Web Inspector: inspector/styles/import-pseudoclass-crash.html intermittently hits an assertion in HTMLLinkElement::startLoadingDynamicSheet
https://bugs.webkit.org/show_bug.cgi?id=116582

Add a flaky crash expectation.

* platform/mac/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150488 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Update Mountain Lion test expectations per bug 116477.

* platform/mac/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150487 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix build.

* WebProcess/Plugins/PluginProcessConnectionManager.cpp:
(WebKit::PluginProcessConnectionManager::getPluginProcessConnection):
Stop using lambdas :(

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150486 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[EFL] Add missing #if guards on drag support methods
https://bugs.webkit.org/show_bug.cgi?id=116574

Reviewed by Christophe Dumez.

* platform/efl/PasteboardEfl.cpp:
(WebCore):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150485 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Provide an API to run sandboxed plug-ins outside of their sandbox
https://bugs.webkit.org/show_bug.cgi?id=116578

Reviewed by Andreas Kling.

* Platform/CoreIPC/HandleMessage.h:
(CoreIPC::callMemberFunction):
Add new overload.

* PluginProcess/EntryPoint/mac/LegacyProcess/PluginProcessMain.mm:
(WebKit::PluginProcessMainDelegate::getExtraInitializationData):
Handle "disable-sandbox".

* PluginProcess/EntryPoint/mac/XPCService/PluginServiceEntryPoint.mm:
(WebKit::PluginServiceInitializerDelegate::getExtraInitializationData):
Handle "disable-sandbox".

* PluginProcess/PluginProcess.h:
Move Type to PluginProcessAttributes and remove hash traits for type.

* PluginProcess/mac/PluginProcessMac.mm:
(WebKit::PluginProcess::platformInitializePluginProcess):
Update for new type name.

(WebKit::PluginProcess::initializeSandbox):
Return early if the sandbox is disabled.

* Shared/Plugins/Netscape/PluginInformation.cpp:
(WebKit::getPluginModuleInformation):
PluginInfoStore::policyForPlugin has been renamed to PluginInfoStore::defaultLoadPolicyForPlugin.

* Shared/Plugins/PluginModuleInfo.h:
Add PluginModuleLoadUnsandboxed.

* Shared/Plugins/PluginProcessAttributes.h: Added.
Add new header with various plug-in process attributes.

* Shared/Plugins/PluginProcessCreationParameters.h:
Change type enum.

* UIProcess/API/C/WKAPICast.h:
(WebKit::toWKPluginLoadPolicy):
(WebKit::toPluginModuleLoadPolicy):
Handle the unsandboxed enum.

* UIProcess/API/C/WKPage.h:
Add kWKPluginLoadPolicyLoadUnsandboxed.

* UIProcess/Plugins/PluginInfoStore.h:
Rename policyForPlugin to defaultLoadPolicyForPlugin.

* UIProcess/Plugins/PluginProcessManager.cpp:
(WebKit::PluginProcessManager::pluginProcessToken):
Add new helper function for getting a token given plug-in attributes.

(WebKit::PluginProcessManager::getPluginProcessConnection):
This now takes a pluginProcessToken.

(WebKit::PluginProcessManager::getSitesWithData):
Make sure to call pluginProcessToken.

(WebKit::PluginProcessManager::clearSiteData):
Ditto.

(WebKit::PluginProcessManager::getOrCreatePluginProcess):
This now takes a token.

* UIProcess/Plugins/PluginProcessProxy.cpp:
(WebKit::PluginProcessProxy::create):
(WebKit::PluginProcessProxy::PluginProcessProxy):
Pass the PluginProcessAttributes as well as the plug-in process token.

(WebKit::PluginProcessProxy::getLaunchOptions):
Pass the plug-in process attributes to platformGetLaunchOptions.

(WebKit::PluginProcessProxy::didClose):
Pass the token to PluginProcessCrashed.

(WebKit::PluginProcessProxy::didFinishLaunching):
Update for type change.

* UIProcess/Plugins/mac/PluginInfoStoreMac.mm:
(WebKit::shouldBlockPlugin):
(WebKit::PluginInfoStore::defaultLoadPolicyForPlugin):
Rename policyForPlugin to defaultLoadPolicyForPlugin.

* UIProcess/Plugins/mac/PluginProcessProxyMac.mm:
(WebKit::PluginProcessProxy::platformGetLaunchOptions):
Set the "disable-sandbox" data if needed.

(WebKit::PluginProcessProxy::platformInitializePluginProcess):
(WebKit::PluginProcessProxy::openPluginPreferencePane):
Get the module info from m_pluginProcessAttributes.

* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::findPlugin):
This now returns a token.

* UIProcess/WebPageProxy.messages.in:
FindPlugin now returns a token instead of the path.

* UIProcess/WebProcessProxy.cpp:
(WebKit::WebProcessProxy::getPluginProcessConnection):
Pass the token to the plug-in process manager.

* UIProcess/WebProcessProxy.messages.in:
This now takes a token.

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

* WebProcess/Plugins/PluginProcessConnection.cpp:
(WebKit::PluginProcessConnection::PluginProcessConnection):
This now takes a token.

* WebProcess/Plugins/PluginProcessConnection.h:
(WebKit::PluginProcessConnection::create):
(WebKit::PluginProcessConnection::pluginProcessToken):
Create this with a token instead of a path and process type.

* WebProcess/Plugins/PluginProcessConnectionManager.cpp:
(WebKit::PluginProcessConnectionManager::getPluginProcessConnection):
This now takes a token.

(WebKit::PluginProcessConnectionManager::pluginProcessCrashed):
Ditto.

* WebProcess/Plugins/PluginProcessConnectionManager.messages.in:
PluginProcessCrashed now takes a token.

* WebProcess/Plugins/PluginProxy.cpp:
(WebKit::PluginProxy::create):
(WebKit::PluginProxy::PluginProxy):
These now take tokens.

(WebKit::PluginProxy::initialize):
Pass the token when getting the connection.

* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::createPlugin):
Get the token from the UI process and pass it to the PluginProxy.

(WebKit::WebPage::canPluginHandleResponse):
Update now that FindPlugin returns a token.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150484 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Remove stub HTMLContentElement
https://bugs.webkit.org/show_bug.cgi?id=116580

Reviewed by Andreas Kling.

Inherit DetailsContentElement and DetailsSummaryElement directly from InsertionPoint instead.

* html/HTMLDetailsElement.cpp:
(WebCore::DetailsContentElement::DetailsContentElement):
(WebCore::DetailsSummaryElement::DetailsSummaryElement):
* html/HTMLSummaryElement.cpp:
(WebCore::SummaryContentElement::SummaryContentElement):
* html/shadow/HTMLContentElement.cpp:
(WebCore::HTMLContentElement::contentTagName):
* html/shadow/HTMLContentElement.h:
(WebCore::isHTMLContentElement):
* html/shadow/InsertionPoint.cpp:
(WebCore::InsertionPoint::setResetStyleInheritance):
* html/shadow/InsertionPoint.h:

    Also remove some leftover selector code.

(WebCore::InsertionPoint::insertionPointType):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150483 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Robustify repaint of previous caret node when moving FrameSelection.
<http://webkit.org/b/116558>

Reviewed by Ryosuke Niwa.

As a follow-up to r150396, also repaint the caret rects if the previous node is contenteditable.

* editing/FrameSelection.cpp:
(WebCore::FrameSelection::recomputeCaretRect):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150482 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Unreviewed build fix after r150469. We need to bail out here when DOM is mutated by an event listener such that
the split text nodes no longer exist right next to each other.

* editing/InsertParagraphSeparatorCommand.cpp:
(WebCore::InsertParagraphSeparatorCommand::doApply):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150481 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Remove ScopeContentDistribution
https://bugs.webkit.org/show_bug.cgi?id=116576

Reviewed by Andreas Kling.

With maximum one ShadowRoot per Element this can be smashed into ContentDistributor.

* dom/ShadowRoot.cpp:
(WebCore):
* dom/ShadowRoot.h:
(WebCore):
* html/shadow/ContentDistributor.cpp:
(WebCore::ContentDistributor::ContentDistributor):
(WebCore):
(WebCore::ContentDistributor::~ContentDistributor):
(WebCore::ContentDistributor::invalidateInsertionPointList):
(WebCore::ContentDistributor::ensureInsertionPointList):
(WebCore::ContentDistributor::distribute):
(WebCore::ContentDistributor::invalidate):
* html/shadow/ContentDistributor.h:
(WebCore):
(ContentDistributor):
* html/shadow/InsertionPoint.cpp:
(WebCore::InsertionPoint::InsertionPoint):
(WebCore::InsertionPoint::insertedInto):
(WebCore::InsertionPoint::removedFrom):

    Simplify insertion point list invalidation.

* html/shadow/InsertionPoint.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150480 268f45cc-cd09-0410-ab3c-d52691b4dbfc

REGRESSION(r150393): editing/inserting/typing-at-end-of-line.html fails
https://bugs.webkit.org/show_bug.cgi?id=116516

Reviewed by Alexey Proskuryakov.

Reset states in DumpRenderTree. The fix for WebKitTestRunner should be implemented in the bug 116577.

* DumpRenderTree/mac/DumpRenderTree.mm:
(createWebViewAndOffscreenWindow): Match the ordering.
(resetWebViewToConsistentStateBeforeTesting): Reset text replacement states.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150479 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[CSS Regions][CSS Exclusions] Multiple regions with shape-insides should respect positioned shapes and overflow
https://bugs.webkit.org/show_bug.cgi?id=115001

Reviewed by David Hyatt.

In r150375 I implemented the new overflow behavior for shape-inside on regions, but it covered only the simple cases
when you had only one region. This patch adds support for the new overflow behavior for multiple regions with multiple
positioned shape-insides and overflow.

Source/WebCore:

Tests: fast/regions/shape-inside/shape-inside-on-regions-block-content-overflow-multiple-shapes.html
       fast/regions/shape-inside/shape-inside-on-regions-inline-content-overflow-multiple-shapes.html

* rendering/RenderBlockLineLayout.cpp:
(WebCore::RenderBlock::updateLineBoundariesForExclusions): Cover cases when content flows into the next region, update the
line segments for the overflow.
(WebCore::RenderBlock::adjustLogicalLineTopAndLogicalHeightIfNeeded): In flow thread case we don't want to adjust the height
in two places.

LayoutTests:

* fast/regions/shape-inside/shape-inside-on-regions-block-content-overflow-multiple-shapes-expected.html: Added.
* fast/regions/shape-inside/shape-inside-on-regions-block-content-overflow-multiple-shapes.html: Added.
* fast/regions/shape-inside/shape-inside-on-regions-inline-content-overflow-multiple-shapes-expected.html: Added.
* fast/regions/shape-inside/shape-inside-on-regions-inline-content-overflow-multiple-shapes.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150478 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[Windows] Reenable storage tests
https://bugs.webkit.org/show_bug.cgi?id=116559

Unreviewed update to TestExpectations:
1. Storage features are present on Windows, so test them.
2. Disable (for now) the four websql tests that fail.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150477 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Unreviewed gardening; rebaseline and re-enable media/track/track-cue-rendering-snap-to-lines-not-set.html

* media/track/track-cue-rendering-snap-to-lines-not-set-expected.txt:
* platform/mac/TestExpectations:
* platform/mac/media/track/track-cue-rendering-snap-to-lines-not-set-expected.txt: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150476 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Unreviewed, rolling out r150425.
http://trac.webkit.org/changeset/150425
https://bugs.webkit.org/show_bug.cgi?id=116573

Not the right fix for WebKit2 (Requested by rniwa on #webkit).

Patch by Commit Queue <commit-queue@webkit.org> on 2013-05-21

* testing/Internals.cpp:
(WebCore::Internals::resetToConsistentState):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150475 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    <rdar://problem/13929704> [WK2] WebKitTestRunner always crashes on exit
        https://bugs.webkit.org/show_bug.cgi?id=116568

        Reviewed by Anders Carlsson.

        * WebKitTestRunner/WebNotificationProvider.cpp:
        (WTR::WebNotificationProvider::~WebNotificationProvider):
        * WebKitTestRunner/WebNotificationProvider.h:
        When the provider is destroyed, it should no longer be registered with the manager.

        * WebKitTestRunner/GeolocationProviderMock.cpp:
        (WTR::GeolocationProviderMock::~GeolocationProviderMock):
        * WebKitTestRunner/GeolocationProviderMock.h:
        I wasn't seeing a crash here, but this code has the same issue as notification
        provider, so applying the same fix.

        * WebKitTestRunner/TestController.cpp: (WTR::TestController::~TestController):
        Close icon database. I started seeing ASSERT(!isOpen()) in WebCore IconDatabase
        destructor having fixed the above.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150474 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Update test expectations after r150472 for the bug 116125.

* platform/wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150473 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Unreviewed, rolling out r150386 and r150397.
http://trac.webkit.org/changeset/150386
http://trac.webkit.org/changeset/150397
https://bugs.webkit.org/show_bug.cgi?id=116572

Broke a test; Alex is going to try again later. (Requested by
thorton on #webkit).

Tools:

* WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp:
(WTR::pathSuitableForTestResult):
(WTR::dumpRequestDescriptionSuitableForTestResult):
(WTR::dumpResponseDescriptionSuitableForTestResult):
(WTR::InjectedBundlePage::willPerformClientRedirectForFrame):
(WTR::InjectedBundlePage::didInitiateLoadForResource):
(WTR::InjectedBundlePage::willSendRequestForFrame):
(WTR::InjectedBundlePage::didReceiveResponseForResource):
* WebKitTestRunner/InjectedBundle/InjectedBundlePage.h:
(InjectedBundlePage):

LayoutTests:

* platform/wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150472 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Unreviewed gardening; rebaseline and re-enable media/track/track-cue-rendering-snap-to-lines-not-set.html
https://bugs.webkit.org/show_bug.cgi?id=116567

* media/track/track-cue-rendering-snap-to-lines-not-set-expected.txt:
* platform/mac/TestExpectations:
* platform/mac/media/track/track-cue-rendering-snap-to-lines-not-set-expected.txt: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150471 268f45cc-cd09-0410-ab3c-d52691b4dbfc

dispatchEvent call can execute javascript and blow away endRoot from underneath
https://bugs.webkit.org/show_bug.cgi?id=116483

Source/WebCore:

Reviewed by Andreas Kling.

Merge https://chromium.googlesource.com/chromium/blink/+/798cba0af9b2aff21e475e2e08ea3ca5e97dfc2c.

Test: editing/undo/undo-after-event-edited.html

* editing/Editor.cpp:
(WebCore::dispatchEditableContentChangedEvents):

LayoutTests:

Reviewed by Andreas Kling.

Add a regression test.

* editing/undo/undo-after-event-edited-expected.txt: Added.
* editing/undo/undo-after-event-edited.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150469 268f45cc-cd09-0410-ab3c-d52691b4dbfc

media/track/track-cue-rendering-snap-to-lines-not-set.html is failing
https://bugs.webkit.org/show_bug.cgi?id=116567

Add a failing test expectation to the test added in r150452.

Patch by Ryosuke Niwa <rniwa@webkit.org> on 2013-05-21

* platform/mac/TestExpectations:
* platform/mac/media/track/track-cue-rendering-snap-to-lines-not-set-expected.txt: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150468 268f45cc-cd09-0410-ab3c-d52691b4dbfc

media/track/track-cue-rendering-snap-to-lines-not-set.html is failing
https://bugs.webkit.org/show_bug.cgi?id=116567

Add a failing test expectation to the test added in r150452.

* platform/mac/TestExpectations:
* platform/mac/media/track/track-cue-rendering-snap-to-lines-not-set-expected.txt: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150467 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Remove ContentDistribution
https://bugs.webkit.org/show_bug.cgi?id=116527

Reviewed by Andreas Kling.

Resolving distribution on traversal is simpler.

* WebCore.exp.in:
* dom/ComposedShadowTreeWalker.cpp:
(WebCore::ComposedShadowTreeWalker::traverseNode):
(WebCore::ComposedShadowTreeWalker::traverseDistributedNodes):
(WebCore::ComposedShadowTreeWalker::traverseSiblingOrBackToInsertionPoint):
* html/HTMLDetailsElement.cpp:
* html/shadow/ContentDistributor.cpp:
(WebCore::ContentDistributor::distributeSelectionsTo):
* html/shadow/ContentDistributor.h:
* html/shadow/HTMLContentElement.idl:

    Remove getDistributedNodes().

* html/shadow/InsertionPoint.cpp:
(WebCore::InsertionPoint::InsertionPoint):
(WebCore::InsertionPoint::attach):
(WebCore::InsertionPoint::detach):
(WebCore):
(WebCore::InsertionPoint::firstDistributed):
(WebCore::InsertionPoint::lastDistributed):
(WebCore::InsertionPoint::nextDistributedTo):
(WebCore::InsertionPoint::previousDistributedTo):

    Rename for clarity.

* html/shadow/InsertionPoint.h:
(WebCore::InsertionPoint::hasDistribution):
(WebCore::InsertionPoint::setHasDistribution):
(WebCore::InsertionPoint::clearDistribution):
(WebCore::InsertionPoint::matchTypeFor):
(InsertionPoint):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150464 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[Windows] Workers are crashing tests.
https://bugs.webkit.org/show_bug.cgi?id=116560.

Unreviewed change to TestExpectations while investigating.

* platform/win/TestExpectations: Disable worker tests under Windows.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150463 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Unreviewed. Add myself as committer in contributors.json.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150462 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix a python unittest after r149419. We need to support Python 2.6 on Windows.

* Scripts/webkitpy/tool/commands/suggestnominations.py:
(AbstractCommitLogCommand._recent_commit_messages):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150461 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[GTK] [CMake] Add support for building WebKit2
https://bugs.webkit.org/show_bug.cgi?id=116372

Reviewed by Gustavo Noronha Silva.

.:

* Source/cmake/FindWebP.cmake: Added.
* Source/cmake/OptionsGTK.cmake: Turn on WebKit2 and the plugin process and also look
for WebP. These missing symbols were hidden up until now. ENABLE_TEXTURE_MAPPER was
also incorrect specified.

Source/WebCore:

* PlatformGTK.cmake: Add missing source files, include directories, and WebP properties.

Source/WebKit2:

* CMakeLists.txt: Add common directories to the common include list and remove UIProcess/PageViewportController.cpp
from the source list since it should only be compiled for WebKitEFL. Instead of taking explicit forwarding-header
dependencies accept a list of "tacked-on" dependencies for WebKit2. This allows GTK+ to specify its extra fake
installed header step.
* PlatformEfl.cmake: Remove common include directories and add UIProcess/PageViewportController.cpp to the source
list. Fix up forwarding header dependencies now.
* PlatformGTK.cmake: Added.
* config.h: First check whether this is a CMake build and then fall back to the autotools header.

Tools:

* CMakeLists.txt: Do not build WebKitTestRunner yet for WebKitGTK+.
* MiniBrowser/gtk/CMakeLists.txt: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150460 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[Windows] Unreviewed test expectation change - skip failing tests.
https://bugs.webkit.org/show_bug.cgi?id=116564

1. Skip a number of failing tests, and investigate the sources of the
problems.  Progress will be tracked on the above bug.
2. Correct a typo I made in r150382
3. Remove skip entries from https://webkit.org/b/113908; these were
replaced with new tests added in https://webkit.org/b/113946.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150459 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[CSSRegions] Remove percentages from offsetLeft/offsetTop tests to avoid sub-pixel errors
https://bugs.webkit.org/show_bug.cgi?id=116544

Patch by Radu Stavila <stavila@adobe.com> on 2013-05-21
Reviewed by Andreas Kling.

* fast/regions/offsetLeft-offsetTop-in-multiple-regions.html:
* fast/regions/offsetLeft-offsetTop-in-region-absolute-sticky-fixed.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150458 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Revert r149635, it was too optimistic.

Rubber-stamped by Andreas Kling.

We are not at the point where we can always use more descriptive macros instead of PLATFORM()

* Scripts/webkitpy/style/checkers/cpp.py:
(check_invalid_increment):
(process_line):
(CppChecker):
* Scripts/webkitpy/style/checkers/cpp_unittest.py:
(CppStyleTest.test_plain_integral_bitfields):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150457 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Unreviewed; fix lldb synthetic representation of WTF::Vector after r148891.

* lldb/lldb_webkit.py:
(WTFVectorProvider.num_children):
(WTFVectorProvider.get_child_index):
(WTFVectorProvider.get_child_at_index):
(WTFVectorProvider.update):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150455 268f45cc-cd09-0410-ab3c-d52691b4dbfc

REGRESSION(r101810): Media controls status text missing for live-stream videos.
https://bugs.webkit.org/show_bug.cgi?id=116547

Reviewed by Eric Carlson.

Un-reverse the logic in loadedMetadata() so that the status text isn't
hidden (when it should be shown) for live-stream videos.

* html/shadow/MediaControlsApple.cpp:
(WebCore::MediaControlsApple::loadedMetadata):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150454 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Implement overlap-avoidance for in-band text track cues.
https://bugs.webkit.org/show_bug.cgi?id=116540

Reviewed by Eric Carlson.

In-band (or Generic) cues need special casing for certain features
present in in-band tracks, like paint-on and roll-up modes. To avoid
the problem of overlap avoidance forcing a caption meant to appear
below a cue to appear above it when a larger font size is selected,
impose an additional sort ordering for "generic cues". Instead of
cues being ordered by the order they appear in the track, "generic
cues" further sorted by their position within the video area, such
that cues at the bottom of the video area appear first, and later
cues are pushed up to avoid them, preserving the desired apparent
ordering.

* html/HTMLMediaElement.cpp:
(WebCore::compareCueInterval): Added; wrapper around
    TextTrackCue::isOrderedBefore.
(WebCore::HTMLMediaElement::updateActiveTextTrackCues):
    After creating the list of current cues, sort them.
* html/track/TextTrackCue.cpp:
(WebCore::TextTrackCue::isOrderedBefore): Added; implementation moved
    from TextTrackCueList::add().
* html/track/TextTrackCue.h:
* html/track/TextTrackCueGeneric.cpp:
(WebCore::TextTrackCueGeneric::isOrderedBefore): Added override;
    impose additional oredring on generic cues.
* html/track/TextTrackCueGeneric.h:
* html/track/TextTrackCueList.cpp:
(WebCore::TextTrackCueList::add): Moved ordering test into
    isOrderedBefore().
* rendering/RenderTextTrackCue.cpp:
(WebCore::RenderTextTrackCue::repositionGenericCue):
    Call repositionCueSnapToLinesNotSet() after positioning the cue.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150453 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Implement overlap avoidance for cues with snap-to-lines flag not set
https://bugs.webkit.org/show_bug.cgi?id=84296

Reviewed by Eric Carlson.

Source/WebCore:

Test: media/track/track-cue-overlap-snap-to-lines-not-set.html

Support overlap avoidance for the non-snap-to-lines part of the WebVTT spec.

* rendering/RenderTextTrackCue.cpp:
(WebCore::RenderTextTrackCue::isOutside): Split implementation into rectIsWithinContainer().
(WebCore::RenderTextTrackCue::rectIsWithinContainer): Ditto.
(WebCore::RenderTextTrackCue::isOverlapping): Split into overlappingObject() and overlappingObjectForRect().
(WebCore::RenderTextTrackCue::overlappingObject): Ditto.
(WebCore::RenderTextTrackCue::overlappingObjectForRect): Ditto.
(WebCore::RenderTextTrackCue::moveIfNecessaryToKeepWithinContainer): Added.
(WebCore::RenderTextTrackCue::findNonOverlappingPosition): When an overlapping object is found, move the
    cue to just above or below that object and try again.
(WebCore::RenderTextTrackCue::repositionCueSnapToLinesSet): Move implementation into moveIfNecessaryToKeepWithinContainer().
(WebCore::RenderTextTrackCue::repositionCueSnapToLinesNotSet): Add implementanton based on above.
* rendering/RenderTextTrackCue.h:

LayoutTests:

* media/track/track-cue-overlap-snap-to-lines-not-set-expected.txt: Added.
* media/track/track-cue-overlap-snap-to-lines-not-set.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150452 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Build fix.

* wtf/NeverDestroyed.h:
(NeverDestroyed):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150451 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Add WTF::NeverDestroyed and start using it in WTF
https://bugs.webkit.org/show_bug.cgi?id=116472

Reviewed by Benjamin Poulain.

NeverDestroyed is a class template that can be used for singletons and other objects that we never
want to destroy. It's intended as a replacement for WTF_STATIC_LOCAL with the advantage that it doesn't
fragment the heap.

* GNUmakefile.list.am:
* WTF.vcproj/WTF.vcproj:
* WTF.vcxproj/WTF.vcxproj:
* WTF.vcxproj/WTF.vcxproj.filters:
* WTF.xcodeproj/project.pbxproj:
Add NeverDestroyed.h

* wtf/Compiler.h:
Add a helper macro, WTF_DELETED_FUNCTION.

* wtf/CryptographicallyRandomNumber.cpp:
Use NeverDestroyed.

* wtf/NeverDestroyed.h: Added.

* wtf/Noncopyable.h:
Use WTF_DELETED_FUNCTION.

* wtf/text/WTFString.cpp:
(WTF::emptyString):
Use

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150450 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Added missing assert condition for PositiveOrZero in ARM branch32().
https://bugs.webkit.org/show_bug.cgi?id=116538.

Reviewed by Geoffrey Garen.

* assembler/MacroAssemblerARM.h:
(JSC::MacroAssemblerARM::branchAdd32):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150449 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Remove GraphicsLayerClient::contentsVisible()
https://bugs.webkit.org/show_bug.cgi?id=116523

Patch by Alberto Garcia <agarcia@igalia.com> on 2013-05-21
Reviewed by Darin Adler.

This is dead code from the BlackBerry port.

* platform/graphics/GraphicsLayerClient.h:
* rendering/RenderLayerBacking.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150446 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[GTK] Compile everything in C++11 mode
https://bugs.webkit.org/show_bug.cgi?id=116452

Reviewed by Anders Carlsson.

.:

* Source/autotools/SetupCompilerFlags.m4: Use the C++11 standard by default when compiling C++ source code.
Perform some minor cleanup around the comments and the order of specifying additional CXXFLAGS entries.

Source/WebKit2:

* GNUmakefile.am: Remove the -std=c++11 flags from the various CPPFLAGS lists, the mode is now specified
by default for all the C++ source code in SetupCompilerFlags.m4.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150445 268f45cc-cd09-0410-ab3c-d52691b4dbfc

PLATFORM(*) macros used in Source/WebCore/loader/archive/ArchiveFactory.cpp
https://bugs.webkit.org/show_bug.cgi?id=116453

Reviewed by Alexey Proskuryakov.

* loader/archive/ArchiveFactory.cpp: Remove a redundant PLATFORM(QT) macro when checking whether
to include the LegacyWebArchive.h header that's specific to the WebArchive support. The Qt port
does not enable the WebArchive feature at all (which is enabled only on AppleMac, AppleWin and iOS
platforms) so there's no reason to additionally condition the header include with building on
non-Qt platforms.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150444 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Add BlackBerry definition of NativeImagePtr
https://bugs.webkit.org/show_bug.cgi?id=116526

Patch by Alberto Garcia <agarcia@igalia.com> on 2013-05-21
Reviewed by Darin Adler.

* platform/graphics/NativeImagePtr.h:
(Graphics):
(WebCore):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150443 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[BlackBerry] FontCache::getFontDataForCharacters() returns nullptr
https://bugs.webkit.org/show_bug.cgi?id=116529

Patch by Alberto Garcia <agarcia@igalia.com> on 2013-05-21
Reviewed by Andreas Kling.

This cannot be converted to a PassRefPtr and breaks the build.

* platform/graphics/blackberry/FontCacheBlackBerry.cpp:
(WebCore::FontCache::getFontDataForCharacters):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150442 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Expose a way to know when forms are added to a page or when form controls are added to a form
in the injected bundle
https://bugs.webkit.org/show_bug.cgi?id=116334

Reviewed by Alexey Proskuryakov.

Source/WebKit2:

Add shouldNotifyOnFormChanges and didAssociateFormControls to the WKBundlePageFormClient.

* WebProcess/InjectedBundle/API/c/WKBundlePage.h:
Add the new callbacks as part of version 2 of the WKBundlePageFormClient.

* WebProcess/InjectedBundle/InjectedBundlePageFormClient.cpp:
(WebKit::InjectedBundlePageFormClient::didAssociateFormControls):
Pass the message along to the client if the client has a handler.
(WebKit::InjectedBundlePageFormClient::shouldNotifyOnFormChanges):
Ditto.
* WebProcess/InjectedBundle/InjectedBundlePageFormClient.h:

* WebProcess/WebCoreSupport/WebChromeClient.cpp:
(WebKit::WebChromeClient::didAssociateFormControls):
Tell the injected bundle form client for the page.
(WebKit::WebChromeClient::shouldNotifyOnFormChanges):
Ditto.
* WebProcess/WebCoreSupport/WebChromeClient.h:

Tools:

Add tests for the new callbacks.

* TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj:
Add DidAssociateFormControls/_Bundle.cpp and associate-form-controls.html

* TestWebKitAPI/Tests/WebKit2/DidAssociateFormControls.cpp: Added.
(TestWebKitAPI::nullJavaScriptCallback):
A "null" callback to handle the fact that WKPageRunJavaScriptInMainFrame cannot handle null
being passed in for the callback.
(TestWebKitAPI::didReceiveMessageFromInjectedBundle):
After receiving the message that didAssociateFormControls callback was invoked from adding
the form in the onload handler, tell the page to add a password field to the form, which
should also invoke didAssociateFormControls callback.
(TestWebKitAPI::setInjectedBundleClient):
Register to receive messages.
(TestWebKitAPI::TEST):
Load associate-form-controls.html and wait until the didAssociateFormControls callback has
been invoked for both adding the form and for adding a password field to the form.

* TestWebKitAPI/Tests/WebKit2/DidAssociateFormControls_Bundle.cpp: Added.
(TestWebKitAPI::shouldNotifyOnFormChanges):
Return true so the didAssociateFormControls callback is invoked.
(TestWebKitAPI::didAssociateFormControls):
Tell the UI process.
(TestWebKitAPI::DidAssociateFormControlsTest::DidAssociateFormControlsTest):
(TestWebKitAPI::DidAssociateFormControlsTest::didCreatePage):
Register for the shouldNotifyOnFormChanges and didAssociateFormControls callbacks.

* TestWebKitAPI/Tests/WebKit2/associate-form-controls.html: Added.
Add a form in response to the onload event. Add a button that will add the password field
for manual testing.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150441 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[GTK] Always use EGL to create the GL context when running on Wayland
https://bugs.webkit.org/show_bug.cgi?id=115721

Patch by Iago Toral Quiroga <itoral@igalia.com> on 2013-05-21
Reviewed by Martin Robinson.

* platform/graphics/cairo/GLContext.cpp:
(WebCore::GLContext::createContextForWindow):
(WebCore::GLContext::createOffscreenContext): Always use EGL to create the GL context
instead of GLX when running on Wayland and make createOffScreenContext reuse the code
in createContextWindow.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150440 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[GTK][WK2] accessibility/label-for-control-hittest.html is failing
https://bugs.webkit.org/show_bug.cgi?id=106341

Patch by Krzysztof Czech <k.czech@samsung.com> on 2013-05-21
Reviewed by Martin Robinson.

Tools:

Adds support for retrieving accessibility element at exact point.

* WebKitTestRunner/InjectedBundle/atk/AccessibilityUIElementAtk.cpp:
(WTR::AccessibilityUIElement::elementAtPoint):

LayoutTests:

Fixing failing test. The reason of failure is missing
possibility of getting accessibility object at exact position.

* platform/gtk-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150439 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[EFL] accessibility/accessibility-node-reparent.html is failing
https://bugs.webkit.org/show_bug.cgi?id=112026

Unreviewed EFL gardening.

accessibility/accessibility-node-reparent.html passes after r150431.

Patch by Krzysztof Czech <k.czech@samsung.com> on 2013-05-21

* platform/efl-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150438 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Add FloatRect::normalized() for BlackBerry
https://bugs.webkit.org/show_bug.cgi?id=116531

Patch by Alberto Garcia <agarcia@igalia.com> on 2013-05-21
Reviewed by Carlos Garcia Campos.

This was added to FloatRectBlackBerry.cpp in r111072 when the
graphics platform code was upstreamed, but this change was
missing.

* platform/graphics/FloatRect.h:
(FloatRect):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150437 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[EFL] accessibility/accessibility-node-memory-management.html is failing
https://bugs.webkit.org/show_bug.cgi?id=112025

Unreviewed EFL gardening

accessibility/accessibility-node-memory-management.html passes after r150431.

Patch by Krzysztof Czech <k.czech@samsung.com> on 2013-05-21

* platform/efl-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150436 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[GStreamer] cleanup duration query
https://bugs.webkit.org/show_bug.cgi?id=116228

Reviewed by Philippe Normand.

Covered by existing tests.

* platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.cpp:
(WebCore::MediaPlayerPrivateGStreamer::duration):
We can also cache the duration here if the query succeeds.
(MediaPlayerPrivateGStreamer::updateStates):
Don't query when we are in GST_STATE_READY state because it never succeeds.
Do it instead when we reached a stable state.
(MediaPlayerPrivateGStreamer::cacheDuration): Return early if it is already
cached. Recaching is not necessary and now we call it a lot of times.
Only look into the state if the query failed.
* platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.h:
(MediaPlayerPrivateGStreamer): Made m_mediaDuration mutable to be
able to cache it in duration(). Internal cached values is one of
the sensible uses of mutable.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150435 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[BlackBerry] Make PagePopup implementation independent from WebCore
https://bugs.webkit.org/show_bug.cgi?id=116448

Patch by Carlos Garcia Campos <cgarcia@igalia.com> on 2013-05-21
Reviewed by Anders Carlsson.

.:

* Source/cmake/OptionsBlackBerry.cmake: Do not enable PAGE_POPUP
feature.

Source/WebKit:

* PlatformBlackBerry.cmake: Add
blackberry/WebCoreSupport/PagePopupBlackBerryClient.cpp to
compilation.

Source/WebKit/blackberry:

Add our own implementation of PagePopupClient and make all the
pickers inherit from it. Unused methods have been removed and
common implementation have been moved from the pickers to the
parent class.

* Api/WebPage.cpp:
(BlackBerry::WebKit::WebPagePrivate::~WebPagePrivate): Destroy the
popup if there's one active.
(BlackBerry::WebKit::WebPagePrivate::setVisible): Call
closePagePopup() directly.
(BlackBerry::WebKit::WebPagePrivate::openPagePopup): Create a new
PagePopupBlackBerry for the given client.
(BlackBerry::WebKit::WebPagePrivate::closePagePopup): Close the
active popup if there's one.
(BlackBerry::WebKit::WebPagePrivate::hasOpenedPopup): Return
whether there's an active popup.
* Api/WebPage.h:
* Api/WebPage_p.h:
* WebCoreSupport/ChromeClientBlackBerry.cpp:
(WebCore::ChromeClientBlackBerry::chromeDestroyed): Remove
closePagePopup() call since this is now handled by
WebPagePrivate.
(WebCore::ChromeClientBlackBerry::hasOpenedPopup): Call
WebPagePrivate::hasOpenedPopup().
* WebCoreSupport/ChromeClientBlackBerry.h:
(ChromeClientBlackBerry):
* WebCoreSupport/ColorPickerClient.cpp:
(BlackBerry::WebKit::ColorPickerClient::ColorPickerClient):
(BlackBerry::WebKit::ColorPickerClient::didClosePopup):
* WebCoreSupport/ColorPickerClient.h:
(ColorPickerClient):
* WebCoreSupport/DatePickerClient.cpp:
(BlackBerry::WebKit::DatePickerClient::DatePickerClient):
(BlackBerry::WebKit::DatePickerClient::didClosePopup):
* WebCoreSupport/DatePickerClient.h:
(DatePickerClient):
* WebCoreSupport/PagePopupBlackBerry.cpp:
(BlackBerry::WebKit::PagePopupBlackBerry::PagePopupBlackBerry):
(BlackBerry::WebKit::PagePopupBlackBerry::closePopup):
* WebCoreSupport/PagePopupBlackBerry.h:
(PagePopupBlackBerry):
(BlackBerry::WebKit::PagePopupBlackBerry::SharedClientPointer::SharedClientPointer):
(BlackBerry::WebKit::PagePopupBlackBerry::SharedClientPointer::get):
(SharedClientPointer):
* WebCoreSupport/PagePopupBlackBerryClient.cpp: Added.
(BlackBerry::WebKit::PagePopupBlackBerryClient::PagePopupBlackBerryClient):
(BlackBerry::WebKit::PagePopupBlackBerryClient::closePopup):
(BlackBerry::WebKit::PagePopupBlackBerryClient::didClosePopup):
(BlackBerry::WebKit::PagePopupBlackBerryClient::contentSize):
(BlackBerry::WebKit::PagePopupBlackBerryClient::writeDocument):
* WebCoreSupport/PagePopupBlackBerryClient.h: Added.
(PagePopupBlackBerryClient):
(BlackBerry::WebKit::PagePopupBlackBerryClient::~PagePopupBlackBerryClient):
* WebCoreSupport/SelectPopupClient.cpp:
(BlackBerry::WebKit::SelectPopupClient::SelectPopupClient):
(BlackBerry::WebKit::SelectPopupClient::didClosePopup):
* WebCoreSupport/SelectPopupClient.h:
(SelectPopupClient):
* WebKitSupport/InputHandler.cpp:
(BlackBerry::WebKit::InputHandler::openDatePopup): Do not call
closePagePopup() since this is already done by openPagePopup() and
use openPagePopup() from WebPagePrivate directly.
(BlackBerry::WebKit::InputHandler::openColorPopup): Ditto.
(BlackBerry::WebKit::InputHandler::openSelectPopup): Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150434 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Build fix for !USE(ACCELERATED_COMPOSITING) after r150307.

* rendering/RenderLayerCompositor.h: Added missing #if USE(ACCELERATED_COMPOSITING).

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150433 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[Qt][WK2] Fix Mountain Lion builds after r150277.
https://bugs.webkit.org/show_bug.cgi?id=116467

Reviewed by Jocelyn Turcotte.

* Shared/API/c/WKBase.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150432 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[ATK] [WebKit2] AccessibilityUIElement often leaks its AtkObject
https://bugs.webkit.org/show_bug.cgi?id=112913

Patch by Krzysztof Czech <k.czech@samsung.com> on 2013-05-21
Reviewed by Martin Robinson.

AccessibilityUIElement is often created by passing new reference to AtkObject (transfer full).
AccessibilityUIElement::~AccessibilityUIElement does not decrement the reference of the element,
so the leak appears.

* WebKitTestRunner/InjectedBundle/AccessibilityController.cpp:
(WTR):
* WebKitTestRunner/InjectedBundle/AccessibilityUIElement.h:
* WebKitTestRunner/InjectedBundle/atk/AccessibilityControllerAtk.cpp:
(WTR::AccessibilityController::rootElement):
(WTR):
(WTR::AccessibilityController::focusedElement):
* WebKitTestRunner/InjectedBundle/atk/AccessibilityUIElementAtk.cpp:
(WTR::checkElementState):
(WTR::indexRangeInTable):
(WTR::alterCurrentValue):
(WTR::AccessibilityUIElement::getChildren):
(WTR::AccessibilityUIElement::getChildrenWithRange):
(WTR::AccessibilityUIElement::childrenCount):
(WTR::AccessibilityUIElement::indexOfChild):
(WTR::AccessibilityUIElement::titleUIElement):
(WTR::AccessibilityUIElement::parentElement):
(WTR::AccessibilityUIElement::allAttributes):
(WTR::AccessibilityUIElement::role):
(WTR::AccessibilityUIElement::title):
(WTR::AccessibilityUIElement::description):
(WTR::AccessibilityUIElement::orientation):
(WTR::AccessibilityUIElement::stringValue):
(WTR::AccessibilityUIElement::language):
(WTR::AccessibilityUIElement::x):
(WTR::AccessibilityUIElement::y):
(WTR::AccessibilityUIElement::width):
(WTR::AccessibilityUIElement::height):
(WTR::AccessibilityUIElement::intValue):
(WTR::AccessibilityUIElement::minValue):
(WTR::AccessibilityUIElement::maxValue):
(WTR::AccessibilityUIElement::isEnabled):
(WTR::AccessibilityUIElement::isFocused):
(WTR::AccessibilityUIElement::isSelected):
(WTR::AccessibilityUIElement::isExpanded):
(WTR::AccessibilityUIElement::isChecked):
(WTR::AccessibilityUIElement::rowCount):
(WTR::AccessibilityUIElement::columnCount):
(WTR::AccessibilityUIElement::rowIndexRange):
(WTR::AccessibilityUIElement::columnIndexRange):
(WTR::AccessibilityUIElement::cellForColumnAndRow):
(WTR::AccessibilityUIElement::increment):
(WTR::AccessibilityUIElement::decrement):
(WTR::AccessibilityUIElement::press):
(WTR::AccessibilityUIElement::documentEncoding):
(WTR::AccessibilityUIElement::documentURI):
(WTR::AccessibilityUIElement::isFocusable):
(WTR::AccessibilityUIElement::isSelectable):
(WTR::AccessibilityUIElement::isMultiSelectable):
(WTR::AccessibilityUIElement::isVisible):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150431 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Simplify Shadow DOM distribution code
https://bugs.webkit.org/show_bug.cgi?id=116454

Reviewed by Andreas Kling.

Remove code supporting nested insertion points and distributing to multiple insertion points.

* html/HTMLDetailsElement.cpp:
(DetailsContentElement):

    Make distribution to <details> and <summary> mutually exclusive.

* html/shadow/ContentDistributor.cpp:
(WebCore):
(WebCore::ContentDistributor::distribute):
(WebCore::ContentDistributor::invalidate):
(WebCore::ContentDistributor::distributeSelectionsTo):
* html/shadow/ContentDistributor.h:
* html/shadow/InsertionPoint.cpp:
(WebCore::resolveReprojection):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150430 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[CSSRegions] Constrain auto-height region computation in a different way
https://bugs.webkit.org/show_bug.cgi?id=116310

Reviewed by Alexandru Chiculita.

After https://bugs.webkit.org/show_bug.cgi?id=74132, the region is a render block. We can use RenderBox::constrainContentBoxLogicalHeightByMinMax
instead of RenderBox::computeReplacedLogicalHeightRespectingMinMaxHeight to better reflect the fact that the region is not a replaced element.
Covered by existing regions tests, since when replaced-based the region had an intrinsic height of 0 and now, block-based, it does not have any children.

* rendering/RenderFlowThread.cpp:
(WebCore::RenderFlowThread::addForcedRegionBreak):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150427 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[Mac] svg/batik/paints/patternRegions-positioned-objects.svg is flaky
https://bugs.webkit.org/show_bug.cgi?id=116519

Add a flaky test expectation.

* platform/mac/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150426 268f45cc-cd09-0410-ab3c-d52691b4dbfc

REGRESSION(r150393): editing/inserting/typing-at-end-of-line.html fails
https://bugs.webkit.org/show_bug.cgi?id=116516

Reviewed by Antti Koivisto.

The bug was caused by DRT not resetting various auto correction states.
Do that in Internals::resetToConsistentState.

* testing/Internals.cpp:
(WebCore::Internals::resetToConsistentState):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150425 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Source/WebKit2:     Rollout r150398.

Tools:     Rollout 150398.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150424 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[GTK] Fix concatenations of string literals that are not C++11-compliant
https://bugs.webkit.org/show_bug.cgi?id=116449

Reviewed by Martin Robinson.

Fix the problematic concatenations of string literals so they are compilable under the C++11 standard. This
is simply a matter of separating the string and the string literal that are being concatenated with a space.

Source/WebCore:

* platform/graphics/gtk/FullscreenVideoControllerGtk.cpp:
* platform/gtk/FileSystemGtk.cpp:
(WebCore::sharedResourcesPath):

Source/WebKit2:

* UIProcess/API/gtk/tests/TestPrinting.cpp:
(testPrintOperationPrint):
* UIProcess/API/gtk/tests/WebExtensionTest.cpp:
(methodCallCallback):

Tools:

* ImageDiff/gtk/ImageDiff.cpp:
(printImage):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150421 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[GTK] Pass GTEST_HAS_RTTI=0 cppflag to TestWebKitAPI programs at compilation-time
https://bugs.webkit.org/show_bug.cgi?id=116450

Reviewed by Martin Robinson.

The -fno-rtti flag is used when compiling C++ source code throughout the project. The TestWebKitAPI
programs fail to compile because of this when using the C++11 standard as the GTest headers assume by
default that the RTTI support is available. This can be turned off by passing GTEST_HAS_RTTI=0 to the
TestWebKitAPI programs when compiling them.

* TestWebKitAPI/GNUmakefile.am:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150420 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Use '__typeof__' keyword to avoid problems when compiling in C++11 standard compliance
https://bugs.webkit.org/show_bug.cgi?id=116447

Reviewed by Darin Adler.

The 'typeof' keyword is a GNU C extension which becomes unavailable when compiling in ANSI or specific standard mode
(such as -std=c++11). The '__typeof__' keyword should be used instead as it's available in all modes both under GCC and Clang.

* wtf/FastMalloc.cpp:
(WTF): Replace the 'typeof' keyword with '__typeof__'.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150419 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[GTK] Support application/x-mimearchive mimetype for MHTML archives instead of message/rfc822
https://bugs.webkit.org/show_bug.cgi?id=116442

Reviewed by Martin Robinson.

* loader/archive/ArchiveFactory.cpp:
(WebCore::archiveMIMETypes): Like EFL and Qt, register the application/x-mimearchive mime type as a MHTML archive type
and stop propagating the message/rfc822 mime type as such. This also removes a couple of PLATFORM ifdefs.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150418 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Minor String fixes in CSS
https://bugs.webkit.org/show_bug.cgi?id=116291

Patch by Benjamin Poulain <bpoulain@apple.com> on 2013-05-20
Reviewed by Darin Adler.

Some trival changes to make up for a bad day :)

* css/CSSImageSetValue.cpp:
(WebCore::CSSImageSetValue::customCssText):
* css/CSSImageValue.cpp:
(WebCore::CSSImageValue::customCssText):
* css/CSSImportRule.cpp:
(WebCore::CSSImportRule::cssText):
* css/CSSInheritedValue.cpp:
(WebCore::CSSInheritedValue::customCssText):
* css/CSSInitialValue.cpp:
(WebCore::CSSInitialValue::customCssText):
* css/CSSPrimitiveValue.cpp:
(WebCore::CSSPrimitiveValue::customCssText):
* css/CSSPropertySourceData.cpp:
(WebCore::CSSPropertySourceData::toString):
* css/CSSSelector.cpp:
(WebCore::CSSSelector::selectorText):
* css/CSSStyleRule.cpp:
(WebCore::CSSStyleRule::generateSelectorText):
* css/CSSStyleSheet.h:
(WebCore::CSSStyleSheet::type):
* css/CSSTimingFunctionValue.cpp:
(WebCore::CSSLinearTimingFunctionValue::customCssText):
(WebCore::CSSCubicBezierTimingFunctionValue::customCssText):
* css/CSSValueList.cpp:
(WebCore::CSSValueList::customCssText):
(WebCore::CSSValueList::customSerializeResolvingVariables):
* css/MediaQueryExp.cpp:
(WebCore::MediaQueryExp::serialize):
* css/WebKitCSSFilterValue.cpp:
(WebCore::WebKitCSSFilterValue::customCssText):
* css/WebKitCSSKeyframesRule.cpp:
(WebCore::StyleRuleKeyframes::findKeyframeIndex):
(WebCore::WebKitCSSKeyframesRule::cssText):
* css/WebKitCSSMatFunctionValue.cpp:
(WebCore::WebKitCSSMatFunctionValue::customCssText):
* css/WebKitCSSMixFunctionValue.cpp:
(WebCore::WebKitCSSMixFunctionValue::customCssText):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150417 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Remove DriverProxy._running_drivers from webkitpy.layout_tests.port
https://bugs.webkit.org/show_bug.cgi?id=116492

Reviewed by Dirk Pranke.

Merge https://chromium.googlesource.com/chromium/blink/+/1af2f0be5a1746784ff114952f29f7b8fa9895f3.

The DriverProxy class had the concept of keeping a pool of started DRTs around (indexed by command line)
so that we didn't necessarily need to restart DRT when the command line for a test changed (e.g., when
starting to run tests in a virtual suite). However, the pool was per-worker (so not shared across the machine)
and we didn't cap the number of processes started, so we would end up with a whole bunch of DRTs (3x-4x
the number of workers started). If anything, this seemed to make the tests slightly slower and probably added
to memory pressure on the machine.

This change removes the pool code; now we'll simply restart DRT when the command line changes. With this change,
we can actually fold the restarting code into Driver itself (and get rid of DriverProxy), but I'll leave that
for a subsequent patch just to make things easier to follow.

* Scripts/webkitpy/port/driver.py:
(DriverProxy.__init__):
(DriverProxy.run_test):
(DriverProxy.has_crashed):
(DriverProxy.stop):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150416 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Change EFL Linux 64-bit Release with EFL Linux 64-bit Release WK1
https://bugs.webkit.org/show_bug.cgi?id=116423

Reviewed by Ryosuke Niwa.

EFL Linux 64-bit Release starts to support WK1's build and layout test from now on.

* BuildSlaveSupport/build.webkit.org-config/config.json:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150415 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Remove remaining link-prerender code
https://bugs.webkit.org/show_bug.cgi?id=116429

Patch by Seokju Kwon <seokju.kwon@gmail.com> on 2013-05-20
Reviewed by Darin Adler.

Tools:

Link prerendering code was removed in r150356.

* Scripts/webkitperl/FeatureList.pm:
* qmake/mkspecs/features/features.pri:

LayoutTests:

Remove test cases related to link-prerender.

* fast/dom/HTMLLinkElement/prerender-insert-after-stop-expected.txt: Removed.
* fast/dom/HTMLLinkElement/prerender-insert-after-stop.html: Removed.
* fast/dom/HTMLLinkElement/prerender-remove-after-stop-expected.txt: Removed.
* fast/dom/HTMLLinkElement/prerender-remove-after-stop.html: Removed.
* fast/dom/HTMLLinkElement/resources/empty.html: Removed.
* fast/dom/HTMLLinkElement/resources/empty2.html: Removed.
* fast/dom/HTMLLinkElement/resources/simple-prerender.html: Removed.
* platform/mac/TestExpectations:
* platform/qt/TestExpectations:
* platform/win/TestExpectations:
* platform/wincairo/TestExpectations:
* platform/wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150413 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Unreviewed, rolling out r150384.
http://trac.webkit.org/changeset/150384
https://bugs.webkit.org/show_bug.cgi?id=116513

Not ready for primetime. (Requested by kling on #webkit).

Patch by Commit Queue <commit-queue@webkit.org> on 2013-05-20

* UIProcess/WebProcessProxy.cpp:
(WebKit::WebProcessProxy::WebProcessProxy):
(WebKit::WebProcessProxy::removeWebPage):
* UIProcess/WebProcessProxy.h:
(WebProcessProxy):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150408 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Versioning.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150407 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Unreviewed EFL gardening. Below tests don't come to crash anymore after r150368.
But, pending-activity.html still comes to crash on wk2.

storage/indexeddb/pending-activity.html
storage/indexeddb/transaction-coordination-across-databases.html

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

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150406 268f45cc-cd09-0410-ab3c-d52691b4dbfc

commit-queue should check for oops in changelog entries
https://bugs.webkit.org/show_bug.cgi?id=116395

Reviewed by Martin Robinson.

Make changelog validation fail when it contains oops!.

* Scripts/webkitpy/tool/bot/commitqueuetask.py:
(CommitQueueTask._validate_changelog):
* Scripts/webkitpy/tool/bot/commitqueuetask_unittest.py:
(CommitQueueTaskTest._run_through_task):
* Scripts/webkitpy/tool/commands/queues_unittest.py:
* Scripts/webkitpy/tool/steps/validatechangelogs.py:
(ValidateChangeLogs.options):
(ValidateChangeLogs._check_changelog_diff):
(ValidateChangeLogs._changelog_contains_oops):
(ValidateChangeLogs.run):
* Scripts/webkitpy/tool/steps/validatechangelogs_unittest.py:
(ValidateChangeLogsTest._assert_start_line_produces_output):
(ValidateChangeLogsTest.test_check_changelog_diff):
(ValidateChangeLogsTest):
(ValidateChangeLogsTest.test_changelog_contains_oops):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150404 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Add tests for named access checks on DOMWindow
https://bugs.webkit.org/show_bug.cgi?id=116512

Reviewed by Geoffrey Garen.

Merge https://chromium.googlesource.com/chromium/blink/+/9207625a100abe9b72333ac5dba02b38ca5f4466

* http/tests/security/xss-DENIED-window-name-alert-expected.txt: Added.
* http/tests/security/xss-DENIED-window-name-alert.html: Added.
* http/tests/security/xss-DENIED-window-name-navigator-expected.txt: Added.
* http/tests/security/xss-DENIED-window-name-navigator.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150403 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Null pointer deference in WebCore::AppendNodeCommand::create
https://bugs.webkit.org/show_bug.cgi?id=116479

Source/WebCore:

Reviewed by Andreas Kling.

Merge https://chromium.googlesource.com/chromium/blink/+/5cb43002a44f67a60ecf5a7ed76de2d0bcf89eb2

DeleteSelection::makeStylingElementsDirectChildrenOfEditableRootToPreventStyleLoss() make style and link elements
to be the direct children of the editable root. However, these style and link elements are not necessary editable
and WebKit crashes when they are not.

Test: editing/deleting/delete-uneditable-style.html

* editing/DeleteSelectionCommand.cpp:
(WebCore::DeleteSelectionCommand::makeStylingElementsDirectChildrenOfEditableRootToPreventStyleLoss):

LayoutTests:

Reviewed by Andreas Kling.

Add a regression test.

* editing/deleting/delete-uneditable-style-expected.txt: Added.
* editing/deleting/delete-uneditable-style.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150402 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Disable SuperRegion
https://bugs.webkit.org/show_bug.cgi?id=116362

Rubber stamped by Geoff Garen.

* heap/Region.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150401 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Remove SharedMemoryGtk.cpp
https://bugs.webkit.org/show_bug.cgi?id=116457

Reviewed by Andreas Kling.

* Platform/gtk/SharedMemoryGtk.cpp: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150400 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Scrollbars still show sometimes even when WKPageSetSuppressScrollbarAnimations()
is set
https://bugs.webkit.org/show_bug.cgi?id=116493
-and corresponding-
<rdar://problem/13912871>

Reviewed by Dean Jackson.

We need to be more aggressive with our approach since AppKit may still ask the
scrollbars to paint here.

scrollbarAnimationsAreSuppressed() will allow us to find out if the setting has
been set.
* page/FrameView.cpp:
(WebCore::FrameView::scrollbarAnimationsAreSuppressed):
* page/FrameView.h:
* platform/ScrollableArea.h:
(WebCore::ScrollableArea::scrollbarAnimationsAreSuppressed):
* rendering/RenderLayer.cpp:
(WebCore::RenderLayer::scrollbarAnimationsAreSuppressed):
* rendering/RenderLayer.h:
(RenderLayer):
* rendering/RenderListBox.cpp:
(WebCore::RenderListBox::scrollbarAnimationsAreSuppressed):
* rendering/RenderListBox.h:

Don’t allow animations if scrollbars are suppressed.
* platform/mac/ScrollAnimatorMac.mm:
(-[WebScrollbarPainterDelegate scrollerImp:animateKnobAlphaTo:duration:]):

Call cancelAnimations() to prevent a just-started animation from continuing.
(WebCore::ScrollAnimatorMac::finishCurrentScrollAnimations):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150399 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Expose a way to know when forms are added to a page or when form controls are added to a form
in the injected bundle
https://bugs.webkit.org/show_bug.cgi?id=116334

Reviewed by Alexey Proskuryakov.

Source/WebKit2:

Add shouldNotifyOnFormChanges and didAssociateFormControls to the WKBundlePageFormClient.

* WebProcess/InjectedBundle/API/c/WKBundlePage.h:
Add the new callbacks as part of version 2 of the WKBundlePageFormClient.

* WebProcess/InjectedBundle/InjectedBundlePageFormClient.cpp:
(WebKit::InjectedBundlePageFormClient::didAssociateFormControls):
Pass the message along to the client if the client has a handler.
(WebKit::InjectedBundlePageFormClient::shouldNotifyOnFormChanges):
Ditto.
* WebProcess/InjectedBundle/InjectedBundlePageFormClient.h:

* WebProcess/WebCoreSupport/WebChromeClient.cpp:
(WebKit::WebChromeClient::didAssociateFormControls):
Tell the injected bundle form client for the page.
(WebKit::WebChromeClient::shouldNotifyOnFormChanges):
Ditto.
* WebProcess/WebCoreSupport/WebChromeClient.h:

Tools:

Add tests for the new callbacks.

* TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj:
Add DidAssociateFormControls/_Bundle.cpp and associate-form-controls.html

* TestWebKitAPI/Tests/WebKit2/DidAssociateFormControls.cpp: Added.
(TestWebKitAPI::nullJavaScriptCallback):
A "null" callback to handle the fact that WKPageRunJavaScriptInMainFrame cannot handle null
being passed in for the callback.
(TestWebKitAPI::didReceiveMessageFromInjectedBundle):
After receiving the message that didAssociateFormControls callback was invoked from adding
the form in the onload handler, tell the page to add a password field to the form, which
should also invoke didAssociateFormControls callback.
(TestWebKitAPI::setInjectedBundleClient):
Register to receive messages.
(TestWebKitAPI::TEST):
Load associate-form-controls.html and wait until the didAssociateFormControls callback has
been invoked for both adding the form and for adding a password field to the form.

* TestWebKitAPI/Tests/WebKit2/DidAssociateFormControls_Bundle.cpp: Added.
(TestWebKitAPI::shouldNotifyOnFormChanges):
Return true so the didAssociateFormControls callback is invoked.
(TestWebKitAPI::didAssociateFormControls):
Tell the UI process.
(TestWebKitAPI::DidAssociateFormControlsTest::DidAssociateFormControlsTest):
(TestWebKitAPI::DidAssociateFormControlsTest::didCreatePage):
Register for the shouldNotifyOnFormChanges and didAssociateFormControls callbacks.

* TestWebKitAPI/Tests/WebKit2/associate-form-controls.html: Added.
Add a form in response to the onload event. Add a button that will add the password field
for manual testing.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150398 268f45cc-cd09-0410-ab3c-d52691b4dbfc

REGRESSION(r150386): [WK2] loader/go-back-cached-main-resource.html fails
https://bugs.webkit.org/show_bug.cgi?id=116491

Add a test expectation.

* platform/wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150397 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Avoid caret repaints if we're not showing carets anyway.
<http://webkit.org/b/116489>

Reviewed by Simon Fraser.

When moving the selection, check if we're inside a contenteditable element (or in caret browsing mode)
before repainting the caret's previous location.

* editing/FrameSelection.cpp:
(WebCore::FrameSelection::recomputeCaretRect):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150396 268f45cc-cd09-0410-ab3c-d52691b4dbfc

position:sticky should stick for the enclosing overflow ancestor
https://bugs.webkit.org/show_bug.cgi?id=100054

Source/WebCore:

Reviewed by Beth Dakin.

Make position: -webkit-sticky be constrained by an enclosing ancestor
with non-visible overflow if there is one, rather than being constained
always by the viewport.

Test: fast/css/sticky/sticky-top-overflow.html

* rendering/RenderBoxModelObject.cpp:
(WebCore::RenderBoxModelObject::computeStickyPositionConstraints): Rename
viewportRect to constrainingRect, since it isn't just about the viewport any more.
Fix an issue where just adding stickyLocation to absContainerFrame.location() was
incorrect when the container is scrolled; we also have to take the scroll offset
into account, and that offset is stored in the layer tree.
(WebCore::RenderBoxModelObject::stickyPositionOffset):
Look for an enclosing layer with overflow clipping, and, if found, use it to
compute the sticky constraint rect.
* rendering/RenderLayer.h: Add a helpful IncludeSelfOrNot, and fix several member functions
to use it place of hard-to-read boolean arguments.
Add enclosingOverflowClipLayer(IncludeSelfOrNot).
(WebCore::RenderLayer::ancestorCompositingLayer):
* rendering/RenderLayer.cpp:
(WebCore::RenderLayer::enclosingOverflowClipLayer): New function, finds the enclosing
layer with overflow clip.
(WebCore::RenderLayer::enclosingCompositingLayer): Use IncludeSelfOrNot.
(WebCore::RenderLayer::enclosingCompositingLayerForRepaint): Ditto.
(WebCore::RenderLayer::enclosingFilterLayer): Ditto.
* rendering/RenderLayerCompositor.cpp:
(WebCore::RenderLayerCompositor::repaintInCompositedAncestor): Use IncludeSelfOrNot.
(WebCore::isViewportConstrainedFixedOrStickyLayer): Renamed from isRootmostFixedOrStickyLayer
and moved up so we can use it in requiresCompositingForPosition().
(WebCore::RenderLayerCompositor::requiresCompositingForPosition): Only make sticky composited if
it's viewport-constrained.
(WebCore::RenderLayerCompositor::updateViewportConstraintStatus):
(WebCore::RenderLayerCompositor::computeStickyViewportConstraints): Declare StickyPositionViewportConstraints
one line down to just before it gets used.

LayoutTests:

Reviewed by Beth Dakin.

Reference test for sticky behavior inside overflow:.

* fast/css/sticky/sticky-top-overflow-expected.html: Added.
* fast/css/sticky/sticky-top-overflow.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150395 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Remove DISABLE_ROUNDED_CORNER_CLIPPING
https://bugs.webkit.org/show_bug.cgi?id=115531

Patch by Jeff Rogers <jrogers@blackberry.com> on 2013-05-20
Reviewed by Andreas Kling.

BlackBerry port is no longer using this.

* rendering/RenderLayer.cpp:
(WebCore::inContainingBlockChain):
(WebCore::RenderLayer::clipToRect):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150394 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Add a test that inserting a character inside a URL doesn't move caret per auto correction
https://bugs.webkit.org/show_bug.cgi?id=116458

Reviewed by Darin Adler.

Add a test. This is broken on Safari 6 but fixed on ToT.

* editing/inserting/inserting-slash-inside-url-with-smart-link-expected.txt: Added.
* editing/inserting/inserting-slash-inside-url-with-smart-link.html: Added.
* platform/mac-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150393 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[GTK][WK2] Unflag tests related to layoutTestController.setSmartInsertDeleteEnabled
https://bugs.webkit.org/show_bug.cgi?id=116433

Unreviewed gardening.

Patch by Manuel Rego Casasnovas <rego@igalia.com> on 2013-05-20

* platform/gtk-wk2/TestExpectations: Unflag tests as they are passing
since r150289.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150392 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[EFL][JHBUILD] Make sure eet is built before evas
https://bugs.webkit.org/show_bug.cgi?id=116440

Reviewed by Gyuyoung Kim.

Mark eet as a dependency of evas since evas requires eet to enable some of its
functionality.

* efl/jhbuild.modules:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150391 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[Mac] [WK2] storage/websql/open-database-creation-callback-isolated-world.html, statement-error-callback-isolated-world.html,
and transaction-success-callback-isolated-world.html are flaky
https://bugs.webkit.org/show_bug.cgi?id=116478

Add flaky test expectations.

* platform/mac-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150390 268f45cc-cd09-0410-ab3c-d52691b4dbfc

svg/batik/text/textEffect.svg is flaky on Mountain Lion builders
https://bugs.webkit.org/show_bug.cgi?id=116477

Add a failing test expectation.

* platform/mac/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150389 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Clients should have a way to extend rendering suppression
https://bugs.webkit.org/show_bug.cgi?id=116463
<rdar://problem/13738496>

Reviewed by Andy Estes.

* WebProcess/InjectedBundle/API/c/WKBundlePage.cpp:
(WKBundlePageExtendIncrementalRenderingSuppression): Added.
(WKBundlePageStopExtendingIncrementalRenderingSuppression): Added.
* WebProcess/InjectedBundle/API/c/WKBundlePagePrivate.h:
These only have any effect if SuppressesIncrementalRendering is also enabled.

* WebProcess/WebCoreSupport/WebFrameLoaderClient.cpp:
(WebKit::WebFrameLoaderClient::transitionToCommittedForNewPage):
Inform the FrameView if there are any outstanding rendering suppression tokens.

* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::WebPage):
(WebKit::WebPage::extendIncrementalRenderingSuppression):
Return the next available rendering suppression token.
Add it to the list of active tokens.
Inform the FrameView that we want to suppress rendering.

(WebKit::WebPage::stopExtendingIncrementalRenderingSuppression):
Remove the given token from the list of active tokens.
Un-suppress rendering if appropriate.

* WebProcess/WebPage/WebPage.h:
(WebKit::WebPage::shouldExtendIncrementalRenderingSuppression):
Return if there are any active rendering suppression tokens.

* WebCore.exp.in:
Export FrameView::setVisualUpdatesAllowedByClient.

* dom/Document.cpp:
(WebCore::Document::setVisualUpdatesAllowed):
Move rendering-suppression watchdog into setVisualUpdatesAllowed(bool).
setVisualUpdatesAllowed(ReadyState) should not re-enable visual updates
if the client is itself disabling visual updates.

(WebCore::Document::visualUpdatesSuppressionTimerFired):
If the watchdog fires and the client is still disabling visual
updates, we should not re-enable visual updates, but instead
wait for the client.

(WebCore::Document::setVisualUpdatesAllowedByClient):
Actually re-enable visual updates if we deferred the re-enabling above
(when the page finishes loading, but the client had them disabled).

* dom/Document.h:
(Document): Add setVisualUpdatesAllowedByClient.
* page/FrameView.cpp:
(WebCore::FrameView::FrameView): Visual updates are allowed by default.
(WebCore::FrameView::setVisualUpdatesAllowedByClient):
Forward through to the document, but we also need to persist the state
here in FrameView so that it survives through navigation.

* page/FrameView.h:
(WebCore::FrameView::setVisualUpdatesAllowedByClient): Added.
(WebCore::FrameView::visualUpdatesAllowedByClient): Added.
Added storage for m_visualUpdatesAllowedByClient.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150388 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[CSS Exclusions] Add CSS parsing support for image URI shape-inside and shape-outside values
https://bugs.webkit.org/show_bug.cgi?id=116349

Reviewed by Alexandru Chiculita.

Source/WebCore:

Add support for parsing image URI values for the shape-inside and shape-outside CSS properties.
Extended the ExclusionShapeValue class to support tracking the property's StyleImage value.

The ExclusionShapeValue::type enums were changed from all upper case to "Intercase" by order
of the style bot.

The existing CSS parsing tests were extended to check URI values.

* css/CSSComputedStyleDeclaration.cpp:
(WebCore::CSSComputedStyleDeclaration::getPropertyCSSValue):
* css/CSSParser.cpp:
(WebCore::CSSParser::parseValue):
* css/DeprecatedStyleBuilder.cpp:
(WebCore::ApplyPropertyExclusionShape::applyValue):
* css/StyleResolver.cpp:
(WebCore::StyleResolver::loadPendingImages):
* rendering/style/ExclusionShapeValue.h:
(WebCore::ExclusionShapeValue::createImageValue):
(ExclusionShapeValue):
(WebCore::ExclusionShapeValue::image):
(WebCore::ExclusionShapeValue::setImage):
(WebCore::ExclusionShapeValue::ExclusionShapeValue):

LayoutTests:

Check that shape values specified with a URI are parsed correctly.

* fast/exclusions/parsing-wrap-shape-inside-expected.txt:
* fast/exclusions/parsing-wrap-shape-outside-expected.txt:
* fast/exclusions/script-tests/parsing-wrap-shape-inside.js:
* fast/exclusions/script-tests/parsing-wrap-shape-outside.js:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150387 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WTR::pathSuitableForTestResult should behave the same as _drt_descriptionSuitableForTestResult so we can unskip tests.
https://bugs.webkit.org/show_bug.cgi?id=116125

Patch by Alex Christensen <achristensen@apple.com> on 2013-05-20
Reviewed by Tim Horton.

Tools:

* WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp:
(WTR::pathSuitableForTestResult):
Changed string operations to match _drt_descriptionSuitableForTestResult.
(WTR::dumpRequestDescriptionSuitableForTestResult):
(WTR::dumpResponseDescriptionSuitableForTestResult):
(WTR::InjectedBundlePage::willPerformClientRedirectForFrame):
(WTR::InjectedBundlePage::didInitiateLoadForResource):
(WTR::InjectedBundlePage::willSendRequestForFrame):
(WTR::InjectedBundlePage::didReceiveResponseForResource):
Pass the main frame's URL to pathSuitableForTestResult.
* WebKitTestRunner/InjectedBundle/InjectedBundlePage.h:
Added WKBundlePageRef to willPerformClientRedirectForFrame.

LayoutTests:

* platform/wk2/TestExpectations:
Unskipped working tests.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150386 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Add a failing test expectation to editing/selection/user-drag-element-and-user-select-none.html
after r150354. The failure is tracked by the bug 116473.

* TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150385 268f45cc-cd09-0410-ab3c-d52691b4dbfc

PPT: Closing tab that is hung or chewing 100% CPU leaves abandoned WebProcess.
<http://webkit.org/b/116464>
<rdar://problem/10103795>

Reviewed by Anders Carlsson.

Give the web process a 10 second chance to exit nicely after closing the last tab belonging to it.
This code only runs if there was something on the page (e.g an unload/beforeunload event handler)
preventing the UI process from killing it right away.

* UIProcess/WebProcessProxy.cpp:
(WebKit::WebProcessProxy::WebProcessProxy):
(WebKit::WebProcessProxy::removeWebPage):
(WebKit::WebProcessProxy::forcefulTerminationTimerFired):
* UIProcess/WebProcessProxy.h:
(WebProcessProxy):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150384 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[CSSRegions] Fix offsetLeft / offsetTop for elements inside named flow
https://bugs.webkit.org/show_bug.cgi?id=115899

Patch by Radu Stavila <stavila@adobe.com> on 2013-05-20
Reviewed by David Hyatt.

Source/WebCore:

Elements in named flows that have the body as their offsetParent, need to compute their
offsetLeft and offsetTop values relative to the body.

Tests: fast/regions/offsetLeft-offsetTop-in-multiple-regions.html
       fast/regions/offsetLeft-offsetTop-in-region-absolute-sticky-fixed.html
       fast/regions/offsetLeft-offsetTop-in-region-float-vert-rl.html
       fast/regions/offsetLeft-offsetTop-in-region-float.html
       fast/regions/offsetLeft-offsetTop-inlines-region-in-element.html

* rendering/RenderBoxModelObject.cpp:
(WebCore::RenderBoxModelObject::adjustedPositionRelativeToOffsetParent):
* rendering/RenderFlowThread.cpp:
(WebCore):
(WebCore::RenderFlowThread::adjustedPositionRelativeToOffsetParent):
* rendering/RenderFlowThread.h:

LayoutTests:

Added tests for the implementation of offsetLeft and offsetTop for objects inside named flows
for which the offsetParent is the body.

* fast/regions/offsetLeft-offsetTop-in-multiple-regions-expected.txt: Added.
* fast/regions/offsetLeft-offsetTop-in-multiple-regions.html: Added.
* fast/regions/offsetLeft-offsetTop-in-region-absolute-sticky-fixed-expected.txt: Added.
* fast/regions/offsetLeft-offsetTop-in-region-absolute-sticky-fixed.html: Added.
* fast/regions/offsetLeft-offsetTop-in-region-float-expected.txt: Added.
* fast/regions/offsetLeft-offsetTop-in-region-float-vert-rl-expected.txt: Added.
* fast/regions/offsetLeft-offsetTop-in-region-float-vert-rl.html: Added.
* fast/regions/offsetLeft-offsetTop-in-region-float.html: Added.
* fast/regions/offsetLeft-offsetTop-inlines-region-in-element-expected.txt: Added.
* fast/regions/offsetLeft-offsetTop-inlines-region-in-element.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150383 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[Windows] Unreviewed TestExpectation cleanup. Reordered lines
to match Mac version to make it easier to understand what features
are uniquely missing in Windows port.

* platform/win/TestExpectations: Reordering of lines, adds some
bug ID's to existing skipped tests.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150382 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Make C API more robust against null contexts
https://bugs.webkit.org/show_bug.cgi?id=116462

Reviewed by Anders Carlsson.

Handle null contexts in a non-crashy way.  It's a bug to ever call the
API with a null context, and the absence of a context means we can't
produce a meaningful result, so we still assert in debug builds.

Now where possible we detect and early return, returning null for any
pointer type, NaN for doubles, and false for any boolean result.

* API/JSBase.cpp:
(JSEvaluateScript):
(JSCheckScriptSyntax):
(JSReportExtraMemoryCost):
* API/JSContextRef.cpp:
(JSContextGetGlobalObject):
(JSContextGetGroup):
(JSContextGetGlobalContext):
(JSContextCreateBacktrace):
* API/JSObjectRef.cpp:
(JSObjectMake):
(JSObjectMakeFunctionWithCallback):
(JSObjectMakeConstructor):
(JSObjectMakeFunction):
(JSObjectMakeArray):
(JSObjectMakeDate):
(JSObjectMakeError):
(JSObjectMakeRegExp):
(JSObjectGetPrototype):
(JSObjectSetPrototype):
(JSObjectHasProperty):
(JSObjectGetProperty):
(JSObjectSetProperty):
(JSObjectGetPropertyAtIndex):
(JSObjectSetPropertyAtIndex):
(JSObjectDeleteProperty):
(JSObjectCopyPropertyNames):
* API/JSValueRef.cpp:
(JSValueGetType):
(JSValueIsUndefined):
(JSValueIsNull):
(JSValueIsBoolean):
(JSValueIsNumber):
(JSValueIsString):
(JSValueIsObject):
(JSValueIsObjectOfClass):
(JSValueIsEqual):
(JSValueIsStrictEqual):
(JSValueIsInstanceOfConstructor):
(JSValueMakeUndefined):
(JSValueMakeNull):
(JSValueMakeBoolean):
(JSValueMakeNumber):
(JSValueMakeString):
(JSValueMakeFromJSONString):
(JSValueCreateJSONString):
(JSValueToBoolean):
(JSValueToNumber):
(JSValueToStringCopy):
(JSValueToObject):
(JSValueProtect):
* API/JSWeakObjectMapRefPrivate.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150381 268f45cc-cd09-0410-ab3c-d52691b4dbfc

fix w3c test importer unit tests after r150324
https://bugs.webkit.org/show_bug.cgi?id=116459

Reviewed by Ryosuke Niwa.

This patch fixes the unit tests to not actually call out to
'hg' (by using a mock executive) and also changes the real
directory we crawl from Source/WebCore/css to
Tools/Scripts/webkitpy/w3c to run much more quickly. We should
still change this to a mock filesystem, but that's a separate change.

* Scripts/webkitpy/w3c/test_importer.py:
(main):
(TestImporter.__init__):
(TestImporter.load_changeset):
* Scripts/webkitpy/w3c/test_importer_unittest.py:
(TestImporterTest.test_import_dir_with_no_tests_and_no_hg):
(TestImporterTest.test_import_dir_with_no_tests):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150380 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Add helper function for converting a KeyedCodingValue to a CFTypeRef
https://bugs.webkit.org/show_bug.cgi?id=116461

Reviewed by Tim Horton.

* Shared/cf/KeyedCodingValueCF.cpp: Added.
* Shared/cf/KeyedCodingValueCF.h: Added.

* WebKit2.xcodeproj/project.pbxproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150379 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[Qt] REGRESSION(r149457): some fast/exclusions tests are failing
https://bugs.webkit.org/show_bug.cgi?id=115511

Reviewed by Andreas Kling.

Unskipping the exclusions tests on Qt. They were fixed by https://bugs.webkit.org/show_bug.cgi?id=116115.

* platform/qt/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150378 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Begin stubbing out a new KeyedEncoder class
https://bugs.webkit.org/show_bug.cgi?id=116456

Reviewed by Sam Weinig.

This class (and it's coming counterpart, KeyedDecoder) will be used for serializing
object graphs to a tree-like structure of dictionaries. It is up to individual ports to
serialize them further (for example, on Mac they could be serialized to CFPropertyList objects).

This will be used for serializing session storage state but I hope that it can be used for serializing
back forward state in the future as our current implementation of back forward state serialization
is fragile and depends on our IPC layer's binary encoding(!).

* Shared/KeyedCodingValue.h: Added.
* Shared/KeyedEncoder.cpp: Added.
* Shared/KeyedEncoder.h: Added.

* WebKit2.xcodeproj/project.pbxproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150377 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Unreviewed EFL build fix.

Fix linking errors due to undefined symbols on Ubuntu 13.04.

* DumpRenderTree/efl/CMakeLists.txt:
* EWebLauncher/CMakeLists.txt:
* MiniBrowser/efl/CMakeLists.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150376 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[CSS Regions][CSS Exclusions] shape-inside on regions should respect positioned shapes and overflow
https://bugs.webkit.org/show_bug.cgi?id=116252

Reviewed by David Hyatt.

Regions should respect shape-inside properties which have specified top offset. Since the content which overflows from the
shape should be pushed after the content box (for details check out r148975), I implemented  overflowthe behavior for simple
cases when you have only one region. I'm going to implement it for additional regions in a follow up patch.

Tests: fast/regions/shape-inside/shape-inside-on-regions-block-content-basic-overflow-shape-top-offset.html
       fast/regions/shape-inside/shape-inside-on-regions-inline-content-basic-overflow-shape-top-offset.html

Source/WebCore:

* rendering/RenderBlockLineLayout.cpp:
(WebCore::RenderBlock::updateLineBoundariesForExclusions): Handle the flow thread case, push the overflowing content after the
content box. Respect existing shape-inside on region tests.
(WebCore::RenderBlock::layoutRunsAndFloatsInRange): Set the correct logical top position for the shape-inside in
the case of regions. Respect the region's margins and paddings.

LayoutTests:

* fast/regions/shape-inside/shape-inside-on-regions-block-content-basic-overflow-shape-top-offset-expected.html: Added.
* fast/regions/shape-inside/shape-inside-on-regions-block-content-basic-overflow-shape-top-offset.html: Added.
* fast/regions/shape-inside/shape-inside-on-regions-inline-content-basic-overflow-shape-top-offset-expected.html: Added.
* fast/regions/shape-inside/shape-inside-on-regions-inline-content-basic-overflow-shape-top-offset.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150375 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[GTK] Cleanup TestExpectations to pass --lint-test-files
https://bugs.webkit.org/show_bug.cgi?id=116419

Unreviewed GTK gardening.

Remove one of two http/tests/cache/willsendrequest-returns-null-for-memory-cache-load.html

Patch by Seokju Kwon <seokju.kwon@gmail.com> on 2013-05-20

* platform/gtk/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150374 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Remove custom binding for the Clipboard clearData function
https://bugs.webkit.org/show_bug.cgi?id=116421

Reviewed by Andreas Kling.

* bindings/js/JSClipboardCustom.cpp: Removed the clearData binding.

* dom/Clipboard.idl: Removed [Custom].

* dom/Clipboard.cpp:
(WebCore::Clipboard::clearData): Renamed from clearAllData.
* dom/Clipboard.h: Rename clearAllData to clearData so we just use
overloading for the one with and without a MIME type.
* platform/ios/ClipboardIOS.h: Rename clearAllData to clearData.
* platform/ios/ClipboardIOS.mm:
(WebCore::ClipboardIOS::clearData): Ditto.
* platform/qt/ClipboardQt.cpp:
(WebCore::ClipboardQt::clearData). Ditto.
* platform/qt/ClipboardQt.h: Ditto.
* platform/win/ClipboardWin.cpp:
(WebCore::ClipboardWin::clearData): Ditto.
* platform/win/ClipboardWin.h: Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@150373 268f45cc-cd09-0410-ab3c-d52691b4dbfc