 edits in trunk/Source/WebKit2

Ensure frame creation messages get through to UI process

Reviewed by Anders Carlsson.

If we are middle of handling a synchronous message from UI process a frame creation message back gets delayed.
The subsequent synchronous DecidePolicyForNavigationAction message expects that the frame creation
message has arrived first and fails.

  • WebProcess/WebPage/WebFrame.cpp:


Send messages with DispatchMessageEvenWhenWaitingForSyncReply so they always go through in order.

10:09 AM Changeset in webkit [170756] by andersca@apple.com
  • 4 edits
    4 deletes in trunk/Source/WebKit2

Get rid of DecoderAdapter and EncoderAdapter

Reviewed by Andreas Kling.

  • CMakeLists.txt:
  • WebKit2.xcodeproj/project.pbxproj:
  • WebProcess/WebPage/DecoderAdapter.cpp: Removed.
  • WebProcess/WebPage/DecoderAdapter.h: Removed.
  • WebProcess/WebPage/EncoderAdapter.cpp: Removed.
  • WebProcess/WebPage/EncoderAdapter.h: Removed.
  • WebProcess/WebPage/WebPage.cpp:
10:05 AM Changeset in webkit [170755] by andersca@apple.com
  • 6 edits in trunk/Source

Stop using EncoderAdapter/DecoderAdapter for FormData

Reviewed by Andreas Kling.


  • WebCore.exp.in:
  • platform/network/FormData.cpp:

(WebCore::encodeElement): Deleted.
(WebCore::decodeElement): Deleted.
(WebCore::FormData::encode): Deleted.
(WebCore::FormData::decode): Deleted.

  • platform/network/FormData.h:



  • Shared/Network/NetworkResourceLoadParameters.cpp:


9:19 AM Changeset in webkit [170754] by beidson@apple.com
  • 8 edits in trunk/Source

Possible crash in IconDatabase in WebCore::IconDatabase::dispatchDidRemoveAllIconsOnMainThread
<rdar://problem/17437687> and https://bugs.webkit.org/show_bug.cgi?id=134517

Reviewed by Eric Carlson.

Since WebCore::IconDatabase is not RefCounted there’s no obvious way to keep it alive until it is
truly no longer needed.

This isn’t generally a problem because they are usually a singleton that lasts the lifetime of the process.

In the WebKit2 case, WebCore::IconDatabases can come and go as their owning WebIconDatabases come and go.

So we can rely on WebIconDatabase to handle the lifetime appropriately.

  • loader/icon/IconDatabase.cpp:

(WebCore::IconDatabase::close): If the database is actually closed, notify the client.
(WebCore::IconDatabase::isOpen): Take into account whether there’s any main thread callbacks, plus what

is covered by isOpenBesidesMainThreadCallbacks.

(WebCore::IconDatabase::isOpenBesidesMainThreadCallbacks): Take into account whether the sync thread

is still running and whether or not the database file is still open.

(WebCore::IconDatabase::checkClosedAfterMainThreadCallback): Checks to see if the database has just

become 100% closed and - if so - notify the client.

(WebCore::IconDatabase::dispatchDidImportIconURLForPageURLOnMainThread): Updated to increment the main thread

callback count and to call checkClosedAfterMainThreadCallback when done.

(WebCore::IconDatabase::dispatchDidImportIconDataForPageURLOnMainThread): Ditto.
(WebCore::IconDatabase::dispatchDidRemoveAllIconsOnMainThread): Ditto.
(WebCore::IconDatabase::dispatchDidFinishURLImportOnMainThread): Ditto.

  • loader/icon/IconDatabase.h:
  • loader/icon/IconDatabaseClient.h:

(WebCore::IconDatabaseClient::didClose): Added. For the IconDatabase to tell its client it is 100% closed.


  • UIProcess/WebContext.cpp:

(WebKit::WebContext::~WebContext): Instead of directly deref’ing the WebIconDatabase, ask it to

deref itself when appropriate.

  • UIProcess/WebIconDatabase.cpp:

(WebKit::WebIconDatabase::didClose): If this WebIconDatabase is supposed to deref itself when

appropriate, do so now.

(WebKit::WebIconDatabase::derefWhenAppropriate): If the WebCore::IconDatabase is still open then

defer this deref.

  • UIProcess/WebIconDatabase.h:
8:46 AM Changeset in webkit [170753] by Chris Fleizach
  • 2 edits in trunk/Source/WebCore

AX: VoiceOver does not read aria-expanded attribute on controls in Safari

Reviewed by Mario Sanchez Prada.

Expose whether the ARIA expanded property is present or not.

  • accessibility/ios/WebAccessibilityObjectWrapperIOS.mm:

(-[WebAccessibilityObjectWrapper accessibilitySupportsARIAExpanded]):

7:55 AM Changeset in webkit [170752] by llango.u-szeged@partner.samsung.com
  • 2 edits in trunk/Tools

[JavaScriptCore] Add --cmakeargs option to build-jsc.

Reviewed by Ryosuke Niwa.

  • Scripts/build-jsc:
7:17 AM WebKitGTK/2.4.x edited by berto@igalia.com
6:36 AM Changeset in webkit [170751] by Carlos Garcia Campos
  • 2 edits in trunk

[GTK] The list of features shown by cmake is not accurate

Reviewed by Gustavo Noronha Silva.

The problem is that some options might change after the feature
list has been printed.

  • Source/cmake/OptionsGTK.cmake: Find package dependencies before

setting the default values of features and set WEBGL value
depending on the dependencies, and API_TEST depending on whether
developer mode is enabled or not.

5:49 AM Changeset in webkit [170750] by commit-queue@webkit.org
  • 2 edits in trunk/Tools

Add option to specify jsc args in run-sunspider.

Patch by Gergo Balogh <gbalogh.u-szeged@partner.samsung.com> on 2014-07-03
Reviewed by Csaba Osztrogonác.

  • Scripts/run-sunspider:
1:04 AM Changeset in webkit [170749] by ryuan.choi@samsung.com
  • 4 edits in trunk

[CMAKE] Add WEBKIT_OPTION_DEPEND macro to resolve macro dependency

Reviewed by Gyuyoung Kim.

Some options depend on another option such as ENABLE_VIDEO_TRACK and ENABLE_VIDEO.
This patch adds WEBKIT_OPTION_DEPEND to check the depending option and
disable related option if it is not ON.

  • Source/cmake/OptionsEfl.cmake: Removed hack for option dependency.
  • Source/cmake/OptionsGTK.cmake: Ditto.
  • Source/cmake/WebKitFeatures.cmake: Added WEBKIT_OPTION_DEPEND macro.
12:56 AM Changeset in webkit [170748] by Carlos Garcia Campos
  • 2 edits in trunk/Source/WebKit2

REGRESSION(r170676): [GTK] UI process crashes when the Web Process crashes

Reviewed by Gyuyoung Kim.

It happens when attaching a new web process, because it tries to
encode a null SessionState.

  • UIProcess/LegacySessionStateCodingNone.cpp:

(WebKit::encodeLegacySessionState): Return an empty API::Data
object instead of nullptr.
(WebKit::encodeLegacySessionHistoryEntryData): Ditto.

12:53 AM Changeset in webkit [170747] by Carlos Garcia Campos
  • 3 edits in trunk/Source/WebKit2

REGRESSION(r170743): [GTK] [EFL] Build broken.

Reviewed by Philippe Normand.


  • WebProcess/Plugins/PluginView.cpp:


  • WebProcess/WebPage/WebPage.cpp:


Note: See TracTimeline for information about the timeline view.