⚠ Archived content — this site is no longer maintained.   Current WebKit documentation is at docs.webkit.org.

Timeline



Jul 11, 2015:

11:43 PM Changeset in webkit [186728] by Gyuyoung Kim
  • 2 edits in trunk/LayoutTests

[EFL] Unreviewed, EFL gardening on 12th Jul.

Mark some ax tests and editing tests to failure since r186694 and r185692.

  • platform/efl/TestExpectations:
10:37 PM Changeset in webkit [186727] by bshafiei@apple.com
  • 2 edits in trunk/Source/WebCore

Unreviewed build fix.

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

(WebCore::MediaPlayerPrivateMediaSourceAVFObjC::load):
(WebCore::MediaPlayerPrivateMediaSourceAVFObjC::cancelLoad):

9:28 PM Changeset in webkit [186726] by bshafiei@apple.com
  • 5 edits in tags/Safari-601.1.39.0.1/Source

Versioning.

9:24 PM Changeset in webkit [186725] by bshafiei@apple.com
  • 1 copy in tags/Safari-601.1.39.0.1

New tag.

9:23 PM Changeset in webkit [186724] by Nikita Vasilyev
  • 10 edits in trunk/Source

Web Inspector: Inspector should be able to be docked to the bottom of a narrow window
​https://bugs.webkit.org/show_bug.cgi?id=146871

Reviewed by Timothy Hatcher.

Source/WebCore:

  • inspector/InspectorFrontendClientLocal.cpp:

Source/WebInspectorUI:

  • UserInterface/Views/Toolbar.css:

(@media (max-width: 539px)): Hide download button to fit search field into the screen.

Source/WebKit/mac:

  • WebCoreSupport/WebInspectorClient.mm:

Source/WebKit2:

  • UIProcess/WebInspectorProxy.cpp:
  • UIProcess/mac/WebInspectorProxyMac.mm:

(WebKit::WebInspectorProxy::platformCanAttach):

  • WebProcess/WebPage/WebInspector.cpp:
9:14 PM Changeset in webkit [186723] by bshafiei@apple.com
  • 5 edits in trunk/Source

Versioning.

8:58 PM Changeset in webkit [186722] by bshafiei@apple.com
  • 1 copy in branches/safari-601.1-branch

New Branch.

7:59 PM Changeset in webkit [186721] by timothy@apple.com
  • 1 edit
    3 adds in trunk/Websites/webkit.org

Add some images for a blog post.

  • blog-files/inspector-elements-network-tabs.png: Added.
  • blog-files/inspector-tab-bar.png: Added.
  • blog-files/inspector-tab-icons.svg: Added.
6:54 PM Changeset in webkit [186720] by Joseph Pecoraro
  • 2 edits in trunk/Source/WTF

Unreviewed build fix, restrict APP_LINKS to just iOS.

  • wtf/Platform.h:
5:51 PM WebKitGTK/Releasing edited by Martin Robinson
Warning that contents are out of date (diff)
3:58 PM Changeset in webkit [186719] by Joseph Pecoraro
  • 9 edits in trunk/Source/WebKit2

Allow clients to opt-out of default app link link actions
​https://bugs.webkit.org/show_bug.cgi?id=146883
<rdar://problem/21221902>

Reviewed by Dan Bernstein.

  • UIProcess/API/APIUIClient.h:

(API::UIClient::shouldIncludeAppLinkActionsForElement):

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

(WebKit::UIDelegate::setDelegate):
(WebKit::UIDelegate::UIClient::shouldIncludeAppLinkActionsForElement):
Plumbing for a new delegate message to ask the client if they
would like app link actions or not. Default is yes.

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

(-[WKActionSheetAssistant showImageSheet]):
(-[WKActionSheetAssistant showLinkSheet]):
(-[WKActionSheetAssistant defaultActionsForLinkSheet:]):
(-[WKActionSheetAssistant defaultActionsForImageSheet:]):
Pass elementInfo into methods generating default actions.

(-[WKActionSheetAssistant _appendOpenActionsForURL:actions:elementInfo:]):
When generating open actions, if the process has AppLink capabilities
ask the client if they want to include AppLink actions or fall back
to the basic Open action.

  • UIProcess/ios/WKContentViewInteraction.mm:

(-[WKContentView actionSheetAssistant:shouldIncludeAppLinkActionsForElement:]):
(-[WKContentView previewViewControllerForPosition:inSourceView:]):

  • UIProcess/ios/WKPDFView.mm:

(-[WKPDFView actionSheetAssistant:shouldIncludeAppLinkActionsForElement:]):
Assistant delegate implementations forward to the UIDelegate.

3:58 PM Changeset in webkit [186718] by Joseph Pecoraro
  • 11 edits
    1 move in trunk/Source

Update default link action sheets for app links
​https://bugs.webkit.org/show_bug.cgi?id=146658
<rdar://problem/21221902>

Reviewed by Dan Bernstein.

Source/WebCore:

  • English.lproj/Localizable.strings:
  • platform/spi/ios/LaunchServicesSPI.h:

Source/WebKit2:

  • Shared/mac/SandboxUtilities.h:
  • Shared/mac/SandboxUtilities.mm: Renamed from Source/WebKit2/Shared/mac/SandboxUtilities.cpp.

(WebKit::processHasEntitlement):

  • UIProcess/ApplicationStateTracker.mm:

(WebKit::applicationType):
(WebKit::hasEntitlement): Deleted.

  • WebKit2.xcodeproj/project.pbxproj:

Share a helper for checking if the current process has an entitlement.

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

(+[_WKElementAction _elementActionWithType:title:actionHandler:]):
(+[_WKElementAction _elementActionWithType:customTitle:assistant:]):

  • UIProcess/API/Cocoa/_WKElementActionInternal.h:

Add an internal way to make an element action with a standard type and regular handler.
We use this for the new standard types that don't have a default implementation.

  • UIProcess/ios/WKActionSheetAssistant.mm:

(applicationHasAppLinkEntitlements):
(appLinkForURL):
Helpers checking access to app links and fetching app links.

(-[WKActionSheetAssistant _appendOpenActionsForURL:actions:]):
(-[WKActionSheetAssistant defaultActionsForLinkSheet]):
(-[WKActionSheetAssistant defaultActionsForImageSheet]):
Create "Open in Safari" and "Open in 'App'" default actions for AppLinks,
otherwise add the normal default "Open" action.

3:29 PM Changeset in webkit [186717] by Devin Rousso
  • 3 edits in trunk/Source/WebInspectorUI

Web Inspector: Improve runtime of pseudo-element sidebar style ordering
​https://bugs.webkit.org/show_bug.cgi?id=146866

Reviewed by Timothy Hatcher.

  • UserInterface/Models/CSSRule.js:

(WebInspector.CSSRule.prototype.update): Determines the most specific selector and saves it to a variable.
(WebInspector.CSSRule.prototype.get mostSpecificSelector): Returns the most specific selector.
(WebInspector.CSSRule.prototype.selectorIsGreater): Compares the most specific selector to a given selector.
(WebInspector.CSSRule.prototype._determineMostSpecificSelector):
Searches through the selector list to find and return the selector that is the most specific.
(WebInspector.CSSRule):

  • UserInterface/Views/RulesStyleDetailsPanel.js:

(WebInspector.RulesStyleDetailsPanel.prototype.refresh):

2:30 PM Changeset in webkit [186716] by enrica@apple.com
  • 2 edits in trunk/Source/WebKit2

Remove incorrect overriding of inputAssistantItem in WKContentView.
​https://bugs.webkit.org/show_bug.cgi?id=146863
rdar://problem/21507154

Reviewed by Sam Weinig.

Correction to the patch in r186713.
We need to provide the inputAssistantItem from the responder.

  • UIProcess/ios/WKContentViewInteraction.mm:

(-[WKContentView inputAccessoryView]):

1:35 PM Changeset in webkit [186715] by benjamin@webkit.org
  • 4 edits in trunk/Source/WebCore

Fix CONTENT_EXTENSIONS_STATE_MACHINE_DEBUGGING
​https://bugs.webkit.org/show_bug.cgi?id=146879

Reviewed by Andreas Kling.

Some of my recent changes broke CONTENT_EXTENSIONS_STATE_MACHINE_DEBUGGING :(

  • contentextensions/CombinedURLFilters.cpp:

(WebCore::ContentExtensions::prefixTreeVertexToString):
(WebCore::ContentExtensions::recursivePrint):
(WebCore::ContentExtensions::CombinedURLFilters::print):

  • contentextensions/ContentExtensionsDebugging.h:
  • contentextensions/DFA.cpp:

(WebCore::ContentExtensions::DFA::debugPrintDot):

1:15 PM Changeset in webkit [186714] by Chris Dumez
  • 5 edits in trunk

Unreviewed, rolling out r186689 and r186710.
​https://bugs.webkit.org/show_bug.cgi?id=146880

May have caused a PLT regression and broken a layout test
(Requested by cdumez on #webkit).

Reverted changesets:

"[WK2] Increase the QoS of some of our WorkQueues to match the
one of our processes"
​https://bugs.webkit.org/show_bug.cgi?id=146855
​http://trac.webkit.org/changeset/186689

"REGRESSION (r186689?): fast/frames/frame-limit.html timeout
on Mavericks {Release,Debug} WK2 bots"
​https://bugs.webkit.org/show_bug.cgi?id=146876
​http://trac.webkit.org/changeset/186710

Patch by Commit Queue <​commit-queue@webkit.org> on 2015-07-11

12:02 PM Changeset in webkit [186713] by enrica@apple.com
  • 2 edits in trunk/Source/WebKit2

Remove incorrect overriding of inputAssistantItem in WKContentView.
​https://bugs.webkit.org/show_bug.cgi?id=146863
rdar://problem/21507154

Reviewed by Dan Bernstein.

Removing implementation of inputAssistantItem and moving initialization
of formAccessoryView to inputAccessoryView to avoid interfering with
the keyboard handling of the assistant bar.

  • UIProcess/ios/WKContentViewInteraction.mm:

(-[WKContentView inputAccessoryView]):
(-[WKContentView inputAssistantItem]): Deleted.
(-[WKContentView _inputAssistantItem]): Deleted.

11:19 AM Changeset in webkit [186712] by Simon Fraser
  • 22 edits in trunk/Source

[iOS WK2] Scrolling issues on horizontally scrollable RTL pages
​https://bugs.webkit.org/show_bug.cgi?id=146872
rdar://problem/7569416

Reviewed by Sam Weinig.

Horizontally scrollable RTL pages in WebKit2 had a variety of issues: they had
a gap down the right edge, and unreachable content on the left side, focussing
form fields scrolled to the wrong location, and programmatic scrolls scrolled
to the wrong place.

Fix by plumbing the WebCore notion of scrollOrigin through to the UI process,
and using it in various places. There are three main aspects to the patch:

  1. scroll origin is included in RemoteLayerTreeTransaction, and plumbed through to -[WKWebView _scrollToContentOffset:] for correct programmatic scrolling,

Source/WebCore:

including zooming to focussed form elements.

  1. WebPageProxy::computeCustomFixedPositionRect() uses the actual documentRect() rather than just conjuring up a rect with a zero origin, which makes position:fixed work correctly.
  2. _interactionViewsContainerView (which hosts tap highlights) is positioned to coincide with the origin of the documentRect (i.e. at the scroll origin, which may not be top-left). This allows tap highlights to show in the correct location.
  3. ScrollView::unobscuredContentRect() is fixed to take scroll origin into account; if the returned rect is wrong, RenderLayer::hitTest() incorrectly clips the hit testing area.
  • platform/ios/ScrollViewIOS.mm:

(WebCore::ScrollView::unobscuredContentRect):

Source/WebKit2:

including zooming to focused form elements. The WKContentView's boundsOrigin
is set to the scroll origin so that the view coordinates match document coordinates.

  1. WebPageProxy::computeCustomFixedPositionRect() uses the actual documentRect() rather than just conjuring up a rect with a zero origin, which makes position:fixed work correctly.
  2. _interactionViewsContainerView (which hosts tap highlights) is positioned to coincide with the origin of the documentRect (i.e. at the scroll origin, which may not be top-left). This allows tap highlights to show in the correct location.
  3. ScrollView::unobscuredContentRect() is fixed to take scroll origin into account; if the returned rect is wrong, RenderLayer::hitTest() incorrectly clips the hit testing area.
  • Shared/mac/RemoteLayerTreeTransaction.h:

(WebKit::RemoteLayerTreeTransaction::scrollOrigin):
(WebKit::RemoteLayerTreeTransaction::setScrollOrigin):

  • Shared/mac/RemoteLayerTreeTransaction.mm:

(WebKit::RemoteLayerTreeTransaction::encode):
(WebKit::RemoteLayerTreeTransaction::decode):
(WebKit::RemoteLayerTreeTransaction::description): Dump some more info.

  • UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView _scrollToContentOffset:scrollOrigin:]):
(-[WKWebView _scrollToContentOffset:]): Deleted.

  • UIProcess/API/Cocoa/WKWebViewInternal.h:
  • UIProcess/API/gtk/PageClientImpl.cpp:

(WebKit::PageClientImpl::requestScroll):

  • UIProcess/API/gtk/PageClientImpl.h:
  • UIProcess/CoordinatedGraphics/WebView.cpp:

(WebKit::WebView::requestScroll):

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

(WebKit::WebPageProxy::requestScroll):

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

(WebKit::PageClientImpl::requestScroll):
(WebKit::PageClientImpl::documentRect):
(WebKit::PageClientImpl::contentsSize): Deleted.

  • UIProcess/ios/WKContentView.mm:

(-[WKContentView _didCommitLayerTree:]):

  • UIProcess/ios/WebPageProxyIOS.mm:

(WebKit::WebPageProxy::computeCustomFixedPositionRect):

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

(WebKit::PageClientImpl::requestScroll):

  • UIProcess/mac/RemoteLayerTreeDrawingAreaProxy.mm:

(WebKit::RemoteLayerTreeDrawingAreaProxy::commitLayerTree):

  • WebProcess/WebPage/WebPage.cpp:

(WebKit::WebPage::willCommitLayerTree):

8:44 AM Changeset in webkit [186711] by ddkilzer@apple.com
  • 2 edits in trunk/LayoutTests

http/tests/security/XFrameOptions/x-frame-options-deny-meta-tag-parent-same-origin-deny.html crashes on Windows

<​http://webkit-test-results.appspot.com/dashboards/flakiness_dashboard.html#showAllRuns=true&tests=http%2Ftests%2Fsecurity%2FXFrameOptions%2Fx-frame-options-deny-meta-tag-parent-same-origin-deny.html>

  • platform/win/TestExpectations: Mark tests as crashing:
  • http/tests/security/XFrameOptions/x-frame-options-deny-meta-tag-parent-same-origin-deny.html
8:18 AM Changeset in webkit [186710] by ddkilzer@apple.com
  • 2 edits in trunk/LayoutTests

REGRESSION (r186689?): fast/frames/frame-limit.html timeout on Mavericks {Release,Debug} WK2 bots

Fix tracked by: <​http://webkit.org/b/146876>

  • platform/mac-wk2/TestExpectations: Mark test as timing out:
  • fast/frames/frame-limit.html
7:17 AM Changeset in webkit [186709] by ddkilzer@apple.com
  • 3 edits in trunk/LayoutTests

REGRESSION (r186697): Fix Yosemite, El Capitan test results for js/dom/global-constructors-attributes.html

  • platform/mac-yosemite/js/dom/global-constructors-attributes-expected.txt:
  • platform/mac/js/dom/global-constructors-attributes-expected.txt:
  • Move MediaStream results into alphabetical order to fix test failures.
1:42 AM Changeset in webkit [186708] by Devin Rousso
  • 3 edits in trunk/Source/WebInspectorUI

Web Inspector: Warning icon tooltip for numbers with no units could be improved
​https://bugs.webkit.org/show_bug.cgi?id=146859

Reviewed by Timothy Hatcher.

  • Localizations/en.lproj/localizedStrings.js:
  • UserInterface/Views/CSSStyleDeclarationTextEditor.js:

(WebInspector.CSSStyleDeclarationTextEditor.prototype._createTextMarkerForPropertyIfNeeded):
If the property's value is incorrect and is comprised of only numbers, that must mean that the value needs
to have units (like "px") after the number. Added another warning icon case to support this scenario.

Jul 10, 2015:

11:55 PM Changeset in webkit [186707] by Chris Dumez
  • 11 edits in trunk/Source

[WK2] Diagnostic logging messages are causing too much IPC
​https://bugs.webkit.org/show_bug.cgi?id=146873
<rdar://problem/21779205>

Reviewed by Ryosuke Niwa.

Source/WebCore:

Move shouldLogAfterSampling() utility function to DiagnosticLoggingClient
so it can be easily reused.

  • page/DiagnosticLoggingClient.h:

(WebCore::DiagnosticLoggingClient::shouldLogAfterSampling):

Source/WebKit2:

Diagnostic logging messages are causing too much IPC. To address the
problem, we now do the sampling of the senders' side (WebContent
process and Networking process) instead of the receiver's side
(UIProcess).

  • NetworkProcess/NetworkProcess.cpp:

(WebKit::NetworkProcess::logDiagnosticMessage):
(WebKit::NetworkProcess::logDiagnosticMessageWithResult):
(WebKit::NetworkProcess::logDiagnosticMessageWithValue):

  • UIProcess/Network/NetworkProcessProxy.cpp:

(WebKit::NetworkProcessProxy::logSampledDiagnosticMessage):
(WebKit::NetworkProcessProxy::logSampledDiagnosticMessageWithResult):
(WebKit::NetworkProcessProxy::logSampledDiagnosticMessageWithValue):

  • UIProcess/Network/NetworkProcessProxy.h:
  • UIProcess/Network/NetworkProcessProxy.messages.in:
  • UIProcess/WebPageProxy.cpp:

(WebKit::WebPageProxy::logDiagnosticMessage):
(WebKit::WebPageProxy::logDiagnosticMessageWithResult):
(WebKit::WebPageProxy::logDiagnosticMessageWithValue):
(WebKit::WebPageProxy::logSampledDiagnosticMessage):
(WebKit::WebPageProxy::logSampledDiagnosticMessageWithResult):
(WebKit::WebPageProxy::logSampledDiagnosticMessageWithValue):

  • UIProcess/WebPageProxy.h:
  • UIProcess/WebPageProxy.messages.in:
  • WebProcess/WebCoreSupport/WebDiagnosticLoggingClient.cpp:

(WebKit::WebDiagnosticLoggingClient::logDiagnosticMessage):
(WebKit::WebDiagnosticLoggingClient::logDiagnosticMessageWithResult):
(WebKit::WebDiagnosticLoggingClient::logDiagnosticMessageWithValue):

11:41 PM Changeset in webkit [186706] by fpizlo@apple.com
  • 2 edits in trunk/Source/JavaScriptCore

DFG::DesiredWatchpoints should accept WatchpointSetType's that aren't necessarily pointers
​https://bugs.webkit.org/show_bug.cgi?id=146875

Reviewed by Dan Bernstein.

In the future we'll want to add a desired watchpoint set that's something like "please
watch property 'Foo' for 'deletion' on structure 'S1'", so that the "set type" is struct
like "struct MySet { StringImpl* property; Mode mode; Structure* structure };".

This is a very mechanical change for now - all of the current users happen to use sets
that are pointer typed, so it's just a matter of moving some "*"'s around.

  • dfg/DFGDesiredWatchpoints.h:

(JSC::DFG::GenericSetAdaptor::add):
(JSC::DFG::GenericSetAdaptor::hasBeenInvalidated):
(JSC::DFG::GenericDesiredWatchpoints::GenericDesiredWatchpoints):
(JSC::DFG::GenericDesiredWatchpoints::addLazily):
(JSC::DFG::GenericDesiredWatchpoints::reallyAdd):
(JSC::DFG::GenericDesiredWatchpoints::areStillValid):
(JSC::DFG::GenericDesiredWatchpoints::isWatched):
(JSC::DFG::DesiredWatchpoints::isWatched):

10:52 PM Changeset in webkit [186705] by fpizlo@apple.com
  • 5 edits in trunk/Source/JavaScriptCore

Watchpoints should be allocated with FastMalloc
​https://bugs.webkit.org/show_bug.cgi?id=146874

Reviewed by Dan Bernstein.

This is in preparation for adding new subclasses of Watchpoint. While starting to do this
I realized that the way Watchpoints allocated themselves was unusual.

  • bytecode/CodeBlockJettisoningWatchpoint.h:

(JSC::CodeBlockJettisoningWatchpoint::CodeBlockJettisoningWatchpoint): No need for the default constructor.

  • bytecode/Watchpoint.h:

(JSC::Watchpoint::Watchpoint): This should be noncopyable and fast-allocated.

  • dfg/DFGCommonData.h: Use a Bag<> instead of SegmentedVector<>. Bag means "malloc things but allow me to free them all at once", which is what we are doing here.
  • dfg/DFGDesiredWatchpoints.h:

(JSC::DFG::GenericDesiredWatchpoints::reallyAdd): Use the Bag<> API rather than the very awkward SegmentedVector<> API.

8:45 PM Changeset in webkit [186704] by dbates@webkit.org
  • 2 edits in trunk/Source/WebCore

Cleanup: Remove default constructor for WebCore::Pair()
​https://bugs.webkit.org/show_bug.cgi?id=146856
<rdar://problem/21773212>

Reviewed by Darin Adler.

Remove default constructor for WebCore::Pair() and the setters Pair::set{First, Second}()
since they are not used and it seems weird to keep the latter after the removal of the former.
Should it turn out that we find a use for having a default constructor and/or setters then
we can revert this change.

  • css/Pair.h:

(WebCore::Pair::create): Deleted.
(WebCore::Pair::setFirst): Deleted.
(WebCore::Pair::setSecond): Deleted.
(WebCore::Pair::Pair): Deleted.

8:43 PM Changeset in webkit [186703] by dino@apple.com
  • 2 edits in trunk/Source/WebKit2

Airplay button and placeholder are missing in inline controls
​https://bugs.webkit.org/show_bug.cgi?id=146869
<rdar://problem/21555051>

Reviewed by Simon Fraser.

Fix a typo in the ENABLE that was causing the Airplay
setting (allowsAirPlayForMediaPlayback, which becomes
webkitWirelessVideoPlaybackDisabled) to default to false.

  • UIProcess/API/Cocoa/WKWebViewConfiguration.mm:

(-[WKWebViewConfiguration init]): Change WIRELESS_TARGET_PLAYBACK
to WIRELESS_PLAYBACK_TARGET.

8:01 PM Changeset in webkit [186702] by fpizlo@apple.com
  • 2 edits
    2 adds in trunk/Source/JavaScriptCore

AI folding of IsObjectOrNull is broken for non-object types that may be null
​https://bugs.webkit.org/show_bug.cgi?id=146867

Reviewed by Ryosuke Niwa.

  • dfg/DFGAbstractInterpreterInlines.h:

(JSC::DFG::AbstractInterpreter<AbstractStateType>::executeEffects): Fix the bug and add some text describing what is going on.

  • tests/stress/misc-is-object-or-null.js: Added. Test for the bug.

(foo):

  • tests/stress/other-is-object-or-null.js: Added. Test for a bug I almost introduced.

(foo):

7:30 PM Changeset in webkit [186701] by fpizlo@apple.com
  • 5 edits in trunk/Source/JavaScriptCore

It should be easy to measure total compile times.
​https://bugs.webkit.org/show_bug.cgi?id=146857

Reviewed by Sam Weinig.

This gives DFG::Plan the ability to track total compile times, and return them in a map
of stats that jsc.cpp can display.

I want to do some work to bring down DFG compile times. This will help me measure whether
I'm making a difference or not.

  • dfg/DFGPlan.cpp:

(JSC::DFG::Plan::Plan):
(JSC::DFG::Plan::~Plan):
(JSC::DFG::Plan::computeCompileTimes):
(JSC::DFG::Plan::reportCompileTimes):
(JSC::DFG::Plan::compileInThread):
(JSC::DFG::Plan::compileInThreadImpl):
(JSC::DFG::Plan::cancel):
(JSC::DFG::Plan::compileTimeStats):
(JSC::DFG::dumpAndVerifyGraph): Deleted.
(JSC::DFG::profilerCompilationKindForMode): Deleted.

  • dfg/DFGPlan.h:

(JSC::DFG::Plan::compileTimeStats):

  • jsc.cpp:

(jscmain):

  • runtime/Options.h:
5:40 PM Changeset in webkit [186700] by ryuan.choi@navercorp.com
  • 3 edits in trunk/Source/WebCore

[CoordinatedGraphics] Override primaryLayerID in CoordinatedGraphicsLayer
​https://bugs.webkit.org/show_bug.cgi?id=146826

Reviewed by Gyuyoung Kim.

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

(WebCore::CoordinatedGraphicsLayer::id): Moved id as inline function.

  • platform/graphics/texmap/coordinated/CoordinatedGraphicsLayer.h:

(WebCore::CoordinatedGraphicsLayer::id):

5:40 PM Changeset in webkit [186699] by Brent Fulgham
  • 2 edits in trunk/LayoutTests

Unskip some now-passing tests.

  • platform/mac-wk2/TestExpectations:
5:16 PM Changeset in webkit [186698] by timothy@apple.com
  • 14 edits in trunk/Source/WebInspectorUI

Web Inspector: Use -apple-system instead of -webkit-system-font
​https://bugs.webkit.org/show_bug.cgi?id=146860

Remove explicit uses of font-family where inhertance from the body is enough.
Change -webkit-system-font to -apple-system and use -apple-system-monospaced-numbers
for DataGrid and the toolbar dashboard.

Reviewed by Sam Weinig.

  • UserInterface/Views/CSSStyleDeclarationTextEditor.css:

(.css-style-text-editor > .CodeMirror .CodeMirror-placeholder):

  • UserInterface/Views/DataGrid.css:

(.data-grid td):

  • UserInterface/Views/DefaultDashboardView.css:

(.toolbar .dashboard.default > .item):
(.toolbar .dashboard.default > :matches(.time, .resourcesSize)):
(.toolbar .dashboard.default > .resourcesSize):

  • UserInterface/Views/HierarchicalPathComponent.css:

(.hierarchical-path-component):

  • UserInterface/Views/LogContentView.css:

(.console-messages):

  • UserInterface/Views/Main.css:

(body):

  • UserInterface/Views/ObjectTreeArrayIndexTreeElement.css:

(.object-tree-array-index .index-name):

  • UserInterface/Views/ObjectTreePropertyTreeElement.css:

(.object-tree-property .prototype-name):

  • UserInterface/Views/ObjectTreeView.css:

(.object-tree-outline li .empty-message):

  • UserInterface/Views/RenderingFrameTimelineOverviewGraph.css:

(.timeline-overview-graph.rendering-frame > .divider > .label):

  • UserInterface/Views/SourceCodeTextEditor.css:

(.source-code.text-editor > .CodeMirror .issue-widget > .text):

  • UserInterface/Views/TypeTreeView.css:

(.type-tree):

5:12 PM Changeset in webkit [186697] by commit-queue@webkit.org
  • 7 edits in trunk

Exposing webkitMediaStream as MediaStream
​https://bugs.webkit.org/show_bug.cgi?id=146813
<rdar://problem/21754383>

Patch by Matthew Daiter <​mdaiter@apple.com> on 2015-07-10
Reviewed by Eric Carlson.

Source/WebCore:

  • Modules/mediastream/MediaStream.idl: Changed interface

name from webkitMediaStream to MediaStream

LayoutTests:

  • platform/mac-mavericks/js/dom/global-constructors-attributes-expected.txt:

Changed attributes from webkitMediaStream to MediaStream

  • platform/mac-yosemite/js/dom/global-constructors-attributes-expected.txt:

Ditto

  • platform/mac/js/dom/global-constructors-attributes-expected.txt:

Ditto

5:04 PM Changeset in webkit [186696] by Brent Fulgham
  • 1 edit
    2 adds in trunk/LayoutTests

REGRESSION (r183133-r183138): Secondary clicking in whitespace selects preceding word
​https://bugs.webkit.org/show_bug.cgi?id=146695
<rdar://problem/21441466>

Reviewed by Tim Horton.

Add a new test that checks for correct mac selection behavior for editable and read-only
text fields.

  • platform/mac/editing/selection/context-menu-select-editability-expected.txt: Added.
  • platform/mac/editing/selection/context-menu-select-editability.html: Added.
4:31 PM Changeset in webkit [186695] by Devin Rousso
  • 4 edits in trunk/Source/WebInspectorUI

Web Inspector: Add source links to functions logged in the console
​https://bugs.webkit.org/show_bug.cgi?id=146377

Reviewed by Timothy Hatcher.

  • UserInterface/Protocol/RemoteObject.js:

(WebInspector.RemoteObject.prototype.findFunctionSourceCodeLocation):
Returns a promise that contains the sourceCodeLocation if the object represents a function and has an objectId.
(WebInspector.RemoteObject.prototype._isFunction):

  • UserInterface/Views/ConsoleMessageView.css:

(.console-message .console-message-location):
Added specified values for font sizing and family to ensure that all location links have the same styling.

  • UserInterface/Views/ConsoleMessageView.js:

(WebInspector.ConsoleMessageView):
Now creates a link to the source code of the entered text if the message is of the type "result".
(WebInspector.ConsoleMessageView.prototype._appendLocationLink):
(WebInspector.ConsoleMessageView.prototype._createRemoteObjectIfNeeded):
(WebInspector.ConsoleMessageView.prototype._appendFormattedArguments):
(WebInspector.ConsoleMessageView.prototype._linkifyLocation):
(WebInspector.ConsoleMessageView.prototype._linkifyCallFrameLocation):
(WebInspector.ConsoleMessageView.prototype._linkifyCallFrame):

4:09 PM Changeset in webkit [186694] by Brent Fulgham
  • 5 edits in trunk/Tools

[Mac] Unable to dismiss context menu during test runs
​https://bugs.webkit.org/show_bug.cgi?id=146836

Reviewed by Tim Horton.

Since Context Menus are modal, the test system would block on the displayed context menu,
causing tests to fail. Instead, we should do what WK1 was doing, which is to perform the
mouse click and generate the menu contents, but not ask AppKit to display it.

Drive-by fix: We did not handle 'escape' key presses. This is also fixed.

  • DumpRenderTree/mac/EventSendingController.mm:

(-[EventSendingController keyDown:withModifiers:withLocation:]): Handle the 'escape' character.

  • WebKitTestRunner/InjectedBundle/EventSendingController.cpp:

(WTR::EventSendingController::mouseMoveTo): Keep track of the current mouse position.
(WTR::EventSendingController::contextClick): Don't emit an actual context menu invocation and then
copy the menu items. Instead, use WKBundlePageCopyContentMenuAtPointInWindow to do all the same
work without actually calling on AppKit to display the menu.
ask WebCore to do all the work EXCEPT displaying the menu.

  • WebKitTestRunner/InjectedBundle/EventSendingController.h:
  • WebKitTestRunner/mac/EventSenderProxy.mm:

(WTR::EventSenderProxy::keyDown): Handle the 'escape' character.

3:29 PM Changeset in webkit [186693] by timothy_horton@apple.com
  • 2 edits in trunk/Source/WebKit2

Use CoreAnimation fences instead of synchronous IPC to synchronize resize
​https://bugs.webkit.org/show_bug.cgi?id=146294
<rdar://problem/21090193>

Reviewed by Anders Carlsson.

  • UIProcess/mac/TiledCoreAnimationDrawingAreaProxy.mm:

(WebKit::TiledCoreAnimationDrawingAreaProxy::createFenceForGeometryUpdate):
Tiny followup; don't touch Connection if the connection isn't valid.

3:04 PM Changeset in webkit [186692] by commit-queue@webkit.org
  • 19 edits
    2 adds in trunk

Source/WebCore:
Added ARIA 1.1 "cell" and "table" roles.
​https://bugs.webkit.org/show_bug.cgi?id=146011
<rdar://problem/21398946>

Patch by Nan Wang <​n_wang@apple.com> on 2015-07-10
Reviewed by Chris Fleizach.

Created a new role called GridCellRole to match the gridcell role,
so the previous CellRole and TableRole will match to cell and table role.
Made the changes to make sure that both GridRole and TableRole have same
behavior, as well as the circumstance for GridCellRole and CellRole.

Test: accessibility/roles-table-and-cell.html

  • accessibility/AXObjectCache.cpp:

(WebCore::createFromRenderer):

  • accessibility/AccessibilityNodeObject.cpp:

(WebCore::AccessibilityNodeObject::supportsRequiredAttribute):
(WebCore::AccessibilityNodeObject::canSetSelectedAttribute):

  • accessibility/AccessibilityObject.cpp:

(WebCore::initializeRoleMap):

  • accessibility/AccessibilityObject.h:
  • accessibility/AccessibilityRenderObject.cpp:

(WebCore::AccessibilityRenderObject::setSelectedRows):
(WebCore::AccessibilityRenderObject::inheritsPresentationalRole):
(WebCore::AccessibilityRenderObject::selectedChildren):

  • accessibility/AccessibilityTable.cpp:

(WebCore::AccessibilityTable::roleValue):

  • accessibility/AccessibilityTableCell.cpp:

(WebCore::AccessibilityTableCell::determineAccessibilityRole):

  • accessibility/ios/WebAccessibilityObjectWrapperIOS.mm:

(-[WebAccessibilityObjectWrapper _accessibilityTableAncestor]):
(-[WebAccessibilityObjectWrapper _accessibilityTraitsFromAncestors]):
(-[WebAccessibilityObjectWrapper determineIsAccessibilityElement]):

  • accessibility/mac/WebAccessibilityObjectWrapperMac.mm:

(createAccessibilityRoleMap):

LayoutTests:
Added tests for new role: table and cell.
​https://bugs.webkit.org/show_bug.cgi?id=146011.

Patch by Nan Wang <​n_wang@apple.com> on 2015-07-10
Reviewed by Chris Fleizach.

Added tests for table and cell role. Also modified other tests to fit the changes.

  • accessibility/roles-computedRoleString-expected.txt:
  • accessibility/roles-computedRoleString.html:
  • accessibilit/roles-table-and-cell-expected.txt: Added.
  • accessibility/roles-table-and-cell.html: Added.
  • platform/mac/accessibility/aria-table-hierarchy-expected.txt:
  • platform/mac/accessibility/aria-tables-expected.txt:
  • platform/mac/accessibility/roles-exposed-expected.txt:
2:19 PM Changeset in webkit [186691] by fpizlo@apple.com
  • 45 edits
    4 adds in trunk/Source/JavaScriptCore

DFG fragile frozen values are fundamentally broken
​https://bugs.webkit.org/show_bug.cgi?id=146602

Reviewed by Mark Lam.

This change gets rid of the FragileValue value strength, because it was fundamentally
broken.

FragileValue was a value known to the compiler but not tracked by the GC in any way -
it wasn't marked and it wasn't weak. This was used to support AI bootstrap for OSR
must-handle values. The philosophy was that if the compiler did use the value for
optimization, it would have been strengthened to a weak value (or maybe even a strong
value, though we probably won't do that). But this was too much of a pipe dream. I've
found at least one case where the compiler did use the value, but never strengthened
it: it would happen if the value ended up in an OSR entry data expected value. Then if
we GCed, we might have killed the value, but OSR entry would still try to use it for
validation. That might have sort of just worked, but it's clearly shady.

The reason why we made must-handle values fragile and not weak is that most of the time
the values disappear from the abstract state: they are LUBed to a non-constant. If we
kept them around as weak, we'd have too many cases of the GC killing the code because
it thought that the value was somehow meaningful to the code when it was only used as a
temporary artifact of optimization.

So, it's true that it's very important for must-handle values not to automatically be
weak or strong. It's also true that the values are necessary for AI bootstrap because
we need to know what values OSR entry will require. But we shouldn't accomplish these
goals by having the compiler hold onto what are essentially dangling pointers.

This implements a better solution: instead of having InPlaceAbstractState bootstrap the
AI with must-handle values at the beginning, we now widen the valuesAtHead of the
must-handle block after AI converges. This widening is done in CFAPhase. This allows us
to see if the must-handle values are necessary at all. In most cases, the widening
takes a non-constant abstract value and simply amends something to its type based on
the type of the must-handle value, and so the must-handle value never actually shows up
in either the IR or any abstract value. In the unlikely event that the value at head is
bottom, we freeze the must-handle value. This change removes FragileValue, and this
freezing uses WeakValue as the strength. That makes sense: since the abstract value was
bottom, the must-handle value becomes integral to the IR and so it makes no sense for
the GC to keep the resulting CodeBlock alive if that must-handle value dies. This will
sometimes happen for example if you have a very long-running loop whose pre-header
allocates some object, but that pre-header appears to always exit to the optimizing JIT
because it was only profiled once in the LLInt and that profiling appears insufficient
to the DFG. In that case, we'll effectively constant-fold the references to the object
inside the loop, which is both efficient (yay constant folding!) and necessary
(otherwise we wouldn't know what the type of the variable should have been).

Testing and debugging this is complicated. So, this adds some new capabilities:

  • DFG IR dumps also dump all of the FrozenValues that point to the heap along with their strengths, so that it's easy to see what GC objects the DFG feels are necessary for the compilation.


  • DFG OSR entry preparation prints out the OSR entry data structures, so that it's easy to see what GC pointers (and other things) are used for OSR entry validation. The printouts are quite detailed, and should also help other kinds of OSR entry debugging.


  • DFG::Plan now validates whether all of the GC pointers planted in the various JITCode data structures are also properly registered as either weak or strong pointers in the CodeBlock. This validation check previously failed due to fragile values ending up in the OSR entry data structures, both in the newly added test (dead-osr-entry-value.js) and in some pre-existing tests (like earley-boyer and 3d-raytrace).

(JSC::CodeBlock::stronglyVisitStrongReferences):

  • bytecode/CodeOrigin.cpp:

(JSC::InlineCallFrame::visitAggregate):

  • bytecode/Operands.h:

(JSC::Operands::operand):
(JSC::Operands::hasOperand):

  • bytecode/StructureSet.cpp:

(JSC::StructureSet::dump):
(JSC::StructureSet::validateReferences):

  • bytecode/StructureSet.h:
  • bytecode/TrackedReferences.cpp: Added.

(JSC::TrackedReferences::TrackedReferences):
(JSC::TrackedReferences::~TrackedReferences):
(JSC::TrackedReferences::add):
(JSC::TrackedReferences::check):
(JSC::TrackedReferences::dump):

  • bytecode/TrackedReferences.h: Added.
  • dfg/DFGAbstractValue.cpp:

(JSC::DFG::AbstractValue::observeTransitions):
(JSC::DFG::AbstractValue::set):
(JSC::DFG::AbstractValue::fixTypeForRepresentation):
(JSC::DFG::AbstractValue::mergeOSREntryValue):
(JSC::DFG::AbstractValue::filter):
(JSC::DFG::AbstractValue::dumpInContext):
(JSC::DFG::AbstractValue::validateReferences):
(JSC::DFG::AbstractValue::setOSREntryValue): Deleted.

  • dfg/DFGAbstractValue.h:

(JSC::DFG::AbstractValue::fullTop):
(JSC::DFG::AbstractValue::merge):

  • dfg/DFGByteCodeParser.cpp:

(JSC::DFG::ByteCodeParser::InlineStackEntry::InlineStackEntry):

  • dfg/DFGCFAPhase.cpp:

(JSC::DFG::CFAPhase::run):

  • dfg/DFGCommonData.cpp:

(JSC::DFG::CommonData::invalidate):
(JSC::DFG::CommonData::validateReferences):

  • dfg/DFGCommonData.h:

(JSC::DFG::CommonData::requiredRegisterCountForExecutionAndExit):

  • dfg/DFGFrozenValue.h:

(JSC::DFG::FrozenValue::FrozenValue):
(JSC::DFG::FrozenValue::strengthenTo):
(JSC::DFG::FrozenValue::pointsToHeap):
(JSC::DFG::FrozenValue::strength):
(JSC::DFG::FrozenValue::freeze):

  • dfg/DFGGraph.cpp:

(JSC::DFG::Graph::Graph):
(JSC::DFG::Graph::dump):
(JSC::DFG::Graph::registerFrozenValues):
(JSC::DFG::Graph::visitChildren):
(JSC::DFG::Graph::freeze):
(JSC::DFG::Graph::freezeStrong):
(JSC::DFG::Graph::freezeFragile): Deleted.

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

(JSC::DFG::InPlaceAbstractState::initialize):

  • dfg/DFGJITCode.cpp:

(JSC::DFG::JITCode::setOptimizationThresholdBasedOnCompilationResult):
(JSC::DFG::JITCode::validateReferences):

  • dfg/DFGJITCode.h:
  • dfg/DFGJITCompiler.cpp:

(JSC::DFG::JITCompiler::addressOfDoubleConstant):
(JSC::DFG::JITCompiler::noticeOSREntry):

  • dfg/DFGJITCompiler.h:

(JSC::DFG::JITCompiler::branchStructurePtr):
(JSC::DFG::JITCompiler::jitCode):
(JSC::DFG::JITCompiler::noticeOSREntry): Deleted.

  • dfg/DFGMinifiedGraph.cpp: Added.

(JSC::DFG::MinifiedGraph::prepareAndShrink):
(JSC::DFG::MinifiedGraph::validateReferences):

  • dfg/DFGMinifiedGraph.h:

(JSC::DFG::MinifiedGraph::append):
(JSC::DFG::MinifiedGraph::prepareAndShrink): Deleted.

  • dfg/DFGOSREntry.cpp:

(JSC::DFG::OSREntryData::dumpInContext):
(JSC::DFG::OSREntryData::dump):
(JSC::DFG::prepareOSREntry):

  • dfg/DFGOSREntry.h:

(JSC::DFG::getOSREntryDataBytecodeIndex):

  • dfg/DFGPlan.cpp:

(JSC::DFG::Plan::finalizeWithoutNotifyingCallback):

  • dfg/DFGSpeculativeJIT.cpp:

(JSC::DFG::SpeculativeJIT::linkOSREntries):
(JSC::DFG::SpeculativeJIT::compileDoublePutByVal):

  • dfg/DFGStructureAbstractValue.cpp:

(JSC::DFG::StructureAbstractValue::dump):
(JSC::DFG::StructureAbstractValue::validateReferences):

  • dfg/DFGStructureAbstractValue.h:
  • dfg/DFGValidate.cpp:

(JSC::DFG::Validate::validate):

  • dfg/DFGValueStrength.cpp:

(WTF::printInternal):

  • dfg/DFGValueStrength.h:

(JSC::DFG::merge):

  • ftl/FTLExitPropertyValue.cpp:

(JSC::FTL::ExitPropertyValue::dump):
(JSC::FTL::ExitPropertyValue::validateReferences):

  • ftl/FTLExitPropertyValue.h:
  • ftl/FTLExitTimeObjectMaterialization.cpp:

(JSC::FTL::ExitTimeObjectMaterialization::dump):
(JSC::FTL::ExitTimeObjectMaterialization::validateReferences):

  • ftl/FTLExitTimeObjectMaterialization.h:
  • ftl/FTLExitValue.cpp:

(JSC::FTL::ExitValue::dump):
(JSC::FTL::ExitValue::validateReferences):

  • ftl/FTLExitValue.h:
  • ftl/FTLJITCode.cpp:

(JSC::FTL::JITCode::dfgCommon):
(JSC::FTL::JITCode::validateReferences):

  • ftl/FTLJITCode.h:

(JSC::FTL::JITCode::handles):
(JSC::FTL::JITCode::dataSections):

  • ftl/FTLOSRExit.cpp:

(JSC::FTL::OSRExit::codeLocationForRepatch):
(JSC::FTL::OSRExit::validateReferences):

  • ftl/FTLOSRExit.h:

(JSC::FTL::OSRExit::considerAddingAsFrequentExitSite):

  • jit/JITCode.cpp:

(JSC::JITCode::typeName):
(JSC::JITCode::validateReferences):
(JSC::JITCode::execute):

  • jit/JITCode.h:

(JSC::JITCode::start):

  • tests/stress/dead-osr-entry-value.js: Added.

(foo):

2:17 PM Changeset in webkit [186690] by timothy@apple.com
  • 2 edits in trunk/Source/WebInspectorUI

Web Inspector: Storage tab should have a scope bar in the sidebar
​https://bugs.webkit.org/show_bug.cgi?id=146799

Fix a couple typos found post review.

  • UserInterface/Views/StorageSidebarPanel.js:

(WebInspector.StorageSidebarPanel):

2:15 PM Changeset in webkit [186689] by Chris Dumez
  • 3 edits in trunk/Source/WebKit2

[WK2] Increase the QoS of some of our WorkQueues to match the one of our processes
​https://bugs.webkit.org/show_bug.cgi?id=146855

Reviewed by Anders Carlsson.

Increase the QoS of some of our WorkQueues to match the one of our processes. For
e.g., even though the WebContent and the Networking processes have USER_UNITIATED
QoS, the IPC between them is using a WorkQueue with DEFAULT QoS. Similarly, the
WorkQueue used for network cache reads has DEFAULT QoS.

  • NetworkProcess/cache/NetworkCacheStorage.cpp:

(WebKit::NetworkCache::Storage::Storage):

  • Platform/IPC/Connection.cpp:

(IPC::Connection::Connection):

2:12 PM Changeset in webkit [186688] by fpizlo@apple.com
  • 3 edits
    1 add in trunk

It should be possible to run the OSR exit fuzzer
​https://bugs.webkit.org/show_bug.cgi?id=146814

Reviewed by Mark Lam.

Source/JavaScriptCore:

Fix a silly bug: the option matching was accidentally allowing a prefix match, which
always did the wrong thing for --fireOSRExitFuzzAt and --fireOSRExitFuzzAtStatic. Make
this an exact match instead.

  • runtime/Options.cpp:

(JSC::Options::setOption):

Tools:

This runs a test command to detect how many static OSR exit sites it had, and then for
each exit site, it triggers three exits (earliest opportunity, last opportunity, some
middle opportunity), plus a mode where the check always exits. This is an exhaustive
search, so in the future it'll be very appropriate for cases where we have a simple test
and we want to see what happens when you eventually exit in this test.

Right now this isn't hooked into any specific tests, but I've done some spot checks on it
and it seems to do the right thing.

  • Scripts/jsc-stress-test-helpers/js-osr-exit-fuzz: Added.

(fail):

1:40 PM Changeset in webkit [186687] by dbates@webkit.org
  • 8 edits in trunk

Serialized declaration for background-size/-webkit-mask-size should preserve identical
values instead of coalescing them
​https://bugs.webkit.org/show_bug.cgi?id=143191
<rdar://problem/21759393>

Reviewed by Darin Adler.

Source/WebCore:

Inspired by Blink r157745:
<​http://src.chromium.org/viewvc/blink?view=revision&revision=157745>
And Blink r172548:
<​http://src.chromium.org/viewvc/blink?view=revision&revision=172548>

Fixes an issue where identical values in a CSS declaration for properties background-size
and -webkit-mask-size would be coalesced when serialized.

  • css/CSSParser.cpp:

(WebCore::CSSParser::parseFillSize): Modified to create a Pair object with the appropriate
identical value encoding strategy. We only coalesce identical values when serializing the
declaration of property -webkit-background-size.

  • css/Pair.h:

(WebCore::Pair::create): Added variant that takes an encoding.
(WebCore::Pair::cssText): Moved implementation of Pair::generateCSSString() to here and
modified to conditionally coalesce identical values during serialization.
(WebCore::Pair::Pair): Added variant that takes an encoding.
(WebCore::Pair::generateCSSString): Deleted; moved implementation into Pair::cssText().

LayoutTests:

Inspired by Blink r172548:
<​http://src.chromium.org/viewvc/blink?view=revision&revision=172548>

Update tests and test results to ensure we serialize background-size, -webkit-mask-size
correctly when identical values are present in their declarations.

  • fast/backgrounds/size/parsing-background-size-values-expected.txt:
  • fast/backgrounds/size/resources/parsing-background-size-values.js:
  • fast/masking/parsing-mask-expected.txt:
  • fast/masking/parsing-mask.html:
1:39 PM Changeset in webkit [186686] by mmaxfield@apple.com
  • 11 edits
    2 adds in trunk

Bidi-Isolate inlines break layout with collapsed whitespace
​https://bugs.webkit.org/show_bug.cgi?id=109624
<rdar://problem/21752834>

Reviewed by David Hyatt.

Source/WebCore:

This patch changes the logic in constructBidiRunsForSegment() when it encounters an
isolate. It already has logic to create a BidiResolver for the isolated text;
however, that logic doesn't handle setting up the MidpointState at all.
Specifically, we can set the MidpointState's cursor to point to the context which
we can remember from addPlaceholderRunForIsolatedInline(). This information is
remembered in a HashMap in BidiResolver.

This patch is a partial port of Blink patch
​https://src.chromium.org/viewvc/blink?view=rev&revision=159203

Here is some explanatory text regarding how we collapse spaces:

Collapsing whitespace happens in a series of phases. The first phase occurs when
we perform line breaking. Here, we keep track of sequences of whitespace which
should be collapsed, in the form of a vector of pairs of InlineIterators. We put
this knowledge into a MidpointState object.

Then, once we have a line, we run the bidi algorithm on the line (including the
whitespace). As output, the bidi algorithm calls the BidiResolver::appendRun()
callback with two InlineIterators each time it wants to create a run. Because
each renderer that we create has to be owned by exactly one DOM node,
BidiResolver::appendRun() iterates between its two InlineIterator arguments,
calling RenderBlockFlow::appendRunsForObject() on each interstitial DOM node.

This is the function where whitespace collapsing happens. The MidpointState object
keeps a cursor into its remembered whitespace sequences. Here, we simply make a
bidi run for each region in between adjacent whitespace pairs in the MidpointState
object. These bidi runs eventually get turned into leaf InlineBoxes.

The problem is that the BidiResolver::appendRun() callbacks don't occur in
string-order, but the Midpoint InlineIterator pairs are in string-order. In
particular, within a particular isolate, appendRun() gets called in string
order, but callbacks that occur for inner isolates are deferred. This means that
RenderBlockFlow::appendRunsForObject() gets confused when it looks for relevant
whitespace to skip.

Test: fast/text/bidi-isolate-whitespace-collapse.html

  • platform/text/BidiResolver.h:

(WebCore::MidpointState::numMidpoints): Returning a const unsigned& is silly.
(WebCore::MidpointState::currentMidpoint): Ditto.
(WebCore::MidpointState::setCurrentMidpoint): The isolated MidpointState object
needs to be able to set its current midpoint to point to the first one inside
the isolate.
(WebCore::MidpointState::decrementNumMidpoints): Renamed from "decrease"
(WebCore::MidpointState::betweenMidpoints): This function is true iff
currentMidpoint() % 2. Instead of keeping a member variable, we can just compute
that.
(WebCore::MidpointState::reset): Deleted.
(WebCore::MidpointState::decreaseNumMidpoints): Deleted.
(WebCore::MidpointState::setBetweenMidpoints): Deleted.

  • rendering/InlineIterator.h:

(WebCore::IsolateTracker::addFakeRunIfNecessary): Call
RenderBlockFlow::appendRunsForObject() to keep our MidpointState object in sync
when we pop out of the isolated object. However, we pass in a null run list,
because we don't want to append just yet (that happens when we process the
isolate).
(WebCore::InlineBidiResolver::appendRun): Update for new signature of
appendRunsForObject().

  • rendering/RenderBlock.h:

(WebCore::RenderBlock::shouldSkipCreatingRunsForObject): Take a reference instead
of a pointer.

  • rendering/RenderBlockFlow.h:
  • rendering/RenderBlockLineLayout.cpp:

(WebCore::createRun): Ditto.
(WebCore::RenderBlockFlow::appendRunsForObject): Allow someone passing us a null
BidiRunList. In this case, we will keep the resolver's midpointState() up to date,
but won't actually emit any runs.
(WebCore::notifyResolverToResumeInIsolate): Renamed from setUp.
(WebCore::isolatedResolversMidpointState): Calculate the midpoint state for the
isolated resolver.
(WebCore::setUpResolverToResumeInIsolate): Call isolatedResolversMidpointState().
(WebCore::constructBidiRunsForSegment): Pass in the topResolver, which is
necessary for isolatedResolversMidpointState().

  • rendering/line/BreakingContext.h:

(WebCore::checkMidpoints):

LayoutTests:

  • fast/inline/crash-when-child-renderer-is-removed-and-line-stays-clean-expected.txt:
  • fast/text/bidi-isolate-whitespace-collapse-expected.html: Added.
  • fast/text/bidi-isolate-whitespace-collapse.html: Added.
  • fast/text/international/embed-bidi-style-in-isolate-crash-expected.txt:
  • fast/text/remove-text-node-linebox-not-dirty-crash-expected.txt:
1:34 PM Changeset in webkit [186685] by dbates@webkit.org
  • 4 edits in trunk/Source/WebCore

Cleanup: WebCore::Pair class should use RefPtr&& instead of PassRefPtr
​https://bugs.webkit.org/show_bug.cgi?id=146852

Reviewed by Anders Carlsson.

  • css/CSSParser.cpp:

(WebCore::createPrimitiveValuePair): Write using a variadic template that forwards its
arguments to the appropriate Pair::create constructor.
(WebCore::CSSParser::parse4ValuesFillPosition): Use RefPtr&& instead of PassRefPtr.
(WebCore::CSSParser::parse3ValuesFillPosition): Ditto.
(WebCore::CSSParser::parseFillPosition): Use WTF::move() instead of RefPtr::release()
so that we pass a rvalue reference to RefPtr.
(WebCore::CSSParser::parseFillSize): Ditto.
(WebCore::CSSParser::parseBorderImageRepeat): Ditto.

  • css/CSSParser.h:
  • css/Pair.h:

(WebCore::Pair::create): Use RefPtr&& instead of PassRefPtr.
(WebCore::Pair::setFirst): Ditto.
(WebCore::Pair::setSecond): Ditto.
(WebCore::Pair::Pair): Ditto. Also substitute nullptr for 0 in the member initialization
list of the default constructor.

1:19 PM Changeset in webkit [186684] by Alan Bujtas
  • 6 edits in trunk/Source

Crash at WebCore::WebPage::innerFrameQuad.
​https://bugs.webkit.org/show_bug.cgi?id=146843
rdar://problem/21501819

Reviewed by Andreas Kling.

We may end up with a null rootEditableElement() after calling Document::updateLayout().

Speculative fix. Not reproducible.

Source/WebCore:

  • page/Frame.cpp: Some const cleanup.

(WebCore::Frame::visiblePositionForPoint):

  • page/Frame.h:

Source/WebKit2:

  • WebProcess/WebPage/WebPage.h: Some const cleanup.
  • WebProcess/WebPage/ios/WebPageIOS.mm:

(WebKit::innerFrameQuad): null check assistedNode.rootEditableElement()
(WebKit::constrainPoint):
(WebKit::WebPage::selectWithGesture):
(WebKit::WebPage::visiblePositionInFocusedNodeForPoint):
(WebKit::WebPage::selectPositionAtPoint):
(WebKit::WebPage::selectPositionAtBoundaryWithDirection):
(WebKit::WebPage::selectTextWithGranularityAtPoint):
(WebKit::WebPage::updateSelectionWithExtentPoint):
(WebKit::WebPage::getPositionInformation):

12:01 PM Changeset in webkit [186683] by beidson@apple.com
  • 4 edits in trunk/Source/WebCore

Crash in HistoryController::updateForCommit dereferencing a null HistoryItem.
<rdar://problem/21371589> and ​https://bugs.webkit.org/show_bug.cgi?id=146842

Reviewed by Chris Dumez.

No new tests (Unknown how to reproduce).

This patch basically rolls back part of ​http://trac.webkit.org/changeset/179472.

r179472 changed HistoryController::setCurrentItem() to take a reference instead of a pointer.
Unfortunately, we sometimes call setCurrentItem(nullptr).

We'd like to *not* do that, and there are assertions in place to try to catch when we do,
but in the meantime it is not valid to dereference nullptr.

  • loader/FrameLoader.cpp:

(WebCore::FrameLoader::loadSameDocumentItem):

  • loader/HistoryController.cpp:

(WebCore::HistoryController::updateForCommit):
(WebCore::HistoryController::recursiveUpdateForCommit):
(WebCore::HistoryController::recursiveUpdateForSameDocumentNavigation):
(WebCore::HistoryController::setCurrentItem): Take a ptr instead of a ref.
(WebCore::HistoryController::createItem):

  • loader/HistoryController.h:
10:53 AM Changeset in webkit [186682] by jfernandez@igalia.com
  • 6 edits
    6 adds in trunk

[CSS Grid Layout] Grid item's auto-margins are not applied correctly
​https://bugs.webkit.org/show_bug.cgi?id=146581

Reviewed by Darin Adler.

Source/WebCore:

Grid Layout specification states that auto-margins must prevent to apply
any Box Alignment property in that dimension. Instead, auto-margin is
applied as described in the spec.

​http://dev.w3.org/csswg/css-grid/#auto-margins

Only Flexbox and Grid Layout implement vertical auto-margin alignment;
for other layout modes, auto-margin should compute as 0.

Tests: fast/css-grid-layout/grid-item-auto-margins-alignment-vertical-lr.html

fast/css-grid-layout/grid-item-auto-margins-alignment-vertical-rl.html
fast/css-grid-layout/grid-item-auto-margins-alignment.html

  • rendering/RenderBox.cpp:

(WebCore::RenderBox::styleDidChange):
(WebCore::RenderBox::updateGridAlignmentAfterStyleChange):

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

(WebCore::RenderGrid::layoutGridItems):
(WebCore::RenderGrid::hasAutoMarginsInColumnAxis):
(WebCore::RenderGrid::hasAutoMarginsInRowAxis):
(WebCore::RenderGrid::updateAutoMarginsInColumnAxisIfNeeded):
(WebCore::RenderGrid::rowPositionForChild):
(WebCore::RenderGrid::columnPositionForChild):

  • rendering/RenderGrid.h:

LayoutTests:

Tests to verify auto-margins alignment works as expected with grid items.

  • fast/css-grid-layout/grid-item-auto-margins-alignment-expected.txt: Added.
  • fast/css-grid-layout/grid-item-auto-margins-alignment-vertical-lr-expected.txt: Added.
  • fast/css-grid-layout/grid-item-auto-margins-alignment-vertical-lr.html: Added.
  • fast/css-grid-layout/grid-item-auto-margins-alignment-vertical-rl-expected.txt: Added.
  • fast/css-grid-layout/grid-item-auto-margins-alignment-vertical-rl.html: Added.
  • fast/css-grid-layout/grid-item-auto-margins-alignment.html: Added.
10:41 AM Changeset in webkit [186681] by commit-queue@webkit.org
  • 2 edits in trunk/Tools

Add tests for r186649
​https://bugs.webkit.org/show_bug.cgi?id=146821

Patch by Alex Christensen <​achristensen@webkit.org> on 2015-07-10
Reviewed by Darin Adler.

  • TestWebKitAPI/Tests/WebCore/ContentExtensions.cpp:

(TestWebKitAPI::TEST_F):
(TestWebKitAPI::compareContents):
(TestWebKitAPI::expectedIndex):
Add some larger tests to test large jumps.

10:38 AM Changeset in webkit [186680] by Csaba Osztrogonác
  • 2 edits in trunk/Source/WebCore

Fix overloaded virtual function warning in ScrollbarThemeGtk.h
​https://bugs.webkit.org/show_bug.cgi?id=145920

Reviewed by Darin Adler.

  • platform/gtk/ScrollbarThemeGtk.h:
9:34 AM Changeset in webkit [186679] by Chris Fleizach
  • 2 edits in trunk/Source/WebCore

AX: <details> element should allow expand/close through AX API
​https://bugs.webkit.org/show_bug.cgi?id=146549

Apply review comments from Daniel Bates.

  • accessibility/AccessibilityNodeObject.cpp:

(WebCore::AccessibilityNodeObject::setIsExpanded):

9:33 AM Changeset in webkit [186678] by jhoneycutt@apple.com
  • 2 edits in trunk/Source/WebKit2

[iOS] Document picker has overlapping icons on iPad
​https://bugs.webkit.org/show_bug.cgi?id=146808
<rdar://problem/21251875>

Reviewed by Sam Weinig.

  • UIProcess/ios/forms/WKFileUploadPanel.mm:

(-[WKFileUploadPanel documentMenu:didPickDocumentPicker:]):
Don't show this UI as a popover on iPad; the default is a full screen
form sheet.

8:45 AM Changeset in webkit [186677] by beidson@apple.com
  • 2 edits in trunk/Source/WebCore

ASSERT restoring from page cache as DocumentLoader reattaches to its Frame.
<rdar://problem/21766282> and ​https://bugs.webkit.org/show_bug.cgi?id=146786

Reviewed by NOBODY (Fixing obvious boneheaded mistake in r186642)

No new tests (Covered by existing)

  • loader/DocumentLoader.cpp:

(WebCore::DocumentLoader::attachToFrame): Bail if reattaching to the current Frame,

which happens when restoring from the page cache.

8:28 AM Changeset in webkit [186676] by matthew_hanson@apple.com
  • 10 edits in tags/Safari-601.1.39/Source

Disable non-shipping features.

8:07 AM Changeset in webkit [186675] by matthew_hanson@apple.com
  • 5 edits in tags/Safari-601.1.39/Source

Un-versioning.

7:53 AM Changeset in webkit [186674] by matthew_hanson@apple.com
  • 5 edits in trunk/Source

Versioning.

7:51 AM Changeset in webkit [186673] by matthew_hanson@apple.com
  • 5 edits in tags/Safari-601.1.39/Source

Versioning.

7:37 AM Changeset in webkit [186672] by matthew_hanson@apple.com
  • 1 copy in tags/Safari-601.1.39

New Tag.

5:39 AM Changeset in webkit [186671] by commit-queue@webkit.org
  • 2 edits in trunk/LayoutTests

Fix typo in LayoutTests/platform/wk2/TestExpectation.
​https://bugs.webkit.org/show_bug.cgi?id=146830

Unreviewed, WK2 gardening.

Patch by Hunseop Jeong <Hunseop Jeong> on 2015-07-10

  • platform/wk2/TestExpectations:
5:33 AM Changeset in webkit [186670] by Michael Catanzaro
  • 2 edits in trunk

[GTK] ar warning when linking static libraries
​https://bugs.webkit.org/show_bug.cgi?id=144988

Reviewed by Carlos Garcia Campos.

Do not use 'u' when calling ar. This argument conflicts with 'D' and is ignored if 'D' is
used, and 'D' is used by default if binutils is built with --enable-deterministic-archives.
Using 'u' with 'D' causes a warning to be printed when linking static libraries. At least
Fedora and Debian have recently both chosen to use --enable-deterministic-archives, so we
should either stop using 'u' or else add 'U' as well in order to disable deterministic
archives. Using 'U' should result in a somewhat faster build (at least when using the
Makefile generator), but it's unlikely that the difference is significant, so let's simply
remove 'u' until someone determines otherwise. This seems like a better option than adding
'U' so as not to foil distributions' attempts to perform deterministic builds. This also
aligns us with the behavior of upstream CMake (which has never used 'u'). This is a minor
behavior change on distributions that do not use --enable-deterministic-archives, notably
Arch and openSUSE.

  • Source/cmake/OptionsCommon.cmake:
5:30 AM Changeset in webkit [186669] by commit-queue@webkit.org
  • 2 edits in trunk/Source/WebCore/platform/gtk/po

Updated Swedish translation for WebKitGTK+
​https://bugs.webkit.org/show_bug.cgi?id=146596

Patch by Josef Andersson <​josef.andersson@fripost.org> on 2015-07-10
Reviewed by Carlos Garcia Campos.

  • sv.po:
5:26 AM Changeset in webkit [186668] by loki@webkit.org
  • 2 edits in trunk/Websites/planet.webkit.org

[Planet WebKit] Update the feed url of the University of Szeged's blog.
​https://bugs.webkit.org/show_bug.cgi?id=146835

Reviewed by Csaba Osztrogonác.

  • config.ini:
Note: See TracTimeline for information about the timeline view.