Timeline



May 26, 2014:

11:14 PM Changeset in webkit [169370] by timothy_horton@apple.com
  • 15 edits in trunk/Source

[wk2] RemoteLayerBackingStore front buffers should be purgeable when unparented
https://bugs.webkit.org/show_bug.cgi?id=133020
<rdar://problem/16521736>

Reviewed by Simon Fraser.

  • Shared/mac/RemoteLayerBackingStore.mm:

(WebKit::RemoteLayerBackingStore::display):
Let the context know whenever a RemoteLayerBackingStore is displayed, so that
RemoteLayerBackingStoreCollection can (if needed) note that the backing store
is active once again (because we only display parented backing store).

(WebKit::RemoteLayerBackingStore::setBufferVolatility):
Ensure that we never have live contexts attached to any buffers when
marking them volatile, because checking isInUse() with live contexts is futile.

  • Shared/mac/RemoteLayerBackingStoreCollection.h:
  • Shared/mac/RemoteLayerBackingStoreCollection.mm:

(WebKit::RemoteLayerBackingStoreCollection::RemoteLayerBackingStoreCollection):
(WebKit::RemoteLayerBackingStoreCollection::willFlushLayers):
(WebKit::RemoteLayerBackingStoreCollection::willCommitLayerTree):
(WebKit::RemoteLayerBackingStoreCollection::didFlushLayers):
(WebKit::RemoteLayerBackingStoreCollection::backingStoreWillBeDestroyed):
(WebKit::RemoteLayerBackingStoreCollection::backingStoreWillBeDisplayed):
(WebKit::RemoteLayerBackingStoreCollection::markBackingStoreVolatileImmediately):
(WebKit::RemoteLayerBackingStoreCollection::markBackingStoreVolatile):
(WebKit::RemoteLayerBackingStoreCollection::backingStoreBecameUnreachable):
(WebKit::RemoteLayerBackingStoreCollection::volatilityTimerFired):
(WebKit::RemoteLayerBackingStoreCollection::scheduleVolatilityTimer):
(WebKit::RemoteLayerBackingStoreCollection::purgeabilityTimerFired): Deleted.
(WebKit::RemoteLayerBackingStoreCollection::schedulePurgeabilityTimer): Deleted.
Rename purgeable->volatile for accuracy.
Keep track of two sets of backing store: those which are active/parented, and
those which are not. Backing store is moved to the inactive set after building
the transaction in which its owning layer is unparented.
When backing store is unparented, try to mark it volatile immediately. Also,
mark the backing store property as dirty on the owning layer so that when
said layer is reparented, we encode the backing store in the commit that reparents it,
as the UI process will throw away its reference to the backing store when
the layer is unparented. Mark the front buffers of unparented layers as volatile,
in addition to the others.

  • Shared/mac/RemoteLayerTreeTransaction.h:

(WebKit::RemoteLayerTreeTransaction::layerIDsWithNewlyUnreachableBackingStore):

  • Shared/mac/RemoteLayerTreeTransaction.mm:

(WebKit::RemoteLayerTreeTransaction::encode):
(WebKit::RemoteLayerTreeTransaction::decode):
(WebKit::RemoteLayerTreeTransaction::setLayerIDsWithNewlyUnreachableBackingStore):
Include the list of layers (by ID) with backing store which just became unreachable in the transaction.

  • UIProcess/mac/RemoteLayerTreeHost.mm:

(WebKit::RemoteLayerTreeHost::updateLayerTree):
Clear the contents of layers which now have unreachable backing store.
Otherwise, the UI process would hold a 'use' on the IOSurface, and prevent
the Web process from marking it volatile.

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

(WebKit::RemoteLayerTreeContext::backingStoreWillBeDisplayed):

  • WebProcess/WebPage/mac/RemoteLayerTreeDrawingArea.mm:

(WebKit::RemoteLayerTreeDrawingArea::flushLayers):
Give RemoteLayerBackingStoreCollection a shot at the RemoteLayerTreeTransaction,
so that it can fill in layerIDsWithNewlyUnreachableBackingStore.
Also, let it know when the flush begins and ends, so that it can keep track
of which layers were reached in the flush.

  • WebCore.exp.in:
  • platform/graphics/cg/IOSurfacePool.cpp:

(WebCore::IOSurfacePool::willAddSurface):

  • platform/graphics/cocoa/IOSurface.h:
  • platform/graphics/cocoa/IOSurface.mm:

(IOSurface::releaseGraphicsContext):
Rename clearGraphicsContext to releaseGraphicsContext for clarity.

10:38 PM Changeset in webkit [169369] by gyuyoung.kim@samsung.com
  • 2 edits in trunk/Tools

Fix test failures on python test
https://bugs.webkit.org/show_bug.cgi?id=133297

Reviewed by Dirk Schulze.

2 tests are failing because of unbounded local variable reference in test_importer.py.

  • Scripts/webkitpy/w3c/test_importer.py: Move local variable declaration to top.

(TestImporter.import_tests):

10:23 PM Changeset in webkit [169368] by pdr@google.com
  • 4 edits in trunk/Source/WebCore

Remove special case for transparent SVG root layers
https://bugs.webkit.org/show_bug.cgi?id=116856

Reviewed by Dirk Schulze.

This patch removes a special case from RenderLayer for root SVG layers
with opacity. Instead of checking whether a composited layer exists in
SVGRenderingContext::prepareToRenderSVGContent we now exclude the root
SVG renderobject from the opacity checks in prepareToRenderSVGContent.

This is a merge of https://src.chromium.org/viewvc/blink?view=rev&revision=174102
by Philip Rogers <pdr@chromium.org>.

No new tests as this is covered by svg/custom/composited-svg-with-opacity.html.

  • rendering/RenderLayer.cpp:

(WebCore::RenderLayer::isTransparent): Deleted.

  • rendering/RenderLayer.h:
  • rendering/svg/SVGRenderingContext.cpp:

(WebCore::SVGRenderingContext::prepareToRenderSVGContent):

The opacity case for root svg renderers is now handled by
RenderLayer.

9:52 PM Changeset in webkit [169367] by commit-queue@webkit.org
  • 2 edits in trunk/Source/WebKit2

[EFL][WK2] Fix EWK2BackForwardListTest test fails.
https://bugs.webkit.org/show_bug.cgi?id=133171

Patch by Shivakumar JM <shiva.jm@samsung.com> on 2014-05-26
Reviewed by Gyuyoung Kim.

Use local test html pages instead of EWK2UnitTestServer, call ewk_view_url_set() to load test html pages.
Since even after setting no_proxy="localhost,127.0.0.1", still test server was unable to load the test
html pages for ewk2_back_forward_list tests, so changed to use local html pages.

  • UIProcess/API/efl/tests/test_ewk2_back_forward_list.cpp:

(TEST_F):

9:32 PM Changeset in webkit [169366] by commit-queue@webkit.org
  • 3 edits in trunk/Source/WebKit2

[EFL][WK2] Fix code style check errors in EFL API tests.
https://bugs.webkit.org/show_bug.cgi?id=133170

Patch by Shivakumar JM <shiva.jm@samsung.com> on 2014-05-26
Reviewed by Gyuyoung Kim.

Omit int when using unsigned modifier, as per coding style rules.

  • UIProcess/API/efl/tests/UnitTestUtils/EWK2UnitTestBase.cpp:

(EWK2UnitTest::EWK2UnitTestBase::SetUp):

  • UIProcess/API/efl/tests/UnitTestUtils/EWK2UnitTestEnvironment.h:

(EWK2UnitTest::EWK2UnitTestEnvironment::defaultWidth):
(EWK2UnitTest::EWK2UnitTestEnvironment::defaultHeight):

7:23 PM Changeset in webkit [169365] by gyuyoung.kim@samsung.com
  • 2 edits in trunk/LayoutTests

Fix wrong inclusion depth of js-test-pre.js in is-protocol-handler-registered.html
https://bugs.webkit.org/show_bug.cgi?id=133273

Reviewed by Darin Adler.

Can't use debug() due to wrong inclusion path.

  • fast/dom/NavigatorContentUtils/is-protocol-handler-registered.html:
7:22 PM Changeset in webkit [169364] by gyuyoung.kim@samsung.com
  • 1 edit
    2 moves
    1 add in trunk/LayoutTests

[EFL] Move navigator content utils's to own directory
https://bugs.webkit.org/show_bug.cgi?id=133275

Reviewed by Darin Adler.

Expected results need to be maintained by own directory because navigator content utils have been maintained by
own directory as well.

  • platform/efl/fast/dom/NavigatorContentUtils/register-protocol-handler-expected.txt: Renamed from LayoutTests/platform/efl/fast/dom/register-protocol-handler-expected.txt.
  • platform/efl/fast/dom/NavigatorContentUtils/unregister-protocol-handler-expected.txt: Renamed from LayoutTests/platform/efl/fast/dom/unregister-protocol-handler-expected.txt.
5:19 PM Changeset in webkit [169363] by commit-queue@webkit.org
  • 2 edits in trunk/Source/WebCore

CSS JIT: Fix build error raised when CSS_SELECTOR_JIT_DEBUGGING is 1
https://bugs.webkit.org/show_bug.cgi?id=133266

Patch by Yusuke Suzuki <Yusuke Suzuki> on 2014-05-26
Reviewed by Benjamin Poulain.

When CSS_SELECTOR_JIT_DEBUGGING is 1, build erros occur on x86_64.

  • cssjit/SelectorCompiler.cpp:

(WebCore::SelectorCompiler::SelectorCodeGenerator::generatePrologue):
(WebCore::SelectorCompiler::SelectorCodeGenerator::generateEpilogue):

5:02 PM Changeset in webkit [169362] by barraclough@apple.com
  • 4 edits in trunk/Source/WebKit2

Add UI process watchdog on iOS to ensure WebProcess connections close
https://bugs.webkit.org/show_bug.cgi?id=133200
<rdar://problem/16997983>

Reviewed by Darin Adler.

When the WebProcessProxy wants to disconnect from a WebContent process it just drops the connection,
and hopes the connection closes. There is a watchdog thread in the ChildProcess to try to ensure this
happens.

On iOS the process may not be runnable at the time, preventing termination. Instead add a watchdog in
the UI process to make the process runnable, and to terminate if it doesn't quit in a timely fashion.

  • Platform/IPC/Connection.h:
    • added terminateSoon.
  • Platform/IPC/mac/ConnectionMac.mm:

(IPC::ConnectionTerminationWatchdog::ConnectionTerminationWatchdog):

  • take an assertion to make the process runnable, and start a watchdog timer.

(IPC::ConnectionTerminationWatchdog::watchdogTimerFired):

  • if the process hasn't quit by the timer the watchdog fires, kill it.

(IPC::Connection::terminateSoon):

  • create a ConnectionTerminationWatchdog.
  • UIProcess/WebProcessProxy.cpp:

(WebKit::WebProcessProxy::removeWebPage):

  • when disconnecting from a process, first tell it to terminateSoon.
4:42 PM Changeset in webkit [169361] by commit-queue@webkit.org
  • 2 edits in trunk/Tools

[GTK] GObject DOM bindings API break test fails in the tests bots
https://bugs.webkit.org/show_bug.cgi?id=133211

Patch by Carlos Alberto Lopez Perez <clopez@igalia.com> on 2014-05-26
Reviewed by Martin Robinson.

When generating the built product archive on platform GTK, we need
to copy also the file gtkdoc-webkitdom.cfg and the directory
DerivedSources/webkitdom.

  • BuildSlaveSupport/built-product-archive:

(archiveBuiltProduct):

2:45 PM Changeset in webkit [169360] by benjamin@webkit.org
  • 4 edits
    15 adds in trunk

Fix the quirks mode selector matching of the pseudo classes :hover and :active
https://bugs.webkit.org/show_bug.cgi?id=133063

Reviewed by Antti Koivisto.

Source/WebCore:
Our implementation of the quirks mode of :active and :hover was very wrong. The only
thing it was doing is verify the pseudo class is not the first selector of a fragment
(which was conveniently the only thing that was tested :)).

Since those pseudo class were only checking for the order of the filters, something like

#target:hover

would succeed because :hover is not the first simple selector, while

:hover#target

would fail.

That behavior is also a problem for the CSS JIT as it is an implementation detail of SelectorChecker
and compiling something like that with our out-of-order matching would be nonsense.

This patch update the implementation to follow http://quirks.spec.whatwg.org/#the-:active-and-:hover-quirk
Basically, the only cases that do not work in quirks mode are selectors composed only of "*, :hover and :active".

To implement this behavior, I needed to be able to inspect a complete selector fragment, including
what is before and after :hover/:active.
To do that, I replaced the boolean isSubSelector by a pointer to the first selector of the fragment.
When we need to match :active/:hover in quirks mode, we just go over all the selectors in the fragment
to find one of the qualifying match type.

Tests: fast/selectors/active-hover-quirks.html

fast/selectors/active-quirks.html
fast/selectors/hover-quirks.html

  • css/SelectorChecker.cpp:

(WebCore::SelectorChecker::matchRecursively):
(WebCore::canMatchHoverOrActiveInQuirksMode):
(WebCore::SelectorChecker::checkOne):

  • css/SelectorChecker.h:

(WebCore::SelectorChecker::SelectorCheckingContext::SelectorCheckingContext):

LayoutTests:
The test coverage of :hover and :active was extremly poor.
Those new tests add coverage for the cases fixed by this patch.

  • fast/selectors/active-hover-quirks-expected.txt: Added.
  • fast/selectors/active-hover-quirks.html: Added.
  • fast/selectors/active-hover-strict-expected.txt: Added.
  • fast/selectors/active-hover-strict.html: Added.
  • fast/selectors/active-quirks-expected.txt: Added.
  • fast/selectors/active-quirks.html: Added.
  • fast/selectors/active-strict-expected.txt: Added.
  • fast/selectors/active-strict.html: Added.
  • fast/selectors/hover-quirks-expected.txt: Added.
  • fast/selectors/hover-quirks.html: Added.
  • fast/selectors/hover-strict-expected.txt: Added.
  • fast/selectors/hover-strict.html: Added.
  • fast/selectors/resources/hover-active-quirks-utility.js: Added.
  • selectors/resources/hover-active-strict-utility.js: Added.
1:18 PM Changeset in webkit [169359] by Alan Bujtas
  • 2 edits in trunk/Source/WebCore

Split the call to adjustForLocalZoom out into a separate expression.
https://bugs.webkit.org/show_bug.cgi?id=133286

Reviewed by Darin Adler.

This patch ensures that adjustForLocalZoom() always gets called before we look at
zoomFactor's value. (and not rely on the undefined behavior of the order of function argument evaluation)

  • dom/Element.cpp:

(WebCore::Element::offsetLeft):
(WebCore::Element::offsetTop):

12:58 PM Changeset in webkit [169358] by Darin Adler
  • 11 edits
    4 adds in trunk

Class name matching should use ASCII case-insensitive matching, not Unicode case folding
https://bugs.webkit.org/show_bug.cgi?id=133292

Reviewed by Anders Carlsson.

Source/WebCore:
Tests: fast/dom/getElementsByClassName/ASCII-case-insensitive.html

fast/dom/getElementsByClassName/case-sensitive.html

  • dom/SpaceSplitString.cpp:

(WebCore::hasNonASCIIOrUpper): Deleted.
(WebCore::tokenizeSpaceSplitString): Use a for loop instead of while.
(WebCore::spaceSplitStringTable): Renamed from sharedDataMap; the new name is supposed
to help us see the analogy with the atomic string table.
(WebCore::SpaceSplitString::set): Removed unneeded special case for null and preflight
since AtomicString::convertToASCIILowercase now handles both of those. Changed to call
convertToASCIILowercase instead of foldCase, since we don't want to fold non-ASCII.
(WebCore::SpaceSplitString::spaceSplitStringContainsValue): Ditto.
(WebCore::SpaceSplitStringData::create): Marked this inline since it's only called in
one place and that place is in this file. Also used auto a bit and used get instead of
find since the value type is a simple pointer.
(WebCore::SpaceSplitStringData::destroy): Removed unneeded check for null. We never
create any SpaceSplitStringData with empty strings, and a null is a kind of empty string.

  • dom/SpaceSplitString.h: Removed some unneeded includes and some unneeded uses of the

inline keyword. Changed types from size_t to unsigned in a couple places; we had a mix
of the types and there was no reason to use size_t there.

Source/WTF:

  • wtf/text/AtomicString.cpp:

(WTF::AtomicString::addSlowCase): Change to take references instead of pointers since these
arguments can never be null.
(WTF::AtomicString::lower): Rearranged slightly to use PassRef in a more efficient but
slightly uglier way.
(WTF::AtomicString::convertToASCIILowercase): Added.

  • wtf/text/AtomicString.h: Updated for above changes.
  • wtf/text/StringImpl.cpp:

(WTF::StringImpl::convertToASCIILowercase): Added.

  • wtf/text/StringImpl.h: Updated for above.
  • wtf/text/WTFString.cpp:

(WTF::String::convertToASCIILowercase): Added.

  • wtf/text/WTFString.h: Updated for above.

LayoutTests:

  • fast/dom/getElementsByClassName/ASCII-case-insensitive-expected.txt: Added.
  • fast/dom/getElementsByClassName/ASCII-case-insensitive.html: Added.
  • fast/dom/getElementsByClassName/case-sensitive-expected.txt: Added.
  • fast/dom/getElementsByClassName/case-sensitive.html: Added.
11:01 AM Changeset in webkit [169357] by weinig@apple.com
  • 9 edits in trunk/Source/WebKit2

[WebKit2] Add better default preferences while keeping backward compatibility for the C-SPI
https://bugs.webkit.org/show_bug.cgi?id=133289

Reviewed by Anders Carlsson.

  • Shared/WebPreferencesStore.h:

Change a few defaults:

  • PluginsEnabled -> false
  • JavaEnabled -> false
  • JavaEnabledForLocalFiles -> false
  • StorageBlockingPolicy -> BlockThirdPartyStorage
  • UIProcess/API/C/WKPreferences.cpp:

(WKPreferencesCreate):
(WKPreferencesCreateWithIdentifier):

  • UIProcess/WebPageGroup.cpp:

(WebKit::WebPageGroup::WebPageGroup):
Switch to using WebPreferences::createWithLegacyDefaults().

  • UIProcess/WebPreferences.cpp:

(WebKit::WebPreferences::create):
(WebKit::WebPreferences::createWithLegacyDefaults):

  • UIProcess/WebPreferences.h:

Move create functions to the cpp file and add createWithLegacyDefaults() which keeps
the old defaults;

  • UIProcess/API/C/WKPreferencesRef.h:

Fix the comment.

  • UIProcess/API/Cocoa/WKPreferences.h:

Update the comment to reflect the new defaults.

  • WebProcess/WebPage/WebPage.cpp:

(WebKit::WebPage::updatePreferences):
Unconditionally set setShouldDispatchJavaScriptWindowOnErrorEvents() to true. This setting only
exists for legacy WebKit.

11:00 AM Changeset in webkit [169356] by mitz@apple.com
  • 6 edits in trunk/Source/WebKit2

[Cocoa] Rewrite WK_DESIGNATED_INITIALIZER in installed headers
https://bugs.webkit.org/show_bug.cgi?id=133290

Reviewed by Anders Carlsson.

  • Shared/API/Cocoa/WKFoundation.h: When WK_API_AVAILABILITY_ENABLED is defined, don’t

provide definitions of WK_DESIGNATED_INITIALIZER and WK_UNAVAILABLE, since those macros will
be rewritten out of the headers. When WK_API_AVAILABILITY_ENABLED isn’t defined, provide
definitions of the other two macros as appropriate for the target OS.

  • UIProcess/API/Cocoa/WKProcessPoolPrivate.h: Removed WK_DESIGNATED_INITIALIZER from a

category method declaration.

  • UIProcess/API/Cocoa/WKWebView.h: Mark the superclass designated initializer

-initWithCoder: as unavailable.

  • UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView initWithCoder:]): Override this designated initializer of the superclass by
returning nil.

  • mac/rewrite-availability-macros.sh: Rewrite WK_DESIGNATED_INITIALIZER to

NS_DESIGNATED_INITIALIZER and WK_UNAVAILABLE to NS_UNAVAILABLE. Also make sure to process
all headers if the script has changed since the last time it was run.

10:45 AM Changeset in webkit [169355] by commit-queue@webkit.org
  • 6 edits in trunk

Unreviewed, rolling out r169122.
https://bugs.webkit.org/show_bug.cgi?id=133288

Broke CJK font fallback (Requested by ap on #webkit).

Reverted changeset:

"[Mac] WebProcess doesn't follow localization of UI process
when run as a service"
https://bugs.webkit.org/show_bug.cgi?id=133126
http://trac.webkit.org/changeset/169122

10:43 AM Changeset in webkit [169354] by fpizlo@apple.com
  • 25 edits
    53 adds in trunk

Latest emscripten life benchmark is 4x slower because the DFG doesn't realize that arithmetic on booleans is a thing
https://bugs.webkit.org/show_bug.cgi?id=133136

Reviewed by Oliver Hunt.

Source/JavaScriptCore:

Some key concepts:

  • Except for the prediction propagation and type fixup phases, which are super early in the pipeline, nobody has to know about the fact that booleans may flow into numerical operations because there will just be a BooleanToNumber node that will take a value and, if that value is a boolean, will convert it to the equivalent numerical value. It will have a BooleanUse mode where it will also speculate that the input is a boolean but it can also do UntypedUse in which case it will pass through any non-booleans. This operation is very easy to model in all of the compiler tiers.
  • No changes to the baseline JIT. The Baseline JIT will still believe that boolean inputs require taking the slow path and it will still report that it took slow path for any such operations. The DFG will now be smart enough to ignore baseline JIT slow path profiling on operations that were known to have had boolean inputs. That's a little quirky, but it's probably easier than modifying the baseline JIT to track booleans correctly.


4.1x speed-up on the emscripten "life" benchmark. Up to 10x speed-up on microbenchmarks.

  • bytecode/SpeculatedType.h:

(JSC::isInt32OrBooleanSpeculation):
(JSC::isInt32SpeculationForArithmetic):
(JSC::isInt32OrBooleanSpeculationForArithmetic):
(JSC::isInt32OrBooleanSpeculationExpectingDefined):
(JSC::isInt52Speculation):
(JSC::isMachineIntSpeculation):
(JSC::isFullNumberOrBooleanSpeculation):
(JSC::isFullNumberOrBooleanSpeculationExpectingDefined):
(JSC::isInt32SpeculationExpectingDefined): Deleted.
(JSC::isMachineIntSpeculationExpectingDefined): Deleted.
(JSC::isMachineIntSpeculationForArithmetic): Deleted.
(JSC::isBytecodeNumberSpeculationExpectingDefined): Deleted.
(JSC::isFullNumberSpeculationExpectingDefined): Deleted.

  • dfg/DFGAbstractInterpreterInlines.h:

(JSC::DFG::AbstractInterpreter<AbstractStateType>::executeEffects):

  • dfg/DFGAllocator.h:

(JSC::DFG::Allocator<T>::indexOf):

  • dfg/DFGByteCodeParser.cpp:

(JSC::DFG::ByteCodeParser::makeSafe):
(JSC::DFG::ByteCodeParser::makeDivSafe):
(JSC::DFG::ByteCodeParser::handleIntrinsic):

  • dfg/DFGCSEPhase.cpp:

(JSC::DFG::CSEPhase::performNodeCSE):

  • dfg/DFGClobberize.h:

(JSC::DFG::clobberize):

  • dfg/DFGCommon.h:
  • dfg/DFGConstantFoldingPhase.cpp:

(JSC::DFG::ConstantFoldingPhase::foldConstants):

  • dfg/DFGFixupPhase.cpp:

(JSC::DFG::FixupPhase::fixupNode):
(JSC::DFG::FixupPhase::fixIntConvertingEdge):
(JSC::DFG::FixupPhase::fixIntOrBooleanEdge):
(JSC::DFG::FixupPhase::fixDoubleOrBooleanEdge):
(JSC::DFG::FixupPhase::attemptToMakeIntegerAdd):
(JSC::DFG::FixupPhase::fixIntEdge): Deleted.

  • dfg/DFGGraph.h:

(JSC::DFG::Graph::addSpeculationMode):
(JSC::DFG::Graph::valueAddSpeculationMode):
(JSC::DFG::Graph::arithAddSpeculationMode):
(JSC::DFG::Graph::addShouldSpeculateInt32):
(JSC::DFG::Graph::mulShouldSpeculateInt32):
(JSC::DFG::Graph::mulShouldSpeculateMachineInt):
(JSC::DFG::Graph::negateShouldSpeculateInt32):
(JSC::DFG::Graph::negateShouldSpeculateMachineInt):
(JSC::DFG::Graph::addImmediateShouldSpeculateInt32):
(JSC::DFG::Graph::mulImmediateShouldSpeculateInt32): Deleted.

  • dfg/DFGNode.h:

(JSC::DFG::Node::sawBooleans):
(JSC::DFG::Node::shouldSpeculateInt32OrBoolean):
(JSC::DFG::Node::shouldSpeculateInt32ForArithmetic):
(JSC::DFG::Node::shouldSpeculateInt32OrBooleanForArithmetic):
(JSC::DFG::Node::shouldSpeculateInt32OrBooleanExpectingDefined):
(JSC::DFG::Node::shouldSpeculateMachineInt):
(JSC::DFG::Node::shouldSpeculateDouble):
(JSC::DFG::Node::shouldSpeculateNumberOrBoolean):
(JSC::DFG::Node::shouldSpeculateNumberOrBooleanExpectingDefined):
(JSC::DFG::Node::shouldSpeculateNumber):
(JSC::DFG::Node::canSpeculateInt32):
(JSC::DFG::Node::canSpeculateInt52):
(JSC::DFG::Node::sourceFor):
(JSC::DFG::Node::shouldSpeculateInt32ExpectingDefined): Deleted.
(JSC::DFG::Node::shouldSpeculateMachineIntForArithmetic): Deleted.
(JSC::DFG::Node::shouldSpeculateMachineIntExpectingDefined): Deleted.
(JSC::DFG::Node::shouldSpeculateDoubleForArithmetic): Deleted.
(JSC::DFG::Node::shouldSpeculateNumberExpectingDefined): Deleted.

  • dfg/DFGNodeFlags.cpp:

(JSC::DFG::dumpNodeFlags):

  • dfg/DFGNodeFlags.h:

(JSC::DFG::nodeMayOverflow):
(JSC::DFG::nodeMayNegZero):
(JSC::DFG::nodeCanSpeculateInt32):
(JSC::DFG::nodeCanSpeculateInt52):

  • dfg/DFGNodeType.h:
  • dfg/DFGPredictionPropagationPhase.cpp:

(JSC::DFG::PredictionPropagationPhase::run):
(JSC::DFG::PredictionPropagationPhase::propagateToFixpoint):
(JSC::DFG::PredictionPropagationPhase::speculatedDoubleTypeForPrediction):
(JSC::DFG::PredictionPropagationPhase::propagate):
(JSC::DFG::PredictionPropagationPhase::doDoubleVoting):

  • dfg/DFGSafeToExecute.h:

(JSC::DFG::safeToExecute):

  • dfg/DFGSpeculativeJIT.cpp:

(JSC::DFG::SpeculativeJIT::compileValueToInt32):

  • dfg/DFGSpeculativeJIT32_64.cpp:

(JSC::DFG::SpeculativeJIT::compile):

  • dfg/DFGSpeculativeJIT64.cpp:

(JSC::DFG::SpeculativeJIT::compile):

  • ftl/FTLCapabilities.cpp:

(JSC::FTL::canCompile):

  • ftl/FTLLowerDFGToLLVM.cpp:

(JSC::FTL::LowerDFGToLLVM::compileNode):
(JSC::FTL::LowerDFGToLLVM::compileValueToInt32):
(JSC::FTL::LowerDFGToLLVM::compileBooleanToNumber):

  • runtime/JSCJSValue.h:
  • runtime/JSCJSValueInlines.h:

(JSC::JSValue::asInt32ForArithmetic):

  • tests/stress/max-boolean-exit.js: Added.

(foo):
(test):

  • tests/stress/mul-boolean-exit.js: Added.

(foo):
(test):

  • tests/stress/plus-boolean-exit.js: Added.

(foo):
(test):

  • tests/stress/plus-boolean-or-double.js: Added.

(foo):
(test):

  • tests/stress/plus-boolean-or-int.js: Added.

(foo):
(test):

LayoutTests:

  • js/regress/abs-boolean-expected.txt: Added.
  • js/regress/abs-boolean.html: Added.
  • js/regress/div-boolean-double-expected.txt: Added.
  • js/regress/div-boolean-double.html: Added.
  • js/regress/div-boolean-expected.txt: Added.
  • js/regress/div-boolean.html: Added.
  • js/regress/max-boolean-expected.txt: Added.
  • js/regress/max-boolean.html: Added.
  • js/regress/min-boolean-expected.txt: Added.
  • js/regress/min-boolean.html: Added.
  • js/regress/minus-boolean-double-expected.txt: Added.
  • js/regress/minus-boolean-double.html: Added.
  • js/regress/minus-boolean-expected.txt: Added.
  • js/regress/minus-boolean.html: Added.
  • js/regress/mod-boolean-double-expected.txt: Added.
  • js/regress/mod-boolean-double.html: Added.
  • js/regress/mod-boolean-expected.txt: Added.
  • js/regress/mod-boolean.html: Added.
  • js/regress/mul-boolean-double-expected.txt: Added.
  • js/regress/mul-boolean-double.html: Added.
  • js/regress/mul-boolean-expected.txt: Added.
  • js/regress/mul-boolean.html: Added.
  • js/regress/neg-boolean-expected.txt: Added.
  • js/regress/neg-boolean.html: Added.
  • js/regress/plus-boolean-arith-expected.txt: Added.
  • js/regress/plus-boolean-arith.html: Added.
  • js/regress/plus-boolean-double-expected.txt: Added.
  • js/regress/plus-boolean-double.html: Added.
  • js/regress/plus-boolean-expected.txt: Added.
  • js/regress/plus-boolean.html: Added.
  • js/regress/script-tests/abs-boolean.js: Added.
  • js/regress/script-tests/div-boolean-double.js: Added.
  • js/regress/script-tests/div-boolean.js: Added.
  • js/regress/script-tests/max-boolean.js: Added.
  • js/regress/script-tests/min-boolean.js: Added.
  • js/regress/script-tests/minus-boolean-double.js: Added.
  • js/regress/script-tests/minus-boolean.js: Added.
  • js/regress/script-tests/mod-boolean-double.js: Added.
  • js/regress/script-tests/mod-boolean.js: Added.
  • js/regress/script-tests/mul-boolean-double.js: Added.
  • js/regress/script-tests/mul-boolean.js: Added.
  • js/regress/script-tests/neg-boolean.js: Added.
  • js/regress/script-tests/plus-boolean-arith.js: Added.
  • js/regress/script-tests/plus-boolean-double.js: Added.
  • js/regress/script-tests/plus-boolean.js: Added.
  • js/regress/script-tests/sin-boolean.js: Added.
  • js/regress/sin-boolean-expected.txt: Added.
  • js/regress/sin-boolean.html: Added.
10:30 AM Changeset in webkit [169353] by andersca@apple.com
  • 3 edits in trunk/Source/WebKit2

Convert a couple of loops over to range-based for
https://bugs.webkit.org/show_bug.cgi?id=133287

Reviewed by Simon Fraser.

  • UIProcess/Plugins/PluginInfoStore.cpp:

(WebKit::PluginInfoStore::findPluginForMIMEType):
(WebKit::PluginInfoStore::findPluginForExtension):
(WebKit::PluginInfoStore::infoForPluginWithPath):

  • UIProcess/Plugins/mac/PluginProcessManagerMac.mm:

(WebKit::PluginProcessManager::setProcessSuppressionEnabled):

9:55 AM Changeset in webkit [169352] by Carlos Garcia Campos
  • 6 edits in trunk

[GTK] WebProcess leaked when closing pages with network process enabled
https://bugs.webkit.org/show_bug.cgi?id=129684

Reviewed by Anders Carlsson.

Source/WebKit2:
The problem is that the web process is not notified when the UI
process closes the connection, because when close() is called on
the socket by the UI process, the socket is shared by another web
process launched later, preventing the connection from being
shut down. We need to set the CLOEXEC flag on the sockets file
descriptor to make sure they are not exposed to other processes.

  • Platform/IPC/Connection.h: Add ConnectionOptions parameter to

createPlatformConnection() with a default value compatible with
existing callers.

  • Platform/IPC/unix/ConnectionUnix.cpp:

(IPC::Connection::createPlatformConnection): Set the CLOEXEC flag
on the client and server socket file descriptors depending on the
options passed.

  • UIProcess/Launcher/gtk/ProcessLauncherGtk.cpp:

(WebKit::ProcessLauncher::launchProcess): Use
IPC::Connection::createPlatformConnection() instead of
socketpair() directly, setting the CLOEXEC flag on the server
before spawning the new process and on the client right after
spawning the new process.

Tools:
Enable the test to check that web processes finish when the web
view is destroyed.

  • TestWebKitAPI/Tests/WebKit2Gtk/TestMultiprocess.cpp:
9:40 AM Changeset in webkit [169351] by andersca@apple.com
  • 2 edits in trunk/Source/WebKit2

Initialize WKWebViewConfiguration ivars lazily
https://bugs.webkit.org/show_bug.cgi?id=133270
<rdar://problem/17027606>

Reviewed by Sam Weinig.

Create a LazyInitialized class template and use it to lazily initialize WKWebViewConfiguration properties.

  • UIProcess/API/Cocoa/WKWebViewConfiguration.mm:

(LazyInitialized::get):
Initialize m_value with the given function if it hasn't been initialized already.

(LazyInitialized::set):
Update m_value and set m_isInitialized to true.

(LazyInitialized::peek):
Return m_value without initializing it.

(-[WKWebViewConfiguration description]):
Use getters so we'll initialize variables if needed.

(-[WKWebViewConfiguration copyWithZone:]):
Ditto.

(-[WKWebViewConfiguration processPool]):
(-[WKWebViewConfiguration setProcessPool:]):
(-[WKWebViewConfiguration preferences]):
(-[WKWebViewConfiguration setPreferences:]):
(-[WKWebViewConfiguration userContentController]):
(-[WKWebViewConfiguration setUserContentController:]):
(-[WKWebViewConfiguration _visitedLinkProvider]):
(-[WKWebViewConfiguration _setVisitedLinkProvider:]):
(-[WKWebViewConfiguration _websiteDataStore]):
(-[WKWebViewConfiguration _setWebsiteDataStore:]):
(-[WKWebViewConfiguration _contentProviderRegistry]):
(-[WKWebViewConfiguration _setContentProviderRegistry:]):
Pass initialization code to all getters.

(-[WKWebViewConfiguration _validate]):
Use getters.

(-[WKWebViewConfiguration init]): Deleted.

9:37 AM Changeset in webkit [169350] by Carlos Garcia Campos
  • 2 edits in trunk/Tools

REGRESSION(r164632): [GTK] Crash in generate-gtkdoc when rebasing docs
https://bugs.webkit.org/show_bug.cgi?id=133279

Reviewed by Philippe Normand.

generator.saw_warnings is only set by GTKDoc::generate(), so it
should only be used after generate.

  • gtk/generate-gtkdoc:

(generate_documentation_for_config): Do not check
generator.saw_warnings again, simply return the value returned by
generate_doc() which is generator.saw_warnings. When rebasing,
returns always False since there aren't warnings.

9:25 AM Changeset in webkit [169349] by commit-queue@webkit.org
  • 9 edits
    3 adds in trunk

[CSS Grid Layout] Implementation of the "grid" shorthand.
https://bugs.webkit.org/show_bug.cgi?id=132122

Patch by Javier Fernandez <jfernandez@igalia.com> on 2014-05-26
Reviewed by Darin Adler.

Source/WebCore:
The grid property is a shorthand that sets all of the explicit
grid properties (grid-template-rows, grid-template-columns, and
grid-template-areas) as well as all the implicit grid properties
(grid-auto-rows, grid-auto-columns, and grid-auto-flow) in a
single declaration

Notice that either explicit or implicit grid can be specified,
assigning the initial values to the omitted properties.

Test: fast/css-grid-layout/grid-shorthand-get-set.html

  • css/CSSComputedStyleDeclaration.cpp:

(WebCore::ComputedStyleExtractor::propertyValue):

  • css/CSSParser.cpp:

(WebCore::CSSParser::parseValue):
(WebCore::CSSParser::parseGridShorthand):

  • css/CSSParser.h:
  • css/CSSPropertyNames.in:
  • css/StylePropertyShorthand.cpp:

(WebCore::webkitGridShorthand):

  • css/StylePropertyShorthand.h:

LayoutTests:
Layout Test for the basic functionality of the 'grid' shorthand. It was
also added a new javascript file with some utility functions.

  • fast/css-grid-layout/grid-shorthand-get-set-expected.txt: Added.
  • fast/css-grid-layout/grid-shorthand-get-set.html: Added.
  • fast/css-grid-layout/resources/grid-shorthand-parsing-utils.js: Added.

(testGridDefinitionsValues):
(testGridDefinitionsSetJSValues):
(testNonGridDefinitionsSetJSValues):
(checkGridDefinitionsSetJSValues):
(testGridDefinitionsSetBadJSValues):

9:11 AM Changeset in webkit [169348] by commit-queue@webkit.org
  • 2 edits in trunk/Source/JavaScriptCore

Remove dead code from VM.cpp
https://bugs.webkit.org/show_bug.cgi?id=133284

Patch by Zsolt Borbely <zsborbely.u-szeged@partner.samsung.com> on 2014-05-26
Reviewed by Darin Adler.

This workaround was added in r127505. Since the clang is the
only used compiler in this case, this workaround is obsolete.

  • runtime/VM.cpp:

(JSC::enableAssembler):

9:11 AM Changeset in webkit [169347] by commit-queue@webkit.org
  • 2 edits in trunk/Tools

W3C test importer should catch exceptions coming from HTML/CSS conversion
https://bugs.webkit.org/show_bug.cgi?id=133176

Patch by Youenn Fablet <youenn.fablet@crf.canon.fr> on 2014-05-26
Reviewed by Darin Adler.

Exceptions raised when converting HTML/CSS are caught.
Files for which conversion failed are still copied to the target directory.
Each individual conversion failure is logged.
Total conversion failure number is logged at the end of the import process.

  • Scripts/webkitpy/w3c/test_importer.py:

(TestImporter.import_tests):

9:08 AM Changeset in webkit [169346] by Alan Bujtas
  • 2 edits in trunk/Source/WebCore

Subpixel layout: Legacy Element.offset* client* return values are invalid in certain cases.
https://bugs.webkit.org/show_bug.cgi?id=133272

Reviewed by Simon Fraser.

Element.offset* client* functions applied various rounding strategies on the return values before r168868.
(for example, offsetLeft/Top either floored through implicit integer arithmetics or rounded explicitly depending
whether the zoom scale was 1. see http://trac.webkit.org/changeset/168868/trunk/Source/WebCore/dom/Element.cpp for details)
This patch mimics the legacy behavior by either rounding or flooring the return value.

Not testable (legacy switch is not exposed)

  • dom/Element.cpp:

(WebCore::adjustForLocalZoom):
(WebCore::convertToNonSubpixelValueIfNeeded):
(WebCore::Element::offsetLeft):
(WebCore::Element::offsetTop):

8:04 AM Changeset in webkit [169345] by Michał Pakuła vel Rutka
  • 3 edits in trunk/LayoutTests

Unreviewed EFL gardening

Remove expectations for now passing tests.

  • platform/efl-wk2/TestExpectations:
  • platform/efl/TestExpectations:
7:19 AM Changeset in webkit [169344] by commit-queue@webkit.org
  • 4 edits
    1 add in trunk/Tools

[GTK] Allow to run the tests on the native X display.
https://bugs.webkit.org/show_bug.cgi?id=133157

Patch by Carlos Alberto Lopez Perez <clopez@igalia.com> on 2014-05-26
Reviewed by Benjamin Poulain.

This adds a new driver (xorgdriver) that runs the tests on the
X display referenced by the environment variable DISPLAY.

This new driver will be activated only if the environment variable
USE_NATIVE_XDISPLAY is defined. This can be used both for layout
tests and performance tests.

This patch also makes the script run-perf-tests to check the
system dependencies before starting the tests. Previously this
was not checked, and if the system dependencies were not met,
the script tried to execute the tests anyway, causing massive
failures. For example, if you had Xvfb not installed and you
wanted to use the Xvfb driver (the default on GTK and EFL),
run-perf-tests would not abort.

  • Scripts/webkitpy/performance_tests/perftestsrunner.py:

(PerfTestsRunner._parse_args): Check the system dependencies of
the driver before starting the tests.

  • Scripts/webkitpy/port/driver.py:

(Driver): Implement generic check_driver method.
(Driver.check_driver):
(Driver.check_driver.implementation):

  • Scripts/webkitpy/port/gtk.py:

(GtkPort._driver_class): Check for environment variable
USE_NATIVE_XDISPLAY to decide if the Xorg driver should be used.

  • Scripts/webkitpy/port/xorgdriver.py: Added.

(XorgDriver): Implement Xorg driver.
(XorgDriver.check_driver):
(XorgDriver._start):

7:16 AM Changeset in webkit [169343] by commit-queue@webkit.org
  • 2 edits in trunk/Tools

REGRESSION(r166798): [GTK] [EFL] kill-old-processes should not kill dbus-daemon
https://bugs.webkit.org/show_bug.cgi?id=133215

Patch by Carlos Alberto Lopez Perez <clopez@igalia.com> on 2014-05-26
Reviewed by Csaba Osztrogonác.

This reverts r166798.

  • BuildSlaveSupport/kill-old-processes:

(main):

7:04 AM Changeset in webkit [169342] by Michał Pakuła vel Rutka
  • 2 edits in trunk/Tools

[EFL] Turn on ENABLE_CSS_FILTERS in build script
https://bugs.webkit.org/show_bug.cgi?id=133278

Reviewed by Gyuyoung Kim.

  • Scripts/webkitperl/FeatureList.pm: Add a build script condition missing in r169172.
6:57 AM Changeset in webkit [169341] by commit-queue@webkit.org
  • 2 edits in trunk/Source/JavaScriptCore

JSC CLoop warning fix
https://bugs.webkit.org/show_bug.cgi?id=133259

Patch by Eva Balazsfalvi <evab.u-szeged@partner.samsung.com> on 2014-05-26
Reviewed by Darin Adler.

  • llint/LLIntSlowPaths.cpp:

(JSC::LLInt::LLINT_SLOW_PATH_DECL):

6:55 AM Changeset in webkit [169340] by commit-queue@webkit.org
  • 2 edits in trunk/Tools

run-javascriptcore-tests warning fix on non Mac platforms
https://bugs.webkit.org/show_bug.cgi?id=133260

Patch by Eva Balazsfalvi <evab.u-szeged@partner.samsung.com> on 2014-05-26
Reviewed by Darin Adler.

  • Scripts/webkitdirs.pm:

(argumentsForConfiguration):

5:06 AM Changeset in webkit [169339] by commit-queue@webkit.org
  • 2 edits in trunk/Source/WebKit2

Remove WebKit2 framework from generate-forwarding-headers.pl
https://bugs.webkit.org/show_bug.cgi?id=133214

Since renaming WebKit2.framework to WebKit.framework is done and all WebKit2/*.h includes
were renamed to WebKit/*.h, generating forwarding headers for WebKit2/*.h framework style
headers is not necessary anymore.

Patch by Martin Hodovan <mhodovan.u-szeged@partner.samsung.com> on 2014-05-26
Reviewed by Darin Adler.

  • Scripts/generate-forwarding-headers.pl:
3:09 AM Changeset in webkit [169338] by Carlos Garcia Campos
  • 1 copy in releases/WebKitGTK/webkit-2.4.3

Tagging the WebKitGTK+ 2.4.3 release

3:01 AM Changeset in webkit [169337] by Carlos Garcia Campos
  • 4 edits in releases/WebKitGTK/webkit-2.4

Unreviewed. Update NEWS and Versions.m4 for 2.4.3 release.

.:

  • Source/autotools/Versions.m4: Bump version numbers.

Source/WebKit/gtk:

  • NEWS: Added release notes for 2.4.3.
3:00 AM Changeset in webkit [169336] by Carlos Garcia Campos
  • 3 edits in releases/WebKitGTK/webkit-2.4/Tools

Unreviewed. Fix make distcheck.

  • gtk/GNUmakefile.am: Remove generate-webkitdom-doc-files from

EXTRA_DIST and add webkitdom.py.

  • gtk/generate-gtkdoc:

(generate_documentation_for_config): Do not try to use
generator.saw_warnings when rebasing, it's only set when
generating the docs.

2:50 AM Changeset in webkit [169335] by Michał Pakuła vel Rutka
  • 5 edits in trunk

[EFL] Input fields and text areas are not rendered correctly after r167771
https://bugs.webkit.org/show_bug.cgi?id=133181

Reviewed by Gyuyoung Kim.

Source/WebCore:
r167771 changed arguments of RenderTheme two virtual method arguments which were not reflected
in changes in derived classes, which caused parent method to be called. Arguments in derived class
were updated, also 'override' specifiers were added to all virtual methods to prevent similar errors.

Already covered by fast/forms/textarea-placeholder-wrapping.html

  • platform/efl/RenderThemeEfl.cpp:

(WebCore::RenderThemeEfl::paintTextField):
(WebCore::RenderThemeEfl::paintTextArea):

  • platform/efl/RenderThemeEfl.h:

(WebCore::RenderThemeEfl::supportsHover):
(WebCore::RenderThemeEfl::supportsControlTints):

LayoutTests:

  • platform/efl-wk2/TestExpectations: Remove now passing test from test expectations.
2:48 AM Changeset in webkit [169334] by svillar@igalia.com
  • 9 edits
    2 adds in trunk

[CSS Grid Layout] Children of grid containers must be turned into grid items
https://bugs.webkit.org/show_bug.cgi?id=132991

Reviewed by Darin Adler.

From Blink r150472 by <jchaffraix@chromium.org>

Source/WebCore:
According to specs each child of a grid container must become a
grid item meaning that grid items are grid level boxes and thus,
do not participate in the block formatting context but in the grid
formatting one.

This change updates the grid items' 'display' property after style
resolution so that we match the specification (see section 4. Grid
Items). The spec basically instructs us to compute the value of
'display' by applying the table in CSS2.1 Chapter 9.7
http://www.w3.org/TR/CSS2/visuren.html#dis-pos-flo

Test: fast/css-grid-layout/grid-item-display.html

  • css/StyleResolver.cpp:

(WebCore::isDisplayGridBox):
(WebCore::StyleResolver::adjustRenderStyle):

LayoutTests:
Added a new test case to verify the computed value for 'display'
in grid items. The test also checks that items are correctly
positioned inside the grid, i.e., they properly became grid
items.

Also removed ietestcenter's grid-items-002.html and
grid-items-003.html from the list of failing tests as they should
be working now.

  • fast/css-grid-layout/grid-item-display-expected.txt: Added.
  • fast/css-grid-layout/grid-item-display.html: Added.
  • ietestcenter/css3/grid/grid-items-002.htm: Fixed a typo and

replaced -webkit-grid-definition-* by -webkit-grid-template-*

  • ietestcenter/css3/grid/grid-items-003.htm: Ditto.
  • platform/efl/TestExpectations: Removed the two tests above.
  • platform/gtk/TestExpectations: Ditto.
  • platform/mac/TestExpectations: Ditto.
  • platform/win/TestExpectations: Ditto.
2:26 AM Changeset in webkit [169333] by zandobersek@gmail.com
  • 2 edits in trunk/Source/WebCore

Remove Vector copies in ShorthandPropertyWrapper implementation and use
https://bugs.webkit.org/show_bug.cgi?id=133265

Reviewed by Simon Fraser.

  • page/animation/CSSPropertyAnimation.cpp:

(WebCore::ShorthandPropertyWrapper::ShorthandPropertyWrapper): Move the passed-in Vector
into the member variable instead of using Vector<>::swap().
(WebCore::ShorthandPropertyWrapper::propertyWrappers): Return a const reference of the member
variable instead of a const value.
(WebCore::CSSPropertyAnimationWrapperMap::CSSPropertyAnimationWrapperMap): Move the Vector
object into the ShorthandProperthyWrapper constructor.
(WebCore::gatherEnclosingShorthandProperties): Deploy a range-based for-loop.

2:22 AM Changeset in webkit [169332] by zandobersek@gmail.com
  • 5 edits in trunk/Source/WebKit2

WebPage::setComposition(), WebPageProxy::didFindStringMatches() should take Vector parameters by const reference
https://bugs.webkit.org/show_bug.cgi?id=133264

Reviewed by Darin Adler.

Taking the Vector parameters in the two methods (invoked through the IPC message handling)
by value causes unnecessary copies. The methods don't modify the objects and don't want
or need fresh copies, hence they should take in const references to Vector objects.

  • UIProcess/WebPageProxy.cpp:

(WebKit::WebPageProxy::didFindStringMatches):

  • UIProcess/WebPageProxy.h:
  • WebProcess/WebPage/WebPage.cpp:

(WebKit::WebPage::setComposition):

  • WebProcess/WebPage/WebPage.h:
12:45 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
(diff)
12:44 AM Changeset in webkit [169331] by Carlos Garcia Campos
  • 5 edits in releases/WebKitGTK/webkit-2.4/Source

[Stable] [GTK] GdiObject.h missing in WebKitGTK 2.4.1 tarball
https://bugs.webkit.org/show_bug.cgi?id=132814

Patch by Milan Crha <mcrha@redhat.com> on 2014-05-25
Reviewed by Carlos Garcia Campos.

Source/WebCore:

Add missing files and includes to fix the mingw32 build.

  • GNUmakefile.list.am:
  • platform/graphics/win/GraphicsContextCairoWin.cpp:

Source/WTF:

Add missing files to fix the mingw32 build.

  • GNUmakefile.list.am:
12:31 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
(diff)
12:30 AM Changeset in webkit [169330] by Carlos Garcia Campos
  • 4 edits
    2 adds in releases/WebKitGTK/webkit-2.4

Merge r169200 - Video is resumed with old playback rate.
https://bugs.webkit.org/show_bug.cgi?id=132905

Patch by Piotr Grad <p.grad@samsung.com> on 2014-05-22
Reviewed by Philippe Normand.

Source/WebCore:
Setting '0' playback rate is causing pipeline to pause.
GStreamer player impl. exposed this information to upper layers but it should not.
Solution is to hidden such situation behind m_playbackRatePause flag.

Test: media/video-paused-0-rate.html

  • platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.cpp:

(WebCore::MediaPlayerPrivateGStreamer::MediaPlayerPrivateGStreamer):
(WebCore::MediaPlayerPrivateGStreamer::play):
(WebCore::MediaPlayerPrivateGStreamer::pause):
(WebCore::MediaPlayerPrivateGStreamer::doSeek):
(WebCore::MediaPlayerPrivateGStreamer::updatePlaybackRate):
(WebCore::MediaPlayerPrivateGStreamer::paused):
(WebCore::MediaPlayerPrivateGStreamer::setRate):
(WebCore::MediaPlayerPrivateGStreamer::updateStates):

  • platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.h:

LayoutTests:

  • media/video-paused-0-rate-expected.txt: Added.
  • media/video-paused-0-rate.html: Added.
12:15 AM Changeset in webkit [169329] by commit-queue@webkit.org
  • 2 edits in trunk/Source/WebCore

[EFL] Fix build error in blob.cpp after r168435
https://bugs.webkit.org/show_bug.cgi?id=132678

Patch by Tanay C <tanay.c@samsung.com> on 2014-05-26
Reviewed by Alexey Proskuryakov.

  • fileapi/Blob.cpp:

(WebCore::Blob::size): Using isInBound to check range of local var
actualsize to resolve build error

May 25, 2014:

11:17 PM Changeset in webkit [169328] by ryuan.choi@samsung.com
  • 8 edits
    2 deletes in trunk/Source/WebCore

[EFL] Remove TileCairo and TiledBackingStoreBackendCairo
https://bugs.webkit.org/show_bug.cgi?id=133274

Reviewed by Gyuyoung Kim.

Now, TiledBackingStore is only used for CoordinatedGraphics and the EFL port.
CoordinatedGraphics does not use TileCairo and TiledBackingStoreBackendCairo.

This patch removed them and refactored related files.

  • PlatformEfl.cmake:
  • PlatformGTK.cmake:
  • WebCore.vcxproj/WebCore.vcxproj:
  • WebCore.vcxproj/WebCore.vcxproj.filters:
  • platform/graphics/TiledBackingStore.cpp:

(WebCore::TiledBackingStore::TiledBackingStore): Deleted.

  • platform/graphics/TiledBackingStore.h:
  • platform/graphics/TiledBackingStoreBackend.h:
  • platform/graphics/cairo/TileCairo.cpp: Removed.
  • platform/graphics/cairo/TiledBackingStoreBackendCairo.cpp: Removed.
7:53 PM Changeset in webkit [169327] by jinwoo7.song@samsung.com
  • 11 edits
    2 adds in trunk

setData() of DataTransfer has a void return type
https://bugs.webkit.org/show_bug.cgi?id=133108

Reviewed by Alexey Proskuryakov.

Source/WebCore:
According to HTML5 spec, setData() of DataTranfer interface does not return value.
http://www.w3.org/TR/html/editing.html#the-datatransfer-interface

Test: editing/pasteboard/set_data_typeof_return.html

  • dom/DataTransfer.cpp:

(WebCore::DataTransfer::setData):

  • dom/DataTransfer.h:
  • dom/DataTransfer.idl:
  • platform/Pasteboard.h:
  • platform/efl/PasteboardEfl.cpp:

(WebCore::Pasteboard::writeString):

  • platform/gtk/PasteboardGtk.cpp:

(WebCore::Pasteboard::writeString):

  • platform/ios/PasteboardIOS.mm:

(WebCore::Pasteboard::writeString):

  • platform/mac/PasteboardMac.mm:

(WebCore::Pasteboard::writeString):

  • platform/win/PasteboardWin.cpp:

(WebCore::Pasteboard::writeString):

LayoutTests:

  • editing/pasteboard/set_data_typeof_return-expected.txt: Added.
  • editing/pasteboard/set_data_typeof_return.html: Added.
5:04 PM Changeset in webkit [169326] by benjamin@webkit.org
  • 4 edits in trunk/Source/WebKit2

[iOS][WK2] Fix some state reset on crash on the WKWebView
https://bugs.webkit.org/show_bug.cgi?id=133039

Reviewed by Sam Weinig.

This is in no way complete, but this should reduce the undefined states on crash.

  • UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView _processDidExit]):
If the view is animating, nuke the resize animation. That should be invisible to the user
because the background color is reset to white below.

Reset the contentView frame and scrollview state.

Reset the runtime states. We do not strictly need to reset _needsResetViewStateAfterCommitLoadForMainFrame,
_delayUpdateVisibleContentRects and _hadDelayedUpdateVisibleContentRects but it seems better to have
a clean slate.

(-[WKWebView _beginAnimatedResizeWithUpdates:]):
(-[WKWebView _endAnimatedResize]):
We could have crashes during rotation. To simplify the reset code, the animated resize code
no longer change any state when WKWebView is using a custom content view.

  • UIProcess/API/Cocoa/WKWebViewInternal.h:
  • UIProcess/ios/PageClientImplIOS.mm:

(WebKit::PageClientImpl::processDidExit):
Send a _processDidExit on the WKWebView too.

3:30 PM Changeset in webkit [169325] by andersca@apple.com
  • 4 edits in trunk/Source/WebKit2

Use the right paths for website data
https://bugs.webkit.org/show_bug.cgi?id=133267
<rdar://problem/17027698>

Reviewed by Sam Weinig.

  • UIProcess/API/Cocoa/WKProcessPool.mm:

(-[WKProcessPool _initWithConfiguration:]):
Get the path of the absolute URL.

  • UIProcess/Storage/LocalStorageDatabaseTracker.cpp:

(WebKit::LocalStorageDatabaseTracker::databasePath):
Fix spelling error.

  • WebProcess/Network/NetworkProcessConnection.cpp:

(WebKit::NetworkProcessConnection::didCacheResource):
Ditto.

3:17 PM Changeset in webkit [169324] by benjamin@webkit.org
  • 6 edits in trunk/Source

[iOS][WK2] Use ScrollView's scrollOffset as the unobscuredContentRect
https://bugs.webkit.org/show_bug.cgi?id=133262

Patch by Benjamin Poulain <bpoulain@apple.com> on 2014-05-25
Reviewed by Simon Fraser.

Source/WebCore:
Since VisibleContentRect was fixed, we were no longer sending scroll events when updating the scrollOffset
when updating the visible content rects. The reason is that the scrollOffset was defined as the top left of the
VisibleContentRect, and as such was already at the end position after updating the unobscured rect.

This patch split the unobscuredContentRect in unobscuredContentSize (updated live on zoom) and the position defined
by the ScrollView's scrollOffset (updated when scrolling).

  • WebCore.exp.in:
  • platform/ScrollView.h:
  • platform/ios/ScrollViewIOS.mm:

(WebCore::ScrollView::unobscuredContentRect):
(WebCore::ScrollView::setUnobscuredContentSize):
(WebCore::ScrollView::setUnobscuredContentRect): Deleted.

Source/WebKit2:

  • WebProcess/WebPage/ios/WebPageIOS.mm:

(WebKit::WebPage::dynamicViewportSizeUpdate):
(WebKit::WebPage::viewportConfigurationChanged):
(WebKit::WebPage::updateVisibleContentRects):

12:54 PM Changeset in webkit [169323] by weinig@apple.com
  • 2 edits in trunk/Source/WebKit2

Build fix 2.

  • UIProcess/API/Cocoa/WKPreferences.h:
12:34 PM Changeset in webkit [169322] by weinig@apple.com
  • 2 edits in trunk/Source/WebKit2

Fix build.

  • UIProcess/API/Cocoa/WKPreferences.h:
11:59 AM Changeset in webkit [169321] by weinig@apple.com
  • 4 edits in trunk/Source/WebKit2

[WebKit2] Rework WebPreferencesStore to allow easier overriding of defaults
https://bugs.webkit.org/show_bug.cgi?id=133258

Reviewed by Anders Carlsson.

To support changing default values of preferences based on which API you are using (either
the legacy C-SPI or the Objective-C API), overhaul the WebPreferencesStore to use a single
HashMap of String -> Value (where Value is new type that can be a String, bool, uint32_t or double)
instead of four HashMaps. This allows us to use two of these new HashMaps, one for the actual
set values, and one for overridden defaults. This new Value class will also allow easier incremental
updating of preferences in the future.

Other notable changes:

  • Removed Float kind of preferences. There were none.
  • Move StorageBlockingPolicy preference to the correct group, it's a uint32_t, not bool.
  • Shared/WebPreferencesStore.cpp:
  • Shared/WebPreferencesStore.h:
11:09 AM Changeset in webkit [169320] by ddkilzer@apple.com
  • 21 edits in trunk/Source

Add type-checked casts for TransformOperations
<http://webkit.org/b/133217>

Reviewed by Simon Fraser.

Source/WebCore:

  • platform/graphics/GraphicsLayer.cpp:

(WebCore::GraphicsLayer::validateTransformOperations):

  • platform/graphics/ca/GraphicsLayerCA.cpp:

(WebCore::getTransformFunctionValue):

  • Switch to type-checked casts.
  • platform/graphics/transforms/IdentityTransformOperation.h:
  • platform/graphics/transforms/Matrix3DTransformOperation.h:
  • platform/graphics/transforms/MatrixTransformOperation.h:
  • platform/graphics/transforms/PerspectiveTransformOperation.h:
  • platform/graphics/transforms/RotateTransformOperation.h:
  • platform/graphics/transforms/ScaleTransformOperation.h:
  • platform/graphics/transforms/SkewTransformOperation.h:
  • platform/graphics/transforms/TranslateTransformOperation.h:
  • Add 'final' to class declaration.
  • Add 'override' to overridden methods.
  • Add type-checked cast via TRANSFORMOPERATION_TYPE_CASTS macro.
  • Move implementation of operator==(const TransformOperation&) from header to source file so it is able to use a type-checked cast, and switch back to using a reference instead of a pointer.
  • Add or update ASSERT in private constructor to check for correct OperationType in classes that represent multiple types.
  • platform/graphics/transforms/Matrix3DTransformOperation.cpp:

(WebCore::Matrix3DTransformOperation::operator==): Added.

  • platform/graphics/transforms/MatrixTransformOperation.cpp:

(WebCore::Matrix3DTransformOperation::operator==): Added.
(WebCore::MatrixTransformOperation::blend):

  • Switch to type-checked casts and use a reference.
  • platform/graphics/transforms/PerspectiveTransformOperation.cpp:

(WebCore::Matrix3DTransformOperation::operator==): Added.
(WebCore::PerspectiveTransformOperation::blend):

  • platform/graphics/transforms/RotateTransformOperation.cpp:

(WebCore::Matrix3DTransformOperation::operator==): Added.
(WebCore::RotateTransformOperation::blend):

  • platform/graphics/transforms/ScaleTransformOperation.cpp:

(WebCore::Matrix3DTransformOperation::operator==): Added.
(WebCore::ScaleTransformOperation::blend):

  • platform/graphics/transforms/SkewTransformOperation.cpp:

(WebCore::Matrix3DTransformOperation::operator==): Added.
(WebCore::SkewTransformOperation::blend):

  • platform/graphics/transforms/TranslateTransformOperation.cpp:

(WebCore::Matrix3DTransformOperation::operator==): Added.
(WebCore::TranslateTransformOperation::blend):

  • Switch to type-checked casts.
  • platform/graphics/transforms/TransformOperation.h:

(WebCore::TransformOperation::isRotateTransformOperationType):
(WebCore::TransformOperation::isScaleTransformOperationType):
(WebCore::TransformOperation::isSkewTransformOperationType):
(WebCore::TransformOperation::isTranslateTransformOperationType):

  • Add type-checking methods used in constructors and type-checked casts.
  • Define TRANSFORMOPERATION_TYPE_CASTS macro used by subclasses.

Source/WebKit2:

  • Shared/CoordinatedGraphics/CoordinatedGraphicsArgumentCoders.cpp:

(IPC::ArgumentCoder<TransformOperations>::encode):

  • Switch to type-checked casts.
11:00 AM Changeset in webkit [169319] by ddkilzer@apple.com
  • 2 edits in trunk/Source/WebKit2

Use type-checking FilterOperation casts in CoordinatedGraphicsArgumentCoders.cpp
<http://webkit.org/b/133203>

Reviewed by Simon Fraser.

  • Shared/CoordinatedGraphics/CoordinatedGraphicsArgumentCoders.cpp:

(IPC::ArgumentCoder<WebCore::FilterOperations>::encode):

  • Replace static_cast<>() operators with type-checking FilterOperation casts.
  • Replace 'default' statement with 'case' statements so new filters added later require an explicit decision to be made.

(IPC::ArgumentCoder<WebCore::FilterOperations>::decode):

  • Replace 'default' statement with 'case' statements so new filters added later require an explicit decision to be made.
9:37 AM Changeset in webkit [169318] by ddkilzer@apple.com
  • 3 edits
    2 adds in trunk

Crash in WebCore::TextResourceDecoder::checkForCSSCharset
<http://webkit.org/b/133257>
<rdar://problem/17027109>

Reviewed by Alexey Proskuryakov.

Source/WebCore:

Test: fast/encoding/css-charset-missing-semi-colon-and-newline.html

  • loader/TextResourceDecoder.cpp:

(WebCore::TextResourceDecoder::checkForCSSCharset): Add early
return.

LayoutTests:

  • fast/encoding/css-charset-missing-semi-colon-and-newline-expected.txt: Added.
  • fast/encoding/css-charset-missing-semi-colon-and-newline.html: Added.
12:00 AM Changeset in webkit [169317] by eric.carlson@apple.com
  • 2 edits in trunk/Source/WebCore

[Mac] preload AVAssetTrack properties before asking for them
https://bugs.webkit.org/show_bug.cgi?id=133240

Reviewed by Jer Noble.

  • platform/graphics/avfoundation/objc/MediaPlayerPrivateAVFoundationObjC.mm:

(WebCore::assetTrackMetadataKeyNames): Add preferredTransform and naturalSize to the list

of properties we preload.

May 24, 2014:

8:50 PM Changeset in webkit [169316] by akling@apple.com
  • 4 edits in trunk/Source/JavaScriptCore

Object.prototype.toString() should use cached strings for null/undefined.
<https://webkit.org/b/133261>

Normally, when calling Object.prototype.toString() on a regular object,
we'd cache the result of the stringification on the object's structure,
making repeated calls fast.

For null and undefined, we were not as smart. We'd instead construct a
new string with either "[object Null]" or "[object Undefined]" each time.

This was exposed by Dromaeo's JS library tests, where some prototype.js
subtests generate millions of strings this way.

This patch adds two VM-permanent cached strings to the SmallStrings.
Looks like ~10% speed-up on Dromaeo/jslib-traverse-prototype.html

Reviewed by Darin Adler.

  • runtime/ObjectPrototype.cpp:

(JSC::objectProtoFuncToString):

  • runtime/SmallStrings.cpp:

(JSC::SmallStrings::SmallStrings):
(JSC::SmallStrings::initializeCommonStrings):
(JSC::SmallStrings::visitStrongReferences):

  • runtime/SmallStrings.h:

(JSC::SmallStrings::nullObjectString):
(JSC::SmallStrings::undefinedObjectString):

8:30 PM Changeset in webkit [169315] by jhoneycutt@apple.com
  • 2 edits in trunk/Source/WebKit2

Potential null dereference in
WebFrameLoaderClient::dispatchDidFailProvisionalLoad()
<https://bugs.webkit.org/show_bug.cgi?id=133193>

WebFrameLoaderClient::dispatchDidFailProvisionalLoad() calls the
InjectedBundleLoaderClient's didFailProvisionalLoadWithErrorForFrame()
before sending a message to the UI process that a provisional load has
failed. It's possible for the provisional document loader to become
null while calling into the InjectedBundleLoaderClient (as is the case
with the WebKitTestRunner's injected bundle), leading to a null
dereference when trying to send the DidFailProvisionalLoadForFrame
message.

Reviewed by Darin Adler.

  • WebProcess/WebCoreSupport/WebFrameLoaderClient.cpp:

(WebKit::WebFrameLoaderClient::dispatchDidFailProvisionalLoad):
Get the navigation ID before calling into the injected bundle.

8:12 PM Changeset in webkit [169314] by commit-queue@webkit.org
  • 3 edits in trunk/Source/WebCore

Remove one of the CSSProperty constructor
https://bugs.webkit.org/show_bug.cgi?id=131094

Patch by Zsolt Borbely <zsborbely.u-szeged@partner.samsung.com> on 2014-05-24
Reviewed by Darin Adler.

Remove one of the CSSProperty constructor, because it is obsolete.

  • css/CSSProperty.h:

(WebCore::CSSProperty::CSSProperty): Deleted.

  • css/StyleProperties.h:

(WebCore::StyleProperties::PropertyReference::toCSSProperty):

1:40 PM Changeset in webkit [169313] by Simon Fraser
  • 6 edits in trunk/Source

Rename ScrollingTreeScrollingNode's m_scrollPosition to make it clear that it's the value committed from the state tree
https://bugs.webkit.org/show_bug.cgi?id=133254

Reviewed by Tim Horton.

Source/WebCore:

Make ScrollingTreeScrollingNode::scrollPosition() pure virtual, and rename
the member variable and associated getter to make it clear that they relate
to the last committed scroll position.

  • page/scrolling/ScrollingTreeScrollingNode.cpp:

(WebCore::ScrollingTreeScrollingNode::updateBeforeChildren):

  • page/scrolling/ScrollingTreeScrollingNode.h:

(WebCore::ScrollingTreeScrollingNode::lastCommittedScrollPosition):
(WebCore::ScrollingTreeScrollingNode::scrollPosition): Deleted.

Source/WebKit2:

Override scrollPosition() on ScrollingTreeOverflowScrollingNodeIOS.

  • UIProcess/Scrolling/ios/ScrollingTreeOverflowScrollingNodeIOS.h:
  • UIProcess/Scrolling/ios/ScrollingTreeOverflowScrollingNodeIOS.mm:

(WebKit::ScrollingTreeOverflowScrollingNodeIOS::scrollPosition):

1:40 PM Changeset in webkit [169312] by Simon Fraser
  • 8 edits in trunk/Source/WebCore

Share some more ScrollingTreeScrollingNode code
https://bugs.webkit.org/show_bug.cgi?id=133248

Reviewed by Sam Weinig.

Push scrollBy(), scrollByWithoutContentEdgeConstraints() and setScrollPosition()
down to ScrollingTreeFrameScrollingNode.

This requires that scrollPosition() return the right thing for each class, so make
it virtual. Future patches will reduce the confusion between the committed scroll
position and the one derived from layers.

  • page/scrolling/ScrollingTreeFrameScrollingNode.cpp:

(WebCore::ScrollingTreeFrameScrollingNode::scrollBy):
(WebCore::ScrollingTreeFrameScrollingNode::scrollByWithoutContentEdgeConstraints):
(WebCore::ScrollingTreeFrameScrollingNode::setScrollPosition):

  • page/scrolling/ScrollingTreeFrameScrollingNode.h:
  • page/scrolling/ScrollingTreeScrollingNode.h:

(WebCore::ScrollingTreeScrollingNode::scrollPosition):

  • page/scrolling/ios/ScrollingTreeFrameScrollingNodeIOS.h:
  • page/scrolling/ios/ScrollingTreeFrameScrollingNodeIOS.mm:

(WebCore::ScrollingTreeFrameScrollingNodeIOS::setScrollPosition): Deleted.
(WebCore::ScrollingTreeFrameScrollingNodeIOS::scrollBy): Deleted.
(WebCore::ScrollingTreeFrameScrollingNodeIOS::scrollByWithoutContentEdgeConstraints): Deleted.

  • page/scrolling/mac/ScrollingTreeFrameScrollingNodeMac.h:
  • page/scrolling/mac/ScrollingTreeFrameScrollingNodeMac.mm:

(WebCore::ScrollingTreeFrameScrollingNodeMac::updateBeforeChildren):
(WebCore::ScrollingTreeFrameScrollingNodeMac::immediateScrollBy):
(WebCore::ScrollingTreeFrameScrollingNodeMac::immediateScrollByWithoutContentEdgeConstraints):
(WebCore::ScrollingTreeFrameScrollingNodeMac::scrollPosition):
(WebCore::ScrollingTreeFrameScrollingNodeMac::setScrollPosition):
(WebCore::ScrollingTreeFrameScrollingNodeMac::scrollBy): Deleted.
(WebCore::ScrollingTreeFrameScrollingNodeMac::scrollByWithoutContentEdgeConstraints): Deleted.

1:40 PM Changeset in webkit [169311] by Simon Fraser
  • 13 edits in trunk/LayoutTests

Scrolling tests should not try to dump pixels
https://bugs.webkit.org/show_bug.cgi?id=133245

Reviewed by Tim Horton.

dumpAsText(true) -> dumpAsText().

  • platform/mac-wk2/tiled-drawing/fast-scroll-div-latched-div-with-handler.html:
  • platform/mac-wk2/tiled-drawing/fast-scroll-div-latched-div.html:
  • platform/mac-wk2/tiled-drawing/fast-scroll-div-latched-mainframe-with-handler.html:
  • platform/mac-wk2/tiled-drawing/fast-scroll-div-latched-mainframe.html:
  • platform/mac-wk2/tiled-drawing/fast-scroll-iframe-latched-iframe-with-handler.html:
  • platform/mac-wk2/tiled-drawing/fast-scroll-iframe-latched-iframe.html:
  • platform/mac-wk2/tiled-drawing/fast-scroll-iframe-latched-mainframe-with-handler.html:
  • platform/mac-wk2/tiled-drawing/fast-scroll-iframe-latched-mainframe.html:
  • platform/mac-wk2/tiled-drawing/fast-scroll-select-latched-mainframe-with-handler.html:
  • platform/mac-wk2/tiled-drawing/fast-scroll-select-latched-mainframe.html:
  • platform/mac-wk2/tiled-drawing/fast-scroll-select-latched-select-with-handler.html:
  • platform/mac-wk2/tiled-drawing/fast-scroll-select-latched-select.html:
12:30 PM Changeset in webkit [169310] by Chris Fleizach
  • 23 edits in trunk/Source

AX: fix coordinate mapping for iOS accessibility
https://bugs.webkit.org/show_bug.cgi?id=133188

Reviewed by Sam Weinig.

Source/WebCore:
Make WebCore aware of the accessibility point/rect conversion methods.

  • accessibility/ios/WebAccessibilityObjectWrapperIOS.mm:

(-[WebAccessibilityObjectWrapper convertPointToScreenSpace:]):
(-[WebAccessibilityObjectWrapper convertRectToScreenSpace:]):

  • loader/EmptyClients.h:
  • page/Chrome.cpp:

(WebCore::Chrome::accessibilityScreenToRootView):
(WebCore::Chrome::rootViewToAccessibilityScreen):

  • page/Chrome.h:
  • page/ChromeClient.h:
  • platform/HostWindow.h:

Source/WebKit/mac:

  • WebCoreSupport/WebChromeClient.h:
  • WebCoreSupport/WebChromeClient.mm:

(WebChromeClient::accessibilityScreenToRootView):
(WebChromeClient::rootViewToAccessibilityScreen):

Source/WebKit2:
Add in accessibility conversion methods that will check if the methods are present
(if accessibility is not on, they won't be)
and use those to convert points into the right coordinate space that VoiceOver is expecting.

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

(WebKit::WebPageProxy::accessibilityScreenToRootView):
(WebKit::WebPageProxy::rootViewToAccessibilityScreen):

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

(WebKit::PageClientImpl::accessibilityScreenToRootView):
(WebKit::PageClientImpl::rootViewToAccessibilityScreen):

  • UIProcess/mac/PageClientImpl.h:
  • WebProcess/WebCoreSupport/WebChromeClient.cpp:

(WebKit::WebChromeClient::accessibilityScreenToRootView):
(WebKit::WebChromeClient::rootViewToAccessibilityScreen):

  • WebProcess/WebCoreSupport/WebChromeClient.h:
  • WebProcess/WebPage/WKAccessibilityWebPageObjectIOS.mm:

(-[WKAccessibilityWebPageObject accessibilityHitTest:]):

  • WebProcess/WebPage/WebPage.cpp:

(WebKit::WebPage::accessibilityScreenToRootView):
(WebKit::WebPage::rootViewToAccessibilityScreen):

  • WebProcess/WebPage/WebPage.h:
7:50 AM Changeset in webkit [169309] by Alan Bujtas
  • 4 edits
    2 adds in trunk

Subpixel rendering: Non-compositing transforms with subpixel coordinates paint to wrong position.
https://bugs.webkit.org/show_bug.cgi?id=133184
<rdar://problem/16745606>

Reviewed by Simon Fraser.

Snapping relative negative coordinate values should produce the same position as if they were
positive absolute coordinates.
When a child box gets positioned off of its containers towards top/left, its relative coordinates
become negative. Pixel snapping those negative values should produce the same
final painting position as if the child box was fixed positioned with positive coordinates.
Since halfway values always round away from zero, negative and positive halfway values
produce opposite rounding direction.
This patch ensures that negative halfway values round to the direction as if they were positive.

Source/WebCore:
Test: fast/layers/hidpi-floor-negative-coordinate-values-to-maintain-rounding-direction.html

  • platform/LayoutUnit.h:

(WebCore::roundToDevicePixel):

  • rendering/RenderLayer.cpp:

(WebCore::RenderLayer::paintLayerByApplyingTransform):

LayoutTests:

  • fast/layers/hidpi-floor-negative-coordinate-values-to-maintain-rounding-direction-expected.html: Added.
  • fast/layers/hidpi-floor-negative-coordinate-values-to-maintain-rounding-direction.html: Added.
4:49 AM Changeset in webkit [169308] by fred.wang@free.fr
  • 3 edits in trunk/LayoutTests

Update GTK references for some tests after bug 130322.
https://bugs.webkit.org/show_bug.cgi?id=130322

Unreviewed gardening.

  • platform/gtk/mathml/opentype/vertical-LatinModern-expected.txt:
  • platform/gtk/mathml/presentation/mo-stretch-expected.txt:
4:30 AM Changeset in webkit [169307] by fred.wang@free.fr
  • 2 edits in trunk/LayoutTests

Mark some opentype MATH tests failing due to bad references.
https://bugs.webkit.org/show_bug.cgi?id=130322

Unreviewed gardening.

  • platform/mac/TestExpectations:
12:17 AM Changeset in webkit [169306] by Carlos Garcia Campos
  • 5 edits in trunk/Source/WebKit2

REGRESSION(r165841): Messages sent before the child process is launched are never sent after r165841
https://bugs.webkit.org/show_bug.cgi?id=131675

Reviewed by Anders Carlsson.

Since r165841 the connection is opened after the pending messages
are sent, because connectionWillOpen might send messages that we
want to happen after the ones already pending. The problem is that
Connection::canSendOutgoingMessages() returns false when
connection hasn't been opened. We should ensure no messages are
sent by connectionWillOpen.

  • Shared/ChildProcessProxy.cpp:

(WebKit::ChildProcessProxy::didFinishLaunching): Open the
connection before sending pending messages.

  • UIProcess/WebPageProxy.cpp:

(WebKit::WebPageProxy::connectionWillOpen): Do not call
VisitedLinkProvider::addProcess() here because it tries to send a
message to the web process, but the connection hasn't be opened yet.
(WebKit::WebPageProxy::processDidFinishLaunching): Call
VisitedLinkProvider::addProcess() here instead.

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

(WebKit::WebProcessProxy::didFinishLaunching): Call
WebPageProxy::processDidFinishLaunching() for every web page.

May 23, 2014:

9:56 PM Changeset in webkit [169305] by fred.wang@free.fr
  • 10 edits
    30 adds in trunk

Use size variants and glyph assembly from the MATH data.
https://bugs.webkit.org/show_bug.cgi?id=130322

Reviewed by Chris Fleizach.

Source/WebCore:
This patch modifies the RenderMathMLOperator code to use the MATH table
when one is provided in the current font on the <math> tag. More
precisely, the MathVariants table is used to draw a size variant or
a glyph assembly. The displaystyle attribute is not supported yet, so
for now large operators are always assumed to be in display style. The
MATH support does not work well with all platforms+fonts, so at the
moment the default font-family on the <math> is not changed.

Tests: mathml/opentype/large-operators-LatinModern.html

mathml/opentype/opentype-stretchy.html
mathml/opentype/vertical-LatinModern.html

  • css/mathml.css: We only specify the default font-family on the math root, so that people can easily style the mathematics.

For now, old fonts without the MATH table are still used as the default.
(math):
(math, mfenced > *): Deleted.
(mo, mfenced): Deleted.

  • platform/graphics/SimpleFontData.cpp: don't return the math data if the font is loading.

(WebCore::SimpleFontData::mathData):

  • platform/graphics/opentype/OpenTypeMathData.cpp: update #ifdef so that disabling ENABLE_OPENTYPE_MATH won't lead to errors with unused parameters.

(WebCore::OpenTypeMathData::OpenTypeMathData):
(WebCore::OpenTypeMathData::getMathConstant):
(WebCore::OpenTypeMathData::getItalicCorrection):
(WebCore::OpenTypeMathData::getMathVariants):

  • rendering/mathml/RenderMathMLOperator.cpp:

(WebCore::RenderMathMLOperator::boundsForGlyph):
(WebCore::RenderMathMLOperator::heightForGlyph):
(WebCore::RenderMathMLOperator::advanceForGlyph):
(WebCore::RenderMathMLOperator::computePreferredLogicalWidths): We handle preferred width of size variants.
(WebCore::RenderMathMLOperator::shouldAllowStretching): This function now only returns whether the operator will stretch and no longer has side effect.
(WebCore::RenderMathMLOperator::getGlyphAssemblyFallBack): We add a function to convert from the MathVariant table data to the format supported by RenderMathMLOperator.
(WebCore::RenderMathMLOperator::getDisplayStyleLargeOperator): We add a function to get the glyph that will be used for large operators in display style.
(WebCore::RenderMathMLOperator::findStretchyData): We make this function handle size variants.
(WebCore::RenderMathMLOperator::updateStyle): We handle size variants.
(WebCore::RenderMathMLOperator::paint): We handle size variants.

  • rendering/mathml/RenderMathMLOperator.h:

LayoutTests:
This adds some pixel tests for large operators and vertical stretchy operators.

  • mathml/opentype/LICENSE-LatinModern.txt: Added.
  • mathml/opentype/large-operators-LatinModern.html: Added.
  • mathml/opentype/latinmodern-math.woff: Added.
  • mathml/opentype/opentype-stretchy.html: Added.
  • mathml/opentype/stretchy.woff: Added.
  • mathml/opentype/vertical-LatinModern.html: Added.
  • platform/efl/mathml/opentype/large-operators-LatinModern-expected.png: Added.
  • platform/efl/mathml/opentype/large-operators-LatinModern-expected.txt: Added.
  • platform/efl/mathml/opentype/opentype-stretchy-expected.png: Added.
  • platform/efl/mathml/opentype/opentype-stretchy-expected.txt: Added.
  • platform/efl/mathml/opentype/vertical-LatinModern-expected.png: Added.
  • platform/efl/mathml/opentype/vertical-LatinModern-expected.txt: Added.
  • platform/gtk/mathml/opentype/large-operators-LatinModern-expected.png: Added.
  • platform/gtk/mathml/opentype/large-operators-LatinModern-expected.txt: Added.
  • platform/gtk/mathml/opentype/opentype-stretchy-expected.png: Added.
  • platform/gtk/mathml/opentype/opentype-stretchy-expected.txt: Added.
  • platform/gtk/mathml/opentype/vertical-LatinModern-expected.png: Added.
  • platform/gtk/mathml/opentype/vertical-LatinModern-expected.txt: Added.
  • platform/gtk/mathml/presentation/mo-stretch-expected.png: update reference due to change in mathml.css
  • platform/gtk/mathml/presentation/mo-stretch-expected.txt: ditto
  • platform/mac/mathml/opentype/large-operators-LatinModern-expected.txt: Added.
  • platform/mac/mathml/opentype/opentype-stretchy-expected.txt: Added.
  • platform/mac/mathml/opentype/vertical-LatinModern-expected.txt: Added.
  • platform/mac-wk2/mathml/opentype/large-operators-LatinModern-expected.txt: Added.
  • platform/mac-wk2/mathml/opentype/opentype-stretchy-expected.txt: Added.
  • platform/mac-wk2/mathml/opentype/vertical-LatinModern-expected.txt: Added.
  • platform/win/TestExpectations: Mark the OpenType MATH tests as failing
9:35 PM Changeset in webkit [169304] by benjamin@webkit.org
  • 2 edits in trunk/Source/WebKit2

REGRESSION: Rubberbanding out jumps to top left corner on google.com
https://bugs.webkit.org/show_bug.cgi?id=133253
<rdar://problem/17025664>

Patch by Ian Henderson <ianh@apple.com> on 2014-05-23
Reviewed by Benjamin Poulain.

  • UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView scrollViewWillEndDragging:withVelocity:targetContentOffset:]):
Avoid deceleration during zooming to work around a UIKit bug.

9:31 PM Changeset in webkit [169303] by matthew_hanson@apple.com
  • 5 edits in branches/safari-538.34-branch/Source

Versioning.

9:25 PM Changeset in webkit [169302] by benjamin@webkit.org
  • 4 edits in trunk/Source/WebKit2

[iOS][WK2] Avoid updating the view from the viewport configuration unnecessarily
https://bugs.webkit.org/show_bug.cgi?id=133255
<rdar://problem/16890926>

Patch by Benjamin Poulain <bpoulain@apple.com> on 2014-05-23
Reviewed by Simon Fraser.

All the updates were mostly harmless since they are pretty cheap, but they can be
very annoying for debugging.

This patch skips viewportUpdates in the trivial cases.

  • UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView _setMinimumLayoutSizeOverride:]):
(-[WKWebView _setMinimumLayoutSizeOverrideForMinimalUI:]):

  • WebProcess/WebPage/WebPage.cpp:

(WebKit::WebPage::mainFrameDidLayout):

  • WebProcess/WebPage/ios/WebPageIOS.mm:

(WebKit::WebPage::viewportPropertiesDidChange):

9:07 PM Changeset in webkit [169301] by matthew_hanson@apple.com
  • 1 copy in tags/Safari-538.34.48

Create Safari-538.34.48 tag.

8:45 PM Changeset in webkit [169300] by benjamin@webkit.org
  • 8 edits in trunk/Source/WebKit2

[iOS][WK2] The page scale factor randomly resets to initial scale when editing
https://bugs.webkit.org/show_bug.cgi?id=133244

Patch by Benjamin Poulain <bpoulain@apple.com> on 2014-05-23
Reviewed by Enrica Casucci.

We were never telling the WebProcess that zoom is used triggered. As a result, the ViewportConfiguration was free
to reset the scale whenever it felt like it. The fix is to call willStartUserTriggeredZoom before changing the scale.

This patch also split _willStartUserTriggeredScrollingOrZooming from willStartUserTriggeredZoom
and rename it to willStartPanOrPinchGesture. This is just a tiny cleanup since the only thing left in _willStartUserTriggeredScrollingOrZooming
was gesture related.

  • UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView _zoomToFocusRect:WebCore::selectionRect:WebCore::fontSize:minimumScale:maximumScale:allowScaling:forceScroll:]):
(-[WKWebView scrollViewWillBeginZooming:withView:]):
(-[WKWebView scrollViewWillBeginDragging:]):

  • UIProcess/API/Cocoa/WKWebViewInternal.h:
  • UIProcess/API/ios/WKViewIOS.mm:

(-[WKView scrollViewWillBeginDragging:]):

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

(-[WKContentView willStartUserTriggeredZoom]):
(-[WKContentView willStartUserTriggeredScroll]): Deleted.

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

(-[WKContentView willStartPanOrPinchGesture]):
(-[WKContentView _willStartUserTriggeredScrollingOrZooming]): Deleted.

7:17 PM Changeset in webkit [169299] by timothy_horton@apple.com
  • 22 edits
    9 adds in trunk/Source

REGRESSION (iOS WebKit2): requestAnimationFrame fires more than once between layer tree commits
https://bugs.webkit.org/show_bug.cgi?id=132794
<rdar://problem/16877909>

Reviewed by Simon Fraser.

Virtualize DisplayRefreshMonitor so that WebKit2 can implement its own DisplayRefreshMonitor for UI-side compositing views.
This allows the synchronization of requestAnimationFrame callbacks with UI-side compositing painting.

  • WebCore.exp.in:
  • WebCore.xcodeproj/project.pbxproj:
  • page/ChromeClient.h:

(WebCore::ChromeClient::graphicsLayerFactory):
(WebCore::ChromeClient::createDisplayRefreshMonitor):

  • platform/graphics/DisplayRefreshMonitorFactory.h: Added.
  • platform/graphics/GraphicsLayerUpdater.cpp:

(WebCore::GraphicsLayerUpdater::createDisplayRefreshMonitor):

  • platform/graphics/GraphicsLayerUpdater.h:
  • rendering/RenderLayerCompositor.cpp:

(WebCore::RenderLayerCompositor::createDisplayRefreshMonitor):

  • rendering/RenderLayerCompositor.h:

Give the ChromeClient a chance to provide us with a custom DisplayRefreshMonitor.
If it does not, we'll fall back to making a DisplayRefreshMonitorMac or
DisplayRefreshMonitorIOS depending on the platform.

  • dom/ScriptedAnimationController.cpp:
  • dom/ScriptedAnimationController.h:
  • platform/graphics/DisplayRefreshMonitor.cpp:
  • platform/graphics/DisplayRefreshMonitor.h:
  • platform/graphics/DisplayRefreshMonitorClient.cpp: Added.
  • platform/graphics/DisplayRefreshMonitorClient.h: Added.
  • platform/graphics/DisplayRefreshMonitorManager.cpp: Added.
  • platform/graphics/DisplayRefreshMonitorManager.h: Added.

Split DisplayRefreshMonitor.cpp into one file per class.
Use references in a few places.
Remove some useless comments.

  • platform/graphics/DisplayRefreshMonitor.cpp:

(WebCore::DisplayRefreshMonitor::displayDidRefresh):
Use takeAny() and remove a FIXME about it not existing.

  • platform/graphics/DisplayRefreshMonitorClient.cpp: Added.

(WebCore::DisplayRefreshMonitorClient::fireDisplayRefreshIfNeeded):
Use an early return instead of having the whole function body in an if().

  • platform/graphics/ios/DisplayRefreshMonitorIOS.h: Added.
  • platform/graphics/ios/DisplayRefreshMonitorIOS.mm:
  • platform/graphics/mac/DisplayRefreshMonitorMac.cpp:
  • platform/graphics/mac/DisplayRefreshMonitorMac.h: Added.

Virtualize DisplayRefreshMonitorIOS and DisplayRefreshMonitorMac,
and move things specific to each of them out of DisplayRefreshMonitor itself.

  • WebKit2.xcodeproj/project.pbxproj:
  • WebProcess/WebCoreSupport/WebChromeClient.cpp:

(WebKit::WebChromeClient::createDisplayRefreshMonitor):

  • WebProcess/WebCoreSupport/WebChromeClient.h:
  • WebProcess/WebPage/DrawingArea.h:

(WebKit::DrawingArea::graphicsLayerFactory):
(WebKit::DrawingArea::createDisplayRefreshMonitor):
Plumb the DisplayRefreshMonitor request through to the DrawingArea.
RemoteLayerTreeDrawingArea will implement and return a RemoteLayerTreeDisplayRefreshMonitor.

  • WebProcess/WebPage/Cocoa/RemoteLayerTreeDisplayRefreshMonitor.h: Added.

(WebKit::RemoteLayerTreeDisplayRefreshMonitor::create):

  • WebProcess/WebPage/Cocoa/RemoteLayerTreeDisplayRefreshMonitor.mm: Added.

(WebKit::RemoteLayerTreeDisplayRefreshMonitor::RemoteLayerTreeDisplayRefreshMonitor):
(WebKit::RemoteLayerTreeDisplayRefreshMonitor::~RemoteLayerTreeDisplayRefreshMonitor):
(WebKit::RemoteLayerTreeDisplayRefreshMonitor::requestRefreshCallback):
When requesting a display refresh callback, schedule a layer commit.
It may end up being empty, but will still call us back with a didUpdate,
upon which we'll fire the callback.

(WebKit::RemoteLayerTreeDisplayRefreshMonitor::didUpdateLayers):

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

(WebKit::RemoteLayerTreeDrawingArea::createDisplayRefreshMonitor):
(WebKit::RemoteLayerTreeDrawingArea::willDestroyDisplayRefreshMonitor):
Keep track of the set of DisplayRefreshMonitors.

(WebKit::RemoteLayerTreeDrawingArea::didUpdate):
Fire requestAnimationFrame callbacks once the UI process commits the new layer tree.

7:11 PM Changeset in webkit [169298] by commit-queue@webkit.org
  • 3 edits
    2 adds in trunk

CSS JIT: Apply backtracking optimization to adjacent backtracking
https://bugs.webkit.org/show_bug.cgi?id=132951

Patch by Yusuke Suzuki <Yusuke Suzuki> on 2014-05-23
Reviewed by Benjamin Poulain.

Apply the backtracking optimization to the adjacent backtracking.
This optimization is already done for the descendant backtracking.
We apply this to the adjacent backtracking similarly.

Source/WebCore:
Test: fast/selectors/backtracking-adjacent-optimized.html

  • cssjit/SelectorCompiler.cpp:

(WebCore::SelectorCompiler::SelectorFragment::SelectorFragment):
(WebCore::SelectorCompiler::solveAdjacentBacktrackingActionForDirectAdjacent):
(WebCore::SelectorCompiler::solveBacktrackingAction):
(WebCore::SelectorCompiler::computeBacktrackingStartOffsetInChain):
(WebCore::SelectorCompiler::computeBacktrackingHeightFromDescendant):
(WebCore::SelectorCompiler::computeBacktrackingWidthFromIndirectAdjacent):
(WebCore::SelectorCompiler::SelectorCodeGenerator::computeBacktrackingInformation):
(WebCore::SelectorCompiler::SelectorCodeGenerator::generateIndirectAdjacentTreeWalker):
(WebCore::SelectorCompiler::SelectorCodeGenerator::linkFailures):
(WebCore::SelectorCompiler::computeBacktrackingStartHeightFromDescendant): Deleted.

LayoutTests:

  • fast/selectors/backtracking-adjacent-optimized-expected.txt: Added.
  • fast/selectors/backtracking-adjacent-optimized.html: Added.
6:06 PM Changeset in webkit [169297] by achristensen@apple.com
  • 5 edits in trunk/Source/WebCore

Make CSS JIT run on ARM64.
https://bugs.webkit.org/show_bug.cgi?id=133156

Reviewed by Benjamin Poulain.

  • cssjit/FunctionCall.h:

(WebCore::FunctionCall::saveAllocatedRegisters):
(WebCore::FunctionCall::restoreAllocatedRegisters):
Use StackAllocator's new push and pop functions to push and pop a vector instead of iterating it.

  • cssjit/RegisterAllocator.h:

(WebCore::RegisterAllocator::reserveCalleeSavedRegisters):
(WebCore::RegisterAllocator::restoreCalleeSavedRegisters):
Return a vector of registers to allocate instead of doing the allocation to make the RegisterAllocator
not need to know about the StackAllocator and to use the new vector push and pop functions.
(WebCore::RegisterAllocator::~RegisterAllocator):
Store RegisterIDs instead of StackReferences to avoid needing to know about the stack.

  • cssjit/SelectorCompiler.cpp:

(WebCore::SelectorCompiler::SelectorCodeGenerator::compile):
Removed the requirement for assert to be disabled to print disassembly when debugging css jit.
(WebCore::SelectorCompiler::SelectorCodeGenerator::generatePrologue):
(WebCore::SelectorCompiler::SelectorCodeGenerator::generateEpilogue):
Added to correctly push the link register and frame pointer.
This is required if the jit code calls a function on arm64 and helpful for debugging tools on x86_64.
(WebCore::SelectorCompiler::SelectorCodeGenerator::generateSelectorChecker):
Generate the prologue and epilogue which respectively push and pop
the link register, frame pointer, and callee saved registers if needed.

  • cssjit/StackAllocator.h:

(WebCore::StackAllocator::push):
(WebCore::StackAllocator::pop):
Added new vector push and pop functions to use stp and ldb instructions on arm64.

5:29 PM Changeset in webkit [169296] by matthew_hanson@apple.com
  • 5 edits in branches/safari-538.34-branch/Source

Versioning.

5:23 PM Changeset in webkit [169295] by commit-queue@webkit.org
  • 10 edits in trunk/Source

Hide fullscreen immediately when switching tabs.
https://bugs.webkit.org/show_bug.cgi?id=133239

Patch by Jeremy Jones <jeremyj@apple.com> on 2014-05-23
Reviewed by Eric Carlson.

Source/WebCore:
This change allows fullscreen to disappear immediately while still cleaning
up fullscreen normally.

  • WebCore.exp.in:
  • platform/ios/WebVideoFullscreenControllerAVKit.h:

renamed function to requestHideAndExitFullscreen

  • platform/ios/WebVideoFullscreenControllerAVKit.mm:

renamed function to requestHideAndExitFullscreen

(-[WebVideoFullscreenController requestHideAndExitFullscreen]):
rename of requestExitFullscreen that also hides.

(-[WebVideoFullscreenController requestExitFullscreen]): Deleted.

  • platform/ios/WebVideoFullscreenInterfaceAVKit.h: declare requestHideAndExitFullscreen()
  • platform/ios/WebVideoFullscreenInterfaceAVKit.mm:

(WebVideoFullscreenInterfaceAVKit::requestHideAndExitFullscreen):
this hides the window before requesting exitfullscreen from the model.

Source/WebKit/mac:
Renamed a function to better describe it's new functionality.

  • WebView/WebView.mm:

(-[WebView viewDidMoveToWindow]):
now calls: requestHideAndExitFullscreen

Source/WebKit2:
Renamed a function to better describe it's new functionality.

  • UIProcess/WebPageProxy.cpp:

(WebKit::WebPageProxy::viewStateDidChange):
now calls: requestHideAndExitFullscreen

5:18 PM Changeset in webkit [169294] by enrica@apple.com
  • 12 edits in trunk/Source/WebKit2

REGRESSION (WebKit2): Keyboard disappears/appears automatically between text boxes.
https://bugs.webkit.org/show_bug.cgi?id=133243
<rdar://problem/16761913>

Reviewed by Benjamin Poulain.

We need to ensure that when an element loses focus and another one gets it,
this is handled in one single message in the UI process to avoid seeing
the keyboard animating in and out.
This is accomplished by always postponing the blur notification until all events
have been processed.
If a new node is focused before the delayed blur notification is sent to the UI
process, the message we send includes both blur and focus notification. In this case,
the postponed blur notification is not sent.

  • UIProcess/PageClient.h:
  • UIProcess/WebPageProxy.h:
  • UIProcess/WebPageProxy.messages.in:
  • UIProcess/ios/PageClientImplIOS.h:
  • UIProcess/ios/PageClientImplIOS.mm:

(WebKit::PageClientImpl::startAssistingNode):

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

(-[WKContentView _startAssistingNode:userIsInteracting:blurPreviousNode:userObject:]):
(-[WKContentView _startAssistingNode:userIsInteracting:userObject:]): Deleted.

  • UIProcess/ios/WebPageProxyIOS.mm:

(WebKit::WebPageProxy::startAssistingNode):

  • WebProcess/WebPage/WebPage.cpp:

(WebKit::WebPage::WebPage):

  • WebProcess/WebPage/WebPage.h:
  • WebProcess/WebPage/ios/WebPageIOS.mm:

(WebKit::WebPage::elementDidFocus):
(WebKit::WebPage::elementDidBlur):

5:13 PM Changeset in webkit [169293] by msaboff@apple.com
  • 2 edits in trunk/Tools

REGRESSION(r169265): sh: line 0: cd: ../.vm
https://bugs.webkit.org/show_bug.cgi?id=133242

Rubber stamped by Geoffrey Garen.

Added missing shell escapes '\' required as part of previous fix.

  • Scripts/run-jsc-stress-tests:
5:12 PM Changeset in webkit [169292] by andersca@apple.com
  • 2 edits in trunk/Source/WebKit2

Remove an assertion.

  • WebProcess/WebPage/WebPage.cpp:

(WebKit::WebPage::loadRequest):

4:59 PM Changeset in webkit [169291] by matthew_hanson@apple.com
  • 1 copy in tags/Safari-538.34.47

New Tag.

4:52 PM Changeset in webkit [169290] by timothy_horton@apple.com
  • 8 edits
    2 adds in trunk/Source

[iOS] WKPDFView should have a page indicator
https://bugs.webkit.org/show_bug.cgi?id=133109
<rdar://problem/16383003>

Reviewed by Anders Carlsson.

  • UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView _setHasCustomContentView:loadedMIMEType:WTF::]):
(-[WKWebView _setObscuredInsets:]):
Forward obscured insets changes to the custom content view.

(-[WKWebView _updateScrollViewBackground]):
If we have a custom content view, use its background instead of the
(likely nonexistent) page's when updaing the scroll view's background color.

  • UIProcess/API/Cocoa/WKWebViewInternal.h:
  • UIProcess/API/Cocoa/WKWebViewPrivate.h:
  • UIProcess/Cocoa/WKWebViewContentProvider.h:
  • UIProcess/Cocoa/WKWebViewContentProviderDelegate.h: Added.
  • UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView _setInsetForOverlaidAccessoryViews:]):
Add SPI for clients to inform us about potential overlaid accessory views
which should be taken into account e.g. for positioning the page indicator.

Add a view that sits fixed on top of the WKWebView, ignores interaction,
but provides the WKWebViewContentProvider a place to install views
that don't scroll along with the content.

  • UIProcess/ios/WKPDFPageNumberIndicator.h: Added.
  • UIProcess/ios/WKPDFPageNumberIndicator.mm: Added.

(-[WKPDFPageNumberIndicator initWithFrame:]):
(-[WKPDFPageNumberIndicator dealloc]):
(-[WKPDFPageNumberIndicator setCurrentPageNumber:]):
(-[WKPDFPageNumberIndicator setPageCount:]):
(-[WKPDFPageNumberIndicator show]):
(-[WKPDFPageNumberIndicator hide:]):
(-[WKPDFPageNumberIndicator moveToPoint:animated:]):
(-[WKPDFPageNumberIndicator sizeThatFits:]):
(-[WKPDFPageNumberIndicator _updateLabel]):
(-[WKPDFPageNumberIndicator _makeRoundedCorners]):
Add a page number indicator.

  • UIProcess/ios/WKPDFView.mm:

(-[WKPDFView dealloc]):
(-[WKPDFView web_setScrollView:]):
(-[WKPDFView _revalidateViews]):
(-[WKPDFView _offsetForPageNumberIndicator]):
(-[WKPDFView _updatePageNumberIndicator]):
(-[WKPDFView web_setObscuredInsets:]):
(-[WKPDFView web_setInsetForOverlaidAccessoryViews:]):
(-[WKPDFView web_setFixedOverlayView:]):
(-[WKPDFView _computePageAndDocumentFrames]):
Update the page number indicator when the topmost visible page changes.
Move the page number indicator around when the obscured insets change.

  • WebKit2.xcodeproj/project.pbxproj:
  • English.lproj/Localizable.strings:
4:32 PM Changeset in webkit [169289] by rniwa@webkit.org
  • 2 edits in trunk/Tools

Revert the erroneous change committed in r169286.

  • Scripts/webkitpy/performance_tests/perftest.py:

(PerfTest._run_with_driver):

4:25 PM Changeset in webkit [169288] by andersca@apple.com
  • 2 edits in trunk/Source/WebKit2

Fix test failures on the bot.

  • UIProcess/API/Cocoa/WKProcessPool.mm:

(websiteDataDirectoryURL):
If we have no bundle identifier, use the process name.

3:37 PM Changeset in webkit [169287] by ggaren@apple.com
  • 1 edit
    1 add in trunk/PerformanceTests

Performance testing, diamond-square terrain generation + canvas
https://bugs.webkit.org/show_bug.cgi?id=133054

Reviewed by Ryosuke Niwa.

This test was written by Hunter Loftis. It originally appeared on his
blog @ <http://www.playfuljs.com/realistic-terrain-in-130-lines/>.

I did a bit of editing for format and benchmark suitability.

  • Canvas/terrain.html: Added.
3:37 PM Changeset in webkit [169286] by ggaren@apple.com
  • 2 edits in trunk/Tools

run-perf-tests should ignore whitespace lines when snarfing test output
https://bugs.webkit.org/show_bug.cgi?id=133238

Reviewed by Ryosuke Niwa.

Required for PerformanceTests/Canvas/terrain.html.

  • Scripts/webkitpy/performance_tests/perftest.py:

(PerfTest._run_with_driver):
(PerfTest):

3:32 PM Changeset in webkit [169285] by mhahnenberg@apple.com
  • 4 edits in trunk/Source/JavaScriptCore

Remove operationCallGetter

Rubber stamped by Filip Pizlo.

Nobody calls this function.

3:14 PM Changeset in webkit [169284] by akling@apple.com
  • 5 edits in trunk/Source/JavaScriptCore

Templatize GC's destructor invocation for dtor type.
<https://webkit.org/b/133231>

Get rid of a branch in callDestructor() by templatizing it for
the DestructorType. Removed JSCell::methodTableForDestruction()
since this was the only call site and it was jumping through
a bunch of unnecessary hoops.

Reviewed by Geoffrey Garen.

  • heap/MarkedBlock.cpp:

(JSC::MarkedBlock::callDestructor):
(JSC::MarkedBlock::specializedSweep):

  • heap/MarkedBlock.h:
  • runtime/JSCell.h:
  • runtime/JSCellInlines.h:

(JSC::JSCell::methodTableForDestruction): Deleted.

3:13 PM Changeset in webkit [169283] by akling@apple.com
  • 3 edits in trunk/Source/JavaScriptCore

Support inline caching of RegExpMatchesArray.length
<https://webkit.org/b/133234>

Give RegExpMatchesArray.length the same treatment as JSArray in
repatch so we don't have to go out of line on every access.

~13% speed-up on Octane/regexp.

Reviewed by Geoffrey Garen.

  • jit/Repatch.cpp:

(JSC::tryCacheGetByID):

  • runtime/RegExpMatchesArray.h:

(JSC::isRegExpMatchesArray):

2:49 PM Changeset in webkit [169282] by enrica@apple.com
  • 2 edits in trunk/Source/WebKit2

REGRESSION (iOS WebKit2): Autocorrection is not accurate when changing selection.
https://bugs.webkit.org/show_bug.cgi?id=133237
<rdar://problem/16655776>

Reviewed by Ryosuke Niwa.

  • UIProcess/ios/WKContentViewInteraction.mm:

(-[WKContentView beginSelectionChange]):
(-[WKContentView endSelectionChange]):

When the selection is changed with a gesture in an editable element,
we must notify the keyboard to make sure the autocorrection engine
updates its typing context. WKContentView must implement
beginSelectionChange and endSelectionChange to call into
the keyboard layer.

2:32 PM Changeset in webkit [169281] by Brent Fulgham
  • 4 edits in trunk/Source/WebKit2

[iOS] WK2: Provide implementation for windowFrame
https://bugs.webkit.org/show_bug.cgi?id=133205
<rdar://problem/16894890>

Reviewed by Benjamin Poulain.

  • UIProcess/Cocoa/UIDelegate.h:
  • UIProcess/Cocoa/UIDelegate.mm:

(WebKit::UIDelegate::UIClient::windowFrame): Added.

  • UIProcess/ios/PageClientImplIOS.mm:

(WebKit::PageClientImpl::convertToUserSpace): Added.

2:32 PM Changeset in webkit [169280] by enrica@apple.com
  • 2 edits in trunk/Source/WebCore

Remove invalid ASSERT in wordRangeForPosition.
https://bugs.webkit.org/show_bug.cgi?id=133232

Reviewed by Ryosuke Niwa.

In WK2 for iOS this function is called on every selection change.
The selection could be set by Javascript in an element that is not visible.
In that case we won't be able to create a VisiblePosition therefore wordRangeForPosition
should not ASSERT that the VisiblePosition is not null, but rather
perform an early return if it is null.

  • editing/VisibleUnits.cpp:

(WebCore::wordRangeFromPosition):

1:32 PM Changeset in webkit [169279] by dburkart@apple.com
  • 3 edits
    5 copies
    2 adds in branches/safari-537.77-branch

Merged r167832. <rdar://problem/16975563>

1:32 PM Changeset in webkit [169278] by Simon Fraser
  • 3 edits in trunk/LayoutTests

Rebaseline two tests affected by r169229.

  • platform/mac-wk2/compositing/tiling/rotated-tiled-clamped-expected.txt:
  • platform/mac-wk2/compositing/tiling/rotated-tiled-preserve3d-clamped-expected.txt:
1:31 PM Changeset in webkit [169277] by Manuel Rego Casasnovas
  • 2 edits in trunk/Source/WebCore

[CSS Grid Layout] Use gridRowCount() and gridColumnCount() when possible
https://bugs.webkit.org/show_bug.cgi?id=133213

Reviewed by Sergio Villar Senin.

Use gridRowCount() and gridColumnCount() instead of m_grid.size() and
m_grid[0].size() to make the code more descriptive and easier to
understand.

No new tests (No change in behavior).

  • rendering/RenderGrid.cpp:

(WebCore::RenderGrid::growGrid):
(WebCore::RenderGrid::placeAutoMajorAxisItemOnGrid):

12:51 PM Changeset in webkit [169276] by ap@apple.com
  • 4 edits in trunk/Source/WebCore

REGRESSION (r153917): Chart for the Category Explorer within Quicken Essentials
does not draw until you force a redraw
https://bugs.webkit.org/show_bug.cgi?id=133228
<rdar://problem/15221231>

Reviewed by Simon Fraser.

  • platform/RuntimeApplicationChecks.cpp: (WebCore::applicationIsQuickenEssentials):
  • platform/RuntimeApplicationChecks.h:

Added a check for Quicken Essentials.

  • platform/mac/WidgetMac.mm: (WebCore::Widget::paint): Don't do an early return

for Quicken Essentials, it depends on drawing to update its layer.

12:24 PM Changeset in webkit [169275] by Lucas Forschler
  • 15 edits in branches/safari-537.77-branch/Source

Rollout 169219. <rdar://problem/16975536>

12:10 PM Changeset in webkit [169274] by Lucas Forschler
  • 1 edit in branches/safari-537.77-branch/Source/WebCore/platform/graphics/filters/FilterOperation.h

Build fix after rdar://problem/16975484.

12:01 PM Changeset in webkit [169273] by mihnea@adobe.com
  • 5 edits in trunk

REGRESSION(r169105): fast/regions/cssom/region-range-for-box-crash.html is more crashy than before
https://bugs.webkit.org/show_bug.cgi?id=133124

Reviewed by David Hyatt.

Source/WebCore:
When adjusting the end points of a subtree selection in the case of a document
with named flows, we have to make sure that we pass the computed end
points to the visible selection algorithm only if the end points are
part of the same subtree, otherwise assume a selection clear behavior.

The test fast/regions/cssom/region-range-for-box-crash.html is used to
test this behavior.

  • rendering/SelectionSubtreeRoot.cpp:

(WebCore::SelectionSubtreeRoot::adjustForVisibleSelection):
(WebCore::SelectionSubtreeRoot::selectionStartEndPositions): Moved to header file.

  • rendering/SelectionSubtreeRoot.h:

(WebCore::SelectionSubtreeRoot::selectionStartEndPositions):

LayoutTests:
Unskip the test file that used to crash.

12:00 PM Changeset in webkit [169272] by dburkart@apple.com
  • 3 edits in branches/safari-537.77-branch/Source/WebCore

Merged r165862. <rdar://problem/16975511>

11:59 AM Changeset in webkit [169271] by andersca@apple.com
  • 2 edits in trunk/Source/WebKit2

Use a per app directory structure for website data
https://bugs.webkit.org/show_bug.cgi?id=133125
<rdar://problem/16830143>

Reviewed by Dan Bernstein, Sam Weinig and Tim Horton.

Put website data in subdirectories of ~/Library/WebKit/<bundle ID>/WebsiteData.
We use "LocalStorage", "WebSQL", and "IndexedDB" subdirectories.

  • UIProcess/API/Cocoa/WKProcessPool.mm:

(websiteDataDirectoryURL):
(-[WKProcessPool _initWithConfiguration:]):

11:36 AM Changeset in webkit [169270] by mhahnenberg@apple.com
  • 5 edits in trunk/LayoutTests

Rebaseline some tests after r169260

  • fast/dom/Geolocation/enabled-expected.txt:
  • fast/dom/Geolocation/script-tests/enabled.js:
  • js/dom/delete-syntax-expected.txt:
  • js/dom/script-tests/delete-syntax.js:
11:36 AM Changeset in webkit [169269] by matthew_hanson@apple.com
  • 5 edits in branches/safari-538.34-branch/Source

Versioning.

11:33 AM Changeset in webkit [169268] by dburkart@apple.com
  • 2 edits in branches/safari-537.77-branch/LayoutTests

Merged r161008. <rdar://problem/16975471>

11:29 AM Changeset in webkit [169267] by dburkart@apple.com
  • 2 edits in branches/safari-537.77-branch/LayoutTests

Merged r160896. <rdar://problem/16975471>

11:22 AM Changeset in webkit [169266] by dburkart@apple.com
  • 9 edits
    2 copies in branches/safari-537.77-branch

Merged r160847. <rdar://problem/16975471>

11:05 AM Changeset in webkit [169265] by msaboff@apple.com
  • 2 edits in trunk/Tools

REGRESSION (r169241) Crash running run-jsc-stress-tests on EFL ARM
https://bugs.webkit.org/show_bug.cgi?id=133223

Reviewed by Geoffrey Garen.

Change runAndMonitorTestRunnerCommand() to convert the *cmd argument to a
string before calling IO.popen() to work with older versions of ruby.
Changed the remote path in runShellTestRunner to quote the command
string we want to run via ssh so that it will work correctly with array
to string conversion needed for popen.

  • Scripts/run-jsc-stress-tests:
11:02 AM Changeset in webkit [169264] by commit-queue@webkit.org
  • 2 edits in trunk/Source/WebCore

reset m_havePreparedToPlay when changing media engines.
https://bugs.webkit.org/show_bug.cgi?id=133199

Patch by Jeremy Jones <jeremyj@apple.com> on 2014-05-23
Reviewed by Eric Carlson.

If a play request comes in before the media player is created, m_havePreparedToPlay
is set to true. But when the player is created, this prevents the new player from
getting the prepare call.

  • html/HTMLMediaElement.cpp:

(WebCore::HTMLMediaElement::mediaPlayerEngineUpdated): set m_havePreparedToPlay false.

10:52 AM Changeset in webkit [169263] by roger_fong@apple.com
  • 2 edits in trunk/LayoutTests

Unreviewed. Skip some snapshotting tests failing after r169173.
https://bugs.webkit.org/show_bug.cgi?id=133227.

  • platform/mac-wk2/TestExpectations:
10:22 AM Changeset in webkit [169262] by Michał Pakuła vel Rutka
  • 5 edits in trunk/LayoutTests

Unreviewed EFL gardening

  • platform/efl/TestExpectations: Add test expectation for tests faling after r167840.
  • platform/efl/fast/borders/border-antialiasing-expected.png: Rebaseline after r168046.
  • platform/efl/fast/borders/border-antialiasing-expected.txt: Ditto.
  • platform/efl/fast/forms/fieldset-align-expected.txt: Rebaseline after r168575.
10:11 AM Changeset in webkit [169261] by commit-queue@webkit.org
  • 2 edits in trunk/Tools

Remove 'css-shaders' option from the build-webkit
https://bugs.webkit.org/show_bug.cgi?id=133212

Patch by Zsolt Borbely <zsborbely.u-szeged@partner.samsung.com> on 2014-05-23
Reviewed by Daniel Bates.

Original CSS shaders removed in r162644.

  • Scripts/webkitperl/FeatureList.pm:
10:08 AM Changeset in webkit [169260] by oliver@apple.com
  • 2 edits in trunk/Source/WebCore

Navigator object needs to have properties directly on the instance object
https://bugs.webkit.org/show_bug.cgi?id=133221

Reviewed by Mark Lam.

Flag the Navigator object as requiring properties to be on the instance
as there were a few compatibility issues when on the prototype.

  • bindings/scripts/CodeGeneratorJS.pm:

(InterfaceRequiresAttributesOnInstance):

9:42 AM Changeset in webkit [169259] by mitz@apple.com
  • 2 edits in trunk/Source/WebKit2

[Cocoa] WKNavigationResponsePrivate.h is internal
https://bugs.webkit.org/show_bug.cgi?id=133218

Reviewed by Dean Jackson.

  • WebKit2.xcodeproj/project.pbxproj: Promote WKNavigationResponsePrivate.h to private.
9:13 AM Changeset in webkit [169258] by Simon Fraser
  • 3 edits in trunk/Source/WebCore

Switch CSSGradientValue to use more references
https://bugs.webkit.org/show_bug.cgi?id=133206

Reviewed by Andreas Kling.

Switch from pointers to references in various places.

  • css/CSSGradientValue.cpp:

(WebCore::CSSGradientValue::image):
(WebCore::CSSGradientValue::addStops):
(WebCore::positionFromValue):
(WebCore::CSSGradientValue::computeEndPoint):
(WebCore::CSSLinearGradientValue::createGradient):
(WebCore::CSSRadialGradientValue::resolveRadius):
(WebCore::CSSRadialGradientValue::createGradient):

  • css/CSSGradientValue.h:
7:27 AM Changeset in webkit [169257] by Alan Bujtas
  • 5 edits
    3 adds in trunk

Subpixel rendering: Make webkit-box-shadow painting subpixel aware.
https://bugs.webkit.org/show_bug.cgi?id=133201
<rdar://problem/16072830>

Reviewed by Simon Fraser.

This patch enables webkit-box-shadow to be painted on a subpixel position. However, we
still truncate -webkit-box-shadow property values. Tracked here: http://webkit.org/b/133202

Source/WebCore:
Test: fast/box-shadow/hidpi-webkit-box-shadow-subpixel-position.html

  • platform/graphics/FloatRoundedRect.cpp:

(WebCore::FloatRoundedRect::inflateWithRadii): same as in from RoundedRect.
(WebCore::FloatRoundedRect::adjustRadii): same as in from RoundedRect.

  • platform/graphics/FloatRoundedRect.h:
  • rendering/RenderBoxModelObject.cpp:

(WebCore::RenderBoxModelObject::paintBoxShadow):

LayoutTests:
Currently not ref-testable.

  • fast/box-shadow/hidpi-webkit-box-shadow-subpixel-position.html: Added.
  • platform/mac/fast/box-shadow/hidpi-webkit-box-shadow-subpixel-position-expected.png: Added.
  • platform/mac/fast/box-shadow/hidpi-webkit-box-shadow-subpixel-position-expected.txt: Added.
3:29 AM Changeset in webkit [169256] by Carlos Garcia Campos
  • 2 edits in trunk/Tools

[GTK] GObject DOM bindings API break test should fail in the bots when the command fails to run
https://bugs.webkit.org/show_bug.cgi?id=133209

Reviewed by Sergio Villar Senin.

Return FAILURE when the command failed to run.

  • BuildSlaveSupport/build.webkit.org-config/master.cfg:

(RunGtkWebKitGObjectDOMBindingsAPIBreakTests.evaluateCommand):

2:57 AM Changeset in webkit [169255] by Carlos Garcia Campos
  • 8 edits
    3 copies
    5 moves
    7 adds
    1 delete in trunk/Source/WebKit2

[UNIX] Reorganize and cleanup main functions of GTK and EFL ports
https://bugs.webkit.org/show_bug.cgi?id=131024

Reviewed by Gustavo Noronha Silva.

Move main function implementation files to <process-dir>/EntryPoint/unix/<ProcessName>Main.cpp
files that are now shared by GTK and EFL ports. These files call a
main method that wraps a new method ChildProcessMain that contains
the common code and uses a helper class for the platform specific
code. GTK+ and EFL ports implement that helper class for every process type.

  • NetworkProcess/EntryPoint/unix/NetworkProcessMain.cpp: Renamed from Source/WebKit2/efl/MainEfl.cpp.

(main):

  • NetworkProcess/efl/NetworkProcessMainEfl.cpp: Renamed from Source/WebKit2/unix/NetworkMainUnix.cpp.

(WebKit::NetworkProcessMainUnix):

  • NetworkProcess/gtk/NetworkProcessMainGtk.cpp: Renamed from Source/WebKit2/NetworkProcess/unix/NetworkProcessMainUnix.cpp.

(WebKit::NetworkProcessMainUnix):

  • NetworkProcess/unix/NetworkProcessMainUnix.h:
  • PlatformEfl.cmake:
  • PlatformGTK.cmake:
  • PluginProcess/EntryPoint/unix/PluginProcessMain.cpp: Renamed from Source/WebKit2/unix/PluginMainUnix.cpp.

(main):

  • PluginProcess/unix/PluginProcessMainUnix.cpp:

(WebKit::webkitXError):
(WebKit::PluginProcessMainUnix):

  • PluginProcess/unix/PluginProcessMainUnix.h:
  • Shared/unix/ChildProcessMain.cpp: Copied from Source/WebKit2/WebProcess/efl/WebProcessMainEfl.h.

(WebKit::ChildProcessMainBase::parseCommandLine):

  • Shared/unix/ChildProcessMain.h: Renamed from Source/WebKit2/WebProcess/gtk/WebProcessMainGtk.h.

(WebKit::ChildProcessMainBase::platformInitialize):
(WebKit::ChildProcessMainBase::platformFinalize):
(WebKit::ChildProcessMainBase::initializationParameters):
(WebKit::ChildProcessMain):

  • WebProcess/EntryPoint/unix/WebProcessMain.cpp: Renamed from Source/WebKit2/gtk/MainGtk.cpp.

(main):

  • WebProcess/efl/WebProcessMainEfl.cpp:

(WebKit::WebProcessMainUnix):
(dummyExtensionErrorHandler): Deleted.

  • WebProcess/gtk/WebProcessMainGtk.cpp:

(WebKit::WebProcessMainUnix):

  • WebProcess/unix/WebProcessMainUnix.h: Renamed from Source/WebKit2/WebProcess/efl/WebProcessMainEfl.h.
2:09 AM Changeset in webkit [169254] by berto@igalia.com
  • 2 edits in trunk/Tools

[GTK] Install libharfbuzz-dev in Tools/gtk/install-dependencies
https://bugs.webkit.org/show_bug.cgi?id=133210

Reviewed by Carlos Garcia Campos.

  • gtk/install-dependencies:
12:02 AM Changeset in webkit [169253] by Antti Koivisto
  • 2 edits in trunk/Source/WebKit

Try to fix windows build.

  • WebKit.vcxproj/WebKitExportGenerator/WebKitExports.def.in:

May 22, 2014:

11:45 PM Changeset in webkit [169252] by zandobersek@gmail.com
  • 2 edits in trunk/Source/WTF

Add the partial specialization for VectorTraits<std::unique_ptr<P>>
https://bugs.webkit.org/show_bug.cgi?id=133083

Reviewed by Andreas Kling.

  • wtf/VectorTraits.h: Like for RefPtr<P>, OwnPtr<P> and Ref<P>, the partial

specialization for VectorTraits<std::unique_ptr<P>> should exist and should
inherit from SimpleClassVector.

11:44 PM Changeset in webkit [169251] by zandobersek@gmail.com
  • 2 edits in trunk/Source/WebKit2

Have the decoded element in VectorArgumentCoder::decode() moved into Vector::append()
https://bugs.webkit.org/show_bug.cgi?id=133082

Reviewed by Andreas Kling.

  • Platform/IPC/ArgumentCoders.h: Move the decoded element into the Vector::append() call.

This enforces a move instead of a copy if the decoded element's type is move-constructible.

11:43 PM Changeset in webkit [169250] by zandobersek@gmail.com
  • 2 edits in trunk/Source/WebKit2

Move the passed-in Vector<Attachment> in MessageDecoder constructor down into ArgumentDecoder
https://bugs.webkit.org/show_bug.cgi?id=133080

Reviewed by Andreas Kling.

  • Platform/IPC/MessageDecoder.cpp:

(IPC::MessageDecoder::MessageDecoder): The Vector<Attachment> object was passed in by value.
Move it into the ArgumentDecoder constructor as it is not used elsewhere in this constructor
and produces a copy otherwise.

11:36 PM Changeset in webkit [169249] by commit-queue@webkit.org
  • 2 edits in trunk/Source/WebCore

[Curl] Crash when exceeding maximum cache limit.
https://bugs.webkit.org/show_bug.cgi?id=133185

Patch by peavo@outlook.com <peavo@outlook.com> on 2014-05-22
Reviewed by Brent Fulgham.

When the maximum cache limit is exceeded, I get a crash.
This happens when deleting cache entries, because a reference to the url string object in the LRU list
is used as a parameter to invalidateCacheEntry(), when called from makeRoomForNewEntry().
When the string is removed from the LRU list in makeRoomForNewEntry(), the string is deleted.
Next, the string is accessed again to remove the url from the index, and we crash.

This can be fixed by removing the string from the LRU list after it is removed from the index.

Fixing the crash also revealed an infinite loop problem.
If the url for some reason only exist in the LRU list, and not in the index,
we will inifitely loop in makeRoomForNewEntry(), trying to remove this url from the cache, but never succeeding.
This can be fixed by removing the url from the LRU list, also when it's not in the index.

  • platform/network/curl/CurlCacheManager.cpp:

(WebCore::CurlCacheManager::makeRoomForNewEntry): Avoid infinite loop by checking if there are more cache entries to remove.
(WebCore::CurlCacheManager::invalidateCacheEntry): Avoid crash and infinite loop by removing url from LRU list last.

10:18 PM Changeset in webkit [169248] by Simon Fraser
  • 4 edits
    2 adds in trunk

Make viewport units work in CSS gradients
https://bugs.webkit.org/show_bug.cgi?id=133204
<rdar://problem/17012259>

Source/WebCore:

Reviewed by Tim Horton.

Make viewport percentage lengths work in gradients.

Test: fast/gradients/viewport-units-gradient.html

  • css/CSSGradientValue.cpp:

(WebCore::CSSGradientValue::addStops):
(WebCore::CSSLinearGradientValue::createGradient):
(WebCore::CSSRadialGradientValue::createGradient):

  • css/CSSGradientValue.h:

LayoutTests:

Reviewed by Tim Horton.

Make viewport percentage lengths work in gradients.

  • fast/gradients/viewport-units-gradient-expected.html: Added.
  • fast/gradients/viewport-units-gradient.html: Added.
9:34 PM Changeset in webkit [169247] by benjamin@webkit.org
  • 2 edits in trunk/Tools

Add Yusuke Suzuki to the list of contributors (for real)

Wrong section...

Patch by Benjamin Poulain <bpoulain@apple.com> on 2014-05-22

  • Scripts/webkitpy/common/config/contributors.json:
9:30 PM Changeset in webkit [169246] by benjamin@webkit.org
  • 2 edits in trunk/Tools

Add Yusuke Suzuki to the list of contributors

Patch by Benjamin Poulain <bpoulain@apple.com> on 2014-05-22

  • Scripts/webkitpy/common/config/contributors.json:
9:13 PM Changeset in webkit [169245] by benjamin@webkit.org
  • 27 edits in trunk/Source

[iOS][WK2] Add support for minimal-ui viewports
https://bugs.webkit.org/show_bug.cgi?id=133162

Patch by Benjamin Poulain <bpoulain@apple.com> on 2014-05-22
Reviewed by Simon Fraser.

Source/WebCore:
Add support for minimal-ui directly into the Viewport configuration.

ViewportConfiguration knows about two kinds of layout size in scrollview coordinates:
-normal size
-minimal ui size.

Initially, the page is initialized with resetMinimalUI() and starts with a normal layout.

At any time before the main resource is loaded, the Web page can update its viewport meta tag
to include minimal-ui. The viewport configuration is updated immediately to reflect the minimal-ui
layout size. Any layout after that takes minimal-ui into account.

  • WebCore.exp.in:
  • css/LengthFunctions.cpp:

(WebCore::minimumValueForLength):
(WebCore::floatValueForLength):

  • css/StyleResolver.cpp:

(WebCore::StyleResolver::viewportPercentageValue):

  • dom/ViewportArguments.h:

(WebCore::ViewportArguments::operator==):

  • page/FrameView.cpp:

(WebCore::FrameView::setViewportSizeForCSSViewportUnits):
(WebCore::FrameView::viewportSizeForCSSViewportUnits):
(WebCore::FrameView::setViewportSize): Deleted.
(WebCore::FrameView::viewportSize): Deleted.

  • page/FrameView.h:
  • page/ViewportConfiguration.cpp:

(WebCore::ViewportConfiguration::ViewportConfiguration):
(WebCore::ViewportConfiguration::setMinimumLayoutSize):
(WebCore::ViewportConfiguration::setMinimumLayoutSizeForMinimalUI):
(WebCore::ViewportConfiguration::activeMinimumLayoutSizeInScrollViewCoordinates):
(WebCore::ViewportConfiguration::resetMinimalUI):
(WebCore::ViewportConfiguration::pageWillRenderFirstFrame):
(WebCore::ViewportConfiguration::initialScale):
(WebCore::ViewportConfiguration::minimumScale):
(WebCore::ViewportConfiguration::updateConfiguration):
(WebCore::ViewportConfiguration::layoutWidth):
(WebCore::ViewportConfiguration::layoutHeight):
(WebCore::ViewportConfiguration::description):

  • page/ViewportConfiguration.h:

(WebCore::ViewportConfiguration::minimumLayoutSizeForMinimalUI):
(WebCore::ViewportConfiguration::usesMinimalUI):

  • rendering/RenderView.cpp:

(WebCore::RenderView::viewportSizeForCSSViewportUnits):
(WebCore::RenderView::viewportSize): Deleted.

  • rendering/RenderView.h:

Source/WebKit2:
In the WebKit2 layers, we have two parts to minimal-ui.
-In WebPage, we need to get the values from the UIProcess and setup the ViewportConfiguration.

Another part is freezing the state on page load.

-In the UIProcess, we need to get the right information for layout and styling, and we need to tell

the UI clients about minimal-ui changes.

  • UIProcess/API/Cocoa/WKUIDelegatePrivate.h:
  • UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView _setUsesMinimalUI:]):
(-[WKWebView _usesMinimalUI]):
(-[WKWebView _didCommitLayerTree:WebKit::]):
(-[WKWebView _frameOrBoundsChanged]):
(-[WKWebView _minimumLayoutSizeOverride]):
(-[WKWebView _setMinimumLayoutSizeOverride:]):
(-[WKWebView _minimumLayoutSizeOverrideForMinimalUI]):
(-[WKWebView _setMinimumLayoutSizeOverrideForMinimalUI:]):
(-[WKWebView _largestUnobscuredSizeOverride]):
(-[WKWebView _setMaximumUnobscuredSizeOverride:]):
(-[WKWebView _beginAnimatedResizeWithUpdates:]):
(setViewportConfigurationMinimumLayoutSize): Deleted.

  • UIProcess/API/Cocoa/WKWebViewInternal.h:
  • UIProcess/API/Cocoa/WKWebViewPrivate.h:
  • UIProcess/PageClient.h:
  • UIProcess/WebPageProxy.h:
  • UIProcess/WebPageProxy.messages.in:
  • UIProcess/ios/PageClientImplIOS.h:
  • UIProcess/ios/PageClientImplIOS.mm:

(WebKit::PageClientImpl::setUsesMinimalUI):

  • UIProcess/ios/WebPageProxyIOS.mm:

(WebKit::WebPageProxy::dynamicViewportSizeUpdate):
(WebKit::WebPageProxy::setViewportConfigurationMinimumLayoutSizeForMinimalUI):
(WebKit::WebPageProxy::setMaximumUnobscuredSize):
(WebKit::WebPageProxy::setUsesMinimalUI):
(WebKit::WebPageProxy::setMinimumLayoutSizeForMinimalUI): Deleted.

  • WebProcess/WebPage/WebPage.cpp:

(WebKit::WebPage::WebPage):
(WebKit::WebPage::didCommitLoad):

  • WebProcess/WebPage/WebPage.h:
  • WebProcess/WebPage/WebPage.messages.in:
  • WebProcess/WebPage/ios/WebPageIOS.mm:

(WebKit::WebPage::setViewportConfigurationMinimumLayoutSizeForMinimalUI):
(WebKit::WebPage::setMaximumUnobscuredSize):
(WebKit::WebPage::willFlushLayers):
(WebKit::WebPage::dynamicViewportSizeUpdate):
(WebKit::WebPage::viewportConfigurationChanged):
(WebKit::WebPage::updateViewportSizeForCSSViewportUnits):
(WebKit::WebPage::setMinimumLayoutSizeForMinimalUI): Deleted.

  • WebProcess/WebPage/mac/RemoteLayerTreeDrawingArea.mm:

(WebKit::RemoteLayerTreeDrawingArea::flushLayers):

9:06 PM Changeset in webkit [169244] by rniwa@webkit.org
  • 4 edits
    2 deletes in trunk

Can't type in status in facebook.com on iOS Safari because keyboard disappears
https://bugs.webkit.org/show_bug.cgi?id=133196

Reviewed by Geoffrey Garen.

Source/WebCore:
Fixed the bug by rolling out r156252.

When the user taps on the status on facebook.com, the page focuses a textarea,
which is then made momentarily invisible by setting display:none.
The page then removes display:none later and expects the textarea to be still focused.

With r156252, the focus is removed after the page sets display:none and the keyboard disappears.
Since the focus is never reset on the textarea, the user can never type in anything.

The specification may need to change here given that this (rather odd) behavior/expectation exists
on one of the most popular websites on the Web.

  • dom/Document.cpp:

(WebCore::Document::Document):
(WebCore::Document::recalcStyle):
(WebCore::Document::updateLayout):
(WebCore::Document::resetHiddenFocusElementSoon): Deleted.
(WebCore::Document::resetHiddenFocusElementTimer): Deleted.

  • dom/Document.h:

LayoutTests:
Removed the test added by r156252.

  • fast/dom/HTMLDocument/active-element-gets-unfocusable-expected.txt: Removed.
  • fast/dom/HTMLDocument/active-element-gets-unfocusable.html: Removed.
8:59 PM Changeset in webkit [169243] by mmaxfield@apple.com
  • 9 edits in trunk

http/tests/security/xss-DENIED-xsl-document-redirect.xml fails with NetworkProcess
https://bugs.webkit.org/show_bug.cgi?id=132523

Patch by Myles C. Maxfield <mmaxfield@apple.com> on 2014-05-22
Reviewed by Alexey Proskuryakov.

Source/WebCore:
We should not manufacture a response in the event of an error during a
synchronous XHR. In addition, this test removes two places that are
sensitive to such a manufactured response.

Updates test expectations.

  • loader/DocumentThreadableLoader.cpp:

(WebCore::DocumentThreadableLoader::loadRequest): Don't inspect a
loader response if there is an error

  • platform/network/cf/ResourceHandleCFNet.cpp:

(WebCore::ResourceHandle::platformLoadResourceSynchronously): Do not
manufacture a response

  • platform/network/mac/ResourceHandleMac.mm:

(WebCore::ResourceHandle::platformLoadResourceSynchronously): Do not
manufacture a response

  • xml/XSLTProcessorLibxslt.cpp:

(WebCore::docLoaderFunc): Only use the response's URL if no error
occurred.

LayoutTests:
Two tests have been rebaselined, and re-enable test that now passes. Because of
r23889, synchronous XHRs for file: URLs purposely don't invoke error handlers.

  • fast/xmlhttprequest/xmlhttprequest-nonexistent-file-expected.txt: Updated

incorrect test description

  • fast/xmlhttprequest/resources/xmlhttprequest-nonexistent-file-real.html: Ditto
  • platform/mac-wk2/TestExpectations: Re-enable test that now passes.
8:31 PM Changeset in webkit [169242] by mmaxfield@apple.com
  • 2 edits in trunk/Source/WebCore

Unreviewed typo fix.

Reviewed by NOBODY.

No new tests.

  • platform/graphics/mac/FontMac.mm:

(WebCore::Font::dashesForIntersectionsWithRect): Accidentally said run.length() instead of glyphBuffer.size().

8:27 PM Changeset in webkit [169241] by msaboff@apple.com
  • 3 edits in trunk/Tools

Eliminate n/total progress update from run-jsc-stress-tests output to file
https://bugs.webkit.org/show_bug.cgi?id=133191

Reviewed by Geoffrey Garen.

Changed progressMeter setting to be based on stdout being a tty instead of
stdin. Unified the processing of output from the shell runner to match the
same processing as is used by the makefile runner. As part of this, the
shell runner script was simplified. It now forwards the output of each
test_script just like the Makefile and doesn't provide its own progress.

  • Scripts/jsc-stress-test-helpers/shell-runner.sh:
  • Scripts/run-jsc-stress-tests:
8:16 PM Changeset in webkit [169240] by benjamin@webkit.org
  • 2 edits in trunk/Source/WebKit2

[iOS][WK2] Reset the special action-target of WKSyntheticClickTapGestureRecognizer on destruction
https://bugs.webkit.org/show_bug.cgi?id=133165
<rdar://problem/16283914>

Patch by Benjamin Poulain <bpoulain@apple.com> on 2014-05-22
Reviewed by Andy Estes.

It is unclear why, but the WKSyntheticClickTapGestureRecognizer sometimes survives its own view
and is reset after WKContentView was destructed.
This patch cleans up the target-action on destruction to avoid accessing a destructed object.

  • UIProcess/ios/WKContentViewInteraction.mm:

(-[WKContentView cleanupInteraction]):

7:24 PM Changeset in webkit [169239] by benjamin@webkit.org
  • 2 edits in trunk/Source/WebKit2

Do not force a layout when changing the FixedLayoutSize in WebPage
https://bugs.webkit.org/show_bug.cgi?id=133167

Patch by Benjamin Poulain <bpoulain@apple.com> on 2014-05-22
Reviewed by Andreas Kling.

WebPage::setFixedLayoutSize was forcing a layout immediately after changing the layout size on FrameView.
This is forcing plenty of extra layouts since the fixed layout size is one of many properties updated on the FrameView.

FrameView marks that layout is needed. It is not clear why a layout was performed immediately.

  • WebProcess/WebPage/WebPage.cpp:

(WebKit::WebPage::setFixedLayoutSize):

6:47 PM Changeset in webkit [169238] by dburkart@apple.com
  • 6 edits
    1 delete in branches/safari-537.77-branch

Revert r169216.

6:36 PM Changeset in webkit [169237] by akling@apple.com
  • 5 edits in trunk/Source

[iOS WebKit2] Web process should try to shrink its memory footprint when going into background.
<https://webkit.org/b/133197>
<rdar://problem/17011561>

Source/WebCore:
Make releaseMemory() public so we can call it from the process-will-suspend callback.

Reviewed by Gavin Barraclough.

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

Source/WebKit2:
Try to free up as much memory as possible before going into background.

Reviewed by Gavin Barraclough.

  • WebProcess/WebProcess.cpp:

(WebKit::WebProcess::processWillSuspend):

6:34 PM Changeset in webkit [169236] by akling@apple.com
  • 10 edits in trunk/Source

Hook up a setting for showing detailed logging during memory pressure relief.
<https://webkit.org/b/133194>

Source/WebCore:
Make the pressure relief logger opt-in. Also make it flush memory back to the
OS (for both malloc and FastMalloc) to get more accurate numbers at each step.

Reviewed by Gavin Barraclough.

  • WebCore.exp.in:
  • platform/MemoryPressureHandler.cpp:
  • platform/MemoryPressureHandler.h:

(WebCore::MemoryPressureHandler::ReliefLogger::ReliefLogger):
(WebCore::MemoryPressureHandler::ReliefLogger::~ReliefLogger):
(WebCore::MemoryPressureHandler::ReliefLogger::setLoggingEnabled):

  • platform/cocoa/MemoryPressureHandlerCocoa.mm:

(WebCore::MemoryPressureHandler::ReliefLogger::platformMemoryUsage):

Source/WebKit2:
Plumb through the same setting that we used for detailed logging in WK1.

Reviewed by Gavin Barraclough.

  • Shared/WebProcessCreationParameters.cpp:

(WebKit::WebProcessCreationParameters::WebProcessCreationParameters):
(WebKit::WebProcessCreationParameters::encode):
(WebKit::WebProcessCreationParameters::decode):

  • Shared/WebProcessCreationParameters.h:
  • UIProcess/mac/WebContextMac.mm:

(WebKit::WebContext::platformInitializeWebProcess):

  • WebProcess/cocoa/WebProcessCocoa.mm:

(WebKit::WebProcess::platformInitializeWebProcess):

6:29 PM Changeset in webkit [169235] by beidson@apple.com
  • 2 edits in trunk/Source/WebCore

Don't scan for phone numbers in editable regions
<rdar://problem/16949846> and https://bugs.webkit.org/show_bug.cgi?id=133192

Reviewed by Enrica Casucci.

No new tests (Currently untested WK2-only feature)

  • editing/Editor.cpp:

(WebCore::Editor::scanRangeForTelephoneNumbers): Skip the range if the Node is editable.

6:04 PM Changeset in webkit [169234] by commit-queue@webkit.org
  • 3 edits in trunk/Source/WebCore

Captions layout incorrectly in fullscreen.
https://bugs.webkit.org/show_bug.cgi?id=133175

Patch by Jeremy Jones <jeremyj@apple.com> on 2014-05-22
Reviewed by Eric Carlson.

When doing layout for fullscreen, geometry should be absolute so it is not influenced
by the rest of the page. This change adds style for CSSPropertyPosition, CSSPropertyLeft,
and CSSPropertyTop.

  • html/shadow/MediaControlElements.cpp:

(WebCore::MediaControlTextTrackContainerElement::updateDisplay):
Refactor style changes into updateStyleForTextTrackRepresentation().
(WebCore::MediaControlTextTrackContainerElement::updateTimerFired):
Refactor style changes into updateStyleForTextTrackRepresentation().
(WebCore::MediaControlTextTrackContainerElement::clearTextTrackRepresentation):
Refactor style changes into updateStyleForTextTrackRepresentation().
(WebCore::MediaControlTextTrackContainerElement::updateStyleForTextTrackRepresentation):
Consolidate style changes here. Include the two existing and three new style changes.

  • html/shadow/MediaControlElements.h:

Declare updateStyleForTextTrackRepresentation().

5:40 PM Changeset in webkit [169233] by aestes@apple.com
  • 4 edits in trunk/Source/WebKit2

[iOS] Send shareable resources to QuickLook if enabled
https://bugs.webkit.org/show_bug.cgi?id=133189
<rdar://problem/17003995>

Reviewed by Brady Eidson.

Like we do for didReceiveData(), we need to pass an incoming ShareableResource to QuickLook if a QuickLook
handle exists.

  • Shared/ShareableResource.cpp:

(WebKit::ShareableResource::Handle::tryWrapInCFData): Return the shared resource in a CFDataRef.
(WebKit::ShareableResource::Handle::tryWrapInSharedBuffer): Implemented in terms of tryWrapInCFData().

  • Shared/ShareableResource.h:
  • WebProcess/Network/WebResourceLoader.cpp:

(WebKit::WebResourceLoader::didReceiveResource): If a QuickLook handle exists, send the shareable resource to
it via a CFDataRef.

5:39 PM Changeset in webkit [169232] by Simon Fraser
  • 4 edits in trunk/LayoutTests

Fix the compositing/transitions/transform-on-large-layer.html test
https://bugs.webkit.org/show_bug.cgi?id=133195

Reviewed by Dean Jackson.

This test used window resize when it doesn't need to and was flakey.

  • compositing/transitions/transform-on-large-layer.html:
  • platform/efl/TestExpectations:
  • platform/mac/TestExpectations:
5:31 PM Changeset in webkit [169231] by matthew_hanson@apple.com
  • 1 copy in tags/Safari-538.34.46

Create Safari-538.34.46 tag.

5:02 PM Changeset in webkit [169230] by dburkart@apple.com
  • 3 edits
    4 copies in branches/safari-537.77-branch

Merged r169007. <rdar://problem/16975427>

4:57 PM Changeset in webkit [169229] by Simon Fraser
  • 13 edits in trunk

REGRESSION (r155977): Very stuttery 3D css animation on jtechcommunications.com
https://bugs.webkit.org/show_bug.cgi?id=133179
<rdar://problem/16864666>

Reviewed by Dean Jackson.

Source/WebCore:

Revert code added in r155977 to try to pick a good contentsScale for layers
based on a root-relative transform. This resulted in pages with too much
backing store, and too many cases where layers repainted during animations,
causing stutter.

  • platform/graphics/ca/GraphicsLayerCA.cpp:

(WebCore::GraphicsLayerCA::GraphicsLayerCA):
(WebCore::GraphicsLayerCA::flushCompositingState):
(WebCore::GraphicsLayerCA::recursiveCommitChanges):
(WebCore::GraphicsLayerCA::commitLayerChangesBeforeSublayers):
(WebCore::GraphicsLayerCA::updateContentsOpaque):
(WebCore::GraphicsLayerCA::updateAnimations):
(WebCore::GraphicsLayerCA::appendToUncommittedAnimations):
(WebCore::GraphicsLayerCA::setTransformAnimationEndpoints):
(WebCore::GraphicsLayerCA::setTransformAnimationKeyframes):
(WebCore::GraphicsLayerCA::updateContentsScale):
(WebCore::clampedContentsScaleForScale): Deleted.
(WebCore::maxScaleFromTransform): Deleted.
(WebCore::GraphicsLayerCA::updateRootRelativeScale): Deleted.
(WebCore::GraphicsLayerCA::getTransformFromAnimationsWithMaxScaleImpact): Deleted.

  • platform/graphics/ca/GraphicsLayerCA.h:

LayoutTests:

Rebaseline.

  • compositing/contents-scale/animating-expected.txt:
  • compositing/contents-scale/rounded-contents-scale-expected.txt:
  • compositing/contents-scale/scaled-ancestor-expected.txt:
  • compositing/contents-scale/simple-scale-expected.txt:
  • compositing/contents-scale/z-translate-expected.txt:
  • platform/mac/compositing/overflow/composited-scrolling-paint-phases-expected.txt:
  • platform/mac/compositing/tiling/rotated-tiled-clamped-expected.txt:
  • platform/mac/compositing/tiling/rotated-tiled-preserve3d-clamped-expected.txt:
  • platform/mac/compositing/visible-rect/nested-transform-expected.txt:
4:50 PM Changeset in webkit [169228] by Lucas Forschler
  • 14 edits
    12 copies in branches/safari-537.77-branch

Merged r168915. <rdar://problem/16975506>

4:43 PM Changeset in webkit [169227] by dburkart@apple.com
  • 9 edits in branches/safari-537.77-branch/Source/WebCore

Merged r167856. <rdar://problem/16975541>

4:34 PM Changeset in webkit [169226] by Lucas Forschler
  • 3 edits
    2 copies in branches/safari-537.77-branch

Merged r168636. <rdar://problem/16975423>

3:43 PM Changeset in webkit [169225] by Lucas Forschler
  • 12 edits in branches/safari-537.77-branch/Source

Merged r166741. <rdar://problem/16975484>

3:08 PM Changeset in webkit [169224] by dburkart@apple.com
  • 2 edits in branches/safari-537.77-branch/Source/WebCore

Merged r167851. <rdar://problem/16975552>

2:46 PM Changeset in webkit [169223] by mhock@apple.com
  • 12 edits in trunk/Source

[iOS] Enable -apple-system- styled elements to respond to system font size changes.
https://bugs.webkit.org/show_bug.cgi?id=133186
<rdar://problem/16583782>

Reviewed by Enrica Casucci.

Source/WebCore:

  • WebCore.exp.in:
  • WebCore.xcodeproj/project.pbxproj:
  • rendering/RenderThemeIOS.h:
  • rendering/RenderThemeIOS.mm:

(WebCore::_contentSizeCategory):
(WebCore::RenderThemeIOS::contentSizeCategory):
(WebCore::RenderThemeIOS::setContentSizeCategory):

Source/WebKit2:

  • UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView initWithFrame:configuration:]):
(-[WKWebView _contentSizeCategoryDidChange:]):
(-[WKWebView _contentSizeCategory]):

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

(WebKit::WebPageProxy::contentSizeCategoryDidChange):

  • WebProcess/WebPage/WebPage.h:
  • WebProcess/WebPage/WebPage.messages.in:
  • WebProcess/WebPage/ios/WebPageIOS.mm:

(WebKit::WebPage::contentSizeCategoryDidChange):

2:45 PM Changeset in webkit [169222] by dburkart@apple.com
  • 5 edits in branches/safari-537.77-branch

Merged r168641. <rdar://problem/16975502>

2:40 PM Changeset in webkit [169221] by mark.lam@apple.com
  • 3 edits
    2 adds in trunk

REGRESSION(r154797): Debugger crashes when stepping over an uncaught exception.
<https://webkit.org/b/133182>

Reviewed by Oliver Hunt.

Source/JavaScriptCore:
Before r154797, we used to clear the VM exception before calling into the
debugger. After r154797, we don't. This patch will restore this clearing
of the exception before calling into the debugger.

Also added assertions after returning from calls into the debugger to
ensure that the debugger did not introduce any exceptions.

  • interpreter/Interpreter.cpp:

(JSC::unwindCallFrame):
(JSC::Interpreter::unwind):
(JSC::Interpreter::debug):

  • Fixed the assertion here. Interpreter::debug() should never be called with a pending exception. Debugger callbacks for exceptions should be handled by Interpreter::unwind() and Interpreter::unwindCallFrame().

LayoutTests:

  • inspector-protocol/debugger/regress-133182-expected.txt: Added.
  • inspector-protocol/debugger/regress-133182.html: Added.
2:37 PM Changeset in webkit [169220] by dburkart@apple.com
  • 3 edits
    2 copies in branches/safari-537.77-branch

Merged r167818. <rdar://problem/16975502>

2:35 PM Changeset in webkit [169219] by Lucas Forschler
  • 15 edits in branches/safari-537.77-branch/Source

Merged r167635. <rdar://problem/16975536>

2:16 PM Changeset in webkit [169218] by dburkart@apple.com
  • 4 edits in branches/safari-537.77-branch/Source/WebCore

Merged r167264. <rdar://problem/16975518>

2:11 PM Changeset in webkit [169217] by msaboff@apple.com
  • 2 edits in trunk/Tools

Add option to run-jsc-stress-tests to use installed jsc
https://bugs.webkit.org/show_bug.cgi?id=133102

Reviewed by Geoffrey Garen.

Added --no-copy option to not copy a JavaScriptCore framework, but use the one
provied with the --jsc option.

  • Scripts/run-jsc-stress-tests:
2:10 PM Changeset in webkit [169216] by dburkart@apple.com
  • 6 edits
    1 copy in branches/safari-537.77-branch

Merged r166628. <rdar://problem/16975444>

1:56 PM Changeset in webkit [169215] by dburkart@apple.com
  • 1 edit in branches/safari-537.77-branch/LayoutTests/ChangeLog

Fix borked ChangeLog merge from r169213.

1:35 PM Changeset in webkit [169214] by ap@apple.com
  • 2 edits in trunk/Source/WebKit2

[iOS] Add CFBundleAllowMixedLocalizations to WebContentService
https://bugs.webkit.org/show_bug.cgi?id=133187
<rdar://problem/16561084>

Reviewed by Tim Horton.

  • WebProcess/EntryPoint/mac/XPCService/WebContentService/Info-iOS.plist:

We already had it in WebContentService.Development, just not here.

1:14 PM Changeset in webkit [169213] by dburkart@apple.com
  • 6 edits
    2 copies in branches/safari-537.77-branch

Merged r166628.

1:00 PM Changeset in webkit [169212] by dburkart@apple.com
  • 3 edits in branches/safari-537.77-branch/Source/WebCore

Merged r165206.

12:37 PM Changeset in webkit [169211] by dburkart@apple.com
  • 2 edits in branches/safari-537.77-branch/Source/JavaScriptCore

Merged r165162.

12:06 PM Changeset in webkit [169210] by matthew_hanson@apple.com
  • 8 edits in branches/safari-538.34-branch/Source/WebCore

Merge r169209. <rdar://problem/16933311>

11:25 AM Changeset in webkit [169209] by jer.noble@apple.com
  • 8 edits in trunk/Source/WebCore

[MSE] Stored samples are not freed when SourceBuffer is removed from MediaSource
https://bugs.webkit.org/show_bug.cgi?id=133174

Reviewed by Eric Carlson.

Clear out stored MediaSamples from SourceBuffer's TrackBuffer storage when
aborting loading. Also, report the memory cost of those samples, so that the
SourceBuffer will be GCd more readily.

Add a mechanism for reporting the size of a MediaSample:

  • Modules/mediasource/SourceBuffer.h:
  • platform/MediaSample.h:
  • platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:
  • platform/mock/mediasource/MockSourceBufferPrivate.cpp:

Track the memory usage of a SampleMap when adding and removing samples, as well
as allowing the SampleMap to be cleared wholesale:

  • Modules/mediasource/SampleMap.cpp:

(WebCore::SampleMap::clear): Release all stored MediaSamples.
(WebCore::SampleMap::addSample): Update m_totalSize.
(WebCore::SampleMap::removeSample): Ditto.

  • Modules/mediasource/SampleMap.h:

(WebCore::SampleMap::SampleMap): Initialize m_totalSize.
(WebCore::SampleMap::sizeInBytes): Simple accessor.

Clear the stored samples when loading is aborted, and report the extra memory
cost

  • Modules/mediasource/SourceBuffer.cpp:

(WebCore::SourceBuffer::SourceBuffer): Initialize m_reportedExtraMemoryCost.
(WebCore::SourceBuffer::removedFromMediaSource): Clear all stored samples.
(WebCore::SourceBuffer::sourceBufferPrivateAppendComplete): Call reportExtraMemoryCost().
(WebCore::SourceBuffer::reportExtraMemoryCost): Inform the vm of the new

extra memory cost incurred by the object.

11:23 AM Changeset in webkit [169208] by ap@apple.com
  • 2 edits in trunk/Source/WebKit2

[iOS][WK2] Application cache is broken
https://bugs.webkit.org/show_bug.cgi?id=133172
<rdar://problem/16994593>

Reviewed by Andy Estes.

  • UIProcess/mac/WebContextMac.mm:

(WebKit::WebContext::platformDefaultApplicationCacheDirectory): Use a proper
path for application cache database (same as WebKit1).

11:03 AM Changeset in webkit [169207] by enrica@apple.com
  • 4 edits in trunk/Source

REGRESSION (WebKit2): Keyboard should have Search button in duckduckgo.com.
https://bugs.webkit.org/show_bug.cgi?id=133183
<rdar://problem/17004207>

Reviewed by Geoff Garen.

Source/WebCore:
Adding new export.

  • WebCore.exp.in:

Source/WebKit2:
The keyboard type should be search if the input type is search
or if it is text and it is inside a form with an action and either
name, id or title contain the word search. This matches the
heuristics we have in WK1 for iOS.

  • WebProcess/WebPage/ios/WebPageIOS.mm:

(WebKit::WebPage::getAssistedNodeInformation):

10:09 AM Changeset in webkit [169206] by Lucas Forschler
  • 5 edits
    2 copies in branches/safari-537.77-branch

Merged r164950. <rdar://problem/16975484>

10:00 AM Changeset in webkit [169205] by Lucas Forschler
  • 3 edits in branches/safari-537.77-branch/Source/WebCore

Merged r157816. <rdar://problem/16975484>

9:50 AM Changeset in webkit [169204] by Lucas Forschler
  • 2 edits in branches/safari-537.77-branch/Source/WTF

Merged r157717. <rdar://problem/16975484>

9:23 AM Changeset in webkit [169203] by Carlos Garcia Campos
  • 2 edits in trunk/Source/WebCore

REGRESSION(r163712): [GTK] Misspelling and grammar underline marks are no longer drawn
https://bugs.webkit.org/show_bug.cgi?id=133047

Reviewed by Darin Adler.

Change two conditions changed by mistake in r163712.

  • editing/TextCheckingHelper.cpp:

(WebCore::findMisspellings): Enter the loop also when wordStart is 0.
(WebCore::TextCheckingHelper::findFirstMisspelling): Skip the work
when the text is a single character. Also reworked it to use a for
loop to improve the readability.

8:51 AM Changeset in webkit [169202] by Michał Pakuła vel Rutka
  • 2 edits
    1 add in trunk/LayoutTests

Unreviewed EFL gardening

Add test expectations for failing tests.

  • platform/efl-wk2/TestExpectations:
  • platform/efl/editing/apply-style-iframe-crash-expected.txt: Added after r168641.
8:46 AM Changeset in webkit [169201] by Lucas Forschler
  • 9 edits
    2 deletes in branches/safari-537.77-branch/Source/WebCore

Merged r155315.

8:37 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
(diff)
7:57 AM Changeset in webkit [169200] by commit-queue@webkit.org
  • 4 edits
    2 adds in trunk

Video is resumed with old playback rate.
https://bugs.webkit.org/show_bug.cgi?id=132905

Patch by Piotr Grad <p.grad@samsung.com> on 2014-05-22
Reviewed by Philippe Normand.

Source/WebCore:
Setting '0' playback rate is causing pipeline to pause.
GStreamer player impl. exposed this information to upper layers but it should not.
Solution is to hidden such situation behind m_playbackRatePause flag.

Test: media/video-paused-0-rate.html

  • platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.cpp:

(WebCore::MediaPlayerPrivateGStreamer::MediaPlayerPrivateGStreamer):
(WebCore::MediaPlayerPrivateGStreamer::play):
(WebCore::MediaPlayerPrivateGStreamer::pause):
(WebCore::MediaPlayerPrivateGStreamer::doSeek):
(WebCore::MediaPlayerPrivateGStreamer::updatePlaybackRate):
(WebCore::MediaPlayerPrivateGStreamer::paused):
(WebCore::MediaPlayerPrivateGStreamer::setRate):
(WebCore::MediaPlayerPrivateGStreamer::updateStates):

  • platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.h:

LayoutTests:

  • media/video-paused-0-rate-expected.txt: Added.
  • media/video-paused-0-rate.html: Added.
6:05 AM Changeset in webkit [169199] by Antti Koivisto
  • 6 edits
    4 adds in trunk

Text markers don't paint on simple lines
https://bugs.webkit.org/show_bug.cgi?id=133177

Reviewed by Anders Carlsson.

Source/WebCore:

Marker painting code does not yet support simple lines.

Tests: fast/text/mark-matches-rendering-simple-lines.html

fast/text/mark-matches-rendering.html

  • dom/DocumentMarkerController.cpp:

(WebCore::DocumentMarkerController::addMarker):

Force text blocks with markers to use line boxes.

  • testing/Internals.cpp:

(WebCore::Internals::setMarkedTextMatchesAreHighlighted):

Expose this so we can make reftests for marker rendering.

  • testing/Internals.h:
  • testing/Internals.idl:

LayoutTests:

  • fast/text/mark-matches-rendering-expected.html: Added.
  • fast/text/mark-matches-rendering-simple-lines-expected.html: Added.
  • fast/text/mark-matches-rendering-simple-lines.html: Added.
  • fast/text/mark-matches-rendering.html: Added.
5:16 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
(diff)
5:16 AM Changeset in webkit [169198] by Carlos Garcia Campos
  • 6 edits in releases/WebKitGTK/webkit-2.4/Source/WebCore

Merge r168060 - [GStreamer] Use GstMetaVideo
https://bugs.webkit.org/show_bug.cgi?id=132247

Reviewed by Philippe Normand.

In WebKitVideoSink we announce the usage of GstMetaVideo, but we do
not use it when handling the video frames. This might break
some decoders and filters that rely on buffer's meta, rather
that in the caps structures.

This patch enables the use of GstMetaVideo through the GstVideoFrame
API. And it is used everywhere the buffer mapping is required.

Also this patch changes to nullptr where zeros were used.

Also, compile conditionally the video buffer conversion when it is
ARGB/BGRA, since it is only required for the Cairo backend.

No new tests, already covered by current tests.

  • platform/graphics/gstreamer/GStreamerUtilities.cpp:

(WebCore::getVideoSizeAndFormatFromCaps): init the GstVideoInfo before
used and remove caps fixate check since it is done by
gst_video_info_from_caps().

  • platform/graphics/gstreamer/ImageGStreamer.h:
  • platform/graphics/gstreamer/ImageGStreamerCairo.cpp:

(ImageGStreamer::ImageGStreamer): use GstVideoFrame for buffer mapping
and unmapping.
(ImageGStreamer::~ImageGStreamer): ditto.

  • platform/graphics/gstreamer/MediaPlayerPrivateGStreamerBase.cpp:

(WebCore::MediaPlayerPrivateGStreamerBase::updateTexture): ditto.
(WebCore::MediaPlayerPrivateGStreamerBase::currentVideoSinkCaps):
return nullptr if failed.

  • platform/graphics/gstreamer/VideoSinkGStreamer.cpp:

(webkitVideoSinkRender): rely on GstVideoInfo rather than on the
caps. Use GstVideoFrame for buffer mapping and unmapping. Add guards
for buffer transformation, since it's only used by Cairo.
(webkitVideoSinkDispose): remove glib version guards.
(webkitVideoSinkSetCaps): update the value of the private
GstVideoInfo.

5:01 AM Changeset in webkit [169197] by commit-queue@webkit.org
  • 1 edit
    8 adds
    2 deletes in trunk/LayoutTests

[CSS Grid Layout] Split the grid-item-margin-auto-columns-rows.html test.
https://bugs.webkit.org/show_bug.cgi?id=131064

Patch by Javier Fernandez <jfernandez@igalia.com> on 2014-05-22
Reviewed by Benjamin Poulain.

Splitting the "grid-item-margin-auto-columns-rows" into different writing-modes
related tests, so besides increasing clarity, it makes it fit in the 800x600 viewport.

  • fast/css-grid-layout/grid-item-margin-auto-columns-rows-expected.html: Removed.
  • fast/css-grid-layout/grid-item-margin-auto-columns-rows.html: Removed.
  • fast/css-grid-layout/grid-item-margin-auto-columns-rows-horizontal-ltr-expected.html: Added.
  • fast/css-grid-layout/grid-item-margin-auto-columns-rows-horizontal-ltr.html: Added.
  • fast/css-grid-layout/grid-item-margin-auto-columns-rows-horizontal-rtl-expected.html: Added.
  • fast/css-grid-layout/grid-item-margin-auto-columns-rows-horizontal-rtl.html: Added.
  • fast/css-grid-layout/grid-item-margin-auto-columns-rows-vertical-ltr-expected.html: Added.
  • fast/css-grid-layout/grid-item-margin-auto-columns-rows-vertical-ltr.html: Added.
  • fast/css-grid-layout/grid-item-margin-auto-columns-rows-vertical-rtl-expected.html: Added.
  • fast/css-grid-layout/grid-item-margin-auto-columns-rows-vertical-rtl.html: Added.
5:00 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
(diff)
4:59 AM Changeset in webkit [169196] by Carlos Garcia Campos
  • 5 edits in releases/WebKitGTK/webkit-2.4

Merge r169112 - [GTK] WebKitWebPage::send-request always pass a valid pointer for redirected response
https://bugs.webkit.org/show_bug.cgi?id=133119

Reviewed by Sergio Villar Senin.

Source/WebKit2:
The problem is that we were checking whether the passed in
WKURLResponseRef is NULL or not, but it's always a valid object
that can contain a NULL WebCore::ResourceResponse.

  • WebProcess/InjectedBundle/API/gtk/WebKitWebPage.cpp:

(willSendRequestForFrame): Pass NULL as redirect-response
parameter of WebKitWebPage::send-request signal when the
WebCore::ResourceResponse of the passed in WKURLResponseRef is NULL.

Tools:
Check that redirect response parameter of
WebKitWebPage::send-request signal is NULL when not redirecting
and a valid WebKitURIResponse object when redirecting.

  • TestWebKitAPI/Tests/WebKit2Gtk/TestResources.cpp:

(testWebResourceSendRequest):
(serverCallback):

  • TestWebKitAPI/Tests/WebKit2Gtk/WebExtensionTest.cpp:

(sendRequestCallback):

4:54 AM Changeset in webkit [169195] by Manuel Rego Casasnovas
  • 3 edits in trunk/Source/WebCore

[CSS Grid Layout] Guard RenderObject::isRenderGrid() method
https://bugs.webkit.org/show_bug.cgi?id=132380

Reviewed by Benjamin Poulain.

Guard RenderObject::isRenderGrid() method under ENABLE_CSS_GRID_LAYOUT compilation flag.

  • rendering/RenderBox.cpp:

(WebCore::RenderBox::computeLogicalWidthInRegion):

  • rendering/RenderObject.h:
3:53 AM WebKitGTK/2.4.x edited by Carlos Garcia Campos
(diff)
3:52 AM Changeset in webkit [169194] by Carlos Garcia Campos
  • 2 edits in releases/WebKitGTK/webkit-2.4

[Stable] Typo in ReadCommandLineArguments.m4
https://bugs.webkit.org/show_bug.cgi?id=132816

Patch by Milan Crha <mcrha@redhat.com> on 2014-05-22
Reviewed by Carlos Garcia Campos.

  • Source/autotools/ReadCommandLineArguments.m4:
3:37 AM Changeset in webkit [169193] by Carlos Garcia Campos
  • 2 edits in releases/WebKitGTK/webkit-2.4/Source/JavaScriptCore

Merge r169094 - operationOptimize() should defer the GC for a while.
<https://webkit.org/b/133103>

Reviewed by Filip Pizlo.

Currently, operationOptimize() only defers the GC until its end. As a result,
a GC may be triggered just before we return from operationOptimize(), and it may
jettison the optimize codeBlock that we're planning to OSR enter into when we
return from this function. This is because the OSR entry on-ramp code hasn't
been executed yet, and hence, there is not yet a reference to this new codeBlock
from the stack, and there won't be until we've had a chance to return out of
operationOptimize() to run the OSR entry on-ramp code.

This issue is now fixed by using DeferGCForAWhile instead of DeferGC. This
ensures that the GC will be deferred until after the OSR entry on-ramp can be
executed.

  • jit/JITOperations.cpp:
2:52 AM Changeset in webkit [169192] by Carlos Garcia Campos
  • 4 edits
    2 adds in releases/WebKitGTK/webkit-2.4

Merge r168460 - Dragging text from one paragraph to another does not render as expected
https://bugs.webkit.org/show_bug.cgi?id=132633

Reviewed by Darin Adler and Ryosuke Niwa.

Source/WebCore:
When we are dragging and dropping into a content editable field, we detect
if we are trying to put a <p> into an existing <p>, and if so, split the
outer <p> and insert the new <p> as its sibling. However, the outer <p>
might not be editable, so we don't want to do any splitting and inserting
at that location.

Test: editing/pasteboard/drag-drop-paragraph-crasher.html

  • editing/ReplaceSelectionCommand.cpp:

(WebCore::ReplaceSelectionCommand::makeInsertedContentRoundTrippableWithHTMLTreeBuilder):

LayoutTests:
The problem occurs when dragging text that includes a <p> into an editable
area that has a <p> as a parent.

  • editing/pasteboard/drag-drop-paragraph-crasher-expected.txt: Added.
  • editing/pasteboard/drag-drop-paragraph-crasher.html: Added.
2:22 AM WebKitGTK/2.4.x edited by vjaquez@igalia.com
fix style (diff)
2:15 AM WebKitGTK/2.4.x edited by vjaquez@igalia.com
add "[GStreamer] Use GstMetaVideo" (diff)
1:09 AM Changeset in webkit [169191] by Carlos Garcia Campos
  • 2 edits in releases/WebKitGTK/webkit-2.4/Tools

Unreviewed. Fix the build after r168707.

  • Scripts/webkitdirs.pm:

(buildAutotoolsProject): Fix merge conflicts not handled in r168707.

Note: See TracTimeline for information about the timeline view.