WebKit-https.git
4 years agoUnreviewed, rolling out r181824.
commit-queue@webkit.org [Sat, 21 Mar 2015 07:37:07 +0000 (07:37 +0000)]
Unreviewed, rolling out r181824.
https://bugs.webkit.org/show_bug.cgi?id=142935

Broke Windows build (Requested by ap on #webkit).

Reverted changeset:

"Remove the prefix for CSS Transforms"
https://bugs.webkit.org/show_bug.cgi?id=142927
http://trac.webkit.org/changeset/181824

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181825 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoRemove the prefix for CSS Transforms
dino@apple.com [Sat, 21 Mar 2015 06:54:14 +0000 (06:54 +0000)]
Remove the prefix for CSS Transforms
https://bugs.webkit.org/show_bug.cgi?id=142927
<rdar://problem/11568151>

Reviewed by Simon Fraser.

Source/WebCore:

Support the bare form of:
- transform
- transform-style
- transform-origin
- perspective
- perspective-origin

For transform-style, we also still differentiate the prefixed property
because we may want to use that as an indicator of legacy content.

This is mostly a find and replace, but with some related naming
changes. The DOM interface for WebKitCSSTransformValue remains
unchanged.

Tests: transforms/2d/transform-2d-prefixed.html
       transforms/3d/general/prefixed-3dtransform-values.html

* bindings/js/JSCSSStyleDeclarationCustom.cpp:
(WebCore::propertyChangeMayRepaintNonDescendants):
* css/CSSComputedStyleDeclaration.cpp:
(WebCore::isLayoutDependent):
(WebCore::ComputedStyleExtractor::propertyValue):
* css/CSSParser.cpp:
(WebCore::isValidKeywordPropertyAndValue):
(WebCore::isKeywordPropertyID):
(WebCore::parseTranslateTransformValue):
(WebCore::CSSParser::parseValue):
(WebCore::CSSParser::parseTransformOrigin):
(WebCore::CSSParser::parsePerspectiveOrigin):
* css/CSSPropertyNames.in:
* css/StyleProperties.cpp:
(WebCore::StyleProperties::getPropertyValue):
(WebCore::StyleProperties::asText):
* css/WebKitCSSMatrix.cpp:
(WebCore::WebKitCSSMatrix::setMatrixValue):
* html/track/VTTCue.cpp:
(WebCore::VTTCueBox::applyCSSProperties):
* page/animation/AnimationController.cpp:
(WebCore::AnimationControllerPrivate::computeExtentOfAnimation):
* page/animation/CSSPropertyAnimation.cpp:
(WebCore::PropertyWrapperAcceleratedTransform::PropertyWrapperAcceleratedTransform):
(WebCore::CSSPropertyAnimationWrapperMap::CSSPropertyAnimationWrapperMap):
* page/animation/CompositeAnimation.cpp:
(WebCore::CompositeAnimation::computeExtentOfTransformAnimation):
* page/animation/ImplicitAnimation.cpp:
(WebCore::ImplicitAnimation::computeExtentOfTransformAnimation):
* page/animation/KeyframeAnimation.cpp:
(WebCore::KeyframeAnimation::computeExtentOfTransformAnimation):
(WebCore::KeyframeAnimation::validateTransformFunctionList):
* platform/graphics/GraphicsLayer.cpp:
(WebCore::GraphicsLayer::validateTransformOperations):
* platform/graphics/GraphicsLayerClient.h:
* platform/graphics/ca/GraphicsLayerCA.cpp:
(WebCore::propertyIdToString):
(WebCore::GraphicsLayerCA::moveOrCopyAnimations):
(WebCore::GraphicsLayerCA::addAnimation):
(WebCore::GraphicsLayerCA::isRunningTransformAnimation):
(WebCore::GraphicsLayerCA::createAnimationFromKeyframes):
(WebCore::GraphicsLayerCA::createTransformAnimationsFromKeyframes):
* platform/graphics/texmap/GraphicsLayerTextureMapper.cpp:
(WebCore::GraphicsLayerTextureMapper::addAnimation):
* platform/graphics/texmap/TextureMapperAnimation.cpp:
(WebCore::TextureMapperAnimation::applyInternal):
* platform/graphics/texmap/TextureMapperLayer.cpp:
(WebCore::TextureMapperLayer::syncAnimations):
* platform/graphics/texmap/coordinated/CoordinatedGraphicsLayer.cpp:
(WebCore::CoordinatedGraphicsLayer::selfOrAncestorHasActiveTransformAnimation):
(WebCore::CoordinatedGraphicsLayer::selfOrAncestorHaveNonAffineTransforms):
(WebCore::CoordinatedGraphicsLayer::addAnimation):
* rendering/RenderLayerBacking.cpp:
(WebCore::RenderLayerBacking::updateGeometry):
(WebCore::RenderLayerBacking::startAnimation):
(WebCore::RenderLayerBacking::startTransition):
(WebCore::RenderLayerBacking::graphicsLayerToCSSProperty):
(WebCore::RenderLayerBacking::cssToGraphicsLayerProperty):
* rendering/RenderLayerCompositor.cpp:
(WebCore::RenderLayerCompositor::requiresCompositingForAnimation):
(WebCore::RenderLayerCompositor::isRunningTransformAnimation):
* svg/SVGElement.cpp:
(WebCore::populateAttributeNameToCSSPropertyIDMap):

Source/WebKit2:

Rename AnimatedPropertyWebkitTransform to AnimatedPropertyTransform.

* Shared/CoordinatedGraphics/CoordinatedGraphicsArgumentCoders.cpp:
(IPC::ArgumentCoder<TextureMapperAnimation>::encode):

LayoutTests:

Convert the helper functions for transforms to use the unprefixed
form, and copy some tests to make sure the prefixed form still works.

* fast/css/getComputedStyle/computed-style-expected.txt:
* fast/css/getComputedStyle/computed-style-without-renderer-expected.txt:
* fast/css/getComputedStyle/resources/property-names.js:
* svg/css/getComputedStyle-basic-expected.txt:
* transforms/2d/computed-style-origin-expected.txt:
* transforms/2d/computed-style-origin.html:
* transforms/2d/resources/prefixed-transform-test-utils.js: Copied from LayoutTests/transforms/2d/resources/transform-test-utils.js.
(floatingPointEqual):
(compareMatrices):
(testTransforms):
* transforms/2d/resources/transform-test-utils.js:
(testTransforms):
* transforms/2d/transform-2d-prefixed-expected.txt: Added.
* transforms/2d/transform-2d-prefixed.html: Added.
* transforms/3d/general/prefixed-3dtransform-values-expected.txt: Added.
* transforms/3d/general/prefixed-3dtransform-values.html: Added.
* transitions/transition-end-event-multiple-04-expected.txt:
* transitions/transition-end-event-multiple-04.html:
* transitions/transition-end-event-transform-expected.txt:
* transitions/transition-end-event-transform.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181824 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Win] Additional debug assertion documentation.
bfulgham@apple.com [Sat, 21 Mar 2015 04:20:29 +0000 (04:20 +0000)]
[Win] Additional debug assertion documentation.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181823 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoGCTimer should know keep track of nested GC phases
mhahnenb@gmail.com [Sat, 21 Mar 2015 02:02:44 +0000 (02:02 +0000)]
GCTimer should know keep track of nested GC phases
https://bugs.webkit.org/show_bug.cgi?id=142675

Reviewed by Darin Adler.

Source/JavaScriptCore:

This improves the GC phase timing output in Heap.cpp by linking
phases nested inside other phases together, allowing tools
to compute how much time we're spending in various nested phases.

* heap/Heap.cpp:

Tools:

Adds a tool to aid in parsing the GC phase timing output into a
tree-like structure based on the parent-child relationships
of nested GC phases.

* Scripts/parse-gc-phase-timings: Added.
(Timing):
(Timing.__init__):
(Timing.__unicode__):
(Timing.__str__):
(Timing.__repr__):
(parse_input):
(print_timing_node):
(print_timing_tree):
(link_parents):
(main):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181821 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Win] More debug assertions. Almost there!
bfulgham@apple.com [Sat, 21 Mar 2015 00:07:17 +0000 (00:07 +0000)]
[Win] More debug assertions. Almost there!

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181820 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoWeb Inspector: Fast typing lags in the Styles sidebar or Console
nvasilyev@apple.com [Fri, 20 Mar 2015 23:38:03 +0000 (23:38 +0000)]
Web Inspector: Fast typing lags in the Styles sidebar or Console
https://bugs.webkit.org/show_bug.cgi?id=142919

Asynchronous autocomplete causes a race condition in CodeMirror,
which results in skipped characters in while typing. Completing immediately
fixes the issues.

Reviewed by Timothy Hatcher.

* UserInterface/Controllers/CodeMirrorCompletionController.js:
(WebInspector.CodeMirrorCompletionController.prototype.hideCompletions):
(WebInspector.CodeMirrorCompletionController.prototype._completeAtCurrentPosition):
(WebInspector.CodeMirrorCompletionController.prototype.):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181819 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoFunctionBodyNode should known where its parameters started
ggaren@apple.com [Fri, 20 Mar 2015 23:37:51 +0000 (23:37 +0000)]
FunctionBodyNode should known where its parameters started
https://bugs.webkit.org/show_bug.cgi?id=142926

Reviewed by Ryosuke Niwa.

This will allow us to re-parse parameters instead of keeping the
parameters piece of the AST around forever.

I also took the opportunity to initialize most FunctionBodyNode data
members at construction time, to help clarify that they are set right.

* parser/ASTBuilder.h:
(JSC::ASTBuilder::createFunctionExpr): No need to pass
functionKeywordStart here; we now provide it at FunctionBodyNode
creation time.

(JSC::ASTBuilder::createFunctionBody): Require everything we need at
construction time, including the start of our parameters.

(JSC::ASTBuilder::createGetterOrSetterProperty):
(JSC::ASTBuilder::createFuncDeclStatement):  No need to pass
functionKeywordStart here; we now provide it at FunctionBodyNode
creation time.

(JSC::ASTBuilder::setFunctionNameStart): Deleted.

* parser/Nodes.cpp:
(JSC::FunctionBodyNode::FunctionBodyNode): Initialize everything at
construction time.

* parser/Nodes.h: Added a field for the location of our parameters.

* parser/Parser.cpp:
(JSC::Parser<LexerType>::parseFunctionBody):
(JSC::Parser<LexerType>::parseFunctionInfo):
(JSC::Parser<LexerType>::parseFunctionDeclaration):
(JSC::Parser<LexerType>::parseClass):
(JSC::Parser<LexerType>::parsePropertyMethod):
(JSC::Parser<LexerType>::parseGetterSetter):
(JSC::Parser<LexerType>::parsePrimaryExpression):
* parser/Parser.h: Refactored to match above interface changes.

* parser/SyntaxChecker.h:
(JSC::SyntaxChecker::createFunctionExpr):
(JSC::SyntaxChecker::createFunctionBody):
(JSC::SyntaxChecker::createFuncDeclStatement):
(JSC::SyntaxChecker::createGetterOrSetterProperty): Refactored to match
above interface changes.

(JSC::SyntaxChecker::setFunctionNameStart): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181818 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoObservably effectful nodes in DFG IR should come last in their bytecode instruction...
fpizlo@apple.com [Fri, 20 Mar 2015 23:26:26 +0000 (23:26 +0000)]
Observably effectful nodes in DFG IR should come last in their bytecode instruction (i.e. forExit section), except for Hint nodes
https://bugs.webkit.org/show_bug.cgi?id=142920

Reviewed by Oliver Hunt, Geoffrey Garen, and Mark Lam.

Observably effectful, n.: If we reexecute the bytecode instruction after this node has
executed, then something other than the bytecode instruction's specified outcome will
happen.

We almost never had observably effectful nodes except at the end of the bytecode
instruction.  The exception is a lowered transitioning PutById:

PutStructure(@o, S1 -> S2)
PutByOffset(@o, @o, @v)

The PutStructure is observably effectful: if you try to reexecute the bytecode after
doing the PutStructure, then we'll most likely crash.  The generic PutById handling means
first checking what the old structure of the object is; but if we reexecute, the old
structure will seem to be the new structure.  But the property ensured by the new
structure hasn't been stored yet, so any attempt to load it or scan it will crash.

Intriguingly, however, none of the other operations involved in the PutById are
observably effectful.  Consider this example:

PutByOffset(@o, @o, @v)
PutStructure(@o, S1 -> S2)

Note that the PutStructure node doesn't reallocate property storage; see further below
for an example that does that. Because no property storage is happening, we know that we
already had room for the new property.  This means that the PutByOffset is no observable
until the PutStructure executes and "reveals" the property.  Hence, PutByOffset is not
observably effectful.

Now consider this:

b: AllocatePropertyStorage(@o)
PutByOffset(@b, @o, @v)
PutStructure(@o, S1 -> S2)

Surprisingly, this is also safe, because the AllocatePropertyStorage is not observably
effectful. It *does* reallocate the property storage and the new property storage pointer
is stored into the object. But until the PutStructure occurs, the world will just think
that the reallocation didn't happen, in the sense that we'll think that the property
storage is using less memory than what we just allocated. That's harmless.

The AllocatePropertyStorage is safe in other ways, too. Even if we GC'd after the
AllocatePropertyStorage but before the PutByOffset (or before the PutStructure),
everything could be expected to be fine, so long as all of @o, @v and @b are on the
stack. If they are all on the stack, then the GC will leave the property storage alone
(so the extra memory we just allocated would be safe). The GC will not scan the part of
the property storage that contains @v, but that's fine, so long as @v is on the stack.

The better long-term solution is probably bug 142921.

But for now, this:

- Fixes an object materialization bug, exemplified by the two tests, that previously
  crashed 100% of the time with FTL enabled and concurrent JIT disabled.

- Allows us to remove the workaround introduced in r174856.

* dfg/DFGByteCodeParser.cpp:
(JSC::DFG::ByteCodeParser::handlePutById):
* dfg/DFGConstantFoldingPhase.cpp:
(JSC::DFG::ConstantFoldingPhase::emitPutByOffset):
* dfg/DFGFixupPhase.cpp:
(JSC::DFG::FixupPhase::insertCheck):
(JSC::DFG::FixupPhase::indexOfNode): Deleted.
(JSC::DFG::FixupPhase::indexOfFirstNodeOfExitOrigin): Deleted.
* dfg/DFGInsertionSet.h:
(JSC::DFG::InsertionSet::insertOutOfOrder): Deleted.
(JSC::DFG::InsertionSet::insertOutOfOrderNode): Deleted.
* tests/stress/materialize-past-butterfly-allocation.js: Added.
(bar):
(foo0):
(foo1):
(foo2):
(foo3):
(foo4):
* tests/stress/materialize-past-put-structure.js: Added.
(foo):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181817 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[WK2] NetworkCache retrievals sometimes fail on browser startup
cdumez@apple.com [Fri, 20 Mar 2015 23:09:56 +0000 (23:09 +0000)]
[WK2] NetworkCache retrievals sometimes fail on browser startup
https://bugs.webkit.org/show_bug.cgi?id=142925
<rdar://problem/20245368>

Reviewed by Antti Koivisto.

NetworkCache retrievals sometimes fail on browser startup for resources
that are actually cached. The reason is that we are using a bloom filter
for performance reasons to avoid unnecessary disk I/O and this bloom
filter is populated on start up in a background thread by traversing the
cache files on disk. However, when restoring the tabs on start-up we
sometimes query this bloom filter before it is completely populated and
we thus fail to retrieve cached entries because we think they don't
exist and don't check the disk.

This patch adds an "isPopulatingContentsFilter" flag that is turned ON
on start up while we are populating the bloon filter. We then bypass
the bloom filter and send queries directly to disk on start up if this
flag is ON.

* NetworkProcess/cache/NetworkCacheStorage.cpp:
(WebKit::NetworkCache::Storage::initialize):
(WebKit::NetworkCache::Storage::retrieve):
(WebKit::NetworkCache::Storage::dispatchPendingWriteOperations):
(WebKit::NetworkCache::Storage::dispatchHeaderWriteOperation):
* NetworkProcess/cache/NetworkCacheStorage.h:
(WebKit::NetworkCache::Storage::cacheMayContain):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181816 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[WK2] Allow stale content when restoring the browser's session state
cdumez@apple.com [Fri, 20 Mar 2015 22:45:14 +0000 (22:45 +0000)]
[WK2] Allow stale content when restoring the browser's session state
https://bugs.webkit.org/show_bug.cgi?id=142916
<rdar://problem/20243493>

Reviewed by Darin Adler.

Allow stale content when restoring the browser's session state
(restoring all tabs from previous session), e.g.
- Via History > Reopen All windows from previous session on Safari
- Happens on startup for MobileSafari

I have verified that using "Open in tabs" on a bookmarks folder
still does fresh loads (rdar://problem/8131355) as it is not using
the restoreFromSessionState() code path.

* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::reattachToWebProcessForReload):
(WebKit::WebPageProxy::reattachToWebProcessWithItem):
(WebKit::WebPageProxy::goForward):
(WebKit::WebPageProxy::goBack):
(WebKit::WebPageProxy::goToBackForwardItem):
(WebKit::WebPageProxy::restoreFromSessionState):
* UIProcess/WebPageProxy.h:
* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::goToBackForwardItem):
* WebProcess/WebPage/WebPage.h:
* WebProcess/WebPage/WebPage.messages.in:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181815 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoREGRESSION (r179429): Potential Use after free in JavaScriptCore`WTF::StringImpl...
utatane.tea@gmail.com [Fri, 20 Mar 2015 21:35:17 +0000 (21:35 +0000)]
REGRESSION (r179429): Potential Use after free in JavaScriptCore`WTF::StringImpl::ref + 83
https://bugs.webkit.org/show_bug.cgi?id=142410

Reviewed by Geoffrey Garen.

Source/JavaScriptCore:

Before this patch, added function JSValue::toPropertyKey returns PropertyName.
Since PropertyName doesn't have AtomicStringImpl ownership,
if Identifier is implicitly converted to PropertyName and Identifier is destructed,
PropertyName may refer freed AtomicStringImpl*.

This patch changes the result type of JSValue::toPropertyName from PropertyName to Identifier,
to keep AtomicStringImpl* ownership after the toPropertyName call is done.
And receive the result value as Identifier type to keep ownership in the caller side.

To catch the result of toPropertyKey as is, we catch the result of toPropertyName as auto.

However, now we don't need to have both Identifier and PropertyName.
So we'll merge PropertyName to Identifier in the subsequent patch.

* dfg/DFGOperations.cpp:
(JSC::DFG::operationPutByValInternal):
* jit/JITOperations.cpp:
(JSC::getByVal):
* llint/LLIntSlowPaths.cpp:
(JSC::LLInt::getByVal):
(JSC::LLInt::LLINT_SLOW_PATH_DECL):
* runtime/CommonSlowPaths.cpp:
(JSC::SLOW_PATH_DECL):
* runtime/CommonSlowPaths.h:
(JSC::CommonSlowPaths::opIn):
* runtime/JSCJSValue.h:
* runtime/JSCJSValueInlines.h:
(JSC::JSValue::toPropertyKey):
* runtime/ObjectConstructor.cpp:
(JSC::objectConstructorGetOwnPropertyDescriptor):
(JSC::objectConstructorDefineProperty):
* runtime/ObjectPrototype.cpp:
(JSC::objectProtoFuncPropertyIsEnumerable):

Source/WebCore:

The same issues are found in the existing code; PropertyName does not have ownership.
This patch rewrite the point that should have ownership to Identifier.

* bindings/js/JSDOMWindowCustom.cpp:
(WebCore::JSDOMWindow::getOwnPropertySlotByIndex):
(WebCore::JSDOMWindow::putByIndex):
* bindings/js/ReadableStreamJSSource.cpp:
(WebCore::getInternalSlotFromObject):
* bindings/scripts/CodeGeneratorJS.pm:
(GenerateImplementation):
* bindings/scripts/test/JS/JSTestCustomNamedGetter.cpp:
(WebCore::JSTestCustomNamedGetter::getOwnPropertySlotByIndex):
* bindings/scripts/test/JS/JSTestEventTarget.cpp:
(WebCore::JSTestEventTarget::getOwnPropertySlotByIndex):
* bindings/scripts/test/JS/JSTestInterface.cpp:
(WebCore::JSTestInterface::putByIndex):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181814 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoLog the slow-scrolling reasons (if any) on first commit of a ScrollingTreeFrameScroll...
simon.fraser@apple.com [Fri, 20 Mar 2015 21:32:28 +0000 (21:32 +0000)]
Log the slow-scrolling reasons (if any) on first commit of a ScrollingTreeFrameScrollingNodeMac
https://bugs.webkit.org/show_bug.cgi?id=142917

Reviewed by Beth Dakin.

The scrollperf test would fail to detect transitions between slow and fast scrolling
for new pages, because there was no logging when slow-scrolling scrolling tree was
torn down, and a new fast-scrolling tree constructed.

Fix by always logging the slow-scrolling reasons in the first update of the
ScrollingTreeFrameScrollingNodeMac, even when the slow-scrolling reasons haven't
changed.

* page/scrolling/mac/ScrollingTreeFrameScrollingNodeMac.h:
* page/scrolling/mac/ScrollingTreeFrameScrollingNodeMac.mm:
(WebCore::ScrollingTreeFrameScrollingNodeMac::ScrollingTreeFrameScrollingNodeMac):
(WebCore::ScrollingTreeFrameScrollingNodeMac::updateBeforeChildren):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181813 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Immediate Actions] Should re-enable immediate actions for iBooks
bdakin@apple.com [Fri, 20 Mar 2015 21:23:21 +0000 (21:23 +0000)]
[Immediate Actions] Should re-enable immediate actions for iBooks
https://bugs.webkit.org/show_bug.cgi?id=142915
-and corresponding-
rdar://problem/19585689

Reviewed by Tim Horton.

* UIProcess/API/mac/WKView.mm:
(-[WKView initWithFrame:processPool:configuration:webView:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181812 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Win] Attempting to get EWS green.
bfulgham@apple.com [Fri, 20 Mar 2015 21:16:51 +0000 (21:16 +0000)]
[Win] Attempting to get EWS green.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181811 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoFunction.prototype.toString should not decompile the AST
ggaren@apple.com [Fri, 20 Mar 2015 20:12:10 +0000 (20:12 +0000)]
Function.prototype.toString should not decompile the AST
https://bugs.webkit.org/show_bug.cgi?id=142853

Reviewed by Sam Weinig.

Source/JavaScriptCore:

To recover the function parameter string, Function.prototype.toString
decompiles the function parameters from the AST. This is bad for a few
reasons:

(1) It requires us to keep pieces of the AST live forever. This is an
awkward design and a waste of memory.

(2) It doesn't match Firefox or Chrome (because it changes whitespace
and ES6 destructuring expressions).

(3) It doesn't scale to ES6 default argument parameters, which require
arbitrarily complex decompilation.

(4) It can counterfeit all the line numbers in a function (because
whitespace can include newlines).

(5) It's expensive, and we've seen cases where websites invoke
Function.prototype.toString a lot by accident.

The fix is to do what we do for the rest of the function: Just quote the
original source text.

Since this change inevitably changes some function stringification, I
took the opportunity to make our stringification match Firefox's and
Chrome's.

* API/tests/testapi.c:
(assertEqualsAsUTF8String): Be more informative when this fails.

(main): Updated to match new stringification rules.

* bytecode/UnlinkedCodeBlock.cpp:
(JSC::UnlinkedFunctionExecutable::paramString): Deleted. Yay!
* bytecode/UnlinkedCodeBlock.h:

* parser/Nodes.h:
(JSC::StatementNode::isFuncDeclNode): New helper for constructing
anonymous functions.

* parser/SourceCode.h:
(JSC::SourceCode::SourceCode): Allow zero because WebCore wants it.

* runtime/CodeCache.cpp:
(JSC::CodeCache::getFunctionExecutableFromGlobalCode): Updated for use
of function declaration over function expression.

* runtime/Executable.cpp:
(JSC::FunctionExecutable::paramString): Deleted. Yay!
* runtime/Executable.h:
(JSC::FunctionExecutable::parameterCount):

* runtime/FunctionConstructor.cpp:
(JSC::constructFunctionSkippingEvalEnabledCheck): Added a newline after
the opening brace to match Firefox and Chrome, and a space after the comma
to match Firefox and WebKit coding style. Added the function name to
the text of the function so it would look right when stringify-ing. Switched
from parentheses to braces to produce a function declaration instead of
a function expression because we are required to exclude the function's
name from its scope, and that's what a function declaration does.

* runtime/FunctionPrototype.cpp:
(JSC::functionProtoFuncToString): Removed an old workaround because the
library it worked around doesn't really exist anymore, and the behavior
doesn't match Firefox or Chrome. Use type profiling offsets instead of
function body offsets because we want to include the function name and
the parameter string, rather than stitching them in manually by
decompiling the AST.

(JSC::insertSemicolonIfNeeded): Deleted.

* tests/mozilla/js1_2/function/tostring-1.js:
* tests/mozilla/js1_5/Scope/regress-185485.js:
(with.g): Updated these test results for formatting changes.

Source/WebCore:

* bindings/js/JSLazyEventListener.cpp:
(WebCore::JSLazyEventListener::initializeJSFunction): Adjust the line
number of attribute event listeners to account for the leading newline
now added by JavaScriptCore.

This solution is not perfect, but there are a lot of pre-existing problems
with line and column reporting for attribute event listeners, and this
preserves existing behavior with reasonable reliability.

LayoutTests:

Updated test results to match new rules for Function.prototype.toString.

* fast/dom/TreeWalker/acceptNode-filter-expected.txt: Removed a space
because it was not in the original source.

* fast/events/window-onerror2-expected.txt: Column number changed because
the event listener body starts on its own line now. This was a bit wrong
before and is still a bit wrong now in a different way.

* fast/profiler/dead-time-expected.txt:
* fast/profiler/inline-event-handler-expected.txt:
* fast/profiler/stop-profiling-after-setTimeout-expected.txt: Line number
changed because WebCore shifts line nubmers on attribute event listeners
by one.

* js/class-syntax-default-constructor-expected.txt: Constructor name
is not present now because it is not present in the source text. This
test failed before and it still fails now in a slightly different way.

* js/destructuring-assignment-expected.txt: Destructuring arguments now
match their source text faithfully.

* js/dfg-redundant-load-of-captured-variable-proven-constant-expected.txt:
Removed a space because it was not present in the original source text.

* js/dfg-resolve-global-specific-dictionary-expected.txt: Ditto.

* js/function-toString-semicolon-insertion-expected.txt: Removed.
* js/script-tests/function-toString-semicolon-insertion.js: Removed.
* js/function-toString-semicolon-insertion.html: Removed. This test checked
for a work-around that I have removed.

* js/object-literal-computed-methods-expected.txt:
* js/object-literal-methods-expected.txt: These tests fail because object
literal methods do not register their function names appropriately. This
was a pre-existing failure that is now more explicit.

* js/dom/JSON-parse-expected.txt:
* js/dom/JSON-stringify-expected.txt: Whitespace removed because it was
not present in the original.

* js/dom/dfg-strcat-over-objects-then-exit-on-it-expected.txt: Ditto.

* js/dom/function-prototype-expected.txt:
* js/dom/function-prototype.html: Ditto.

* js/dom/parse-error-external-script-in-new-Function-expected.txt: Line
changed by one due to new extra newline.

* js/dom/script-start-end-locations-expected.txt: Lines and columns
changed due to new extra newline.

* js/dom/toString-and-valueOf-override-expected.txt: Whitespace removed
because it was not present in the original.

* js/dom/script-tests/dfg-strcat-over-objects-then-exit-on-it.js: Ditto.

* js/kde/lval-exceptions-expected.txt: Ditto.

* js/script-tests/dfg-redundant-load-of-captured-variable-proven-constant.js: Ditto.

* js/script-tests/dfg-resolve-global-specific-dictionary.js: Ditto.

* platform/mac/http/tests/media/media-source/mediasource-sourcebuffer-mode-expected.txt: Ditto.

* storage/domstorage/localstorage/string-conversion-expected.txt: Ditto.

* storage/domstorage/sessionstorage/string-conversion-expected.txt: Ditto.

* userscripts/window-onerror-for-isolated-world-1-expected.txt:
* userscripts/window-onerror-for-isolated-world-2-expected.txt: Line numbers
changed because of new anonymous function formatting. These line numbers
were wrong before and they are still wrong now.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181810 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Win] Add some skips to support EWS use.
bfulgham@apple.com [Fri, 20 Mar 2015 19:14:00 +0000 (19:14 +0000)]
[Win] Add some skips to support EWS use.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181809 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoProgress towards CMake on Mac.
achristensen@apple.com [Fri, 20 Mar 2015 18:18:41 +0000 (18:18 +0000)]
Progress towards CMake on Mac.
https://bugs.webkit.org/show_bug.cgi?id=136003

Reviewed by Chris Dumez.

Source/WebCore:

* PlatformMac.cmake:
Added some directories to create forwarding headers from.
Generate ObjC bindings.

Source/WebKit:

* CMakeLists.txt:
* PlatformMac.cmake: Added.

Source/WebKit2:

* PlatformMac.cmake:
Add include directories and forwarding headers.
* Shared/API/APIObject.h:
(API::ObjectImpl::operator new):
* UIProcess/DrawingAreaProxy.h:
(WebKit::DrawingAreaProxy::intrinsicContentSizeDidChange):
* UIProcess/mac/TiledCoreAnimationDrawingAreaProxy.h:
Fix some unused parameter warnings.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181808 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoSyntaxChecker assertion is trapped with computed property name and getter
commit-queue@webkit.org [Fri, 20 Mar 2015 18:10:17 +0000 (18:10 +0000)]
SyntaxChecker assertion is trapped with computed property name and getter
https://bugs.webkit.org/show_bug.cgi?id=142863

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2015-03-20
Reviewed by Ryosuke Niwa.

Source/JavaScriptCore:

* parser/SyntaxChecker.h:
(JSC::SyntaxChecker::getName):
Remove invalid assert. Computed properties will not have a name
and the calling code is checking for null expecting it. The
AST path (non-CheckingPath) already does this without the assert
so it is well tested.

LayoutTests:

* js/basic-computed-property-name-expected.txt:
* js/script-tests/basic-computed-property-name.js:
(a.string_appeared_here.runTest):
Add tests that wrap the expression in a function, this
will test the syntax checker.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181807 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoJSCallbackObject<JSGlobalObject> should not destroy its JSCallbackObjectData before...
mark.lam@apple.com [Fri, 20 Mar 2015 18:08:29 +0000 (18:08 +0000)]
JSCallbackObject<JSGlobalObject> should not destroy its JSCallbackObjectData before all its finalizers have been called.
<https://webkit.org/b/142846>

Reviewed by Geoffrey Garen.

Currently, JSCallbackObject<JSGlobalObject> registers weak finalizers via 2 mechanisms:
1. JSCallbackObject<Parent>::init() registers a weak finalizer for all JSClassRef
   that a JSCallbackObject references.
2. JSCallbackObject<JSGlobalObject>::create() registers a finalizer via
   vm.heap.addFinalizer() which destroys the JSCallbackObject.

The first finalizer is implemented as a virtual function of a JSCallbackObjectData
instance that will be destructed if the 2nd finalizer is called.  Hence, if the
2nd finalizer if called first, the later invocation of the 1st finalizer will
result in a crash.

This patch fixes the issue by eliminating the finalizer registration in init().
Instead, we'll have the JSCallbackObject destructor call all the JSClassRef finalizers
if needed.  This ensures that these finalizers are called before the JSCallbackObject
is destructor.

Also added assertions to a few Heap functions because JSCell::classInfo() expects
all objects that are allocated from MarkedBlock::Normal blocks to be derived from
JSDestructibleObject.  These assertions will help us catch violations of this
expectation earlier.

* API/JSCallbackObject.cpp:
(JSC::JSCallbackObjectData::finalize): Deleted.
* API/JSCallbackObject.h:
(JSC::JSCallbackObjectData::~JSCallbackObjectData):
* API/JSCallbackObjectFunctions.h:
(JSC::JSCallbackObject<Parent>::~JSCallbackObject):
(JSC::JSCallbackObject<Parent>::init):
* API/tests/GlobalContextWithFinalizerTest.cpp: Added.
(finalize):
(testGlobalContextWithFinalizer):
* API/tests/GlobalContextWithFinalizerTest.h: Added.
* API/tests/testapi.c:
(main):
* JavaScriptCore.vcxproj/testapi/testapi.vcxproj:
* JavaScriptCore.vcxproj/testapi/testapi.vcxproj.filters:
* JavaScriptCore.xcodeproj/project.pbxproj:
* heap/HeapInlines.h:
(JSC::Heap::allocateObjectOfType):
(JSC::Heap::subspaceForObjectOfType):
(JSC::Heap::allocatorForObjectOfType):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181806 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[WinCairo] Crash when leaving page while video is playing.
peavo@outlook.com [Fri, 20 Mar 2015 17:53:31 +0000 (17:53 +0000)]
[WinCairo] Crash when leaving page while video is playing.
https://bugs.webkit.org/show_bug.cgi?id=142578

Reviewed by Alex Christensen.

This can happen when there is an asynchronous event invoked from a video worker
thread, accessing the media player object after it has been deleted.
This is solved by resetting the media player pointer in the event object,
when the media player is deleted.

* platform/graphics/win/MediaPlayerPrivateMediaFoundation.cpp:
(WebCore::MediaPlayerPrivateMediaFoundation::~MediaPlayerPrivateMediaFoundation):
(WebCore::MediaPlayerPrivateMediaFoundation::addListener):
(WebCore::MediaPlayerPrivateMediaFoundation::removeListener):
(WebCore::MediaPlayerPrivateMediaFoundation::notifyDeleted):
(WebCore::MediaPlayerPrivateMediaFoundation::AsyncCallback::AsyncCallback):
(WebCore::MediaPlayerPrivateMediaFoundation::AsyncCallback::~AsyncCallback):
(WebCore::MediaPlayerPrivateMediaFoundation::AsyncCallback::Invoke):
(WebCore::MediaPlayerPrivateMediaFoundation::AsyncCallback::onMediaPlayerDeleted):
* platform/graphics/win/MediaPlayerPrivateMediaFoundation.h:
(WebCore::MediaPlayerPrivateMediaFoundation::MediaPlayerListener::MediaPlayerListener):
(WebCore::MediaPlayerPrivateMediaFoundation::MediaPlayerListener::~MediaPlayerListener):
(WebCore::MediaPlayerPrivateMediaFoundation::MediaPlayerListener::onMediaPlayerDeleted):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181805 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoRegression(r181782): Caused WebKit2.RestoreSessionStateContainingFormData API test...
cdumez@apple.com [Fri, 20 Mar 2015 17:18:52 +0000 (17:18 +0000)]
Regression(r181782): Caused WebKit2.RestoreSessionStateContainingFormData API test to crash
https://bugs.webkit.org/show_bug.cgi?id=142911

Reviewed by Antti Koivisto.

Drop the isBackForwardLoadType(m_loadType) assertion in
FrameLoader::retryAfterFailedCacheOnlyMainResourceLoad(). This assertion
used to hold true when goToBackForwardItem() was only called with
history navigation load types. However, after r181782, we use a Standard
load in the cases where we do not wish to display stale content (e.g.
restoring session state).

The assertion comment says "We only use cache-only loads to avoid
resubmitting forms". However, we then call loadDifferentDocumentItem()
which already takes care of using the ReturnCacheDataElseLoad cache
policy as the HistoryItem has form data and we pass
MayNotAttemptCacheOnlyLoadForFormSubmissionItem as parameter.

No new tests, covered by WebKit2.RestoreSessionStateContainingFormData
API test.

* loader/FrameLoader.cpp:
(WebCore::FrameLoader::retryAfterFailedCacheOnlyMainResourceLoad):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181804 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoplatform/mac/fast/loader/file-url-mimetypes-3.html fails when MS Office is installed
ap@apple.com [Fri, 20 Mar 2015 17:07:18 +0000 (17:07 +0000)]
platform/mac/fast/loader/file-url-mimetypes-3.html fails when MS Office is installed
https://bugs.webkit.org/show_bug.cgi?id=142884

Reviewed by Tim Horton.

Disable a test for .xls until rdar://problem/20232251 is fixed. It is not particularly
relevant to a web browser which MIME type local Excel files get anyway.

* platform/mac/fast/loader/file-url-mimetypes-3-expected.txt:
* platform/mac/fast/loader/file-url-mimetypes-3.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181803 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Mac] svg/text/non-bmp-positioning-lists.svg fails when MS Office is installed
ap@apple.com [Fri, 20 Mar 2015 16:47:38 +0000 (16:47 +0000)]
[Mac] svg/text/non-bmp-positioning-lists.svg fails when MS Office is installed
https://bugs.webkit.org/show_bug.cgi?id=142861

Reviewed by Tim Horton.

Converted into a reftest. It would be nice to make tests not depend on installed
fonts in general, but that doesn't seem very feasible (rdar://problem/20000350).

* platform/efl/svg/text/non-bmp-positioning-lists-expected.png: Removed.
* platform/efl/svg/text/non-bmp-positioning-lists-expected.txt: Removed.
* platform/gtk/svg/text/non-bmp-positioning-lists-expected.png: Removed.
* platform/gtk/svg/text/non-bmp-positioning-lists-expected.txt: Removed.
* platform/ios-sim-deprecated/svg/text/non-bmp-positioning-lists-expected.txt: Removed.
* platform/ios-simulator/svg/text/non-bmp-positioning-lists-expected.txt: Removed.
* platform/mac/svg/text/non-bmp-positioning-lists-expected.png: Removed.
* platform/mac/svg/text/non-bmp-positioning-lists-expected.txt: Removed.
* platform/win/svg/text/non-bmp-positioning-lists-expected.txt: Removed.
* svg/text/non-bmp-positioning-lists-expected.svg: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181802 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Win] Temporarily turn of EWS Windows tests while I debug the test failure.
bfulgham@apple.com [Fri, 20 Mar 2015 15:57:44 +0000 (15:57 +0000)]
[Win] Temporarily turn of EWS Windows tests while I debug the test failure.

* Scripts/webkitpy/common/config/ews.json:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181801 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[CMake] Remove duplicated cmake function calls in WebKit2/CMakeLists.txt
gyuyoung.kim@samsung.com [Fri, 20 Mar 2015 11:26:13 +0000 (11:26 +0000)]
[CMake] Remove duplicated cmake function calls in WebKit2/CMakeLists.txt
https://bugs.webkit.org/show_bug.cgi?id=142901

Reviewed by Csaba Osztrogon√°c.

add_depencencies(), install() can support multiple targets or location.
We don't need to call it twice.

* CMakeLists.txt: Additionaly move -DBUILDING_WEBKIT to PlatformGTK.cmake,
because it is only used by GTK port.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181800 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[GTK] Disable accelerated compositing on Wayland
zandobersek@gmail.com [Fri, 20 Mar 2015 10:38:19 +0000 (10:38 +0000)]
[GTK] Disable accelerated compositing on Wayland
https://bugs.webkit.org/show_bug.cgi?id=142877

Reviewed by Carlos Garcia Campos.

* UIProcess/API/gtk/WebKitWebViewBase.cpp:
(webkitWebViewBaseCreateWebPage): As was done in the past, we should disable
accelerated compositing on Wayland until the proper support for it lands.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181798 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUnreviewed EFL gardening. Update attachment-foo tests to failure.
commit-queue@webkit.org [Fri, 20 Mar 2015 10:09:26 +0000 (10:09 +0000)]
Unreviewed EFL gardening. Update attachment-foo tests to failure.
https://bugs.webkit.org/show_bug.cgi?id=142653

Patch by Hunseop Jeong <hs85.jeong@samsung.com> on 2015-03-20

* platform/efl/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181797 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUnreviewed EFL gardening. Remove duplicate or ambiguous entry lines.
commit-queue@webkit.org [Fri, 20 Mar 2015 09:40:38 +0000 (09:40 +0000)]
Unreviewed EFL gardening. Remove duplicate or ambiguous entry lines.
https://bugs.webkit.org/show_bug.cgi?id=142775

Patch by Hunseop Jeong <hs85.jeong@samsung.com> on 2015-03-20

* platform/efl/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181796 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Content Filtering] Adopt NEFilterSource unblock handling
aestes@apple.com [Fri, 20 Mar 2015 09:36:21 +0000 (09:36 +0000)]
[Content Filtering] Adopt NEFilterSource unblock handling
https://bugs.webkit.org/show_bug.cgi?id=142904

Reviewed by Andreas Kling.

* platform/cocoa/NetworkExtensionContentFilter.mm:
(WebCore::NetworkExtensionContentFilter::unblockHandler):
Returned an unblock handler that calls -[NEFilterSource remediateWithDecisionHandler:].
* platform/cocoa/ParentalControlsContentFilter.mm:
(WebCore::ParentalControlsContentFilter::unblockHandler): Used ASCIILiteral.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181795 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[GTK] Search for the Wayland dependency when enabling Wayland target
zandobersek@gmail.com [Fri, 20 Mar 2015 09:23:40 +0000 (09:23 +0000)]
[GTK] Search for the Wayland dependency when enabling Wayland target
https://bugs.webkit.org/show_bug.cgi?id=142876

Reviewed by Carlos Garcia Campos.

* Source/cmake/OptionsGTK.cmake: The Wayland dependency isn't a public
requirement of either the GTK+ or GDK pkg-config files, so we have to
search for it ourselves when WebKitGTK+ has been configured to support
the Wayland windowing target.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181793 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[GTK][jhbuild] Fix flakey failures due to unstable googlecode repository
philn@webkit.org [Fri, 20 Mar 2015 09:00:49 +0000 (09:00 +0000)]
[GTK][jhbuild] Fix flakey failures due to unstable googlecode repository
https://bugs.webkit.org/show_bug.cgi?id=142867

Rubber-stamped by Csaba Osztrogon.

* gtk/jhbuild.modules: Use Ossy's libusrsctp Github fork. Also use
a fixed revision for the openwebrtc-gst-plugins.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181792 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Content Filtering] Add tests for unblock requests
aestes@apple.com [Fri, 20 Mar 2015 08:42:59 +0000 (08:42 +0000)]
[Content Filtering] Add tests for unblock requests
https://bugs.webkit.org/show_bug.cgi?id=142900

Reviewed by Andreas Kling.

Source/WebCore:

Currently the iOS Parental Controls content filter has a mechanism for requesting that a page be unblocked.
WebKit implements this by listening for navigations originating from the filter's error page to a special URL,
and requesting the page be unblocked via platform API, which might cause UI to be displayed. If the unblock is
successful then we schedule a reload of the frame in order to display the unblocked document.

NetworkExtension also supports unblock requests, so in preparation for adopting its API, this patch allows
content filters to specify their own unblock request method, teaches MockContentFilter to provide such a method,
and writes tests to cover both allowed and denied unblock requests.

The content filter that blocks a load creates a ContentFilterUnblockHandler, passing it a lambda that is executed
when a navigation matches the filter's special unblock URL. Filters can also specify that a script be executed in
the context of its error page if the unblock is denied.

All platform content filters can handle unblock requests like this with the exception of iOS Parental Controls in WebKit2.
Since UI can be displayed by the system in this case, the request must be made from within the UI process. Therefore the
existing method is retained of serializing a WebFilterEvaluator and intercepting navigation policy calls in the UI process.

Tests: contentfiltering/allow-after-unblock-request.html
       contentfiltering/block-after-unblock-request.html

* bindings/js/JSMockContentFilterSettingsCustom.cpp:
(WebCore::JSMockContentFilterSettings::decisionPoint): Added some using statements for clarity.
(WebCore::JSMockContentFilterSettings::setDecisionPoint): Ditto.
(WebCore::toJSValue): Returns a JSValue from a Decision.
(WebCore::toDecision): Returns a Decision from a JSValue.
(WebCore::JSMockContentFilterSettings::decision): Used toJSValue.
(WebCore::JSMockContentFilterSettings::setDecision): Used toDecision.
(WebCore::JSMockContentFilterSettings::unblockRequestDecision): Used toJSValue.
(WebCore::JSMockContentFilterSettings::setUnblockRequestDecision): Used toDecision.
* loader/ContentFilter.cpp:
(WebCore::ContentFilter::createIfNeeded): Passed a reference to the owning DocumentLoader.
(WebCore::ContentFilter::ContentFilter): Ditto.
(WebCore::ContentFilter::unblockHandler): If the unblockHandler requests that a script be executed when an
unblock request is denied, create a wrapper unblockHandler that executes that script in m_documentLoader's frame.
* loader/ContentFilter.h:
* loader/DocumentLoader.cpp:
(WebCore::DocumentLoader::responseReceived): Passed this to ContentFilter::createIfNeeded.
* loader/FrameLoader.cpp:
(WebCore::FrameLoader::prepareForLoadStart): Called PolicyChecker::prepareForLoadStart.
* loader/PolicyChecker.cpp:
(WebCore::PolicyChecker::prepareForLoadStart): Reset m_contentFilterUnblockHandler.
(WebCore::PolicyChecker::checkNavigationPolicy): Moved logic to here from WebKit1's WebFrameLoaderClient.
Placing it here allows it to be shared between WebKit1 and WebKit2 (when the unblock handler does not need to
be called in the UI process).
* loader/PolicyChecker.h:
(WebCore::PolicyChecker::setContentFilterUnblockHandler): Added.
* page/Frame.h: Made Frame ThreadSafeRefCounted, since RefPtr<Frames> are captured in lambdas that can be
copied by background threads managed by the underlying platform.
* platform/ContentFilterUnblockHandler.h:
(WebCore::ContentFilterUnblockHandler::unblockURLScheme): Returned the Apple content filter scheme.
(WebCore::ContentFilterUnblockHandler::unblockURLHost): Returned the unblock URL host.
(WebCore::ContentFilterUnblockHandler::clear): Deleted.
* platform/PlatformContentFilter.h:
(WebCore::PlatformContentFilter::unblockRequestDeniedScript): Returned the unblock request denied script.
* platform/cocoa/ContentFilterUnblockHandlerCocoa.mm:
(WebCore::ContentFilterUnblockHandler::ContentFilterUnblockHandler): Added a constructor that takes an
unblockURLHost and a UnblockRequesterFunction. Added an alternate constructor for iOS Parental Controls on
WebKit2 that takes an unblockURLHost and a WebFilterEvaluator.
(WebCore::ContentFilterUnblockHandler::needsUIProcess): Returned true if m_webFilterEvaluator is non-null.
(WebCore::ContentFilterUnblockHandler::encode): Encoded m_unblockURLHost in addition to m_webFilterEvaluator.
(WebCore::ContentFilterUnblockHandler::decode): Decoded m_unblockURLHost in addition to m_webFilterEvaluator.
(WebCore::ContentFilterUnblockHandler::canHandleRequest): Returned true if there is a either a m_unblockRequester
or a m_webFilterEvaluator and the request's host and scheme match those of the unblock request URL.
(WebCore::dispatchToMainThread): Added a helper to dispatch a block to the main thread. Then if the web thread
is enabled on iOS, dispatch it there.
(WebCore::ContentFilterUnblockHandler::requestUnblockAsync): Renamed from handleUnblockRequestAndDispatchIfSuccessful.
Requested an unblock using either m_unblockRequester or m_webFilterEvaluator, then called decisionHandler with the response.
(WebCore::scheme): Moved to ContentFilterUnblockHandler::unblockURLScheme.
(WebCore::ContentFilterUnblockHandler::handleUnblockRequestAndDispatchIfSuccessful): Renamed to requestUnblockAsync.
* platform/cocoa/ParentalControlsContentFilter.mm:
(WebCore::ParentalControlsContentFilter::unblockHandler): Returned an unblock handler using the WebFilterEvaluator constructor.
* testing/MockContentFilter.cpp: Added using statments for clarity.
(WebCore::settings): Added a helper to get MockContentFilterSettings::singleton().
(WebCore::MockContentFilter::canHandleResponse): Used the helper.
(WebCore::MockContentFilter::MockContentFilter): Took advantage of the using statements.
(WebCore::MockContentFilter::addData): Ditto.
(WebCore::MockContentFilter::finishedAddingData): Ditto.
(WebCore::MockContentFilter::unblockHandler): Returned a ContentFilterUnblockHandler that checks settings() for its decision.
(WebCore::MockContentFilter::unblockRequestDeniedScript): Returned the script to execute in MockContentFilter's
error page when an unblock request is denied.
(WebCore::MockContentFilter::maybeDetermineStatus): Took advantage of settings() and using statements.
* testing/MockContentFilterSettings.cpp:
(WebCore::MockContentFilterSettings::unblockRequestURL): Constructed a static unblock URL and returned it.
* testing/MockContentFilterSettings.h:
(WebCore::MockContentFilterSettings::unblockURLHost): Returned the filter's unblock URL host.
(WebCore::MockContentFilterSettings::unblockRequestDecision): Returns the decision to make for an unblock request.
(WebCore::MockContentFilterSettings::setUnblockRequestDecision): Sets the decision to make for an unblock request.
* testing/MockContentFilterSettings.idl: Added the unblockRequestDecision and unblockRequestURL attributes.

Source/WebKit/mac:

* WebCoreSupport/WebFrameLoaderClient.mm:
(WebFrameLoaderClient::dispatchDidStartProvisionalLoad): This now happens in PolicyChecker.
(WebFrameLoaderClient::dispatchDecidePolicyForNavigationAction): Ditto.
* WebView/WebFrame.mm:
(-[WebFrame _contentFilterDidHandleNavigationAction:]): Deleted.
* WebView/WebFrameInternal.h: Removed contentFilterUnblockHandler from WebFramePrivate.

Source/WebKit2:

* UIProcess/Cocoa/WebPageProxyCocoa.mm:
(WebKit::WebPageProxy::contentFilterDidBlockLoadForFrame): Called WebFrameProxy::contentFilterDidBlockLoad.
* UIProcess/WebFrameProxy.cpp:
(WebKit::WebFrameProxy::didStartProvisionalLoad): Assigned a default-constructed ContentFilterUnblockHandler instead of calling clear().
(WebKit::WebFrameProxy::didHandleContentFilterUnblockNavigation): Renamed from contentFilterDidHandleNavigationAction.
Updated to use ContentFilterUnblockHandler's new API.
(WebKit::WebFrameProxy::contentFilterDidHandleNavigationAction): Deleted.
* UIProcess/WebFrameProxy.h:
(WebKit::WebFrameProxy::contentFilterDidBlockLoad): Renamed from setContentFilterUnblockHandler.
(WebKit::WebFrameProxy::setContentFilterUnblockHandler): Deleted.
* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::decidePolicyForNavigationAction): Called WebFrameProxy::didHandleContentFilterUnblockNavigation.
* WebProcess/WebCoreSupport/WebFrameLoaderClient.cpp:
(WebKit::WebFrameLoaderClient::contentFilterDidBlockLoad): If the unblock handler needs the UI process,
send WebPageProxy::ContentFilterDidBlockLoadForFrame. Oterwise, call PolicyChecker::setContentFilterUnblockHandler.

LayoutTests:

Taught contentfiltering.js how to perform an unblock request test, and added tests for both allowed and blocked requests.

* contentfiltering/allow-after-unblock-request-expected.html: Added.
* contentfiltering/allow-after-unblock-request.html: Added.
* contentfiltering/block-after-unblock-request-expected.html: Added.
* contentfiltering/block-after-unblock-request.html: Added.
* contentfiltering/resources/contentfiltering.js:
(testContentFiltering): Added an argument specifying if the decision applies to the initial load or the unblock request.
(_doTest): When testing unblock handling, navigate the test iframe to settings.unblockRequestURL when the error page is displayed.
If the unblock is denied, the test harness will call window.unblockRequestDenied(). If the unblock is successful,
the iframe will reload, which we detect by listening for its load event.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181791 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[GTK] Properly guard X11-specific code in BackingStore::createBackend()
zandobersek@gmail.com [Fri, 20 Mar 2015 08:24:42 +0000 (08:24 +0000)]
[GTK] Properly guard X11-specific code in BackingStore::createBackend()
https://bugs.webkit.org/show_bug.cgi?id=142875

Reviewed by Martin Robinson.

* UIProcess/cairo/BackingStoreCairo.cpp:
(WebKit::BackingStore::createBackend): Guard the GTK- and X11-specific
bit of code with PLATFORM(GTK) and PLATFORM(X11). Testing GDK_WINDOWING_X11
ensures that the GTK+ dependency has X11 support, but does not ensure
that WebKitGTK+ has been configured to build for X11 environments (which
is what PLATFORM(X11) ensures).

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181789 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[GTK] Crash due to empty drag image during drag-and-drop
carlosgc@webkit.org [Fri, 20 Mar 2015 08:04:37 +0000 (08:04 +0000)]
[GTK] Crash due to empty drag image during drag-and-drop
https://bugs.webkit.org/show_bug.cgi?id=142671

Reviewed by Philippe Normand.

Return early from ImageBuffer constructor if an empty size is
given. This is a speculative fix for a crash while starting a drag
and drop operation, that I haven't been able to reproduce.

* platform/graphics/cairo/ImageBufferCairo.cpp:
(WebCore::ImageBuffer::ImageBuffer):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181787 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUnreviewed, kick the GTK bots to regenerate makefile.
ossy@webkit.org [Fri, 20 Mar 2015 08:02:13 +0000 (08:02 +0000)]
Unreviewed, kick the GTK bots to regenerate makefile.
https://bugs.webkit.org/show_bug.cgi?id=137394

* Source/cmake/OptionsEfl.cmake: Revert my last accidenatal change.
* Source/cmake/OptionsGTK.cmake: Really kick the GTK bots.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181786 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUnreviewed, kick the GTK bots to regenerate makefile.
ossy@webkit.org [Fri, 20 Mar 2015 07:52:25 +0000 (07:52 +0000)]
Unreviewed, kick the GTK bots to regenerate makefile.
https://bugs.webkit.org/show_bug.cgi?id=137394

* Source/cmake/OptionsEfl.cmake:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181785 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoFixed a ChangeLog misspelling pointed out by Andreas.
aestes@apple.com [Fri, 20 Mar 2015 06:01:11 +0000 (06:01 +0000)]
Fixed a ChangeLog misspelling pointed out by Andreas.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181784 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Content Filtering] Give contentfiltering tests a JavaScript harness
aestes@apple.com [Fri, 20 Mar 2015 05:58:32 +0000 (05:58 +0000)]
[Content Filtering] Give contentfiltering tests a JavaScript harness
https://bugs.webkit.org/show_bug.cgi?id=142899

Reviewed by Andreas Kling.

Create a JavaScript function to run the contentfiltering/ tests. This consolodates shared logic and will make
it easier to write new tests.

* TestExpectations: Skipped a test that times out with the new harness (but it also failed before).
* contentfiltering/allow-after-add-data.html: Called testContentFiltering().
* contentfiltering/allow-after-finished-adding-data.html: Ditto.
* contentfiltering/allow-after-response.html: Ditto.
* contentfiltering/block-after-add-data.html: Ditto.
* contentfiltering/block-after-finished-adding-data.html: Ditto.
* contentfiltering/block-after-response.html: Ditto.
* contentfiltering/resources/contentfiltering.js: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181783 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[WK2] We should allow stale content when restoring a killed / crashed WebProcess
cdumez@apple.com [Fri, 20 Mar 2015 05:45:38 +0000 (05:45 +0000)]
[WK2] We should allow stale content when restoring a killed / crashed WebProcess
https://bugs.webkit.org/show_bug.cgi?id=142893
<rdar://problem/19633737>
<rdar://problem/19869257>

Reviewed by Antti Koivisto.

Source/WebCore:

Always use ReturnCacheDataElseLoad cache policy for history navigations.
Previously, we would use the default protocol cache policy for history
navigations if m_stateMachine.committedFirstRealDocumentLoad() returned
false so that we do not display stale content when restoring the session
state of Safari statup (rdar://problem/8131355). This is now handled
before reaching this method by no longer using "IndexedBackForward"
FrameLoadType when restoring session tabs, and using "Standard" load
type instead.

* loader/FrameLoader.cpp:
(WebCore::FrameLoader::loadDifferentDocumentItem):

Source/WebKit2:

Allow stale content when restoring a killed / crashed WebProcess
(similarly to a history navigation). This is actually what was intended
as we call WebPage::goToBackForwardItem() in this case, which uses
IndexedBackForward load type. However, this value was then ignored in
FrameLoader::loadDifferentDocumentItem().

This patch gets rid of the special casing in
FrameLoader::loadDifferentDocumentItem() and makes it clear at call
sites when calling goToBackForwardItem() if we should allow stale
content or not. If goToBackForwardItem() is called with the
allowStaleContent flag off, we'll keep using the IndexedBackForward
frame load type. However, if the flag is on, we'll use the Standard
frame load type and do a fresh load.

* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::reattachToWebProcessForReload):
Set allowStaleContent flag to true as we want to allow stale content
when reloading a WebProcess after it crashed or was killed (behavior
change).

(WebKit::WebPageProxy::reattachToWebProcessWithItem):
Add allowStaleContent parameter so that call sites can decide which
policy to use.

(WebKit::WebPageProxy::goForward):
(WebKit::WebPageProxy::goBack):
Set allowStaleContent flag to true as this is our policy for history
navigations.

(WebKit::WebPageProxy::goToBackForwardItem):
Add allowStaleContent parameter so that the call site can let us know
what to do. This is useful as this is called not only for indexed
back / forward navigations (allowing stale content) but also for
restoring session state (forbidding state content:
<rdar://problem/8131355>).

(WebKit::WebPageProxy::restoreFromSessionState):
Call goToBackForwardItem() with allowStaleContent flag set to false
as users expect fresh contents in this case. This method is called
when:
- Tabs are restored on Safari / MobileSafari startup
- Tabs are restored via "History > Reopen All windows from last session"

* UIProcess/WebPageProxy.h:
* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::goToBackForwardItem):
Add allowStaleContent parameter and use "Standard" frame load type
instead of "IndexedBackForward" if it is false. This is needed as
"IndexedBackForward" frame load type allows using stale content.

* WebProcess/WebPage/WebPage.h:
* WebProcess/WebPage/WebPage.messages.in:
Add allowStaleContent flag to GoToBackForwardItem message.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181782 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Content Filtering] Move ContentFilter to from platform/ to loader/
aestes@apple.com [Fri, 20 Mar 2015 05:22:22 +0000 (05:22 +0000)]
[Content Filtering] Move ContentFilter to from platform/ to loader/
https://bugs.webkit.org/show_bug.cgi?id=142897

Reviewed by Dan Bernstein.

ContentFilterCollection never contained any platform code, so it doesn't make sense for it to be in platform.
Move it to loader/, since ContentFilters are owned by DocumentLoaders, and just call it ContentFilter. Extract
the abstract part of ContentFilter into a new base class called PlatformContentFilter that each of the
individual filters inherit from.

* WebCore.xcodeproj/project.pbxproj:
* loader/ContentFilter.cpp: Renamed from Source/WebCore/platform/ContentFilter.cpp.
(WebCore::ContentFilter::types): Moved from ContentFilterCollection.
(WebCore::ContentFilter::createIfNeeded): Ditto.
(WebCore::ContentFilter::ContentFilter): Ditto.
(WebCore::ContentFilter::addData): Ditto.
(WebCore::ContentFilter::finishedAddingData): Ditto.
(WebCore::ContentFilter::needsMoreData): Ditto.
(WebCore::ContentFilter::didBlockData): Ditto.
(WebCore::ContentFilter::getReplacementData): Ditto.
(WebCore::ContentFilter::unblockHandler): Ditto.
* loader/ContentFilter.h: Copied from Source/WebCore/platform/ContentFilter.h.
(WebCore::ContentFilter::type): Replaced a use of ContentFilterCollection.
* platform/PlatformContentFilter.h: Renamed from Source/WebCore/platform/ContentFilter.h.
* platform/cocoa/NetworkExtensionContentFilter.h: Inherited from PlatformContentFilter.
* platform/cocoa/ParentalControlsContentFilter.h: Ditto.
* testing/MockContentFilter.h: Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181781 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Win] Try to get the Debug bots green.
bfulgham@apple.com [Fri, 20 Mar 2015 05:06:40 +0000 (05:06 +0000)]
[Win] Try to get the Debug bots green.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181780 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Win] Skip an <attachment> test that times out on Windows.
bfulgham@apple.com [Fri, 20 Mar 2015 05:02:39 +0000 (05:02 +0000)]
[Win] Skip an <attachment> test that times out on Windows.

Also mark two flakes I noticed on the test bots.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181779 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoCSS Animations with triggers should map scroll position to duration
dino@apple.com [Fri, 20 Mar 2015 04:47:20 +0000 (04:47 +0000)]
CSS Animations with triggers should map scroll position to duration
https://bugs.webkit.org/show_bug.cgi?id=142870
<rdar://problem/20227244>

Reviewed by Simon Fraser.

Source/WebCore:

Expose a prototype implementation of what will eventually be called
animation-timebase, mapping the scroll location to the duration of an
animation. This only applies if the animation has a start and end
trigger.

Test: animations/trigger-container-scroll-boundaries.html
Test: animations/trigger-container-scroll-empty.html

* css/CSSComputedStyleDeclaration.cpp:
(WebCore::createAnimationTriggerValue): Change to downcast.

* css/CSSToStyleMap.cpp:
(WebCore::CSSToStyleMap::mapAnimationTrigger): Fix a bug where I wasn't
telling the ScrollAnimationTrigger object whether or not it had
an end value. This wasn't visible before because we were not using
the end value.

* page/animation/AnimationBase.cpp:
(WebCore::AnimationBase::fireAnimationEventsIfNeeded): Add a comment and
change to using a downcast.
(WebCore::AnimationBase::timeToNextService): Change to downcast.
(WebCore::AnimationBase::freezeAtTime): Whitespace cleanup.
(WebCore::AnimationBase::getElapsedTime): Calculate the elapsedTime based
on the scroll position relative to the start and end trigger spots.

* page/animation/AnimationController.cpp:
(WebCore::AnimationControllerPrivate::updateAnimations): Since this
can potentially call into beginAnimationUpdateTime, it should have
an update blocker. This fixes the assert we were seeing in tests.
(WebCore::AnimationControllerPrivate::scrollWasUpdated): Keep track of the scroll
position so we don't need to ask for it all the time.
* page/animation/AnimationControllerPrivate.h:
(WebCore::AnimationControllerPrivate::scrollPosition): New accessor.

* platform/animation/AnimationTrigger.h: Add downcast macros.
(WebCore::ScrollAnimationTrigger::create): Calculate hasEndValue from the passed
parameters rather than a separate value.
(WebCore::ScrollAnimationTrigger::hasEndValue):
(WebCore::ScrollAnimationTrigger::ScrollAnimationTrigger):
(WebCore::ScrollAnimationTrigger::setHasEndValue): Deleted.

* platform/graphics/ca/GraphicsLayerCA.cpp:
(WebCore::GraphicsLayerCA::animationCanBeAccelerated): If we have an
animation of this sort, we can't use a CAAnimation.

LayoutTests:

Add a new test that exercises an animation trigger over a scroll
region, and a test of the boundary condition. Also, unskip a test that
was failing due to a bug fixed in this patch.

* animations/trigger-container-scroll-boundaries-expected.txt: Added.
* animations/trigger-container-scroll-boundaries.html: Added.
* animations/trigger-container-scroll-empty-expected.txt: Added.
* animations/trigger-container-scroll-empty.html: Added.
* platform/efl/TestExpectations:
* platform/gtk/TestExpectations:
* platform/mac/TestExpectations:
* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181778 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Mac] Move MediaPlaybackTargetPicker
eric.carlson@apple.com [Fri, 20 Mar 2015 03:51:45 +0000 (03:51 +0000)]
[Mac] Move MediaPlaybackTargetPicker
https://bugs.webkit.org/show_bug.cgi?id=142889

Source/WebCore:

Move WebMediaPlaybackTargetPickerProxy from WK2 and rename as MediaPlaybackTargetPicker so it
can be used by both WK2 and WK1.

Reviewed by Tim Horton.

* WebCore.xcodeproj/project.pbxproj:
* platform/graphics/MediaPlaybackTargetPicker.cpp: Copied from Source/WebKit2/UIProcess/WebMediaPlaybackTargetPickerProxy.cpp.
* platform/graphics/MediaPlaybackTargetPicker.h: Copied from Source/WebKit2/UIProcess/WebMediaPlaybackTargetPickerProxy.h.
* platform/graphics/avfoundation/objc/MediaPlaybackTargetPickerMac.h: Copied from Source/WebKit2/UIProcess/mac/WebMediaPlaybackTargetPickerProxyMac.h.
* platform/graphics/avfoundation/objc/MediaPlaybackTargetPickerMac.mm: Copied from Source/WebKit2/UIProcess/mac/WebMediaPlaybackTargetPickerProxyMac.mm.

Source/WebKit2:

Move WebMediaPlaybackTargetPickerProxy to WebCore so it can be used by
both WK2 and WK1.

Reviewed by Tim Horton.

* UIProcess/PageClient.h:
* UIProcess/WebMediaPlaybackTargetPickerProxy.cpp: Removed.
* UIProcess/WebMediaPlaybackTargetPickerProxy.h: Removed.
* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::devicePickerProxy):
* UIProcess/WebPageProxy.h:
* UIProcess/mac/PageClientImpl.h:
* UIProcess/mac/PageClientImpl.mm:
(WebKit::PageClientImpl::createPlaybackTargetPicker):
* UIProcess/mac/WebMediaPlaybackTargetPickerProxyMac.h: Removed.
* UIProcess/mac/WebMediaPlaybackTargetPickerProxyMac.mm: Removed.
* WebKit2.xcodeproj/project.pbxproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181777 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago32-bit build fix after r181770.
jberlin@webkit.org [Fri, 20 Mar 2015 01:52:57 +0000 (01:52 +0000)]
32-bit build fix after r181770.

* UIProcess/mac/WKFullScreenWindowController.mm:
(-[WKFullScreenWindowController initWithWindow:webView:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181776 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoAdd support for deleting individual cache entries
andersca@apple.com [Fri, 20 Mar 2015 01:37:14 +0000 (01:37 +0000)]
Add support for deleting individual cache entries
https://bugs.webkit.org/show_bug.cgi?id=142886

Reviewed by Antti Koivisto.

* NetworkProcess/NetworkProcess.cpp:
(WebKit::fetchDiskCacheEntries):
Remove an unnecessary call to cfURLCacheOrigins().

(WebKit::clearDiskCacheEntries):
Helper function for clearing disk cache entries. This handles both the old and new cache implementations.

(WebKit::NetworkProcess::deleteWebsiteDataForOrigins):
Call clearDiskCacheEntries if needed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181775 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUnreviewed build fix. Don't try to report stats for all benchmarks.
rniwa@webkit.org [Fri, 20 Mar 2015 00:53:33 +0000 (00:53 +0000)]
Unreviewed build fix. Don't try to report stats for all benchmarks.
Just report the geometric mean of preferred means.

* Scripts/run-jsc-benchmarks:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181774 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoSource/WebCore:
enrica@apple.com [Fri, 20 Mar 2015 00:47:32 +0000 (00:47 +0000)]
Source/WebCore:
REGRESSION (r109593): Clicking after last inline element could cause a crash.
https://bugs.webkit.org/show_bug.cgi?id=142880
rdar://problem/17222294

Reviewed by Ryosuke Niwa.

Test: editing/selection/click-after-last-inline-crash.html

* rendering/RootInlineBox.cpp:
(WebCore::RootInlineBox::closestLeafChildForLogicalLeftPosition):

LayoutTests:
Web Inspector: Adopt ES6 Class Syntax for all Model Objects
https://bugs.webkit.org/show_bug.cgi?id=142858

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2015-03-19
Reviewed by Timothy Hatcher.

* inspector/model/parse-script-syntax-tree.html:
This test was calling a constructor without "new". Class
syntax enforces "new" and threw an exception.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181773 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoOne more build fix after r181760.
enrica@apple.com [Fri, 20 Mar 2015 00:44:12 +0000 (00:44 +0000)]
One more build fix after r181760.

* page/EventHandler.cpp:
(WebCore::EventHandler::handleDrag):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181772 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoBuild fix after r181760.
jberlin@webkit.org [Fri, 20 Mar 2015 00:38:14 +0000 (00:38 +0000)]
Build fix after r181760.

* rendering/RenderAttachment.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181771 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[WK2][Mac] Fullscreen animations with mismatched aspect ratios are "squished".
jer.noble@apple.com [Fri, 20 Mar 2015 00:20:52 +0000 (00:20 +0000)]
[WK2][Mac] Fullscreen animations with mismatched aspect ratios are "squished".
https://bugs.webkit.org/show_bug.cgi?id=142132

Reviewed by Tim Horton.

Source/WebCore:

Add a new utility method to calculate a rect with a specific aspect ratio wrapping a smaller one.

* platform/graphics/GeometryUtilities.cpp:
(WebCore::smallestRectWithAspectRatioAroundRect):
* platform/graphics/GeometryUtilities.h:

Source/WebKit2:

Use CALayer animations for the transition into and out of fullscreen.

The fullscreen transition consists of three separate animations:
1. An opacity animation for the black backdrop behind the fullscreen content
2. A scale/translation animation from fullscreen element's initial screen to its final one.
3. A clip animation from the fullscreen element's initial aspect ratio to its final one.

The opacity animation will apply to the fullscreen window's content view's layer's
background color. To separately animate the transform and mask of the web view's content, a
layer-backed subview is added to the content view, and the scale/translation & clip
animations are applied to its transform property and mask layer respectively.

Utility methods to create each animation have been added, and each includes a parameter for
the direction of the animation, so that the same methods can be used for entering and
exiting fullscreen transitions.

The user-visible changes to this new transition are when the aspect ratios of the initial
and final screen locations are different: previously the transition would use a scale
transform to "squish" the fullscreen content into the initial aspect ratio. The new
transition instead "clips" the fullscreen content to the initial aspect ratio. For common
operations such a <video> element with a different aspect ratio than the screen, this makes
the black letterbox "grow" during the transition, and makes the transition overall much
smoother.

* UIProcess/mac/WKFullScreenWindowController.h:
* UIProcess/mac/WKFullScreenWindowController.mm:
(-[WKFullScreenWindowController initWithWindow:webView:]): Create and initialze the clipping view.
(-[WKFullScreenWindowController applicationDidChangeScreenParameters:]): _backgroundWindow was removed.
(-[WKFullScreenWindowController enterFullScreen:]): Add the webView to the _clipView, not the contentView.
(-[WKFullScreenWindowController beganEnterFullScreenWithInitialFrame:finalFrame:]): _backgroundWindow,
        _fadeAnimation, and _scaleAnimation are all removed.
(-[WKFullScreenWindowController finishedEnterFullScreenAnimation:]): Ditto.
(-[WKFullScreenWindowController finishedExitFullScreenAnimation:]): Ditto.  Hide the contentView's
        layer. Pause visibility updates.
(-[WKFullScreenWindowController completeFinishExitFullScreenAnimationAfterRepaint]): Resume visibility updates.
(-[WKFullScreenWindowController close]): _fadeAnimation and _scaleAnimation are removed.
(-[WKFullScreenWindowController customWindowsToEnterFullScreenForWindow:]): Return only the fullscreen
        window.
(-[WKFullScreenWindowController customWindowsToExitFullScreenForWindow:]): Ditto.
(zoomAnimation): Added.
(maskAnimation): Added.
(fadeAnimation): Added.
(-[WKFullScreenWindowController _startEnterFullScreenAnimationWithDuration:]): Use the utility
        methods above to set up the animation.
(-[WKFullScreenWindowController _startExitFullScreenAnimationWithDuration:]): Ditto.
(createBackgroundFullscreenWindow): Deleted.
(windowFrameFromApparentFrames): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181770 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoWeb Inspector: Adopt ES6 Class Syntax for all Model Objects
joepeck@webkit.org [Fri, 20 Mar 2015 00:18:15 +0000 (00:18 +0000)]
Web Inspector: Adopt ES6 Class Syntax for all Model Objects
https://bugs.webkit.org/show_bug.cgi?id=142858

Reviewed by Timothy Hatcher.

Source/WebInspectorUI:

  - Convert WebInspector.Object to a class
  - Convert all UserInterface/Models objects to classes
  - Convert constructor functions to constructor methods
  - Convert "constructor.method" to class static methods where possible
  - Convert all methods to method syntax, eliminate commas between methods
  - Convert all superclass calls in classes to use "super"
  - Workaround <https://webkit.org/b/142862> and add empty constructors
  - Added "deprecated" prefix to addConstructorFunctions, since it is not needed with classes
  - Added many missing calls to super in constructors
  - Added FIXME to WebInspector.Object subclasses not yet moved to classes.
  - Cleaned up SourceMap global properties, moved to constructor instead of prototype
  - Cleaned up Timeline factory constructor to static "create" factory method
  - Fixed any style issues noticed in the mass edit
  - Fixed strict mode issues now that classes enforce strict mode
      - RunLoopTimelineRecord.js was missing a `var` for a local variable
      - "const" is not allowed, converted to "var"
      - "arguments.callee" is not allowed in strict mode

* UserInterface/**/*.js:
Many files modified mostly mechanically.

LayoutTests:

* inspector/model/parse-script-syntax-tree.html:
This test was calling a constructor without "new". Class
syntax enforces "new" and threw an exception.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181769 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoWeb Inspector: Debugger sidebar header should say "Scripts" instead of "Breakpoints...
jonowells@apple.com [Thu, 19 Mar 2015 23:52:13 +0000 (23:52 +0000)]
Web Inspector: Debugger sidebar header should say "Scripts" instead of "Breakpoints", appear only on pause
https://bugs.webkit.org/show_bug.cgi?id=142847

Reviewed by Timothy Hatcher.

Add a `paused` class on the debugger sidebar panel. Use that class to adjust the sidebar styles so that, when
the debugger is not paused: there is no header for the scripts section, there is no border beneath the scripts
section, and the scripts section cannot be collapsed (which could be done while the debugger is paused). The
header has changed to say "Scripts" instead of "Breakpoints" when it is showing.

* UserInterface/Views/DebuggerSidebarPanel.css:
(.sidebar > .panel.navigation.debugger .details-section.scripts):
(.sidebar > .panel.navigation.debugger .details-section.scripts .header):
(.sidebar > .panel.navigation.debugger .details-section.scripts.collapsed > .content):
(.sidebar > .panel.navigation.debugger.paused .details-section.scripts):
(.sidebar > .panel.navigation.debugger.paused .details-section.scripts .header):
(.sidebar > .panel.navigation.debugger.paused .details-section.scripts.collapsed > .content):
Adjust styles to hide header and border.

* UserInterface/Views/DebuggerSidebarPanel.js:
(WebInspector.DebuggerSidebarPanel):
(WebInspector.DebuggerSidebarPanel.prototype._debuggerDidPause):
(WebInspector.DebuggerSidebarPanel.prototype._debuggerDidResume):
Add a `paused` class to the debugger sidebar when it is paused.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181768 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoGet rid of "CachePolicyCache" cache policy
cdumez@apple.com [Thu, 19 Mar 2015 23:26:08 +0000 (23:26 +0000)]
Get rid of "CachePolicyCache" cache policy
https://bugs.webkit.org/show_bug.cgi?id=142881

Reviewed by Antti Koivisto.

Get rid of "CachePolicyCache" cache policy as it is unused.

* loader/cache/CachePolicy.h:
* loader/cache/CachedResource.cpp:
(WebCore::CachedResource::mustRevalidateDueToCacheHeaders):
* page/DiagnosticLoggingKeys.cpp:
(WebCore::DiagnosticLoggingKeys::mustRevalidateIsExpiredKey): Deleted.
* page/DiagnosticLoggingKeys.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181766 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoJSCallee unnecessarily overrides a bunch of things in the method table.
akling@apple.com [Thu, 19 Mar 2015 23:01:14 +0000 (23:01 +0000)]
JSCallee unnecessarily overrides a bunch of things in the method table.
<https://webkit.org/b/142855>

Reviewed by Geoffrey Garen.

Remove JSCallee method table overrides that simply call to base class.
This makes JSFunction property slot lookups slightly more efficient since
they can take the fast path when passing over JSCallee in the base class chain.

* runtime/JSCallee.cpp:
(JSC::JSCallee::getOwnPropertySlot): Deleted.
(JSC::JSCallee::getOwnNonIndexPropertyNames): Deleted.
(JSC::JSCallee::put): Deleted.
(JSC::JSCallee::deleteProperty): Deleted.
(JSC::JSCallee::defineOwnProperty): Deleted.
* runtime/JSCallee.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181765 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agorun-jsc-benchmarks should only report preferred mean
rniwa@webkit.org [Thu, 19 Mar 2015 22:46:56 +0000 (22:46 +0000)]
run-jsc-benchmarks should only report preferred mean
https://bugs.webkit.org/show_bug.cgi?id=142874

Reviewed by Geoffrey Garen.

Just collect statistics for the preferred mean.

* Scripts/run-jsc-benchmarks:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181764 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Content Extensions] Test regular expression parse failures.
enrica@apple.com [Thu, 19 Mar 2015 22:39:55 +0000 (22:39 +0000)]
[Content Extensions] Test regular expression parse failures.
https://bugs.webkit.org/show_bug.cgi?id=142872

Patch by Alex Christensen <achristensen@webkit.org> on 2015-03-19
Reviewed by Benjamin Poulain.

* contentextensions/URLFilterParser.cpp:
(WebCore::ContentExtensions::GraphBuilder::finalize):
(WebCore::ContentExtensions::GraphBuilder::quantifyAtom):
(WebCore::ContentExtensions::URLFilterParser::addPattern):
(WebCore::ContentExtensions::URLFilterParser::statusString):
* contentextensions/URLFilterParser.h:
Changed failures that can never happen to assertions.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181763 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Content Extensions] Test regular expression parse failures.
achristensen@apple.com [Thu, 19 Mar 2015 22:37:35 +0000 (22:37 +0000)]
[Content Extensions] Test regular expression parse failures.
https://bugs.webkit.org/show_bug.cgi?id=142872

Reviewed by Benjamin Poulain.

Source/WebCore:

* contentextensions/URLFilterParser.cpp:
(WebCore::ContentExtensions::GraphBuilder::finalize):
(WebCore::ContentExtensions::GraphBuilder::quantifyAtom):
(WebCore::ContentExtensions::URLFilterParser::addPattern):
(WebCore::ContentExtensions::URLFilterParser::statusString):
* contentextensions/URLFilterParser.h:
Changed failures that can never happen to assertions.

Tools:

* TestWebKitAPI/Tests/WebCore/ContentExtensions.cpp:
(TestWebKitAPI::testPatternStatus):
(TestWebKitAPI::TEST_F):
Added tests for each parsing status.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181762 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUnreviewed build fixes.
rniwa@webkit.org [Thu, 19 Mar 2015 22:36:43 +0000 (22:36 +0000)]
Unreviewed build fixes.

* public/include/manifest.php:
(Manifest::generate): These should be {} instead of [] when they're empty.
* public/v2/data.js:
(Measurement.prototype.formattedRevisions): Don't assume previousRevisions[repositoryId] exits.
* public/v2/manifest.js:
(App.Metric.fullName): Fixed the typo.
* tests/admin-regenerate-manifest.js: Fixed the test.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181761 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago<attachment> should put URLs on the pasteboard so that Finder can accept drops.
enrica@apple.com [Thu, 19 Mar 2015 21:59:23 +0000 (21:59 +0000)]
<attachment> should put URLs on the pasteboard so that Finder can accept drops.
https://bugs.webkit.org/show_bug.cgi?id=142801
rdar://problem/19982527

Reviewed by Tim Horton.

Source/WebCore:

Test: editing/pasteboard/drag-and-drop-attachment-contenteditable.html

This patch adds all the necessary support to write attachment elements into
the pasteboard, including the promised type.

* WebCore.xcodeproj/project.pbxproj:
* page/DragActions.h:
* page/DragClient.h:
(WebCore::DragClient::declareAndWriteAttachment):
* page/DragController.cpp:
(WebCore::DragController::draggableElement):
(WebCore::DragController::startDrag):
* page/DragController.h:
(WebCore::DragController::draggingAttachmentURL):
* page/EventHandler.cpp:
(WebCore::EventHandler::dragHysteresisExceeded):
(WebCore::EventHandler::handleDrag):
* page/mac/DragControllerMac.mm:
(WebCore::DragController::declareAndWriteAttachment):
* page/win/DragControllerWin.cpp:
(WebCore::DragController::declareAndWriteAttachment):
* platform/URL.cpp:
(WebCore::URL::fileURLWithFileSystemPath):
* platform/URL.h:
* rendering/HitTestResult.cpp:
(WebCore::HitTestResult::absoluteAttachmentURL):
* rendering/HitTestResult.h:

Source/WebKit/mac:

This patch adds all the necessary support to write attachment elements into
the pasteboard, including the promised type in WK1.

* Misc/WebNSPasteboardExtras.mm:
(-[NSPasteboard _web_declareAndWriteDragImageForElement:URL:title:archive:source:]):
* WebCoreSupport/WebDragClient.h:
* WebCoreSupport/WebDragClient.mm:
(WebDragClient::declareAndWriteAttachment):
* WebView/WebHTMLView.mm:
(-[WebHTMLView namesOfPromisedFilesDroppedAtDestination:]):
* WebView/WebUIDelegate.h:

Source/WebKit2:

This patch adds all the necessary support to write attachment elements into
the pasteboard, including the promised type in WK2.

* UIProcess/API/mac/WKView.mm:
(-[WKView _setPromisedDataForImage:withFileName:withExtension:withTitle:withURL:withVisibleURL:withArchive:forPasteboard:]):
(-[WKView _setPromisedDataForAttachment:withExtension:withTitle:withURL:withVisibleURL:forPasteboard:]):
(-[WKView namesOfPromisedFilesDroppedAtDestination:]):
(-[WKView _setPromisedData:withFileName:withExtension:withTitle:withURL:withVisibleURL:withArchive:forPasteboard:]): Deleted.
* UIProcess/API/mac/WKViewInternal.h:
* UIProcess/PageClient.h:
* UIProcess/WebPageProxy.h:
* UIProcess/WebPageProxy.messages.in:
* UIProcess/mac/PageClientImpl.h:
* UIProcess/mac/PageClientImpl.mm:
(WebKit::PageClientImpl::setPromisedDataForImage):
(WebKit::PageClientImpl::setPromisedDataForAttachment):
(WebKit::PageClientImpl::setPromisedData): Deleted.
* UIProcess/mac/WebPageProxyMac.mm:
(WebKit::WebPageProxy::setPromisedDataForImage):
(WebKit::WebPageProxy::setPromisedDataForAttachment):
(WebKit::WebPageProxy::setPromisedData): Deleted.
* WebProcess/WebCoreSupport/WebDragClient.h:
* WebProcess/WebCoreSupport/mac/WebDragClientMac.mm:
(WebKit::WebDragClient::declareAndWriteAttachment):
(WebKit::WebDragClient::declareAndWriteDragImage):

LayoutTests:

* editing/pasteboard/drag-and-drop-attachment-contenteditable-expected.txt: Added.
* editing/pasteboard/drag-and-drop-attachment-contenteditable.html: Added.
* platform/mac-wk2/TestExpectations: Skipping new test on WK2 since it uses eventSender.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181760 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoWeb Inspector: FilterBar for debugger sidebar hides breakpoints for displayed resources
jonowells@apple.com [Thu, 19 Mar 2015 21:44:59 +0000 (21:44 +0000)]
Web Inspector: FilterBar for debugger sidebar hides breakpoints for displayed resources
https://bugs.webkit.org/show_bug.cgi?id=142777

Reviewed by Brian Burg.

Typing into the text input on the filter bar for the debugger sidebar no longer incorrectly hides the
breakpoints for scripts that match the text input.

* UserInterface/Views/BreakpointTreeElement.js:
(WebInspector.BreakpointTreeElement.prototype.get filterableData): Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181759 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoDFGAllocator should use bmalloc's aligned allocator.
akling@apple.com [Thu, 19 Mar 2015 21:19:32 +0000 (21:19 +0000)]
DFGAllocator should use bmalloc's aligned allocator.
<https://webkit.org/b/142871>

Reviewed by Geoffrey Garen.

Source/JavaScriptCore:

Switch DFGAllocator to using bmalloc through fastAlignedMalloc().

* dfg/DFGAllocator.h:
(JSC::DFG::Allocator<T>::allocateSlow):
(JSC::DFG::Allocator<T>::freeRegionsStartingAt):
* heap/CopiedSpace.h:
* heap/MarkedBlock.h:
* heap/MarkedSpace.h:

Source/WTF:

Remove PageAllocationAligned since it has no more clients.

* WTF.vcxproj/WTF.vcxproj:
* WTF.vcxproj/WTF.vcxproj.filters:
* WTF.xcodeproj/project.pbxproj:
* wtf/CMakeLists.txt:
* wtf/PageAllocationAligned.cpp: Removed.
* wtf/PageAllocationAligned.h: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181758 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUnreviewed, fix linking error after r181753.
cdumez@apple.com [Thu, 19 Mar 2015 21:17:53 +0000 (21:17 +0000)]
Unreviewed, fix linking error after r181753.

Drop WEBCORE_EXPORT for Settings::minimumDOMTimerInterval() as it is
now inlined.

* page/Settings.h:
(WebCore::Settings::minimumDOMTimerInterval):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181757 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoCache::remove should take a Key instead of an entry
andersca@apple.com [Thu, 19 Mar 2015 21:16:47 +0000 (21:16 +0000)]
Cache::remove should take a Key instead of an entry
https://bugs.webkit.org/show_bug.cgi?id=142878

Reviewed by Antti Koivisto.

* NetworkProcess/cache/NetworkCache.cpp:
(WebKit::NetworkCache::Cache::remove):
* NetworkProcess/cache/NetworkCache.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181756 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoFix API tests after r181754.
achristensen@apple.com [Thu, 19 Mar 2015 20:50:18 +0000 (20:50 +0000)]
Fix API tests after r181754.

* TestWebKitAPI/Tests/WebCore/ContentExtensions.cpp:
(TestWebKitAPI::testRequest):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181755 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoPart of content extensions should apply css selectors
beidson@apple.com [Thu, 19 Mar 2015 19:36:22 +0000 (19:36 +0000)]
Part of content extensions should apply css selectors
https://bugs.webkit.org/show_bug.cgi?id=142604

Reviewed by Alex Christensen.

Source/WebCore:

Add a new action for applying the cached global stylesheet.

* contentextensions/CompiledContentExtension.cpp:
(WebCore::ContentExtensions::CompiledContentExtension::globalDisplayNoneSelectors):

* contentextensions/ContentExtensionActions.h:

* contentextensions/ContentExtensionCompiler.cpp:
(WebCore::ContentExtensions::serializeActions):

* contentextensions/ContentExtensionParser.cpp:
(WebCore::ContentExtensions::loadAction):

* contentextensions/ContentExtensionRule.cpp:
(WebCore::ContentExtensions::Action::deserialize):
* contentextensions/ContentExtensionRule.h:
(WebCore::ContentExtensions::Action::Action):
(WebCore::ContentExtensions::Action::operator==):
(WebCore::ContentExtensions::Action::stringArgument):
(WebCore::ContentExtensions::Action::cssSelector): Deleted.

* contentextensions/ContentExtensionsBackend.cpp:
(WebCore::ContentExtensions::ContentExtensionsBackend::actionsForResourceLoad):

* loader/cache/CachedResourceLoader.cpp:
(WebCore::CachedResourceLoader::requestResource):

Tools:

* TestWebKitAPI/Tests/WebCore/ContentExtensions.cpp:
(WebCore::ContentExtensions::operator<<):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181754 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoClean up DOMTimer related settings
cdumez@apple.com [Thu, 19 Mar 2015 19:22:22 +0000 (19:22 +0000)]
Clean up DOMTimer related settings
https://bugs.webkit.org/show_bug.cgi?id=142837

Reviewed by Darin Adler.

Source/WebCore:

Clean up DOMTimer related settings:
- Make default minimum timer interval 4ms instead of 10ms as all ports
  were setting it to 4ms and the specifications says 4ms.
- Get rid of setters for default DOMTimer interval values as they are
  not set anymore.
- Move default DOMTimer interval values from Settings to DOMTimer as
  those are global (not per-page) and not really settings.
- Stop using abbreviations in the names (e.g. min -> minimum).
- Move DOMTimer settings members from Page to Settings. There is no
  reason for those to be stored on the Page. The getters were private
  so call-sites had to go via Settings anyway. Also, call-sites already
  had to do a null check on the page to get the settings. It seems
  unfortunate for the settings to do another null-check on the page for
  retrieving the setting value from the page.
- Inline the DOMTimer settings getters as these are trivial.

* dom/Document.cpp:
(WebCore::Document::minimumTimerInterval):
(WebCore::Document::timerAlignmentInterval):
* dom/ScriptExecutionContext.cpp:
(WebCore::ScriptExecutionContext::minimumTimerInterval):
(WebCore::ScriptExecutionContext::timerAlignmentInterval):
* page/DOMTimer.h:
* page/Page.cpp:
(WebCore::Page::Page):
(WebCore::Page::setTimerThrottlingEnabled):
(WebCore::Page::setMinimumTimerInterval): Deleted.
(WebCore::Page::minimumTimerInterval): Deleted.
* page/Page.h:
(WebCore::Page::timerAlignmentInterval): Deleted.
* page/Settings.cpp:
(WebCore::Settings::Settings):
(WebCore::Settings::setMinimumDOMTimerInterval):
(WebCore::Settings::setDOMTimerAlignmentInterval):
(WebCore::Settings::setHiddenPageDOMTimerAlignmentInterval): Deleted.
(WebCore::Settings::hiddenPageDOMTimerAlignmentInterval): Deleted.
(WebCore::Settings::setDefaultMinDOMTimerInterval): Deleted.
(WebCore::Settings::defaultMinDOMTimerInterval): Deleted.
(WebCore::Settings::setMinDOMTimerInterval): Deleted.
(WebCore::Settings::minDOMTimerInterval): Deleted.
(WebCore::Settings::setDefaultDOMTimerAlignmentInterval): Deleted.
(WebCore::Settings::defaultDOMTimerAlignmentInterval): Deleted.
(WebCore::Settings::domTimerAlignmentInterval): Deleted.
* page/Settings.h:
(WebCore::Settings::minimumDOMTimerInterval):
(WebCore::Settings::domTimerAlignmentInterval):
* testing/InternalSettings.cpp:
(WebCore::InternalSettings::Backup::Backup):
(WebCore::InternalSettings::Backup::restoreTo):
(WebCore::InternalSettings::setMinimumTimerInterval):

Source/WebKit/mac:

Stop setting the DOMTimers' default minimum interval to 4ms as this
is now the default.

* WebView/WebView.mm:
(-[WebView _commonInitializationWithFrameName:groupName:]):

Source/WebKit/win:

Stop setting the DOMTimers' default minimum interval to 4ms as this
is now the default.

* WebView.cpp:
(WebView::initWithFrame):
(WebView::defaultMinimumTimerInterval):

Source/WebKit2:

Stop setting the DOMTimers' default minimum interval to 4ms as this
is now the default.

* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::WebPage):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181753 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agohttp://webkit.org/b/142790
dino@apple.com [Thu, 19 Mar 2015 17:14:59 +0000 (17:14 +0000)]
http://webkit.org/b/142790

Temporarily skip animations/trigger-container-scroll-simple.html
while it is crashing.

* platform/mac/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181749 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Win] Activate tests on EWS
bfulgham@apple.com [Thu, 19 Mar 2015 16:42:56 +0000 (16:42 +0000)]
[Win] Activate tests on EWS
https://bugs.webkit.org/show_bug.cgi?id=142850

Reviewed by Alexey Proskuryakov.

* Scripts/webkitpy/common/config/ews.json: Add me as
the EWS bot watcher, and activate tests.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181748 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Win] Test gardening.
bfulgham@apple.com [Thu, 19 Mar 2015 16:40:36 +0000 (16:40 +0000)]
[Win] Test gardening.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181747 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[GTK] Skip Legacy WebNotifications test
commit-queue@webkit.org [Thu, 19 Mar 2015 13:25:25 +0000 (13:25 +0000)]
[GTK] Skip Legacy WebNotifications test
https://bugs.webkit.org/show_bug.cgi?id=141962

The revision r177073 skips all legacy tests for notifications except from
http/tests/notifications/legacy/notification-request-permission-then-navigate.html
which was failing. We skip this test now too.

Patch by Marcos Chavarr√≠a Teijeiro <chavarria1991@gmail.com> on 2015-03-19
Reviewed by Carlos Garcia Campos.

* platform/gtk/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181746 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[GTK] Scrollbars look bad with GTK+ 3.16
carlosgc@webkit.org [Thu, 19 Mar 2015 10:24:50 +0000 (10:24 +0000)]
[GTK] Scrollbars look bad with GTK+ 3.16
https://bugs.webkit.org/show_bug.cgi?id=140800

Reviewed by Sergio Villar Senin.

Take margin into account when rendering scrollbars. This fixes the
huge scrollbars rendered with GTK+ 3.16. We don't need to check
the GTK+ version because in previous versions the marging were 0,
so the same code just works.

* platform/gtk/ScrollbarThemeGtk.cpp:
(WebCore::adjustRectAccordingToMargin):
(WebCore::ScrollbarThemeGtk::paintTrackBackground):
(WebCore::ScrollbarThemeGtk::paintThumb):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181744 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Streams API] Update ReadableStream API according new version of the specification
youenn.fablet@crf.canon.fr [Thu, 19 Mar 2015 08:27:27 +0000 (08:27 +0000)]
[Streams API] Update ReadableStream API according new version of the specification
https://bugs.webkit.org/show_bug.cgi?id=142822

Reviewed by Benjamin Poulain.

Source/WebCore:

This patch updates the IDL of ReadableStream according the new version of the spec, which splits functionality between ReadableStream and ReadableStreamReader.
In particular, this patch removes read(), ready, closed and state from ReadableStream and it adds the getReader method.

Covered by updated readablestream-constructor.html test.

* Modules/streams/ReadableStream.cpp:
(WebCore::ReadableStream::ReadableStream):
(WebCore::ReadableStream::state): Deleted.
(WebCore::ReadableStream::closed): Deleted.
(WebCore::ReadableStream::ready): Deleted.
* Modules/streams/ReadableStream.h:
(WebCore::ReadableStream::internalState): Added to make mac build system happy, to be used by ReadableStreamReader.
* Modules/streams/ReadableStream.idl:
* bindings/js/JSReadableStreamCustom.cpp:
(WebCore::JSReadableStream::cancel):
(WebCore::JSReadableStream::getReader):
(WebCore::JSReadableStream::read): Deleted.
(WebCore::getOrCreatePromiseDeferredFromObject): Deleted.
(WebCore::readyPromiseSlotName): Deleted.
(WebCore::JSReadableStream::ready): Deleted.
(WebCore::closedPromiseSlotName): Deleted.
(WebCore::JSReadableStream::closed): Deleted.

LayoutTests:

Removing tests checking ready and closed.
Removing assertions checking read(), ready, closed and state.
Adding assertions to test getReader() and parameters of remaining methods.

* streams/readablestream-constructor-expected.txt:
* streams/readablestream-constructor.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181736 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[GTK] Fix inspector userinterface related incremental build issue
ossy@webkit.org [Thu, 19 Mar 2015 08:19:20 +0000 (08:19 +0000)]
[GTK] Fix inspector userinterface related incremental build issue
https://bugs.webkit.org/show_bug.cgi?id=142849

Reviewed by Carlos Garcia Campos.

.:

* Source/cmake/OptionsGTK.cmake: Revert r181733.

Tools:

* Scripts/webkitdirs.pm:
(shouldRemoveCMakeCache):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181735 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoThe network cache should ignore "cache-control: no-cache, must-revalidate" on history...
cdumez@apple.com [Thu, 19 Mar 2015 07:54:06 +0000 (07:54 +0000)]
The network cache should ignore "cache-control: no-cache, must-revalidate" on history navigation
https://bugs.webkit.org/show_bug.cgi?id=142831
<rdar://problem/20209957>

Reviewed by Antti Koivisto.

Source/WebKit2:

The network cache should ignore "no-cache" and "must-revalidate" as
Cache-Control policy for history navigations to avoid unnecessary
revalidation in this case. Our memory cache already behaves this way.

On history navigation, our disk cache implementation currently
revalidates unconditionally resources if "cache-control: no-cache" is
used. It also revalidates expired resources if "cache-control:
must-revalidate" is used. This is sub-optimal.

RFC 7234 does not require us to do so [1]:
"""
  User agents often have history mechanisms, such as "Back" buttons and
  history lists, that can be used to redisplay a representation
  retrieved earlier in a session.

  The freshness model (Section 4.2) does not necessarily apply to
  history mechanisms.  That is, a history mechanism can display a
  previous representation even if it has expired.
"""

Chrome and IE9+ [2] already ignore "cache-control: must-revalidate" on history navigation.

[1] https://tools.ietf.org/html/rfc7234#page-32
[2] http://blogs.msdn.com/b/ie/archive/2010/07/14/caching-improvements-in-internet-explorer-9.aspx

Test: http/tests/cache/disk-cache-validation-back-navigation-policy.html

* NetworkProcess/cache/NetworkCache.cpp:
(WebKit::NetworkCache::decodeStorageEntry):

LayoutTests:

Rebaseline disk-cache-validation-back-navigation-policy.html test now
that we ignore "cache-control: no-cache, must-revalidate" on history
navigation. This is a progression.

* http/tests/cache/disk-cache-validation-back-navigation-policy-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181734 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUnreviewed, kick the GTK bots to regenerate makefile.
ossy@webkit.org [Thu, 19 Mar 2015 06:30:11 +0000 (06:30 +0000)]
Unreviewed, kick the GTK bots to regenerate makefile.

* Source/cmake/OptionsGTK.cmake:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181733 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Win] Windows gardening for debug bots.
bfulgham@apple.com [Thu, 19 Mar 2015 06:22:29 +0000 (06:22 +0000)]
[Win] Windows gardening for debug bots.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181732 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUnreviewed. GTK build fix after r181720.
rego@igalia.com [Thu, 19 Mar 2015 06:09:04 +0000 (06:09 +0000)]
Unreviewed. GTK build fix after r181720.

* rendering/RenderImage.cpp:
(WebCore::RenderImage::styleDidChange):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181729 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoTest disk cache behavior when using back navigation cache policy
antti@apple.com [Thu, 19 Mar 2015 05:21:40 +0000 (05:21 +0000)]
Test disk cache behavior when using back navigation cache policy
https://bugs.webkit.org/show_bug.cgi?id=142848

Reviewed by Chris Dumez.

Source/WebCore:

Add a way to override the normal cache policy so we can test reload behavior of cache without actually reloading.

Test: http/tests/cache/disk-cache-validation-back-navigation-policy.html

* loader/FrameLoader.cpp:
(WebCore::FrameLoader::addExtraFieldsToRequest):
* loader/FrameLoader.h:
(WebCore::FrameLoader::setOverrideCachePolicyForTesting):
* testing/Internals.cpp:
(WebCore::Internals::setOverrideCachePolicy):
* testing/Internals.h:
* testing/Internals.idl:

LayoutTests:

* http/tests/cache/disk-cache-validation-back-navigation-policy-expected.txt: Added.
* http/tests/cache/disk-cache-validation-back-navigation-policy.html: Added.

    Test permutations of cache headers on reload including Cache-control: must-revalidate.

* platform/mac-wk1/TestExpectations:
* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181728 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoTextFragment#start() is always >= 0 since its type is unsigned
utatane.tea@gmail.com [Thu, 19 Mar 2015 05:16:41 +0000 (05:16 +0000)]
TextFragment#start() is always >= 0 since its type is unsigned
https://bugs.webkit.org/show_bug.cgi?id=142860

Reviewed by Andreas Kling.

Since TextFragment::m_start is unsigned, the result of
TextFragment::start() is always >= 0 and assertion is not effective.
This patch removes this assertion to suppress warnings.

* rendering/SimpleLineLayoutTextFragmentIterator.cpp:
(WebCore::SimpleLineLayout::TextFragmentIterator::revertToFragment):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181727 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[ContentExtensions] Prepare for compiling stylesheets of selectors to be used on...
commit-queue@webkit.org [Thu, 19 Mar 2015 04:57:20 +0000 (04:57 +0000)]
[ContentExtensions] Prepare for compiling stylesheets of selectors to be used on every page.
https://bugs.webkit.org/show_bug.cgi?id=142799

Patch by Alex Christensen <achristensen@webkit.org> on 2015-03-18
Reviewed by Brady Eidson.

Source/WebCore:

* WebCore.xcodeproj/project.pbxproj:
Make private headers to use with API tests.
* contentextensions/CompiledContentExtension.cpp:
(WebCore::ContentExtensions::CompiledContentExtension::globalDisplayNoneSelectors):
* contentextensions/CompiledContentExtension.h:
Added method to get only the selectors from the root of the DFA, which apply to all URLs.
* contentextensions/ContentExtensionCompiler.cpp:
(WebCore::ContentExtensions::compileRuleList):
Added checking if the trigger will match everything.
These actions can be put directly on the root of the DFA without adding extra epsilon transitions to the NFA.
* contentextensions/DFA.h:
(WebCore::ContentExtensions::DFA::nodeAt):
* contentextensions/DFABytecodeInterpreter.cpp:
(WebCore::ContentExtensions::DFABytecodeInterpreter::actionsFromDFARoot):
(WebCore::ContentExtensions::DFABytecodeInterpreter::interpret):
* contentextensions/DFABytecodeInterpreter.h:
* contentextensions/NFA.h:
* contentextensions/URLFilterParser.cpp:
(WebCore::ContentExtensions::Term::quantifier):
Sink terms to a vector then add nodes to NFA when finalizing after checking for regexes that match everything.
(WebCore::ContentExtensions::GraphBuilder::GraphBuilder):
(WebCore::ContentExtensions::GraphBuilder::finalize):
(WebCore::ContentExtensions::GraphBuilder::parseStatus):
(WebCore::ContentExtensions::GraphBuilder::atomPatternCharacter):
(WebCore::ContentExtensions::GraphBuilder::atomBuiltInCharacterClass):
(WebCore::ContentExtensions::GraphBuilder::quantifyAtom):
(WebCore::ContentExtensions::GraphBuilder::atomBackReference):
(WebCore::ContentExtensions::GraphBuilder::assertionBOL):
(WebCore::ContentExtensions::GraphBuilder::assertionWordBoundary):
(WebCore::ContentExtensions::GraphBuilder::atomCharacterClassAtom):
(WebCore::ContentExtensions::GraphBuilder::atomCharacterClassRange):
(WebCore::ContentExtensions::GraphBuilder::atomCharacterClassBuiltIn):
(WebCore::ContentExtensions::GraphBuilder::atomParentheticalAssertionBegin):
(WebCore::ContentExtensions::GraphBuilder::disjunction):
(WebCore::ContentExtensions::GraphBuilder::hasError):
(WebCore::ContentExtensions::GraphBuilder::fail):
(WebCore::ContentExtensions::GraphBuilder::sinkFloatingTermIfNecessary):
(WebCore::ContentExtensions::URLFilterParser::addPattern):
(WebCore::ContentExtensions::URLFilterParser::statusString):
(WebCore::ContentExtensions::GraphBuilder::errorMessage): Deleted.
* contentextensions/URLFilterParser.h:
Use an enum instead of strings for the status to avoid checking strings when we have a regex that matches everything.

Tools:

* TestWebKitAPI/Tests/WebCore/ContentExtensions.cpp:
(TestWebKitAPI::testPattern):
(TestWebKitAPI::TEST_F):
Start testing regex failures.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181726 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoFix build failure due to FALLTHROUGH in unreachable code
utatane.tea@gmail.com [Thu, 19 Mar 2015 04:05:51 +0000 (04:05 +0000)]
Fix build failure due to FALLTHROUGH in unreachable code
https://bugs.webkit.org/show_bug.cgi?id=142703

Reviewed by Benjamin Poulain.

FALLTHROUGH should not exist in unreachable code marked by
ASSERT_NOT_REACHABLE.

When !ENABLE(CSS_SELECTOR_JIT) and Debug mode is enabled,
both ASSERT_NOT_REACHED and FALLTHROUGH are instantiated
and it cause build failure.

Here, since CompiledSingle/CompiledSingleWithRootFilter are
not set when !ENABLE(CSS_SELECTOR_JIT). So dropping FALLTHROUGH
and leave ASSERT_NOT_REACHED.
And since ASSERT_NOT_ReACHED is stripped when Release build,
I've inserted FALLTHROUGH() if ASSERT_DISABLED.
This fix is the same to r162906.

* dom/SelectorQuery.cpp:
(WebCore::SelectorDataList::execute):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181725 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoES6 Classes: Extends should accept an expression without parenthesis
joepeck@webkit.org [Thu, 19 Mar 2015 03:27:00 +0000 (03:27 +0000)]
ES6 Classes: Extends should accept an expression without parenthesis
https://bugs.webkit.org/show_bug.cgi?id=142840

Reviewed by Ryosuke Niwa.

Source/JavaScriptCore:

* parser/Parser.cpp:
(JSC::Parser<LexerType>::parseClass):
"extends" allows a LeftHandExpression (new expression / call expression,
which includes a member expression), not a primary expression. Our
parseMemberExpression does all of these.

LayoutTests:

* js/class-syntax-extends-expected.txt:
* js/script-tests/class-syntax-extends.js:
Extend the test to test cases of extends with different types of expressions.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181724 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoWeb Inspector: Replace last use of ObjectPropertiesSection with ObjectTreeView
joepeck@webkit.org [Thu, 19 Mar 2015 03:26:58 +0000 (03:26 +0000)]
Web Inspector: Replace last use of ObjectPropertiesSection with ObjectTreeView
https://bugs.webkit.org/show_bug.cgi?id=142834

Reviewed by Timothy Hatcher.

* UserInterface/Views/DOMNodeDetailsSidebarPanel.js:
(WebInspector.DOMNodeDetailsSidebarPanel.prototype._refreshProperties):
Use properties only ObjectTreeView instead of ObjectPropertiesSection.
This doesn't fix the functionality issues, but fixes the appearance.

* UserInterface/Views/DetailsSection.css:
(.details-section > .content > .group:first-child > .row.simple:first-child > *): Deleted.
(body.mac-platform.legacy .details-section > .content > .group:first-child > .row.simple:first-child > *): Deleted.
(.details-section > .content > .group > .row.properties:not(.empty)): Deleted.
(body.mac-platform.legacy .details-section > .content > .group > .row.properties:not(.empty)): Deleted.
Re-add back the small padding at the top of sections. This actually
broke padding in a few sections (Event Listeners) at the expense
of eliminating a few pixels of whitespace at the top of other sections.
We should focus on addressing the extra whitespace separately.

* UserInterface/Main.html:
* UserInterface/Views/ObjectPropertiesSection.js: Removed.
* UserInterface/Views/TypePropertiesSection.js:
* WebInspectorUI.vcxproj/WebInspectorUI.vcxproj:
* WebInspectorUI.vcxproj/WebInspectorUI.vcxproj.filters:
Remove new unused class.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181723 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoWeb Inspector: Debugger Popovers and Probes should use FormattedValue/ObjectTreeView...
joepeck@webkit.org [Thu, 19 Mar 2015 03:26:55 +0000 (03:26 +0000)]
Web Inspector: Debugger Popovers and Probes should use FormattedValue/ObjectTreeView instead of Custom/ObjectPropertiesSection
https://bugs.webkit.org/show_bug.cgi?id=142830

Reviewed by Timothy Hatcher.

Source/JavaScriptCore:

* inspector/agents/InspectorDebuggerAgent.cpp:
(Inspector::InspectorDebuggerAgent::breakpointActionProbe):
Give Probe Samples object previews.

Source/WebInspectorUI:

* UserInterface/Views/ProbeSetDataGrid.css:
(.details-section.probe-set .data-grid .object-tree > :matches(.title, .object-preview)::before):
Another line-height fix for object tree disclosure triangles.

* UserInterface/Views/ProbeSetDataGridNode.js:
(WebInspector.ProbeSetDataGridNode.prototype.createCellContent):
Create an ObjectTree / FormattedValue for the RemoteObject.

* UserInterface/Views/SourceCodeTextEditor.css:
(.popover .debugger-popover-content > .title):
(.popover .debugger-popover-content > .body):
(.popover .debugger-popover-content.function > .body):
Be more specific and don't accidentally style ".title" within the body.

* UserInterface/Views/SourceCodeTextEditor.js:
(WebInspector.SourceCodeTextEditor.prototype._showPopoverForObject):
Show a properties only ObjectTree instead of an ObjectPropertiesSection.

(WebInspector.SourceCodeTextEditor.prototype._showPopoverForString): Deleted.
(WebInspector.SourceCodeTextEditor.prototype._showPopoverForRegExp): Deleted.
(WebInspector.SourceCodeTextEditor.prototype._showPopoverForNumber): Deleted.
(WebInspector.SourceCodeTextEditor.prototype._showPopoverForBoolean): Deleted.
(WebInspector.SourceCodeTextEditor.prototype._showPopoverForNull): Deleted.
(WebInspector.SourceCodeTextEditor.prototype._showPopoverForUndefined): Deleted.
(WebInspector.SourceCodeTextEditor.prototype._showPopoverWithFormattedValue):
Reduce most of these to a single popover for formatted values.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181722 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoDon't paint PDFs on every scroll
simon.fraser@apple.com [Thu, 19 Mar 2015 02:16:12 +0000 (02:16 +0000)]
Don't paint PDFs on every scroll
https://bugs.webkit.org/show_bug.cgi?id=142857
rdar://problem/20130207

Reviewed by Tim Horton.

Don't push a scale factor onto the PDF layer controller if it hasn't
changed, since doing so causes a repaint.

* WebProcess/Plugins/PDF/PDFPlugin.mm:
(WebKit::PDFPlugin::updatePageAndDeviceScaleFactors):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181721 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoSwitching between two SVG images with no intrinsic sizes causes them to get the defau...
commit-queue@webkit.org [Thu, 19 Mar 2015 02:15:00 +0000 (02:15 +0000)]
Switching between two SVG images with no intrinsic sizes causes them to get the default SVG size instead of the container size.
https://bugs.webkit.org/show_bug.cgi?id=142805.

Patch by Said Abou-Hallawa <sabouhallawa@apple.com> on 2015-03-18
Reviewed by Darin Adler.
Source/WebCore:

The bug happens due to wrong logic in RenderImage::imageDimensionsChanged().
This function decides to setNeedsLayout() if the intrinsic size of the image
changes. If the size does not change, it only repaints the image rectangle.
When switching the src of the an image between two SVG images and both of
them have no intrinsic size, we do not updateInnerContentRect() and this
means an SVGImageForContainer is not going to be created for this image.
When the image is drawn, it is drawn directly from the SVGImage. And this
means the drawing has to be scaled by container_size / SVG_default_intrinsic_size

After figuring out that I need to updateInnerContentRect() to fix this bug,
I found out Blink has already changed this code to do the same thing. But
they also did more clean-up in this function. Here is the link
https://codereview.chromium.org/114323004. I think their change seems correct
although they did not say what exactly they were trying to fix.

The plan for repaintOrMarkForLayout(), which is the new name of this function,
is the following:
    -- setNeedLayout() if the intrinsic size changes and it affects the size
       of the image.
    -- updateInnerContentRect() if the intrinsic size did not change but the
       image has exiting layout.
    -- repaint the image rectangle if layout is not needed.

This change also removes the call to computeLogicalWidthInRegion(), which is
almost running a layout for the image. This call figures out whether the image
needs to setNeedsLayout(). This call is unnecessary; the image needs to run a
layout if the intrinsic size has changed and it affects the size of the image.

Test: svg/as-image/svg-no-intrinsic-size-switching.html

* rendering/RenderImage.cpp:
(WebCore::RenderImage::styleDidChange): Change the function call.
(WebCore::RenderImage::imageChanged): Rename local variable and change the
function call.

(WebCore::RenderImage::updateIntrinsicSizeIfNeeded): Simplify this function.
Call setIntrinsicSize() with the new size unless the image is in error state.

(WebCore::RenderImage::repaintOrMarkForLayout): This a better name for this
function since it is called even if the intrinsic size was not changed.
(WebCore::RenderImage::imageDimensionsChanged): Deleted.

* rendering/RenderImage.h: Rename imageDimensionsChanged() and change the
updateIntrinsicSizeIfNeeded() to return void.

* rendering/svg/RenderSVGForeignObject.cpp:
(WebCore::RenderSVGForeignObject::paint): Code cleanup. This function can
only handle the paint phases PaintPhaseForeground and PaintPhaseSelection.
Use this information to simplify the logic and order of painting there.

LayoutTests:

* svg/as-image/svg-no-intrinsic-size-switching-expected.html: Added.
* svg/as-image/svg-no-intrinsic-size-switching.html: Added.
Ensure that switching the source of an <img> element between two SVG images,
which have no intrinsic sizes, gets the image the size of the container and
not the default SVG intrinsic size which is 300x150 pixels.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181720 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoSource/WTF:
diorahman@rockybars.com [Thu, 19 Mar 2015 01:10:24 +0000 (01:10 +0000)]
Source/WTF:
Fix StringView misplaced implementations after r181525 and r181558
https://bugs.webkit.org/show_bug.cgi?id=142772

Reviewed by Darin Adler.

Fix misplaced implementations in StringView::startsWithIgnoringASCIICase and
StringView::endsWith.

Add some tests in [1] to cover the StringView::startsWith,
StringView::startsWithIgnoringASCIICase, StringView::endsWith and
StringView::endsWithIgnoringASCIICase implementations.

[1] Tools/TestWebKitAPI/Tests/WTF/StringView.cpp.

* wtf/text/StringView.cpp:
(WTF::StringView::startsWithIgnoringASCIICase):
(WTF::StringView::endsWith):
(WTF::StringView::endsWithIgnoringASCIICase):

Tools:
Fix StringView typos after r181525 and r181558
https://bugs.webkit.org/show_bug.cgi?id=142772

Reviewed by Darin Adler.

Add some tests in [1] to cover the StringView::startsWith,
StringView::startsWithIgnoringASCIICase, StringView::endsWith and
StringView::endsWithIgnoringASCIICase implementations.

* TestWebKitAPI/Tests/WTF/StringView.cpp:
(TestWebKitAPI::TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181717 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoRemove extraneous import to fix the build.
dburkart@apple.com [Thu, 19 Mar 2015 00:14:08 +0000 (00:14 +0000)]
Remove extraneous import to fix the build.

Unreviewed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181715 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoASAN_OPTIONS=allocator_may_return_null=1 needs to be set
dburkart@apple.com [Thu, 19 Mar 2015 00:01:42 +0000 (00:01 +0000)]
ASAN_OPTIONS=allocator_may_return_null=1 needs to be set
https://bugs.webkit.org/show_bug.cgi?id=142547

Reviewed by Alexey Proskuryakov.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181714 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoFix typo in playerViewControllerWillCancelOptimizedFullscree.
jeremyj-wk@apple.com [Wed, 18 Mar 2015 23:35:06 +0000 (23:35 +0000)]
Fix typo in playerViewControllerWillCancelOptimizedFullscree.
https://bugs.webkit.org/show_bug.cgi?id=142745

Reviewed by Darin Adler.

* platform/ios/WebVideoFullscreenInterfaceAVKit.mm:
(-[WebAVPlayerController playerViewControllerWillCancelOptimizedFullscreen:]):
(-[WebAVPlayerController playerViewControllerWillCancelOptimizedFullscree:]): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181713 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agowebaudio/convolution-mono-mono.html fails on some machines
ap@apple.com [Wed, 18 Mar 2015 23:29:34 +0000 (23:29 +0000)]
webaudio/convolution-mono-mono.html fails on some machines
rdar://problem/19996807

Rubber-stampted by Jer Noble.

* webaudio/resources/convolution-testing.js: (checkTriangularPulse): Slightly increase
the tolerance. It's still extremely low.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181712 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUse && instead of & when deciding if the network cache should be cleared
andersca@apple.com [Wed, 18 Mar 2015 22:39:46 +0000 (22:39 +0000)]
Use && instead of & when deciding if the network cache should be cleared
https://bugs.webkit.org/show_bug.cgi?id=142845

Reviewed by Antti Koivisto.

* NetworkProcess/NetworkProcess.cpp:
(WebKit::NetworkProcess::deleteWebsiteData):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181711 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoAvoid repaints when changing transform on an element with multiple background images
simon.fraser@apple.com [Wed, 18 Mar 2015 22:17:34 +0000 (22:17 +0000)]
Avoid repaints when changing transform on an element with multiple background images
https://bugs.webkit.org/show_bug.cgi?id=142841

Reviewed by Zalan Bujtas.

Source/WebCore:

Replace the cheap test for changed images in RenderElement::updateFillImages()
with an exhaustive test that walks the entire list of background images,
since any ensuing repaint is way more expensive than a slightly more expensive check here.

Test: fast/repaint/multiple-backgrounds-style-change.html

* rendering/RenderElement.cpp:
(WebCore::RenderElement::updateFillImages):
* rendering/style/FillLayer.cpp:
(WebCore::layerImagesIdentical): See if both images are the same (either none
or both mask images, and same image pointer).
(WebCore::FillLayer::imagesIdentical): Walk the two FillLayer lists, checking the images
on each one. Returns false if we reach the end of one list before the other, or the images
are different.
* rendering/style/FillLayer.h: New static function; static because
it compares two FillLayer lists, and I think that makes more sense than
a member function.

LayoutTests:

Test that changes transform on a composited element with 2 background images,
and tests for no repaints.

* fast/repaint/multiple-backgrounds-style-change-expected.txt: Added.
* fast/repaint/multiple-backgrounds-style-change.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181710 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoPass cookies by reference in CookieHash functions
andersca@apple.com [Wed, 18 Mar 2015 21:06:15 +0000 (21:06 +0000)]
Pass cookies by reference in CookieHash functions
https://bugs.webkit.org/show_bug.cgi?id=142839

Reviewed by Sam Weinig.

* platform/Cookie.h:
(WebCore::CookieHash::hash):
(WebCore::CookieHash::equal):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181709 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoTweak how AppleSystemFontOSSubversion default is added
ap@apple.com [Wed, 18 Mar 2015 20:42:11 +0000 (20:42 +0000)]
Tweak how AppleSystemFontOSSubversion default is added
https://bugs.webkit.org/show_bug.cgi?id=142835

Reviewed by Dan Bernstein.

* DumpRenderTree/mac/DumpRenderTree.mm:
(setDefaultsToConsistentValuesForTesting):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181708 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoAdd WKContextGetWebsiteDataStore API
andersca@apple.com [Wed, 18 Mar 2015 19:54:40 +0000 (19:54 +0000)]
Add WKContextGetWebsiteDataStore API
https://bugs.webkit.org/show_bug.cgi?id=142832
rdar://problem/16544715

Reviewed by Sam Weinig.

Also add WKWebsiteDataStoreRef which is toll-free bridged to _WKWebsiteDataStore.

* Shared/API/c/WKBase.h:
Add new declaration.

* UIProcess/API/APIWebsiteDataStore.cpp:
(API::WebsiteDataStore::create):
* UIProcess/API/APIWebsiteDataStore.h:
Add a new function that takes a configuration.

* UIProcess/API/C/WKAPICast.h:
Add cast.

* UIProcess/API/C/WKContext.cpp:
(WKContextGetWebsiteDataStore):
Return the context's website data store.

* UIProcess/API/C/WKContext.h:
Add new declaration.

* UIProcess/API/C/WKWebsiteDataStoreRef.cpp: Added.
* UIProcess/API/C/WKWebsiteDataStoreRef.h: Added.
Add new files.

* UIProcess/WebKeyValueStorageManager.cpp:
(WebKit::WebKeyValueStorageManager::getKeyValueStorageOrigins):
(WebKit::WebKeyValueStorageManager::getStorageDetailsByOrigin):
(WebKit::WebKeyValueStorageManager::deleteEntriesForOrigin):
(WebKit::WebKeyValueStorageManager::deleteAllEntries):
Update now that WebProcessPool holds on to an API::WebsiteDataStore.

* UIProcess/WebProcessPool.cpp:
(WebKit::WebProcessPool::WebProcessPool):
(WebKit::WebProcessPool::createWebPage):
* UIProcess/WebProcessPool.h:
WebProcessPool should hold on to an API::WebsiteDataStore instead of a WebKit::WebsiteDataStore.

* WebKit2.xcodeproj/project.pbxproj:
Add new files.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181707 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoTemporarily fix the !ENABLE(CSS_SELECTOR_JIT) and assertions-enabled build
timothy_horton@apple.com [Wed, 18 Mar 2015 19:07:30 +0000 (19:07 +0000)]
Temporarily fix the !ENABLE(CSS_SELECTOR_JIT) and assertions-enabled build

* dom/SelectorQuery.cpp:
(WebCore::SelectorDataList::execute):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181706 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoCannot invoke action menus anymore
timothy_horton@apple.com [Wed, 18 Mar 2015 18:40:48 +0000 (18:40 +0000)]
Cannot invoke action menus anymore
https://bugs.webkit.org/show_bug.cgi?id=142797
<rdar://problem/20032670>

* UIProcess/API/mac/WKView.mm:
Need one more respondsToSelector check.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181705 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoWeb Inspector: Scopes sidebar should use new ObjectTreeView and not ObjectPropertiesS...
joepeck@webkit.org [Wed, 18 Mar 2015 18:30:47 +0000 (18:30 +0000)]
Web Inspector: Scopes sidebar should use new ObjectTreeView and not ObjectPropertiesSection
https://bugs.webkit.org/show_bug.cgi?id=142808

Reviewed by Timothy Hatcher.

* UserInterface/Models/PropertyPath.js:
(WebInspector.PropertyPath):
(WebInspector.PropertyPath.emptyPropertyPathForScope):
Allow a special property empty path for "Scopes". This way for a
"<scopeObject>.property" we can show just the tooltip "property".

* UserInterface/Views/ObjectTreeView.css:
(.object-tree.properties-only > :matches(.title, .object-preview)):
(.object-tree.properties-only .object-tree-outline):
(.object-tree.properties-only .object-tree-property .property-name):
Tweak styles for only properties view, which won't have a top-level
preview and doesn't fade out enumerable properties.

* UserInterface/Views/ObjectTreeView.js:
(WebInspector.ObjectTreeView.prototype.get treeOutline):
Access the TreeOutline.

(WebInspector.ObjectTreeView.prototype.showOnlyProperties):
Properties only view modifies the display slightly.

(WebInspector.ObjectTreeView.prototype.appendExtraPropertyDescriptor):
(WebInspector.ObjectTreeView.prototype._updateProperties):
Allow the client to add its own property descriptors to display
as a property in this ObjectTreeView.

* UserInterface/Views/ScopeChainDetailsSidebarPanel.js:
(WebInspector.ScopeChainDetailsSidebarPanel.prototype.refresh):
Switch to using an ObjectTreeView.

(WebInspector.ScopeChainDetailsSidebarPanel.prototype._propertyPathIdentifierForTreeElement):
(WebInspector.ScopeChainDetailsSidebarPanel.prototype._objectTreeAddHandler):
(WebInspector.ScopeChainDetailsSidebarPanel.prototype._objectTreeExpandHandler):
(WebInspector.ScopeChainDetailsSidebarPanel.prototype._objectTreeCollapseHandler):
Keep track of what properties were expanded so we can auto-expand
them again when the sidebar refreshes.

* UserInterface/Main.html:
* UserInterface/Views/ScopeVariableTreeElement.js: Removed.
* WebInspectorUI.vcxproj/WebInspectorUI.vcxproj:
* WebInspectorUI.vcxproj/WebInspectorUI.vcxproj.filters:
Remove the now unused ScopeVariableTreeElement.js.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@181704 268f45cc-cd09-0410-ab3c-d52691b4dbfc