WebKit-https.git
5 years agoWeb Inspector: IndexedDB data not showing for MDN example page
commit-queue@webkit.org [Thu, 11 Jun 2015 21:40:22 +0000 (21:40 +0000)]
Web Inspector: IndexedDB data not showing for MDN example page
https://bugs.webkit.org/show_bug.cgi?id=145851

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2015-06-11
Reviewed by Anders Carlsson.

* UserInterface/Views/StorageSidebarPanel.js:
(WebInspector.StorageSidebarPanel.prototype._indexedDatabaseWasAdded):
Fix typo in Storage tab refactoring that was causing an
uncaught exception and not showing Indexed DBs.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185476 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[OS X] Decoding fonts may return nil when using force click
mmaxfield@apple.com [Thu, 11 Jun 2015 21:29:56 +0000 (21:29 +0000)]
[OS X] Decoding fonts may return nil when using force click
https://bugs.webkit.org/show_bug.cgi?id=145890

Reviewed by Beth Dakin.

[NSFontDescriptor fontDescriptorWithFontAttributes] might return nil, which
causes the encoding of an NSFont to erroneously return true.

* Shared/mac/ArgumentCodersMac.mm:
(IPC::decode):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185475 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoReimplement WKOriginDataManager in terms of API::WebsiteDataStore
andersca@apple.com [Thu, 11 Jun 2015 21:11:45 +0000 (21:11 +0000)]
Reimplement WKOriginDataManager in terms of API::WebsiteDataStore
https://bugs.webkit.org/show_bug.cgi?id=145889

Reviewed by Dan Bernstein.

* UIProcess/API/C/WKAPICast.h:
* UIProcess/API/C/WKContext.cpp:
(WKContextGetOriginDataManager):
* UIProcess/API/C/WKOriginDataManager.cpp:
(WKOriginDataManagerGetTypeID):
(WKOriginDataManagerGetOrigins):
(WKOriginDataManagerDeleteEntriesForOrigin):
(WKOriginDataManagerDeleteEntriesModifiedBetweenDates):
(WKOriginDataManagerDeleteAllEntries):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185474 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoFull screen video space can be resized to 1pt wide
bfulgham@apple.com [Thu, 11 Jun 2015 21:06:23 +0000 (21:06 +0000)]
Full screen video space can be resized to 1pt wide
https://bugs.webkit.org/show_bug.cgi?id=145895
<rdar://problem/20907914>

Reviewed by Simon Fraser.

Prior to the upcoming OS X, full screen mode was not resizable. Now that it is,
we need to set a minimum window width that is large enough to hold the media
controls. This makes our full-screen video behave the same as the QuickTimeX player.

* UIProcess/mac/WKFullScreenWindowController.mm:
(-[WKFullScreenWindowController finishedEnterFullScreenAnimation:]): Active that new
kWKLayoutModeDynamicSizeWithMinimumViewSize style, and set the minimum view size to
be the width of our media controls, with 20 pixels of padding on either side.
(-[WKFullScreenWindowController finishedExitFullScreenAnimation:]): Set the layout
mode back to the default style.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185473 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUpdate media controls JS and CSS to use picture-in-picture
jonlee@apple.com [Thu, 11 Jun 2015 20:53:49 +0000 (20:53 +0000)]
Update media controls JS and CSS to use picture-in-picture
https://bugs.webkit.org/show_bug.cgi?id=145827
<rdar://problem/21311576>

Reviewed by Dean Jackson.

Source/WebCore:

* English.lproj/mediaControlsLocalizedStrings.js: Update localized strings.
Add a string for the placeholder text, and for the aria-label of the
placard.

* Modules/mediacontrols/MediaControlsHost.idl: Remove uncalled functions.
* Modules/mediacontrols/MediaControlsHost.h: Also remove optimizedFullscreenSupported(),
which was never used.
* Modules/mediacontrols/MediaControlsHost.cpp: Remove uncalled functions and media part IDs.
(WebCore::MediaControlsHost::enterFullscreenOptimized): Deleted.
(WebCore::MediaControlsHost::mediaUIImageData): Deleted.

* Modules/mediacontrols/mediaControlsApple.css:
(audio::-webkit-media-controls-picture-in-picture-button): Update pseudo-element name.
(audio::-webkit-media-controls-optimized-fullscreen-button): Deleted.
* Modules/mediacontrols/mediaControlsApple.js: Update global variable name, and classes.
(Controller.prototype.createControls): Update button name. Add aria-label to the placeholder.
(Controller.prototype.updateFullscreenButtons):
(Controller.prototype.updateLayoutForDisplayedWidth):
(Controller.prototype.handlePictureInPictureButtonClicked): Update handler names.
(Controller.prototype.handleOptimizedFullscreenButtonClicked): Deleted.

* Modules/mediacontrols/mediaControlsiOS.css: Rename pseudo-element.
(audio::-webkit-media-controls-picture-in-picture-button): Add mask images.
(audio::-webkit-media-controls-picture-in-picture-button:active):
(video::-webkit-media-controls-picture-in-picture-button):
(video::-webkit-media-controls-picture-in-picture-button.return-from-picture-in-picture):
(audio::-webkit-media-controls-wireless-playback-status.picture-in-picture):
(audio::-webkit-media-controls-wireless-playback-text-top.picture-in-picture):
(audio::-webkit-media-controls-wireless-playback-text-bottom.picture-in-picture):
(audio::-webkit-media-controls-optimized-fullscreen-button): Deleted.
(audio::-webkit-media-controls-optimized-fullscreen-button:active): Deleted.
(video::-webkit-media-controls-optimized-fullscreen-button): Deleted.
(audio::-webkit-media-controls-wireless-playback-status.optimized): Deleted. The placard
pseudo-element should be updated from wireless-playback-status to just playback-status
since it serves both for pip and AirPlay.
(audio::-webkit-media-controls-wireless-playback-text-top.optimized): Deleted.
(audio::-webkit-media-controls-wireless-playback-text-bottom.optimized): Deleted.
* Modules/mediacontrols/mediaControlsiOS.js:
(ControllerIOS.prototype.createControls): Update button names.
(ControllerIOS.prototype.configureInlineControls):
(ControllerIOS.prototype.handlePictureInPictureButtonClicked): Update handler name.
(ControllerIOS.prototype.handlePictureInPictureTouchStart):
(ControllerIOS.prototype.handlePictureInPictureTouchEnd):
(ControllerIOS.prototype.handlePictureInPictureTouchCancel):
(ControllerIOS.prototype.handlePresentationModeChange): Remove changes to the inline bg
image style, and rely on the CSS to fill those in. Use localized strings for the placard.
Move the aria-label to createControls().
(ControllerIOS.prototype.handleOptimizedFullscreenButtonClicked): Deleted.
(ControllerIOS.prototype.handleOptimizedFullscreenTouchStart): Deleted.
(ControllerIOS.prototype.handleOptimizedFullscreenTouchEnd): Deleted.
(ControllerIOS.prototype.handleOptimizedFullscreenTouchCancel): Deleted.

* html/HTMLMediaElement.cpp: Remove enterFullscreenOptimized since it is never called.
(WebCore::HTMLMediaElement::enterFullscreenOptimized): Deleted.
* html/HTMLMediaElement.h:
* platform/ios/WebCoreSystemInterfaceIOS.h: Remove wkGetMediaUIImageData.
* platform/ios/WebCoreSystemInterfaceIOS.mm:
* rendering/RenderThemeIOS.mm:
(WebCore::RenderThemeIOS::mediaControlsStyleSheet): Remove calls to wkGetMediaUIImageData.

Source/WebKit/mac:

* WebCoreSupport/WebSystemInterface.mm:
(InitWebCoreSystemInterface): Deleted GetMediaUIImageData.

Source/WebKit2:

* WebProcess/WebCoreSupport/mac/WebSystemInterface.mm:
(InitWebCoreSystemInterface): Deleted GetMediaUIImageData.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185472 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoWeb Inspector: font-family names interpreted as color
commit-queue@webkit.org [Thu, 11 Jun 2015 20:51:11 +0000 (20:51 +0000)]
Web Inspector: font-family names interpreted as color
https://bugs.webkit.org/show_bug.cgi?id=123468

Patch by Devin Rousso <drousso@apple.com> on 2015-06-11
Reviewed by Timothy Hatcher.

* UserInterface/Views/CodeMirrorAdditions.js: Added logic to prevent color picker buttons from appearing before colors that come after a quote character (both " and ').

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185471 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAdd an appearance keyword for wireless playback / airplay icon
dino@apple.com [Thu, 11 Jun 2015 20:50:34 +0000 (20:50 +0000)]
Add an appearance keyword for wireless playback / airplay icon
https://bugs.webkit.org/show_bug.cgi?id=145892
<rdar://problem/21344872>

Reviewed by Simon Fraser.

Source/WebCore:

Add a new "-webkit-appearance" value named "-apple-airplay"
which will be used to render an icon. This change simply
adds support for the value, and puts in hooks for drawing.

At the moment this is only going to be used by Airplay, so
the public-facing name is "-apple-airplay". However, the
implementation in Theme has a more generic name, in case
we hook it up for other ports later.

Test: fast/css/appearance-airplay.html

* css/CSSParser.cpp:
(WebCore::cssValueKeywordID): This new keyword should not get
translated into having a -webkit prefix.
* css/CSSPrimitiveValueMappings.h:
(WebCore::CSSPrimitiveValue::CSSPrimitiveValue): New mapping from
the Theme to the CSS value.
* css/CSSValueKeywords.in: Add "-apple-airplay".
* platform/ThemeTypes.h: New ControlPart.
* rendering/RenderTheme.cpp: Stub implementation of the rendering and
style update for the new appearance.
(WebCore::RenderTheme::adjustStyle):
(WebCore::RenderTheme::paint):
(WebCore::RenderTheme::adjustWirelessPlaybackIconStyle):
(WebCore::RenderTheme::paintWirelessPlaybackIcon):
* rendering/RenderTheme.h:
* rendering/RenderThemeMac.h:
* rendering/RenderThemeMac.mm:
(WebCore::RenderThemeMac::adjustWirelessPlaybackIconStyle):
(WebCore::RenderThemeMac::paintWirelessPlaybackIcon):

LayoutTests:

Simple test that makes sure we parse and evaluate
the new appearance value.

* fast/css/appearance-airplay-expected.txt: Added.
* fast/css/appearance-airplay.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185470 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUpdate plugins/embed-attributes-style.html to not use the QuickTime plug-in
ap@apple.com [Thu, 11 Jun 2015 20:23:34 +0000 (20:23 +0000)]
Update plugins/embed-attributes-style.html to not use the QuickTime plug-in
https://bugs.webkit.org/show_bug.cgi?id=145891

Reviewed by Anders Carlsson.

* platform/mac/plugins/embed-attributes-style-expected.png:
* plugins/embed-attributes-style.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185469 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed, rolling out r185464.
commit-queue@webkit.org [Thu, 11 Jun 2015 20:21:07 +0000 (20:21 +0000)]
Unreviewed, rolling out r185464.
https://bugs.webkit.org/show_bug.cgi?id=145894

"This patch is breaking binding tests" (Requested by youenn on
#webkit).

Reverted changeset:

"Bindings generator should generate code for Promise-based
APIs"
https://bugs.webkit.org/show_bug.cgi?id=145833
http://trac.webkit.org/changeset/185464

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185468 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed, rolling out r185465.
commit-queue@webkit.org [Thu, 11 Jun 2015 20:12:39 +0000 (20:12 +0000)]
Unreviewed, rolling out r185465.
https://bugs.webkit.org/show_bug.cgi?id=145893

"This patch is breaking 32bit mac build" (Requested by youenn
on #webkit).

Reverted changeset:

"[Streams API] ReadableJSStream should handle promises
returned by JS source start callback"
https://bugs.webkit.org/show_bug.cgi?id=145792
http://trac.webkit.org/changeset/185465

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185467 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoSpeculative build fix for GTK after r185463.
achristensen@apple.com [Thu, 11 Jun 2015 20:09:37 +0000 (20:09 +0000)]
Speculative build fix for GTK after r185463.

* CMakeLists.txt:
Remove WebPluginSiteDataManager.cpp from CMake build.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185466 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[Streams API] ReadableJSStream should handle promises returned by JS source start...
youenn.fablet@crf.canon.fr [Thu, 11 Jun 2015 19:30:25 +0000 (19:30 +0000)]
[Streams API] ReadableJSStream should handle promises returned by JS source start callback
https://bugs.webkit.org/show_bug.cgi?id=145792

Source/JavaScriptCore:

Reviewed by Darin Adler.

Added support for JSFunction implemented by std::function.

* runtime/JSFunction.cpp:
(JSC::getNativeExecutable): Refactored code to share it with the two JSFunction::create
(JSC::JSFunction::create):
(JSC::runStdFunction):
* runtime/JSFunction.h: Added std::function based JSFunction::create prototype.
* runtime.JSPromise.h:

Source/WebCore:

Reviewed by Darin Adler.

Covered by rebased tests.

When calling start callback, the returned value is checked.
If it is not a promise, we do as if it is a resolved promise.
If it is a promise, we call its then() method with two resolve/reject JS functions.

* Modules/streams/ReadableStream.cpp:
* bindings/js/ReadableJSStream.cpp:
(WebCore::ReadableJSStream::invoke): Returns a JSPromise* if any is returned by the JS source callback.
(WebCore::thenPromise): Utility method to call the promise.
(WebCore::createStartResultFulfilledFunction): The promise resolve callback.
(WebCore::ReadableJSStream::doStart): Calls thenPromise if a JSPromise* is returned by invoke.
(WebCore::ReadableJSStream::ReadableJSStream):
* bindings/js/ReadableJSStream.h:

LayoutTests:

Reviewed by Darin Adler.

Rebasing expectations, and removing timeouts for tests that no longer timeout.

* streams/reference-implementation/readable-stream-expected.txt:
* streams/reference-implementation/readable-stream.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185465 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoBindings generator should generate code for Promise-based APIs
youenn.fablet@crf.canon.fr [Thu, 11 Jun 2015 19:26:47 +0000 (19:26 +0000)]
Bindings generator should generate code for Promise-based APIs
https://bugs.webkit.org/show_bug.cgi?id=145833

Reviewed by Darin Adler.

Covered by existing tests.

* Modules/webaudio/AudioContext.idl: Removing custom binding for resume, suspend and close.
* bindings/js/JSAudioContextCustom.cpp: Ditto.
* bindings/scripts/CodeGeneratorJS.pm:
(GenerateHeader): Refactoring to use IsReturningPromise.
(GenerateImplementation): Disabling include for return type if it is a promise.
(GenerateParametersCheck): Adding DeferredWrapper() as argument to the DOM method if JS method returns a promise.
(GenerateImplementationFunctionCall): Added support for promise-returning API.
(IsReturningPromise): Checking whether function is returning a promise.
* bindings/scripts/test/GObject/WebKitDOMTestObj.cpp:
(webkit_dom_test_obj_test_promise_function):
* bindings/scripts/test/GObject/WebKitDOMTestObj.h:
* bindings/scripts/test/JS/JSTestObj.cpp:
(WebCore::jsTestObjPrototypeFunctionTestPromiseFunction):
* bindings/scripts/test/ObjC/DOMTestObj.h:
* bindings/scripts/test/ObjC/DOMTestObj.mm:
(-[DOMTestObj testPromiseFunction]):
* bindings/scripts/test/TestObj.idl: Adding promise returning function.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185464 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoRemove WebPluginSiteDataManager
andersca@apple.com [Thu, 11 Jun 2015 18:33:51 +0000 (18:33 +0000)]
Remove WebPluginSiteDataManager
https://bugs.webkit.org/show_bug.cgi?id=145884

Reviewed by Andreas Kling.

* PluginProcess/PluginProcess.cpp:
(WebKit::PluginProcess::clearSiteData): Deleted.
* PluginProcess/PluginProcess.h:
* PluginProcess/PluginProcess.messages.in:
* UIProcess/API/C/WKAPICast.h:
* UIProcess/API/C/WKContext.cpp:
* UIProcess/API/C/WKPluginSiteDataManager.cpp:
* UIProcess/Plugins/PluginProcessManager.cpp:
(WebKit::PluginProcessManager::clearSiteData): Deleted.
* UIProcess/Plugins/PluginProcessManager.h:
* UIProcess/Plugins/PluginProcessProxy.cpp:
(WebKit::PluginProcessProxy::clearSiteData): Deleted.
(WebKit::PluginProcessProxy::pluginProcessCrashedOrFailedToLaunch): Deleted.
(WebKit::PluginProcessProxy::didFinishLaunching): Deleted.
(WebKit::PluginProcessProxy::didClearSiteData): Deleted.
* UIProcess/Plugins/PluginProcessProxy.h:
* UIProcess/Plugins/PluginProcessProxy.messages.in:
* UIProcess/Plugins/WebPluginSiteDataManager.cpp: Removed.
(WebKit::WebPluginSiteDataManager::GetSitesWithDataState::GetSitesWithDataState): Deleted.
(WebKit::WebPluginSiteDataManager::GetSitesWithDataState::getSitesWithDataForNextPlugin): Deleted.
(WebKit::WebPluginSiteDataManager::GetSitesWithDataState::didGetSitesWithDataForSinglePlugin): Deleted.
(WebKit::WebPluginSiteDataManager::ClearSiteDataState::ClearSiteDataState): Deleted.
(WebKit::WebPluginSiteDataManager::ClearSiteDataState::clearSiteDataForNextPlugin): Deleted.
(WebKit::WebPluginSiteDataManager::ClearSiteDataState::didClearSiteDataForSinglePlugin): Deleted.
(WebKit::WebPluginSiteDataManager::create): Deleted.
(WebKit::WebPluginSiteDataManager::WebPluginSiteDataManager): Deleted.
(WebKit::WebPluginSiteDataManager::~WebPluginSiteDataManager): Deleted.
(WebKit::WebPluginSiteDataManager::invalidate): Deleted.
(WebKit::WebPluginSiteDataManager::getSitesWithData): Deleted.
(WebKit::WebPluginSiteDataManager::didGetSitesWithData): Deleted.
(WebKit::WebPluginSiteDataManager::clearSiteData): Deleted.
(WebKit::WebPluginSiteDataManager::didClearSiteData): Deleted.
(WebKit::WebPluginSiteDataManager::didGetSitesWithDataForSinglePlugin): Deleted.
(WebKit::WebPluginSiteDataManager::didGetSitesWithDataForAllPlugins): Deleted.
(WebKit::WebPluginSiteDataManager::didClearSiteDataForSinglePlugin): Deleted.
(WebKit::WebPluginSiteDataManager::didClearSiteDataForAllPlugins): Deleted.
* UIProcess/Plugins/WebPluginSiteDataManager.h: Removed.
(WebKit::WebPluginSiteDataManager::processPool): Deleted.
(WebKit::WebPluginSiteDataManager::clearProcessPool): Deleted.
* UIProcess/WebProcessPool.cpp:
(WebKit::WebProcessPool::~WebProcessPool): Deleted.
* UIProcess/WebProcessPool.h:
* UIProcess/WebProcessProxy.cpp:
* WebKit2.xcodeproj/project.pbxproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185463 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoMissing content in UIWebView snapshots sometimes
simon.fraser@apple.com [Thu, 11 Jun 2015 18:01:44 +0000 (18:01 +0000)]
Missing content in UIWebView snapshots sometimes
https://bugs.webkit.org/show_bug.cgi?id=145779
rdar://problem/20744708

Reviewed by Darin Adler.

When expanding collapsed parts of quoted text on iOS, sometimes the "concertina"
images would be blank. This happened because we'd fail to create tiles for regions
outside the LegacyTileCache's visibleRect.

Fix by giving LegacyTileCache an optional override visibleRect, which is set
and cleared during -renderInContext: using the context clip rect.

* platform/ios/LegacyTileCache.h:
* platform/ios/LegacyTileCache.mm:
(WebCore::LegacyTileCache::visibleRectInLayer):
(WebCore::LegacyTileCache::setOverrideVisibleRect):
* platform/ios/LegacyTileLayer.mm:
(-[LegacyTileHostLayer renderInContext:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185461 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoMake WKWebsiteDataStoreRef backed by an API::WebsiteDataStore
andersca@apple.com [Thu, 11 Jun 2015 17:45:35 +0000 (17:45 +0000)]
Make WKWebsiteDataStoreRef backed by an API::WebsiteDataStore
https://bugs.webkit.org/show_bug.cgi?id=145882

Reviewed by Dan Bernstein.

* UIProcess/API/C/WKAPICast.h:
* UIProcess/API/C/WKContext.cpp:
(WKContextGetPluginSiteDataManager):
* UIProcess/API/C/WKPluginSiteDataManager.cpp:
(WKPluginSiteDataManagerGetTypeID):
(WKPluginSiteDataManagerGetSitesWithData):
(WKPluginSiteDataManagerClearSiteData):
(WKPluginSiteDataManagerClearAllSiteData):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185460 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoSimplify event handling logic for search cancel button
dbates@webkit.org [Thu, 11 Jun 2015 17:20:53 +0000 (17:20 +0000)]
Simplify event handling logic for search cancel button
https://bugs.webkit.org/show_bug.cgi?id=145780

Reviewed by Darin Adler.

Following the patch for <https://bugs.webkit.org/show_bug.cgi?id=145774> we can take advantage
of the DOM click event to simplify the event handling logic for the search cancel button.
It is sufficient to listen for mousedown and click events instead of listening for mousedown
and mouseup events, capturing all events during drag, and using the hover state of the button
on mouse release together with a instance variable set on mouse press to determine whether
the cancel button was clicked.

As a side effect of removing the requirement that the search cancel button be hovered (a
workaround to ensure that the mouse was pressed and released on the button in absence of
listening for an explicit DOM click event) we can remove the accessibility-specific class,
AccessibilitySearchFieldCancelButton, that was primarily used to fake the hover state of
the button so that the accessibility machinery could simulate a click on it. Instead we
can add use the default accessibility machinery for clicking on an element.

No new tests since no functionality was changed.

* CMakeLists.txt: Remove reference to file AccessibilitySearchFieldButtons.cpp.
* WebCore.vcxproj/WebCore.vcxproj: Remove reference to files AccessibilitySearchFieldButtons.{cpp, h}.
* WebCore.vcxproj/WebCore.vcxproj.filters: Ditto.
* WebCore.xcodeproj/project.pbxproj: Ditto.
* accessibility/AXObjectCache.cpp:
(WebCore::createFromRenderer): Remove special-purpose logic that instantiated
AccessibilitySearchFieldCancelButton for the search cancel button. Instead we can make
use of the existing accessibility machinery to handle this button.
* accessibility/AccessibilityAllInOne.cpp: Remove #include of AccessibilitySearchFieldButtons.cpp
since the file is no longer needed.
* accessibility/AccessibilitySearchFieldButtons.cpp: Removed.
* accessibility/AccessibilitySearchFieldButtons.h: Removed.
* dom/Element.h:
(WebCore::Element::isSearchFieldCancelButtonElement): Deleted.
* html/shadow/TextControlInnerElements.cpp:
(WebCore::SearchFieldCancelButtonElement::SearchFieldCancelButtonElement): Set the ARIA role
and label attributes so that the accessibility system identifies the search cancel button as
a button and shows a description for the button, respectively.
(WebCore::SearchFieldCancelButtonElement::defaultEventHandler): Simplify the event handling logic
such that we clear the field when we receive a DOM click event (i.e. the mouse was pressed
and released on the button). Remove unused variable, oldValue. Substitute emptyString() for the
empty string literal in the call to HTMLInputElement::setValueForUser() to avoid an implicit
conversion to String. Also remove out-of-date comment that repeated what the code does.
(WebCore::SearchFieldCancelButtonElement::willDetachRenderers): Deleted. We no longer need to use
custom style resolve callbacks to reset the state of the EventHandler since we no longer need to
have the EventHandler dispatch all mouse events to us on mouse press.
* html/shadow/TextControlInnerElements.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185459 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoNetwork process crashes decoding invalid cache entry on 32bit system
antti@apple.com [Thu, 11 Jun 2015 17:03:02 +0000 (17:03 +0000)]
Network process crashes decoding invalid cache entry on 32bit system
https://bugs.webkit.org/show_bug.cgi?id=145842
rdar://problem/21228334

Reviewed by Anders Carlsson.

After cache scheme changes we may end up decoding invalid cache entries. This is by design,
we should just fail decoding and delete these entries.

However Decoder::bufferIsLargeEnoughToContain test in some cases would allow corrupted large
sizes due to overflow in 32bit pointer math and we would crash when allocating a string.

* NetworkProcess/cache/NetworkCacheCoders.cpp:
(WebKit::NetworkCache::Coder<CString>::decode):
(WebKit::NetworkCache::decodeStringText):
(WebKit::NetworkCache::Coder<WebCore::CertificateInfo>::decode):
(WebKit::NetworkCache::Coder<MD5::Digest>::encode):
* NetworkProcess/cache/NetworkCacheCoders.h:
* NetworkProcess/cache/NetworkCacheDecoder.cpp:
(WebKit::NetworkCache::Decoder::Decoder):
(WebKit::NetworkCache::Decoder::bufferIsLargeEnoughToContain):

    Reshuffle to avoid sum.

(WebKit::NetworkCache::Decoder::decodeFixedLengthData):
* NetworkProcess/cache/NetworkCacheDecoder.h:
(WebKit::NetworkCache::Decoder::bufferSize):
(WebKit::NetworkCache::Decoder::currentOffset):
(WebKit::NetworkCache::Decoder::length): Deleted.
(WebKit::NetworkCache::Decoder::isInvalid): Deleted.
(WebKit::NetworkCache::Decoder::markInvalid): Deleted.

    Remove these, they are not really used or needed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185458 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoRewrite WKPluginSiteDataManager using WebsiteDataStore functions
andersca@apple.com [Thu, 11 Jun 2015 17:01:46 +0000 (17:01 +0000)]
Rewrite WKPluginSiteDataManager using WebsiteDataStore functions
https://bugs.webkit.org/show_bug.cgi?id=145868

Reviewed by Antti Koivisto.

* UIProcess/API/C/WKPluginSiteDataManager.cpp:
(WKPluginSiteDataManagerGetSitesWithData):
(WKPluginSiteDataManagerClearSiteData):
(WKPluginSiteDataManagerClearAllSiteData):
(toNPClearSiteDataFlags): Deleted.
* UIProcess/Plugins/WebPluginSiteDataManager.h:
(WebKit::WebPluginSiteDataManager::processPool):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185457 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed, rolling out r185453.
commit-queue@webkit.org [Thu, 11 Jun 2015 16:00:40 +0000 (16:00 +0000)]
Unreviewed, rolling out r185453.
https://bugs.webkit.org/show_bug.cgi?id=145881

it broke the 32-bit build (Requested by clopez on #webkit).

Reverted changeset:

"[GTK] [Wayland] Should be possible to build with support for
both X11 and Wayland."
https://bugs.webkit.org/show_bug.cgi?id=145701
http://trac.webkit.org/changeset/185453

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185456 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoWeb Inspector: Rendering Frames timeline should distinguish between layout and painting
mattbaker@apple.com [Thu, 11 Jun 2015 15:42:02 +0000 (15:42 +0000)]
Web Inspector: Rendering Frames timeline should distinguish between layout and painting
https://bugs.webkit.org/show_bug.cgi?id=145856

Reviewed by Timothy Hatcher.

This patch formalizes the runloop task concept in the frontend, adds new Paint task type to the Rendering
Frames timeline, and reorders UI elements such as chart sections and frame "segments" to match the order in
which tasks are executed within the runloop.

We will need to make UI changes to the standard Timelines view in a follow up patch, since tree element icons
for Paint records now use a different color that those of other Layout records.

* UserInterface/Images/TimelineRecordPaint.svg:
Changed to use green color.

* UserInterface/Models/RenderingFrameTimelineRecord.js:
(WebInspector.RenderingFrameTimelineRecord):
(WebInspector.RenderingFrameTimelineRecord.displayNameForTaskType):
(WebInspector.RenderingFrameTimelineRecord.prototype.durationForTask.get validRecordForTaskType):
(WebInspector.RenderingFrameTimelineRecord.prototype.durationForTask.set return):
(WebInspector.RenderingFrameTimelineRecord.prototype.durationForTask):
(WebInspector.RenderingFrameTimelineRecord.prototype.get durationRemainder): Deleted.
(WebInspector.RenderingFrameTimelineRecord.prototype.durationForRecords.get var): Deleted.
Added TaskType enum to decouple runloop task types from timeline record types, and removed duration remainder
support now that "Other" has an associated task type.

* UserInterface/Views/RenderingFrameTimelineDataGridNode.js:
(WebInspector.RenderingFrameTimelineDataGridNode.prototype.get data):
(WebInspector.RenderingFrameTimelineDataGridNode.prototype.createCellContent):
* UserInterface/Views/RenderingFrameTimelineView.js:
(WebInspector.RenderingFrameTimelineView):
Added new grid column for Paint task, reordered columns to be consistent with the rest of the UI.

* UserInterface/Views/TimelineRecordFrame.css:
(.timeline-record-frame > .frame > .duration.rendering-frame-timeline-record-script):
(.timeline-record-frame > .frame > .duration.rendering-frame-timeline-record-layout):
(.timeline-record-frame > .frame > .duration.rendering-frame-timeline-record-paint):
(.timeline-record-frame > .frame > .duration): Deleted.
(.timeline-record-frame > .frame > .duration.timeline-record-type-network): Deleted.
(.timeline-record-frame > .frame > .duration.timeline-record-type-layout): Deleted.
(.timeline-record-frame > .frame > .duration.timeline-record-type-script): Deleted.
Updated styles for Render Frame tasks.

* UserInterface/Views/TimelineRecordFrame.js:
(WebInspector.TimelineRecordFrame.prototype._updateChildElements.createDurationElement):
Updated to use Render Frame task types. Frame segment insertion order is now consistent with the rest of the UI.

* UserInterface/Views/TimelineSidebarPanel.js:
(WebInspector.TimelineSidebarPanel._refreshFrameSelectionChart.chartData.Object.keys.map):
(WebInspector.TimelineSidebarPanel.set this):
(WebInspector.TimelineSidebarPanel._refreshFrameSelectionChart.get this):
(WebInspector.TimelineSidebarPanel): Deleted.
(WebInspector.TimelineSidebarPanel.prototype._refreshFrameSelectionChart.durationForRecordType): Deleted.
(WebInspector.TimelineSidebarPanel.prototype._refreshFrameSelectionChart): Deleted.
Updated to use Render Frame task types.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185455 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoWeb Inspector: Wrong overview graph shown after switching from Frames to Timelines
mattbaker@apple.com [Thu, 11 Jun 2015 15:24:55 +0000 (15:24 +0000)]
Web Inspector: Wrong overview graph shown after switching from Frames to Timelines
https://bugs.webkit.org/show_bug.cgi?id=145778

Reviewed by Timothy Hatcher.

TimelineSidebarPanel stores a reference to the selected timeline before switching to the Rendering Frames view,
which is used to restore the tree selection, overview graph, and content views when switching back to the
standard Timelines view. However if another recording is created before switching, the previously selected
timeline tracked by the sidebar cannot be restored, since it doesn't belong to the active recording.

This is addressed by storing the previously selected timeline type, rather than the timeline object itself.
The type is also updated whenever the selected timeline in the tree outline changes, ensuring that the correct
timeline will always be restored.

* UserInterface/Views/TimelineSidebarPanel.js:
(WebInspector.TimelineSidebarPanel.prototype.treeElementForRepresentedObject.get if.get if):
(WebInspector.TimelineSidebarPanel.get else): Deleted.
(WebInspector.TimelineSidebarPanel.prototype.treeElementForRepresentedObject.get if): Deleted.
(WebInspector.TimelineSidebarPanel.prototype.treeElementForRepresentedObject): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185454 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[GTK] [Wayland] Should be possible to build with support for both X11 and Wayland.
clopez@igalia.com [Thu, 11 Jun 2015 15:21:07 +0000 (15:21 +0000)]
[GTK] [Wayland] Should be possible to build with support for both X11 and Wayland.
https://bugs.webkit.org/show_bug.cgi?id=145701

Reviewed by Žan Doberšek.

.:

* Source/cmake/OptionsGTK.cmake: Remove conflicting options.

Source/WebCore:

No new tests, no behavior changes.

When building both targets, we have to include the wayland-egl
headers in order to build the Wayland target. This causes that
EGLNativePixmapType and EGLNativeWindowType get defined as
different types than when building only the X11 target.

By type casting them to the ones that are expected, we are able
to build both targets at the same time.

I have done tests (building each target alone as also both targets
at the same time), and everything seems to works as expected.

Once built for both targets, if you try to launch the MiniBrowser
from inside a Wayland compositor (Weston on top of X for example),
it will trigger the X11 target if the DISPLAY environment variable
is set and the environment variable GDK_BACKEND is not set to wayland,
otherwise it will trigger the Wayland target.

* platform/graphics/GLContext.cpp:
(WebCore::GLContext::createContextForWindow): Add type cast.
* platform/graphics/egl/GLContextEGL.cpp: Add missing include when
building both targets that is required for defining DefaultRootWindow().
(WebCore::GLContextEGL::createPixmapContext): Add type cast.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185453 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago3% cold PLT regression from network cache on iOS
antti@apple.com [Thu, 11 Jun 2015 12:05:25 +0000 (12:05 +0000)]
3% cold PLT regression from network cache on iOS
https://bugs.webkit.org/show_bug.cgi?id=145694
rdar://problem/21158245

Reviewed by Chris Dumez.

Cache does not help in cold page loads but it shouldn't' be hurting either. Write I/O needs to be toned down a bit.

* NetworkProcess/cache/NetworkCacheIOChannelCocoa.mm:
(WebKit::NetworkCache::IOChannel::IOChannel):
(WebKit::NetworkCache::IOChannel::open):

    Dispatch channels inherit their I/O priority from the target queue. Use background queue for write I/O.

* NetworkProcess/cache/NetworkCacheStorage.cpp:
(WebKit::NetworkCache::Storage::Storage):
(WebKit::NetworkCache::Storage::dispatchPendingReadOperations):
(WebKit::NetworkCache::Storage::dispatchPendingWriteOperations):

    Only write one file at a time instead of maximum of three.

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

    For consistency with store prepend new entries here too.

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

    Delay start of the first write operation by 1s.
    Prepend instead of append to the pending write deque so retrieveFromMemory lookup finds newest entries first in case of duplicates

(WebKit::NetworkCache::Storage::traverse):
* NetworkProcess/cache/NetworkCacheStorage.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185452 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[EFL] Fix the debug build after r185417.
commit-queue@webkit.org [Thu, 11 Jun 2015 09:30:55 +0000 (09:30 +0000)]
[EFL] Fix the debug build after r185417.
https://bugs.webkit.org/show_bug.cgi?id=145876

Patch by Hunseop Jeong <hs85.jeong@samsung.com> on 2015-06-11
Reviewed by Gyuyoung Kim.

* platform/graphics/efl/ImageBufferEfl.cpp: added the missing "MIMETypeRegistry.h".

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185451 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[iOS] Tahoma isn't installed on iOS, so there is no need to blacklist it
mmaxfield@apple.com [Thu, 11 Jun 2015 07:00:17 +0000 (07:00 +0000)]
[iOS] Tahoma isn't installed on iOS, so there is no need to blacklist it
https://bugs.webkit.org/show_bug.cgi?id=145865

Reviewed by Dan Bernstein.

No new tests because I can't copy Tahoma into our open source repository.

* platform/graphics/cocoa/FontCocoa.mm:
(WebCore::fontFamilyShouldNotBeUsedForArabic):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185450 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[EFL] dimmed area for popup does not cover the fullscreen mode.
hyuki.kim@samsung.com [Thu, 11 Jun 2015 06:55:48 +0000 (06:55 +0000)]
[EFL] dimmed area for popup does not cover the fullscreen mode.
https://bugs.webkit.org/show_bug.cgi?id=145133

Reviewed by Gyuyoung Kim.

By calling elm_win_resize_object_add(), popup was added as a resize object of elm_win.
as a result, popup can cover the fullscreen area when it enters a fullscreen mode.

* MiniBrowser/efl/main.c:
(on_fullscreen_accept):
(on_fullscreen_deny):
(on_fullscreen_enter):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185449 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoDon't send touch events on link preview.
commit-queue@webkit.org [Thu, 11 Jun 2015 05:48:02 +0000 (05:48 +0000)]
Don't send touch events on link preview.
https://bugs.webkit.org/show_bug.cgi?id=145863

Disable web touch event recognizer on link preview. We do the same when finding nodes in inspector or when long press gesture
is recognized (by _cancelAllTouches).

Patch by Yongjun Zhang <yongjun_zhang@apple.com> on 2015-06-10
Reviewed by Beth Dakin.

* UIProcess/ios/WKContentViewInteraction.mm:
(-[WKContentView willPresentPreviewViewController:forPosition:inSourceView:]): Remove touch event gesture recognizer on preview start.
(-[WKContentView didDismissPreviewViewController:committing:]): Add the gesture recognizer back on dismiss.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185448 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed, rolling out r185442.
commit-queue@webkit.org [Thu, 11 Jun 2015 04:16:24 +0000 (04:16 +0000)]
Unreviewed, rolling out r185442.
https://bugs.webkit.org/show_bug.cgi?id=145871

seems to have broken WTF.StringOperators test (Requested by
alexchristensen on #webkit).

Reverted changeset:

"[Content Extensions] Limit number of rules."
https://bugs.webkit.org/show_bug.cgi?id=145663
http://trac.webkit.org/changeset/185442

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185447 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUpdate presentation mode JS API to using 'picture-in-picture'
jonlee@apple.com [Thu, 11 Jun 2015 02:57:08 +0000 (02:57 +0000)]
Update presentation mode JS API to using 'picture-in-picture'
https://bugs.webkit.org/show_bug.cgi?id=145826
rdar://problem/2131153

Reviewed by Anders Carlsson.

Update presentation modes to use 'picture-in-picture' in lieu of 'optimized'.

* html/HTMLVideoElement.idl: Update to "picture-in-picture".
* html/HTMLVideoElement.cpp:
(WebCore::presentationModePictureInPicture): Rename static function returning string.
(WebCore::presentationModeOptimized): Deleted.
(WebCore::HTMLVideoElement::webkitSupportsPresentationMode): Use the new static function name.
(WebCore::HTMLVideoElement::webkitSetPresentationMode): Ditto.
(WebCore::HTMLVideoElement::webkitPresentationMode): Ditto.

* Modules/mediacontrols/mediaControlsiOS.js: Update to 'picture-in-picture'.
(ControllerIOS.prototype.configureInlineControls):
(ControllerIOS.prototype.isFullScreen):
(ControllerIOS.prototype.handleOptimizedFullscreenButtonClicked):
(ControllerIOS.prototype.handlePresentationModeChange):
(ControllerIOS.prototype.controlsAlwaysVisible):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185446 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoWKWebsiteDataStore should handle removing plug-in data for individual records
andersca@apple.com [Thu, 11 Jun 2015 02:34:03 +0000 (02:34 +0000)]
WKWebsiteDataStore should handle removing plug-in data for individual records
https://bugs.webkit.org/show_bug.cgi?id=145864

Reviewed by Beth Dakin.

* PluginProcess/PluginProcess.cpp:
(WebKit::PluginProcess::deleteWebsiteDataForHostNames):
* PluginProcess/PluginProcess.h:
* PluginProcess/PluginProcess.messages.in:
* UIProcess/Plugins/PluginProcessManager.cpp:
(WebKit::PluginProcessManager::deleteWebsiteDataForHostNames):
* UIProcess/Plugins/PluginProcessManager.h:
* UIProcess/Plugins/PluginProcessProxy.cpp:
(WebKit::PluginProcessProxy::deleteWebsiteDataForHostNames):
(WebKit::PluginProcessProxy::pluginProcessCrashedOrFailedToLaunch):
(WebKit::PluginProcessProxy::didFinishLaunching):
(WebKit::PluginProcessProxy::didDeleteWebsiteDataForHostNames):
* UIProcess/Plugins/PluginProcessProxy.h:
* UIProcess/Plugins/PluginProcessProxy.messages.in:
* UIProcess/WebsiteData/WebsiteDataStore.cpp:
(WebKit::WebsiteDataStore::removeData):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185445 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[Mobile Safari, WKWebView] increase DeviceOrientationEvent events emission frequency
benjamin@webkit.org [Thu, 11 Jun 2015 02:01:46 +0000 (02:01 +0000)]
[Mobile Safari, WKWebView] increase DeviceOrientationEvent events emission frequency
https://bugs.webkit.org/show_bug.cgi?id=145814

Patch by Boris Smus <boris@smus.com> on 2015-06-10
Reviewed by Benjamin Poulain.

Increased the sampling rate of DeviceOrientationEvent from 20 Hz to 60
Hz to make good head tracking possible.

* platform/ios/WebCoreMotionManager.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185444 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[Mobile Safari, WKWebView] increase DeviceOrientationEvent events emission frequency
benjamin@webkit.org [Thu, 11 Jun 2015 01:57:49 +0000 (01:57 +0000)]
[Mobile Safari, WKWebView] increase DeviceOrientationEvent events emission frequency
https://bugs.webkit.org/attachment.cgi?bugid=145814

Patch by Boris Smus <boris@smus.com> on 2015-06-10
Reviewed by Benjamin Poulain.

Increased the sampling rate of DeviceOrientationEvent from 20 Hz to 60
Hz to make good head tracking possible.

* platform/ios/WebCoreMotionManager.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185443 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoSource/WebCore:
achristensen@apple.com [Thu, 11 Jun 2015 00:52:22 +0000 (00:52 +0000)]
Source/WebCore:
[Content Extensions] Limit number of rules.
https://bugs.webkit.org/show_bug.cgi?id=145663
rdar://problem/21242407

Reviewed by Benjamin Poulain.

Added an API test to make sure that parsing fails when there are too many rules.

* contentextensions/ContentExtensionError.cpp:
(WebCore::ContentExtensions::contentExtensionErrorCategory):
* contentextensions/ContentExtensionError.h:
* contentextensions/ContentExtensionParser.cpp:
(WebCore::ContentExtensions::loadEncodedRules):
Fail to parse a content extension with more than 50000 rules.

Tools:
[Content Extensions] Make max NFA size and max rule count user defaults.
https://bugs.webkit.org/show_bug.cgi?id=145663
rdar://problem/21242407

Reviewed by Benjamin Poulain.

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

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185442 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoASSERTION FAILED: s.length() > 1 on LayoutTests/js/regexp-flags.html
utatane.tea@gmail.com [Thu, 11 Jun 2015 00:40:17 +0000 (00:40 +0000)]
ASSERTION FAILED: s.length() > 1 on LayoutTests/js/regexp-flags.html
https://bugs.webkit.org/show_bug.cgi?id=145599

Unreviewed, simple follow up patch.

use jsString instead of jsMakeNontrivialString
since the flag string may be trivial (0 or 1 length).

* runtime/RegExpPrototype.cpp:
(JSC::regExpProtoGetterFlags):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185440 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoWKWebsiteDataStore should handle removing plug-in data
andersca@apple.com [Thu, 11 Jun 2015 00:27:39 +0000 (00:27 +0000)]
WKWebsiteDataStore should handle removing plug-in data
https://bugs.webkit.org/show_bug.cgi?id=145862

Reviewed by Sam Weinig.

* PluginProcess/PluginProcess.cpp:
(WebKit::PluginProcess::deleteWebsiteData):
* PluginProcess/PluginProcess.h:
* PluginProcess/PluginProcess.messages.in:
* UIProcess/Plugins/PluginProcessManager.cpp:
(WebKit::PluginProcessManager::deleteWebsiteData):
* UIProcess/Plugins/PluginProcessManager.h:
* UIProcess/Plugins/PluginProcessProxy.cpp:
(WebKit::PluginProcessProxy::~PluginProcessProxy):
(WebKit::PluginProcessProxy::deleteWebsiteData):
(WebKit::PluginProcessProxy::pluginProcessCrashedOrFailedToLaunch):
(WebKit::PluginProcessProxy::didFinishLaunching):
(WebKit::PluginProcessProxy::didDeleteWebsiteData):
* UIProcess/Plugins/PluginProcessProxy.h:
* UIProcess/Plugins/PluginProcessProxy.messages.in:
* UIProcess/WebsiteData/WebsiteDataStore.cpp:
(WebKit::WebsiteDataStore::removeData):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185439 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoREGRESSION (r184895): Vertical border elements ([-webkit]-border-image set to 'repeat...
commit-queue@webkit.org [Wed, 10 Jun 2015 23:50:33 +0000 (23:50 +0000)]
REGRESSION (r184895): Vertical border elements ([-webkit]-border-image set to 'repeat') that used to render perfectly are now rendering incorrectly.
https://bugs.webkit.org/show_bug.cgi?id=145801

Patch by Said Abou-Hallawa <sabouhallawa@apple.com> on 2015-06-10
Reviewed by Simon Fraser.

Source/WebCore:

When using the tiling to draw the sides of an image-border, the image slice
is repeated only in one direction. For top and bottom sides, the slice is
repeated horizontally and for left and right sides, it's repeated vertically.
The tile might be scaled in the other direction of the tiling if the border
extent and the slice extent are different in this direction.

The bug happens because we were scaling the tile in the tiling direction.

Test: fast/borders/border-image-repeat-stretch.html

* rendering/style/NinePieceImage.cpp:
(WebCore::NinePieceImage::scaleSlicesIfNeeded): Fix the type and the name
of deviceScaleFactor.

(WebCore::NinePieceImage::computeIntrinsicSideTileScale): Fix the scaling
direction of the border image side tiling.

* rendering/style/NinePieceImage.h:

LayoutTests:

* fast/borders/border-image-repeat-stretch-expected.html: Added.
* fast/borders/border-image-repeat-stretch.html: Added.
* fast/borders/resources/border-image-repeat-stretch-expected.svg: Added.
* fast/borders/resources/border-image.svg: Added.
Ensure that the sides of a border-images are drawn correctly in the cases
of 'repeat' and 'stretch'.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185438 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoJavaScript: Drop the “escaped reserved words as identifiers” compatibility measure
utatane.tea@gmail.com [Wed, 10 Jun 2015 23:43:20 +0000 (23:43 +0000)]
JavaScript: Drop the “escaped reserved words as identifiers” compatibility measure
https://bugs.webkit.org/show_bug.cgi?id=90678

Reviewed by Darin Adler.

After ES6, escaped reserved words in identifiers are prohibited.
After parsing Identifier, we should perform `m_buffer16.shrink(0)`.

* parser/Lexer.cpp:
(JSC::Lexer<CharacterType>::parseIdentifierSlowCase):
* tests/mozilla/ecma_3/Unicode/uc-003.js:
(test): Deleted.
* tests/stress/reserved-word-with-escape.js: Added.
(testSyntax):
(testSyntaxError):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185437 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoMeta refresh preference is not respected on fast path WebView
jhoneycutt@apple.com [Wed, 10 Jun 2015 23:35:36 +0000 (23:35 +0000)]
Meta refresh preference is not respected on fast path WebView
initialization

<https://bugs.webkit.org/show_bug.cgi?id=145861>
<rdar://problem/21309911>

Reviewed by Brady Eidson.

* WebView/WebView.mm:
(-[WebView initSimpleHTMLDocumentWithStyle:frame:preferences:groupName:]):
Set the WebCore setting from the WebPreferences preference.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185436 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoASSERT_WITH_SECURITY_IMPLICATION in WebCore::DocumentOrderedMap::getElementById
cdumez@apple.com [Wed, 10 Jun 2015 23:07:55 +0000 (23:07 +0000)]
ASSERT_WITH_SECURITY_IMPLICATION in WebCore::DocumentOrderedMap::getElementById
https://bugs.webkit.org/show_bug.cgi?id=145857
<rdar://problem/16798440>

Reviewed by Darin Adler.

Source/WebCore:

Make sure Node::insertedInto() gets called on the inserted node and its
descendants after its insertion into the tree but *before*
ContainerNode::childrenChanged() is called on the parent node. This is
needed so that the descendants know they've been inserted into the tree
(and their InDocumentFlag flag gets set) before the parent node does
anything with them in childrenChanged().

In the case of <rdar://problem/16798440>, executing HTMLScriptElement's
childrenChanged() after appending a child to a script element was causing
the script to be executed. The script would call getElementBy() which
would traverse the DOM tree and find a matching Element in the newly
inserted subtree. However, the matching Element's InDocumentFlag flag was
not set yet because the element's insertedInto() method has not been called
yet at this point. This would cause us to hit an assertion as
DocumentOrderedMap::getElementById() is only supposed to return elements
that are in a Document.

This patch is based on Blink r178976 by <esprehn@chromium.org>:
https://src.chromium.org/viewvc/blink?view=rev&revision=178976

Tests: fast/dom/script-getElementById-during-insertion.html
       fast/dom/script-remove-child-id-map.html

* dom/ContainerNode.cpp:
(WebCore::ContainerNode::notifyChildInserted):
(WebCore::ContainerNode::notifyChildRemoved):
(WebCore::ContainerNode::removeChildren):
(WebCore::ContainerNode::parserInsertBefore): Deleted.
(WebCore::ContainerNode::removeChild): Deleted.
(WebCore::ContainerNode::parserRemoveChild): Deleted.
(WebCore::ContainerNode::parserAppendChild): Deleted.
(WebCore::ContainerNode::childrenChanged): Deleted.
(WebCore::ContainerNode::setAttributeEventListener): Deleted.
(WebCore::ContainerNode::querySelector): Deleted.
* dom/ContainerNodeAlgorithms.cpp:
(WebCore::ChildNodeInsertionNotifier::notifyDescendantInsertedIntoDocument):
(WebCore::ChildNodeInsertionNotifier::notifyDescendantInsertedIntoTree):
* dom/ContainerNodeAlgorithms.h:
(WebCore::ChildNodeInsertionNotifier::notifyNodeInsertedIntoDocument):
(WebCore::ChildNodeInsertionNotifier::notifyNodeInsertedIntoTree):
(WebCore::ChildNodeInsertionNotifier::notify):
(WebCore::ChildNodeRemovalNotifier::notifyNodeRemovedFromDocument): Deleted.
* dom/Element.cpp:
(WebCore::Element::addShadowRoot):

LayoutTests:

Add layout tests covering different crashes caused by the same bug.

* fast/dom/script-getElementById-during-insertion-expected.txt: Added.
* fast/dom/script-getElementById-during-insertion.html: Added.

Reduction test case for <rdar://problem/16798440>.

* fast/dom/script-remove-child-id-map-expected.txt: Added.
* fast/dom/script-remove-child-id-map.html: Added.

Test imported from Blink r178976.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185435 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[Web Timing] Fix flaky test.
commit-queue@webkit.org [Wed, 10 Jun 2015 22:53:42 +0000 (22:53 +0000)]
[Web Timing] Fix flaky test.
https://bugs.webkit.org/show_bug.cgi?id=145846

Patch by Alex Christensen <achristensen@webkit.org> on 2015-06-10
Reviewed by Alexey Proskuryakov.

The timing data is gathered in ResourceHandle::getConnectionTimingData as
millisecond deltas from the fetch start time, not the navigation start time.
The difference between navigation and fetch start time is usually so small that
it only caused one flaky test, but this should fix that flakiness. This patch
corrects how the millisecond deltas are used.

* page/PerformanceTiming.cpp:
(WebCore::PerformanceTiming::domainLookupStart):
(WebCore::PerformanceTiming::domainLookupEnd):
(WebCore::PerformanceTiming::connectStart):
(WebCore::PerformanceTiming::connectEnd):
(WebCore::PerformanceTiming::secureConnectionStart):
(WebCore::PerformanceTiming::requestStart):
(WebCore::PerformanceTiming::responseStart):
(WebCore::PerformanceTiming::responseEnd):
(WebCore::PerformanceTiming::documentLoadTiming):
(WebCore::PerformanceTiming::resourceLoadTimeRelativeToFetchStart):
(WebCore::PerformanceTiming::monotonicTimeToIntegerMilliseconds):
(WebCore::PerformanceTiming::resourceLoadTimeRelativeToAbsolute): Deleted.
* page/PerformanceTiming.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185434 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed, fix an typo in unused template code.
basile_clement@apple.com [Wed, 10 Jun 2015 22:44:12 +0000 (22:44 +0000)]
Unreviewed, fix an typo in unused template code.

* wtf/TinyPtrSet.h:
(WTF::TinyPtrSet::iterator::operator*): s/at(index)/at(m_index)/

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185433 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoImplement RegExp.prototype.flags
commit-queue@webkit.org [Wed, 10 Jun 2015 22:32:36 +0000 (22:32 +0000)]
Implement RegExp.prototype.flags
https://bugs.webkit.org/show_bug.cgi?id=145599

Patch by Jordan Harband <ljharb@gmail.com> on 2015-06-10
Reviewed by Geoffrey Garen.
Source/JavaScriptCore:

Per https://people.mozilla.org/~jorendorff/es6-draft.html#sec-get-regexp.prototype.flags

* runtime/CommonIdentifiers.h:
* runtime/RegExpPrototype.cpp:
(JSC::flagsString):
(JSC::regExpProtoFuncToString):
(JSC::regExpProtoGetterFlags):
* tests/stress/static-getter-in-names.js:

LayoutTests:

* js/Object-getOwnPropertyNames-expected.txt:
* js/regexp-flags-expected.txt: Added.
* js/regexp-flags.html: Added.
* js/script-tests/Object-getOwnPropertyNames.js:
* js/script-tests/regexp-flags.js: Added.
(.get shouldBe):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185432 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoOverriding the overlay scrollbar style on WKView doesn't take effect immediately
bdakin@apple.com [Wed, 10 Jun 2015 21:53:42 +0000 (21:53 +0000)]
Overriding the overlay scrollbar style on WKView doesn't take effect immediately
https://bugs.webkit.org/show_bug.cgi?id=145855
-and corresponding-
rdar://problem/20948706

Reviewed by Simon Fraser.

Since we opt into layer-per-part scrollbars in order to move the knob on the
secondary thread, AppKit is creating its own layers for the knob and track. To
invalidate them, we need to call into the ScrollbarPainter’s setNeedsDisplay.

New virtual function on ScrollAnimator to call into ScrollbarPainter’s
setNeedsDispay.
* platform/ScrollAnimator.h:

Call the new function when setting the overlay style.
* platform/ScrollableArea.cpp:
(WebCore::ScrollableArea::setScrollbarOverlayStyle):

Add setNeedsDisplay to the list of ScrollbarPainter methods.
* platform/mac/NSScrollerImpDetails.h:

Implement invalidateScrollbarPartLayers.
* platform/mac/ScrollAnimatorMac.h:
* platform/mac/ScrollAnimatorMac.mm:
(WebCore::ScrollAnimatorMac::invalidateScrollbarPartLayers):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185431 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoMove all content extension tests to same directory to fix flakiness.
achristensen@apple.com [Wed, 10 Jun 2015 21:30:23 +0000 (21:30 +0000)]
Move all content extension tests to same directory to fix flakiness.
https://bugs.webkit.org/show_bug.cgi?id=145854

Reviewed by Alexey Proskuryakov.

* http/tests/contentextensions/loading/main-resource-redirect-blocked-expected.txt: Removed.
* http/tests/contentextensions/loading/main-resource-redirect-blocked.php: Removed.
* http/tests/contentextensions/loading/main-resource-redirect-blocked.php.json: Removed.
* http/tests/contentextensions/loading/resources/main-resource-redirect-blocked-target.html: Removed.
* http/tests/contentextensions/main-resource-redirect-blocked-expected.txt: Copied from http/tests/contentextensions/loading/main-resource-redirect-blocked-expected.txt.
* http/tests/contentextensions/main-resource-redirect-blocked.php: Copied from http/tests/contentextensions/loading/main-resource-redirect-blocked.php.
* http/tests/contentextensions/main-resource-redirect-blocked.php.json: Copied from http/tests/contentextensions/loading/main-resource-redirect-blocked.php.json.
* http/tests/contentextensions/resources/main-resource-redirect-blocked-target.html: Copied from http/tests/contentextensions/loading/resources/main-resource-redirect-blocked-target.html.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185430 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoMediaRemoteControls should be marked JSGenerateToJSObject.
commit-queue@webkit.org [Wed, 10 Jun 2015 20:50:48 +0000 (20:50 +0000)]
MediaRemoteControls should be marked JSGenerateToJSObject.
https://bugs.webkit.org/show_bug.cgi?id=145849

Patch by Matt Rajca <mrajca@apple.com> on 2015-06-10
Reviewed by Eric Carlson.

Since MediaRemoteControls has a parent interface, we need to add the JSGenerateToJSObject attribute so the
toJS(...) function gets generated. This is required for event handling to work.

* Modules/mediasession/MediaRemoteControls.idl:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185429 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[WK2] Do not indicate Media Keys are present when they are not
bfulgham@apple.com [Wed, 10 Jun 2015 20:47:15 +0000 (20:47 +0000)]
[WK2] Do not indicate Media Keys are present when they are not
https://bugs.webkit.org/show_bug.cgi?id=145713
<rdar://problem/19752566>

Reviewed by Dean Jackson.

Correct the reporting of Media Key presence by checking for the actual
media key file, not just the directory. Sometimes the directory will exist
with no media key files (e.g., if a user navigates to the directory using
the Finder).

* UIProcess/WebsiteData/WebsiteDataStore.cpp:
(WebKit::computeMediaKeyFile): Moved earlier in file so it could be used by
the 'mediaKeyOrigins' method.
(WebKit::WebsiteDataStore::mediaKeyOrigins): Only indicate we have a media key
origin if the folder actually contains the media key file.
(WebKit::WebsiteDataStore::removeMediaKeys): Use the existing 'computeMediaKeyFile',
rather than manually building the file from strings.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185428 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoDFG ASSERTION FAILED: !iterate() on stress/singleton-scope-then-overwrite.js.ftl...
fpizlo@apple.com [Wed, 10 Jun 2015 20:35:14 +0000 (20:35 +0000)]
DFG ASSERTION FAILED: !iterate() on stress/singleton-scope-then-overwrite.js.ftl-eager
https://bugs.webkit.org/show_bug.cgi?id=145853

Unreviewed, remove the assertion.

* dfg/DFGCSEPhase.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185427 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUser can not scroll on page when dragging inside video
dino@apple.com [Wed, 10 Jun 2015 18:56:32 +0000 (18:56 +0000)]
User can not scroll on page when dragging inside video
https://bugs.webkit.org/show_bug.cgi?id=145848
<rdar://problem/20997158>

Reviewed by Brent Fulgham.

Revert the change made in http://trac.webkit.org/changeset/183797
which breaks panning on pages if the touch starts inside the video.

* Modules/mediacontrols/mediaControlsiOS.js:
(ControllerIOS.prototype.handleWrapperTouchStart): Do not return true, which stops
the handleEvent call in the superclass from preventing the default action.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185426 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoJetStream should have a more rational story for jitter-oriented latency tests
fpizlo@apple.com [Wed, 10 Jun 2015 18:44:50 +0000 (18:44 +0000)]
JetStream should have a more rational story for jitter-oriented latency tests
https://bugs.webkit.org/show_bug.cgi?id=145762

Reviewed by Geoffrey Garen.

JetStream has some latency tests that are meant to measure jitter.  Prior to this change, they
did this by computing the RMS.  But the RMS is a pretty bad metric.  The thing that it rewards
isn't really the thing that you'd want your browser to do.  These RMS-based tests involve taking
the geomean of the RMS of some samples and the sample average.  The lower the geomean, the better
(in the JetStream harness we then invert the scores so that higher is better, but let's ignore
that for this discussion and assume that lower is better).  Here's an example of how this can go
bad.  A browser that always computes a task in some horribly long time (say, 1000ms) but never
varies that time will perform better than a browser that usually computes the task super quickly
(say, 10ms) and sometimes just a little bit less quickly (say, 15ms).  The former browser will
have an RMS of 0 and an average of 1000.  The latter will have a RMS somewhere around 3.5 and an
average of 12.5 (assuming equal probability of 10ms and 15ms).  The geomean of (0, 1000) is 0.
The geomean of (3.5, 12.5) is 6.6.  Lower is better, so the former browser scores higher - even
though it's obviously never better to have a browser always complete a task in 1000ms when a
different browser can do it in 15ms in the worst case.

JetStream should not have this pathology.  The right way of avoiding it is to replace RMS with
some other metric of how bad things get.  A good metric is the average of the worst percentile.
The worst 1% or the worst 5% would be good things to average.  This will catch cases where the VM
jittered due to JIT or GC, but it never have the pathology that we end up giving the better score
to a VM whose best case is worst than another VM's worst case.

For now, this change uses the highest samples above the 95% percentile. I'm not yet sure if that
is the best thing - it might include too many scores that are around the best-case performance -
but it's certainly better than RMS and it might be good enough to keep. But because of that
uncertainty, I'm setting the version to be "1.1-alpha1" to indicate that we aren't ready to
release this yet.

* JetStream/Octane2/base.js:
(.this.Setup.setup.setup):
(.this.TearDown.tearDown.tearDown):
(BenchmarkSuite.GeometricMeanTime):
(BenchmarkSuite.AverageAbovePercentile):
(BenchmarkSuite.GeometricMeanLatency):
(BenchmarkSuite.prototype.NotifyStep):
(BenchmarkSuite.prototype.RunSingleBenchmark):
* JetStream/Octane2/mandreel.js:
(setupMandreel):
(updateMandreelStats):
(startMandreelTimer):
(latencyMandreel):
(tearDownMandreel):
(RMSMandreel): Deleted.
* JetStream/Octane2/splay.js:
(GenerateKey):
(SplayUpdateStats):
(InsertNewNode):
(SplayTearDown):
(SplayRMS): Deleted.
* JetStream/create.rb:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185425 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoExplicitly keep track of active HTMLMediaElements in MediaSessions.
commit-queue@webkit.org [Wed, 10 Jun 2015 18:43:52 +0000 (18:43 +0000)]
Explicitly keep track of active HTMLMediaElements in MediaSessions.
https://bugs.webkit.org/show_bug.cgi?id=145829

Patch by Matt Rajca <mrajca@apple.com> on 2015-06-10
Reviewed by Eric Carlson.

* Modules/mediasession/MediaSession.cpp: Add support for keeping track of active media elements.
(WebCore::MediaSession::addActiveMediaElement):
* Modules/mediasession/MediaSession.h:
* html/HTMLMediaElement.cpp:
(WebCore::HTMLMediaElement::playInternal): If the paused attribute is true and the readyState attribute has the
  value HAVE_FUTURE_DATA or HAVE_ENOUGH_DATA, the media element becomes an active participating element of the
  media session.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185424 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoDrop unused argument for Node::didNotifySubtreeInsertions()
cdumez@apple.com [Wed, 10 Jun 2015 18:15:23 +0000 (18:15 +0000)]
Drop unused argument for Node::didNotifySubtreeInsertions()
https://bugs.webkit.org/show_bug.cgi?id=145845

Reviewed by Andreas Kling.

* dom/ContainerNodeAlgorithms.h:
(WebCore::ChildNodeInsertionNotifier::notify):
* dom/Node.h:
(WebCore::Node::didNotifySubtreeInsertions):
* html/HTMLFrameElementBase.cpp:
(WebCore::HTMLFrameElementBase::didNotifySubtreeInsertions):
* html/HTMLFrameElementBase.h:
* svg/SVGFEImageElement.cpp:
(WebCore::SVGFEImageElement::didNotifySubtreeInsertions):
* svg/SVGFEImageElement.h:
* svg/SVGMPathElement.cpp:
(WebCore::SVGMPathElement::didNotifySubtreeInsertions):
* svg/SVGMPathElement.h:
* svg/SVGTRefElement.cpp:
(WebCore::SVGTRefElement::didNotifySubtreeInsertions):
* svg/SVGTRefElement.h:
* svg/SVGTextPathElement.cpp:
(WebCore::SVGTextPathElement::didNotifySubtreeInsertions):
* svg/SVGTextPathElement.h:
* svg/animation/SVGSMILElement.cpp:
(WebCore::SVGSMILElement::didNotifySubtreeInsertions):
* svg/animation/SVGSMILElement.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185423 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoSupport fetching website data
andersca@apple.com [Wed, 10 Jun 2015 16:57:20 +0000 (16:57 +0000)]
Support fetching website data
https://bugs.webkit.org/show_bug.cgi?id=145828

Reviewed by Darin Adler.

* UIProcess/WebsiteData/WebsiteDataStore.cpp:
(WebKit::WebsiteDataStore::fetchData):
(WebKit::WebsiteDataStore::plugins):
* UIProcess/WebsiteData/WebsiteDataStore.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185420 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed, rolling out r185414.
commit-queue@webkit.org [Wed, 10 Jun 2015 16:43:13 +0000 (16:43 +0000)]
Unreviewed, rolling out r185414.
https://bugs.webkit.org/show_bug.cgi?id=145844

broke debug and jsc tests (Requested by alexchristensen on
#webkit).

Reverted changeset:

"JavaScript: Drop the “escaped reserved words as identifiers”
compatibility measure"
https://bugs.webkit.org/show_bug.cgi?id=90678
http://trac.webkit.org/changeset/185414

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185419 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed. Fixing the build for the GTK port with Clang.
zandobersek@gmail.com [Wed, 10 Jun 2015 15:58:40 +0000 (15:58 +0000)]
Unreviewed. Fixing the build for the GTK port with Clang.

* UIProcess/gtk/InputMethodFilter.h: Add the <functional> header inclusion
to avoid compilation errors due to missing std::function<>.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185418 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[EFL] Jpeg image export implementation for Canvas.
hyuki.kim@samsung.com [Wed, 10 Jun 2015 15:58:13 +0000 (15:58 +0000)]
[EFL] Jpeg image export implementation for Canvas.
https://bugs.webkit.org/show_bug.cgi?id=145457

Reviewed by Gyuyoung Kim.

Add implementation of jpeg image export on Webkit EFL by using JPEGImageEncoder.

No new tests, fast/canvas/toDataURL-supportedTypes.html can be reused.

* PlatformEfl.cmake:
* platform/MIMETypeRegistry.cpp:
(WebCore::initializeSupportedImageMIMETypesForEncoding):
* platform/graphics/cairo/ImageBufferCairo.cpp:
* platform/graphics/efl/ImageBufferEfl.cpp: Added.
(WebCore::writeFunction):
(WebCore::encodeImage):
(WebCore::ImageBuffer::toDataURL):
* platform/image-encoders/JPEGImageEncoder.cpp:
(WebCore::compressRGBABigEndianToJPEG):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185417 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[GTK] Move KeyBindingTranslator from platform to WebKit2 layer
carlosgc@webkit.org [Wed, 10 Jun 2015 15:17:53 +0000 (15:17 +0000)]
[GTK] Move KeyBindingTranslator from platform to WebKit2 layer
https://bugs.webkit.org/show_bug.cgi?id=145840

Reviewed by Martin Robinson.

It's currently used only by WebKit2 in the UI process and it's
another file built twice because of the GTK2 plugin process.

* PlatformGTK.cmake:
Source/WebKit2:

* UIProcess/API/gtk/WebKitWebViewBase.cpp:
* UIProcess/gtk/KeyBindingTranslator.cpp: Renamed from Source/WebCore/platform/gtk/KeyBindingTranslator.cpp.
(WebKit::backspaceCallback):
(WebKit::selectAllCallback):
(WebKit::cutClipboardCallback):
(WebKit::copyClipboardCallback):
(WebKit::pasteClipboardCallback):
(WebKit::toggleOverwriteCallback):
(WebKit::popupMenuCallback):
(WebKit::showHelpCallback):
(WebKit::deleteFromCursorCallback):
(WebKit::moveCursorCallback):
(WebKit::KeyBindingTranslator::KeyBindingTranslator):
(WebKit::KeyBindingTranslator::commandsForKeyEvent):
* UIProcess/gtk/KeyBindingTranslator.h: Renamed from Source/WebCore/platform/gtk/KeyBindingTranslator.h.
(WebKit::KeyBindingTranslator::addPendingEditorCommand):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185416 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[GTK] Get rid of GetEditorCommandsForKeyEvent sync message
carlosgc@webkit.org [Wed, 10 Jun 2015 13:32:22 +0000 (13:32 +0000)]
[GTK] Get rid of GetEditorCommandsForKeyEvent sync message
https://bugs.webkit.org/show_bug.cgi?id=145598

Reviewed by Žan Doberšek.

Source/WebCore:

* platform/PlatformKeyboardEvent.h:
(WebCore::PlatformKeyboardEvent::PlatformKeyboardEvent):
(WebCore::PlatformKeyboardEvent::commands):
* platform/gtk/KeyBindingTranslator.cpp: Use a single list of
custom key bindings. We don't need to distinguish between key down and
key press commands, since the web editor client already does that
when executing the commands.
(WebCore::KeyBindingTranslator::commandsForKeyEvent): This is
getEditorCommandsForKeyEvent() renamed as commandsForKeyEvent(),
since it now returns the list of commands to simplify the
code. Also simplify the custom key bindings handling, by using the
global list instead of moving it to a HashMap.
(WebCore::KeyBindingTranslator::getEditorCommandsForKeyEvent): Deleted.
* platform/gtk/KeyBindingTranslator.h:

Source/WebKit2:

We are sending GetEditorCommandsForKeyEvent sync message from web
process to the UI process for every key pressed. And if the
keydown event doesn't handle the key, the message is sent again
for the keypress event, so in many cases it happens twice per
keypress. We can get the list of commands when the key press event
happens in the web view, and send it to the web process as part of
the keyboard event like mac port does. In the web process,
commands not inserting text will be handled by keydown and the
rest in keypress without having to use any other IPC message for that.

* Shared/NativeWebKeyboardEvent.h: Add command list parameter to
the constructor.
* Shared/WebEvent.h: Add GTK specific constructor that receives a
list of commands and whether the event was handled by input methods.
(WebKit::WebKeyboardEvent::commands): Returns the list of commands.
* Shared/WebEventConversion.cpp:
(WebKit::WebKit2PlatformKeyboardEvent::WebKit2PlatformKeyboardEvent):
Also copy the commands and handledByInputMethod member.
* Shared/WebKeyboardEvent.cpp:
(WebKit::WebKeyboardEvent::WebKeyboardEvent): Implement GTK
specific constructor that receives a list of commands and whether
the event was handled by input methods.
(WebKit::WebKeyboardEvent::encode): Encode commands and handledByInputMethod.
(WebKit::WebKeyboardEvent::decode): Decode commands and handledByInputMethod.
* Shared/gtk/NativeWebKeyboardEventGtk.cpp:
(WebKit::NativeWebKeyboardEvent::NativeWebKeyboardEvent): Add
command list parameter to the constructor.
* Shared/gtk/WebEventFactory.cpp:
(WebKit::WebEventFactory::createWebKeyboardEvent): Use the new
WebKeyboardEvent constructor and simplify the code. We don't need
to explicitly set VK_PROCESSKEY as windows virtual key code when
the event was handled by input methods. The WebCore event handler
already does that.
* Shared/gtk/WebEventFactory.h:
* UIProcess/API/gtk/PageClientImpl.cpp:
(WebKit::PageClientImpl::getEditorCommandsForKeyEvent): Deleted.
* UIProcess/API/gtk/PageClientImpl.h:
* UIProcess/API/gtk/WebKitWebViewBase.cpp: Add a
KeyBindingTranslator member.
(webkitWebViewBaseKeyPressEvent): Use the new
InputMethodFilter::filterKeyEvent() API and get the list of
commands for the key when events was not handled by input methods.
(webkitWebViewBaseKeyReleaseEvent): Use the new
InputMethodFilter::filterKeyEvent() API.
* UIProcess/PageClient.h:
* UIProcess/WebPageProxy.messages.in: Remove GetEditorCommandsForKeyEvent message.
* UIProcess/gtk/InputMethodFilter.cpp:
(WebKit::InputMethodFilter::handleKeyboardEvent): Call the
completation handler.
(WebKit::InputMethodFilter::handleKeyboardEventWithCompositionResults): Ditto.
(WebKit::InputMethodFilter::filterKeyEvent): Use a completion
handler for the task, so that the caller (the web view) can handle
the keyboard event.
* UIProcess/gtk/InputMethodFilter.h:
* UIProcess/gtk/WebPageProxyGtk.cpp:
(WebKit::WebPageProxy::getEditorCommandsForKeyEvent): Deleted.
* WebProcess/WebCoreSupport/gtk/WebEditorClientGtk.cpp:
(WebKit::WebEditorClient::handleKeyboardEvent): Return earlier if
the event was handled by input methods. And get the list of
commands from the PlatformKeyboardEvent instead of sending a sync
message to the UI process.
(WebKit::WebEditorClient::handleInputMethodKeydown): If the event
was handled by inpout methods, set the event as default handled so
that the WebCore event handler sets VK_PROCESSKEY as windows
virtual key code.
(WebKit::WebEditorClient::getEditorCommandsForKeyEvent): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185415 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoJavaScript: Drop the “escaped reserved words as identifiers” compatibility measure
utatane.tea@gmail.com [Wed, 10 Jun 2015 13:25:38 +0000 (13:25 +0000)]
JavaScript: Drop the “escaped reserved words as identifiers” compatibility measure
https://bugs.webkit.org/show_bug.cgi?id=90678

Reviewed by Darin Adler.

After ES6, escaped reserved words in identifiers are prohibited.

* parser/Lexer.cpp:
(JSC::Lexer<CharacterType>::parseIdentifierSlowCase):
* tests/stress/reserved-word-with-escape.js: Added.
(testSyntax):
(testSyntaxError):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185414 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed. Fix GTK build after r185412.
carlosgc@webkit.org [Wed, 10 Jun 2015 13:07:36 +0000 (13:07 +0000)]
Unreviewed. Fix GTK build after r185412.

Add missing include.

* NetworkProcess/cache/NetworkCacheFileSystem.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185413 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoNetworkCache: Delete old cache versions
antti@apple.com [Wed, 10 Jun 2015 13:00:49 +0000 (13:00 +0000)]
NetworkCache: Delete old cache versions
https://bugs.webkit.org/show_bug.cgi?id=145800

Reviewed by Darin Adler.

* CMakeLists.txt:
* NetworkProcess/cache/NetworkCacheBlobStorage.cpp:
(WebKit::NetworkCache::BlobStorage::synchronize):
* NetworkProcess/cache/NetworkCacheFileSystem.cpp: Added.
(WebKit::NetworkCache::directoryEntryType):
(WebKit::NetworkCache::traverseDirectory):

    Pass entry type as callback argument.

(WebKit::NetworkCache::deleteDirectoryRecursively):

    Add helper.

(WebKit::NetworkCache::fileTimes):
(WebKit::NetworkCache::updateFileModificationTimeIfNeeded):
* NetworkProcess/cache/NetworkCacheFileSystem.h: Copied from Source/WebKit2/NetworkProcess/cache/NetworkCacheFileSystemPosix.h.

    Move code to cpp.

(WebKit::NetworkCache::traverseDirectory): Deleted.
(WebKit::NetworkCache::traverseCacheFiles): Deleted.
(WebKit::NetworkCache::fileTimes): Deleted.
(WebKit::NetworkCache::updateFileModificationTimeIfNeeded): Deleted.
* NetworkProcess/cache/NetworkCacheFileSystemPosix.h: Removed.
* NetworkProcess/cache/NetworkCacheIOChannelCocoa.mm:
* NetworkProcess/cache/NetworkCacheStatistics.cpp:
(WebKit::NetworkCache::Statistics::bootstrapFromNetworkCache):
* NetworkProcess/cache/NetworkCacheStorage.cpp:
(WebKit::NetworkCache::makeBlobDirectoryPath):
(WebKit::NetworkCache::traverseRecordsFiles):

    Move cache hierarchy aware traversal code here from NetworkCacheFileSystem (for better layering).
    Rename for clarity.

(WebKit::NetworkCache::deleteEmptyRecordsDirectories):

    Factor to a function.

(WebKit::NetworkCache::Storage::Storage):
(WebKit::NetworkCache::Storage::synchronize):
(WebKit::NetworkCache::Storage::traverse):
(WebKit::NetworkCache::Storage::clear):

    Use traverseRecordsFiles for traversal.

(WebKit::NetworkCache::Storage::shrink):
(WebKit::NetworkCache::Storage::deleteOldVersions):

    Delete version subdirectories with version number less than the current version.
    Remove code for clearing unversioned V1 caches. No one should have them.

* NetworkProcess/cache/NetworkCacheStorage.h:
* UIProcess/API/APIUserContentExtensionStore.cpp:
* WebKit2.xcodeproj/project.pbxproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185412 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[WinCairo] Crash on exit when terminating egl
peavo@outlook.com [Wed, 10 Jun 2015 11:29:33 +0000 (11:29 +0000)]
[WinCairo] Crash on exit when terminating egl
https://bugs.webkit.org/show_bug.cgi?id=145832

Reviewed by Carlos Garcia Campos.

Avoid terminating egl on exit for WinCairo.

* platform/graphics/PlatformDisplay.cpp:
(WebCore::PlatformDisplay::~PlatformDisplay):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185411 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[GTK] Web Inspector is broken
zandobersek@gmail.com [Wed, 10 Jun 2015 10:33:13 +0000 (10:33 +0000)]
[GTK] Web Inspector is broken
https://bugs.webkit.org/show_bug.cgi?id=145836

Reviewed by Carlos Garcia Campos.

* PlatformGTK.cmake: Thread the InspectorBackendCommands.js file as a separate
dependency of InspectorGResourceBundle.xml. It can't be listed under the InspectorFiles
variable and then searched via the CMake glob command because it probably won't exist
at the time the command is run (since it's copied into the specified path), so it won't
be listed among the dependencies at all.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185410 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[JSC] InlineCallFrame::arguments should be sized-to-fit.
akling@apple.com [Wed, 10 Jun 2015 10:15:12 +0000 (10:15 +0000)]
[JSC] InlineCallFrame::arguments should be sized-to-fit.
<https://webkit.org/b/145782>

Reviewed by Darin Adler.

I spotted this Vector<ValueRecovery> looking a bit chubby in Instruments,
with 354 kB of memory allocated on cnet.com.

Use resizeToFit() instead of resize() since we know the final size up front.

* dfg/DFGByteCodeParser.cpp:
(JSC::DFG::ByteCodeParser::InlineStackEntry::InlineStackEntry):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185409 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[EFL] Unreviewed, EFL gardening on Jun 10th
gyuyoung.kim@webkit.org [Wed, 10 Jun 2015 08:54:33 +0000 (08:54 +0000)]
[EFL] Unreviewed, EFL gardening on Jun 10th

* platform/efl/TestExpectations: Mark fast tests to flaky since r183540.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185408 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoRefactor AudioContext implementation to enable automatic binding generation of promis...
youenn.fablet@crf.canon.fr [Wed, 10 Jun 2015 08:10:44 +0000 (08:10 +0000)]
Refactor AudioContext implementation to enable automatic binding generation of promise-based methods
https://bugs.webkit.org/show_bug.cgi?id=145223

Reviewed by Darin Adler.

Introducing DOMPromise as a typed wrapper above DeferredWrapper.
This DOMPromise allows constraining Promise resolution/rejection to a single resolution type and rejection type.
This might be useful for typed callback usage of promise based APIs, like getUserMedia() or ReadableStreamReader.read().

Applying DOMPromise to AudioContext close, suspend and resume.
Changed binding code to represent what could be automatically generated by binding generator.
Fixing unneeded copies of std::function callbacks.

Disabling DOMPromise copy constructors to lower chances that resolution/rejection is done twice on the same object.

Covered by existing tests.

* Modules/webaudio/AudioContext.cpp:
(WebCore::AudioContext::addReaction):
(WebCore::AudioContext::setState):
(WebCore::AudioContext::suspend):
(WebCore::AudioContext::resume):
(WebCore::AudioContext::close):
* Modules/webaudio/AudioContext.h:
* bindings/js/JSAudioContextCustom.cpp:
(WebCore::JSAudioContext::suspend):
(WebCore::JSAudioContext::resume):
(WebCore::JSAudioContext::close):
* bindings/js/JSDOMPromise.cpp:
(WebCore::DeferredWrapper::callFunction):
* bindings/js/JSDOMPromise.h:
(WebCore::DeferredWrapper::resolve):
(WebCore::DeferredWrapper::reject):
(WebCore::DOMPromise::DOMPromise):
(WebCore::DOMPromise::resolve):
(WebCore::DOMPromise::reject):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185407 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[Streams API] Implement pulling of a source by a ReadableStream
youenn.fablet@crf.canon.fr [Wed, 10 Jun 2015 08:09:02 +0000 (08:09 +0000)]
[Streams API] Implement pulling of a source by a ReadableStream
https://bugs.webkit.org/show_bug.cgi?id=145262

Reviewed by Darin Adler

Source/WebCore:

Introduced abstract ReadableStream::doPull() which is overriden in ReadableJSStream.
Added support to call the "pull" JS callback in ReadableJSStream::doPull().
Added calls to pull as requested by the spec (when resolving a read callback, at start time...).

Fixed issue in ReadableStreamReader::read() (use of successCallback(JSValue()) in lieu of endCallback())

Covered by rebased tests.

* Modules/streams/ReadableStream.cpp:
(WebCore::ReadableStream::start): calling pull() once start.
(WebCore::ReadableStream::pull): calling doPull() if readableStream states requires to.
(WebCore::ReadableStream::read): calling pull() after resolving a read callback.
* Modules/streams/ReadableStream.h:
* Modules/streams/ReadableStreamReader.cpp:
(WebCore::ReadableStreamReader::read): fixed JSValue() bug.
* bindings/js/ReadableJSStream.cpp:
(WebCore::ReadableJSStream::doPull): calling of JS callback.
(WebCore::ReadableJSStream::storeException): catches exception and store them.
(WebCore::ReadableJSStream::storeError): refactoring for checkForException.
(WebCore::ReadableJSStream::enqueue):
* bindings/js/ReadableJSStream.h:

LayoutTests:

Rebased expectations, removed some "timeout: 50" parameters.
Removed a test from streams/reference-implementation/readable-stream.html that cannot pass
until promises returned to start and pull JS callbacks are handled.
Fixed bug in streams-utils.js (was using the old API replaced by controller).

* streams/reference-implementation/bad-underlying-sources-expected.txt:
* streams/reference-implementation/bad-underlying-sources.html:
* streams/reference-implementation/readable-stream-expected.txt:
* streams/reference-implementation/readable-stream.html:
* streams/reference-implementation/resources/streams-utils.js:
(.stream.new.ReadableStream.):
(.stream.new.ReadableStream):
(sequentialReadableStream):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185406 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[EFL] Execute Javascript method in mini browser address bar.
commit-queue@webkit.org [Wed, 10 Jun 2015 07:29:36 +0000 (07:29 +0000)]
[EFL] Execute Javascript method in mini browser address bar.
https://bugs.webkit.org/show_bug.cgi?id=145725

Patch by Hyungwook Lee <hyungwook.lee@navercorp.com> on 2015-06-10
Reviewed by Gyuyoung Kim.

Support 'javascript:' scheme in WebKit EFL Mini Browser.

* MiniBrowser/efl/main.c:
(url_from_user_input):
(url_load_from_user_input):
(on_url_bar_activated):
(elm_main):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185405 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoDeferredWrapper should clear its JS strong references once its promise is resolved...
youenn.fablet@crf.canon.fr [Wed, 10 Jun 2015 06:26:57 +0000 (06:26 +0000)]
DeferredWrapper should clear its JS strong references once its promise is resolved/rejected
https://bugs.webkit.org/show_bug.cgi?id=145753

Reviewed by Darin Adler.

Clear strong references at the end of DeferredWrapper::resolve and DeferredWrapper::reject.
Added assertions to check that resolve or reject is called only once.

Removed DeferredWrapper constructor that creates internally its promise.
Reason is DeferredWrapper can be resolved synchronously and promise
will be lost when returning it at the end of the binding promise function.
Updated all custom bindings accordingly.

Covered by existing tests.

* bindings/js/JSAudioContextCustom.cpp:
(WebCore::JSAudioContext::suspend): Updated DeferredWrapper constructor to properly return the promise.
(WebCore::JSAudioContext::resume): Ditto.
(WebCore::JSAudioContext::close): Ditto.
* bindings/js/JSDOMPromise.cpp:
(WebCore::DeferredWrapper::resolve): Cleared strong references after promise resolution.
(WebCore::DeferredWrapper::reject): Cleared strong references after promise rejection.
* bindings/js/JSDOMPromise.h:
(WebCore::DeferredWrapper::resolve): Added ASSERT to check promise is not yet rejected/resolved.
(WebCore::DeferredWrapper::reject): Ditto.
(WebCore::DeferredWrapper::reject<ExceptionCode>): Ditto.
(WebCore::DeferredWrapper::resolve<String>): Ditto.
(WebCore::DeferredWrapper::resolve<bool>): Ditto.
(WebCore::DeferredWrapper::resolve<JSC::JSValue>): Ditto.
(WebCore::char>>):
(WebCore::DeferredWrapper::reject<String>): Ditto.
* bindings/js/JSMediaDevicesCustom.cpp:
(WebCore::JSMediaDevices::getUserMedia): Updated DeferredWrapper constructor to properly return the promise.
* bindings/js/JSReadableStreamReaderCustom.cpp:
(WebCore::JSReadableStreamReader::read): Ditto.
(WebCore::JSReadableStreamReader::closed): Ditto.
* bindings/js/JSSubtleCryptoCustom.cpp:
(WebCore::JSSubtleCrypto::encrypt): Ditto.
(WebCore::JSSubtleCrypto::decrypt): Ditto.
(WebCore::JSSubtleCrypto::sign): Ditto.
(WebCore::JSSubtleCrypto::verify): Ditto.
(WebCore::JSSubtleCrypto::digest): Ditto.
(WebCore::JSSubtleCrypto::generateKey): Ditto.
(WebCore::JSSubtleCrypto::importKey): Ditto.
(WebCore::JSSubtleCrypto::exportKey): Ditto.
(WebCore::JSSubtleCrypto::wrapKey): Ditto.
(WebCore::JSSubtleCrypto::unwrapKey): Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185404 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoProtect FrameView from being destroyed in Document::recalcStyle()
zalan@apple.com [Wed, 10 Jun 2015 05:59:54 +0000 (05:59 +0000)]
Protect FrameView from being destroyed in Document::recalcStyle()
https://bugs.webkit.org/show_bug.cgi?id=143033
rdar://problem/20326871

Reviewed by Andreas Kling.

This patch ensures that FrameView stays valid in Document::recalcStyle().
It follows the defensive pattern we use to deal with the refcounted FrameView (see EventDispatcher::dispatchEvent)

When the iframe destroys itself in the onBeforeLoad callback (as the result of
PostResolutionCallbackDisabler -> HTMLObjectElement::updateWidget -> guardedDispatchBeforeLoadEvent),
we detach the frame and release the FrameView. However Document::recalcStyle() expects
the FrameView to stay valid.

Covered by fast/frames/flattening/crash-remove-iframe-during-object-beforeload.html.

* dom/Document.cpp:
(WebCore::Document::recalcStyle):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185403 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago3D-transformed video does not display on platforms without accelerated video rendering
commit-queue@webkit.org [Wed, 10 Jun 2015 05:38:03 +0000 (05:38 +0000)]
3D-transformed video does not display on platforms without accelerated video rendering
https://bugs.webkit.org/show_bug.cgi?id=144782

Patch by Daegyu Lee <daegyu.lee@navercorp.com> on 2015-06-09
Reviewed by Simon Fraser.

Source/WebCore:

Video element does not get GraphicsLayer when MediaPlayerPrivate::supportsAcceleratedRendering() returns false
which means not using accelerated video decoding.
Although the video element gets GraphicsLayer by changing the CSS style of video element by javascript,
there is no way to set GraphicsLayer::setDrawsContent(true) for updating the video content.
As a result, after changing the CSS style, the video content does not show.
To avoid missing setDrawsContent(true), add conditions(!supportsAcceleratedRendering() && m_requiresOwnBackingStore).

Test: media/video-transformed-by-javascript.html
      media/video-transformed-by-javascript-expected.html

* rendering/RenderLayerBacking.cpp:
(WebCore::RenderLayerBacking::containsPaintedContent):

LayoutTests:

Add layout test that is simply checking video transform by javascript works properly.

* media/video-transformed-by-javascript-expected.txt: Added.
* media/video-transformed-by-javascript.html: Added.
* platform/gtk/TestExpectations:
* platform/ios-simulator-wk1/TestExpectations:
* platform/ios-simulator-wk2/TestExpectations:
* platform/ios-simulator/TestExpectations:
* platform/mac-wk1/TestExpectations:
* platform/mac-wk2/TestExpectations:
* platform/mac/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185402 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUpdate iOS TestExpectations files
dbates@webkit.org [Wed, 10 Jun 2015 04:00:13 +0000 (04:00 +0000)]
Update iOS TestExpectations files

* platform/ios-simulator-wk1/TestExpectations:
* platform/ios-simulator-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185401 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoImplement dumpProgressFinishedCallback() for Win layoutTestController.
commit-queue@webkit.org [Wed, 10 Jun 2015 03:34:24 +0000 (03:34 +0000)]
Implement dumpProgressFinishedCallback() for Win layoutTestController.
https://bugs.webkit.org/show_bug.cgi?id=66773

Patch by Hyungwook Lee <hyungwook.lee@navercorp.com> on 2015-06-09
Reviewed by Darin Adler.

Source/WebKit/win:

* WebView.cpp:
(WebView::setFrameLoadDelegatePrivate):

Tools:

* DumpRenderTree/win/FrameLoadDelegate.cpp:
* DumpRenderTree/win/FrameLoadDelegate.h:

LayoutTests:

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185400 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed build fix. Some builder names are really long.
rniwa@webkit.org [Wed, 10 Jun 2015 02:00:40 +0000 (02:00 +0000)]
Unreviewed build fix. Some builder names are really long.

* init-database.sql:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185399 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoREGRESSION(r180867): Tabbing to login field on iCloud.com doesn't have highlight...
rniwa@webkit.org [Wed, 10 Jun 2015 00:57:30 +0000 (00:57 +0000)]
REGRESSION(r180867): Tabbing to login field on iCloud.com doesn't have highlight for text
https://bugs.webkit.org/show_bug.cgi?id=145830

Reviewed by Darin Adler.

Source/WebCore:

The bug was caused by setSelection not updating RenderView when there is a style recalc scheduled
that doesn't trigger a layout. Fixed the bug by explicitly updating the selection in
Document::recalcStyle in that case.

Test: editing/selection/update-selection-by-style-change.html

* dom/Document.cpp:
(WebCore::Document::recalcStyle):
* editing/FrameSelection.cpp:
(WebCore::FrameSelection::updateAppearanceAfterLayout): Renamed from didLayout.
* editing/FrameSelection.h:
* page/FrameView.cpp:
(WebCore::FrameView::performPostLayoutTasks):

LayoutTests:

Added a regression test.

* editing/selection/update-selection-by-style-change-expected.html: Added.
* editing/selection/update-selection-by-style-change.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185398 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoMediaControls: Reenable resize of controls on pinch zoom
dino@apple.com [Wed, 10 Jun 2015 00:53:13 +0000 (00:53 +0000)]
MediaControls: Reenable resize of controls on pinch zoom
https://bugs.webkit.org/show_bug.cgi?id=145824
<rdar://problem/21212778>

Reviewed by Darin Adler.

Reinstate the code that updated the scaling of the
controls in response to changes in page scale.
This time around we have to change both the controls
panel, and its blurry background.

* Modules/mediacontrols/mediaControlsApple.js:
(Controller.prototype.set pageScaleFactor): Deleted a comment.
* Modules/mediacontrols/mediaControlsiOS.css:
(video::-webkit-media-controls-panel-background): Set the background to pin
to the bottom of its view.
* Modules/mediacontrols/mediaControlsiOS.js:
(ControllerIOS.prototype.get pageScaleFactor): Basic getter, copied from mediaControlsApple.
(ControllerIOS.prototype.set pageScaleFactor): The setter that reacts to the page scale
and applies an inverse scaling on the control panel using a transform, and adjusts the
height on the background similarly.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185397 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoGraphicsContext state stack wasting lots of memory when empty.
akling@apple.com [Wed, 10 Jun 2015 00:46:51 +0000 (00:46 +0000)]
GraphicsContext state stack wasting lots of memory when empty.
<https://webkit.org/b/145817>

Reviewed by Geoffrey Garen.

Give the GraphicsContextState stack an inline capacity of 1, and make sure
to free any heap-allocated backing store when the stack goes empty.

The 1 is because HTMLCanvasElement keeps one "save" on the underlying
GraphicsContext at all times, and this prevents those canvases from always
sitting on an empty stack with 16 capacity.

This saves ~520 kB on cnet.com video pages.

* platform/graphics/GraphicsContext.cpp:
(WebCore::GraphicsContext::restore):
* platform/graphics/GraphicsContext.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185396 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoSVG Fragment is not rendered if it is the css background image of an HTML element
commit-queue@webkit.org [Wed, 10 Jun 2015 00:45:58 +0000 (00:45 +0000)]
SVG Fragment is not rendered if it is the css background image of an HTML element
https://bugs.webkit.org/show_bug.cgi?id=91790

Patch by Said Abou-Hallawa <sabouhallawa@apple.com> on 2015-06-09
Reviewed by Darin Adler.

Source/WebCore:

To show an SVG fragment, the SVGImage has to scrollToFragment() using
the resource url. The changes http://trac.webkit.org/changeset/164804
and http://trac.webkit.org/changeset/164983 set the url of SVGImage to
to be used later in SVGImage::draw(). The problem is the SVGImage url
is only set when it is the src of an <img> tag. We did not do the same
thing when the SVGImage is the css background image of an HTML element.

The fix is to set the url of the SVGImage always when it's created by
the CachedImage. The CachedImage must have a valid url when the SVGImage
is created.

Test: svg/css/svg-resource-fragment-identifier-background.html

* loader/cache/CachedImage.cpp:
(WebCore::CachedImage::load):
(WebCore::CachedImage::checkShouldPaintBrokenImage):
Replace the calls resourceRequest().url() and m_resourceRequest.url() by
calling url() since they are all the same.

(WebCore::CachedImage::createImage): Pass the resource url to SVGImage
and change ImageObserver& by ImageObserver*, since null is not legal.

* svg/graphics/SVGImage.cpp:
(WebCore::SVGImage::SVGImage):
* svg/graphics/SVGImage.h: Add a url parameter to SVGImage constructor.

* svg/graphics/SVGImageCache.cpp:
(WebCore::SVGImageCache::findImageForRenderer): Add a new helper function.

(WebCore::SVGImageCache::imageSizeForRenderer):
(WebCore::SVGImageCache::imageForRenderer): Code clean up.

* svg/graphics/SVGImageCache.h: Make imageForRenderer() const.

* svg/graphics/SVGImageForContainer.cpp: Remove unneeded header file.

LayoutTests:

* svg/css/svg-resource-fragment-identifier-background-expected.html: Added.
* svg/css/svg-resource-fragment-identifier-background.html: Added.
Ensure that the SVG fragment is displayed correctly when it's used as a
css background image.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185395 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAdd support for toggling playback in MediaSessions and MediaSessionManagers.
commit-queue@webkit.org [Wed, 10 Jun 2015 00:43:05 +0000 (00:43 +0000)]
Add support for toggling playback in MediaSessions and MediaSessionManagers.
https://bugs.webkit.org/show_bug.cgi?id=145822

Patch by Matt Rajca <mrajca@apple.com> on 2015-06-09
Reviewed by Eric Carlson.

* Modules/mediasession/MediaSession.cpp: Toggle playback of all active media elements.
(WebCore::MediaSession::togglePlayback):
* Modules/mediasession/MediaSession.h:
* Modules/mediasession/MediaSessionManager.cpp: Toggle playback of all media sessions as described in the
  Media Session spec.
(WebCore::MediaSessionManager::togglePlayback):
* Modules/mediasession/MediaSessionManager.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185394 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoTakes two delete key presses to delete pasted emoji up-pointing index finger with...
darin@apple.com [Tue, 9 Jun 2015 23:49:49 +0000 (23:49 +0000)]
Takes two delete key presses to delete pasted emoji up-pointing index finger with skin tone
https://bugs.webkit.org/show_bug.cgi?id=145823

Reviewed by Anders Carlsson.

Source/WebCore:

Tests: editing/deleting/delete-emoji.html

* rendering/RenderText.cpp:
(WebCore::isHangulLVT): Use constants instead of macros. Also changed to take a UChar since
the Hangul processing can work on UTF-16 code unit at a time and doesn't have to handle
surrogate pairs.
(WebCore::isMark): Use U_GC_M_MASK instead of writing the algorithm out another way.
(WebCore::isInArmenianToLimbuRange): Added.
(WebCore::RenderText::previousOffsetForBackwardDeletion): Refactored for clarity and to use
the U16_PREV macro instead of doing what it does in a sloppier way. Added code to allow a
variation selector before an emoji modifier to fix the bug. Changed Hangul logic to work a
code unit at a time, since it can, to use an enum class, and to use constants rather than
all capital macros. Also changed the "dumb" case to use a more appropriate ICU macro.

LayoutTests:

* editing/deleting/delete-emoji-expected.txt: Updated to expect a little more testing.
* editing/deleting/delete-emoji.html: Added a test case and streamlined the test a bit.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185393 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agofeComposite filter does not clip the paint rect to its effect rect when the operator...
commit-queue@webkit.org [Tue, 9 Jun 2015 23:24:12 +0000 (23:24 +0000)]
feComposite filter does not clip the paint rect to its effect rect when the operator is 'in' or 'atop'
https://bugs.webkit.org/show_bug.cgi?id=137856

Patch by Said Abou-Hallawa <sabouhallawa@apple.com> on 2015-06-09
Reviewed by Darin Adler.

Source/WebCore:

There was bug in calculating the absolutePaintRect of the feComposite filter
when the operator is equal to 'in' or 'atop'. The absolutePaintRect was set
to the absolutePaintRect of the background FilterEffect which is correct.
What was missing is clipping this rectangle to the maxEffectRect of the
filter which we do for other operators.

Tests: svg/filters/feComposite-background-rect-control-operators.svg

* platform/graphics/IntRect.h:
(WebCore::operator-=):
(WebCore::operator-): Add new operators to IntRect.

* platform/graphics/filters/FEComposite.cpp:
(WebCore::FEComposite::determineAbsolutePaintRect): Make sure the filter
absolutePaintRect is clipped to maxEffectRect for all operators.

(WebCore::FEComposite::platformApplySoftware): Code clean-up.

* platform/graphics/filters/FilterEffect.cpp:
(WebCore::FilterEffect::determineAbsolutePaintRect): Move the clipping
part to a separate function.

(WebCore::FilterEffect::clipAbsolutePaintRect): Clip the absolutePaintRect
to the maxEffectRect of the filter.

* platform/graphics/filters/FilterEffect.h:

LayoutTests:

* svg/filters/feComposite-background-rect-control-operators-expected.svg: Added.
* svg/filters/feComposite-background-rect-control-operators.svg: Added.
Ensure the painting rect of the feComposite filter with operator 'in' or
'atop' is clipped to its bounding rectangle

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185392 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoFix build.
andersca@apple.com [Tue, 9 Jun 2015 23:22:19 +0000 (23:22 +0000)]
Fix build.

* UIProcess/Plugins/WebPluginSiteDataManager.cpp:
(WebKit::WebPluginSiteDataManager::GetSitesWithDataState::getSitesWithDataForNextPlugin):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185391 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoChange the PluginProcessManager data fetching function to take a completion handler
andersca@apple.com [Tue, 9 Jun 2015 23:09:38 +0000 (23:09 +0000)]
Change the PluginProcessManager data fetching function to take a completion handler
https://bugs.webkit.org/show_bug.cgi?id=145821

Reviewed by Sam Weinig.

* UIProcess/Plugins/PluginProcessManager.cpp:
(WebKit::PluginProcessManager::fetchWebsiteData):
(WebKit::PluginProcessManager::getSitesWithData): Deleted.
* UIProcess/Plugins/PluginProcessManager.h:
* UIProcess/Plugins/PluginProcessProxy.cpp:
(WebKit::generateCallbackID):
(WebKit::PluginProcessProxy::~PluginProcessProxy):
(WebKit::PluginProcessProxy::fetchWebsiteData):
(WebKit::PluginProcessProxy::pluginProcessCrashedOrFailedToLaunch):
(WebKit::PluginProcessProxy::didFinishLaunching):
(WebKit::PluginProcessProxy::didGetSitesWithData):
(WebKit::PluginProcessProxy::getSitesWithData): Deleted.
* UIProcess/Plugins/PluginProcessProxy.h:
* UIProcess/Plugins/WebPluginSiteDataManager.cpp:
(WebKit::WebPluginSiteDataManager::GetSitesWithDataState::getSitesWithDataForNextPlugin):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185390 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoImplement MediaSessionManager to keep track of all MediaSessions.
commit-queue@webkit.org [Tue, 9 Jun 2015 22:57:03 +0000 (22:57 +0000)]
Implement MediaSessionManager to keep track of all MediaSessions.
https://bugs.webkit.org/show_bug.cgi?id=145806

Patch by Matt Rajca <mrajca@apple.com> on 2015-06-09
Reviewed by Eric Carlson.

* Modules/mediasession/MediaSession.cpp:
(WebCore::MediaSession::MediaSession): Add ourselves to the MediaSessionManager upon construction.
(WebCore::MediaSession::~MediaSession): Remove ourselves from the MediaSessionManager before destruction.
* Modules/mediasession/MediaSessionManager.cpp: Added to keep track of all living MediaSessions.
(WebCore::MediaSessionManager::singleton):
(WebCore::MediaSessionManager::addMediaSession):
(WebCore::MediaSessionManager::removeMediaSession):
* Modules/mediasession/MediaSessionManager.h: Added.
* WebCore.xcodeproj/project.pbxproj: Added new MediaSessionManager sources.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185389 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[CoordinatedGraphics] Remove RefCounted from Tile
ryuan.choi@navercorp.com [Tue, 9 Jun 2015 22:20:56 +0000 (22:20 +0000)]
[CoordinatedGraphics] Remove RefCounted from Tile
https://bugs.webkit.org/show_bug.cgi?id=145788

Reviewed by Darin Adler.

Since r185140, Tile does not need to be RefCounted.

No new tests, no behavior changes.

* platform/graphics/texmap/coordinated/Tile.cpp:
(WebCore::Tile::create): Deleted.
* platform/graphics/texmap/coordinated/Tile.h:
* platform/graphics/texmap/coordinated/TiledBackingStore.cpp:
(WebCore::TiledBackingStore::invalidate):
(WebCore::TiledBackingStore::updateTileBuffers): Simplifies not to use unnecessary Vector.
(WebCore::TiledBackingStore::coverageRatio):
(WebCore::TiledBackingStore::createTiles):
(WebCore::TiledBackingStore::resizeEdgeTiles):
(WebCore::TiledBackingStore::setKeepRect):
(WebCore::TiledBackingStore::tileAt): Deleted one line wrapper.
(WebCore::TiledBackingStore::setTile): Ditto.
(WebCore::TiledBackingStore::removeTile): Ditto.
* platform/graphics/texmap/coordinated/TiledBackingStore.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185388 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAllow one sync GC per gcTimer interval on critical memory pressure warning
cdumez@apple.com [Tue, 9 Jun 2015 22:15:36 +0000 (22:15 +0000)]
Allow one sync GC per gcTimer interval on critical memory pressure warning
https://bugs.webkit.org/show_bug.cgi?id=145773

Reviewed by Geoffrey Garen.

Source/JavaScriptCore:

On critical memory pressure warning, we were calling GCController::garbageCollectSoon(),
which does not offer any guarantee on when the garbage collection will actually take
place.

On critical memory pressure, we need to free up memory as soon as possible to avoid
getting killed so this is an issue. Also, the fact that we clear the PageCache on
critical memory pressure means a GC would likely be useful, even if the last
collection did not free much memory.

This patch adds a new GCController::garbageCollectNowIfNotDoneRecently() API that allows
one synchronous GC per gcTimer interval on critical memory pressure warning. This makes
us more responsive to critical memory pressure and avoids doing synchronous GCs too
often.

* heap/FullGCActivityCallback.cpp:
(JSC::FullGCActivityCallback::doCollection):
* heap/FullGCActivityCallback.h:
(JSC::GCActivityCallback::createFullTimer):
* heap/GCActivityCallback.h:
* heap/Heap.cpp:
(JSC::Heap::collectAllGarbageIfNotDoneRecently):
* heap/Heap.h:

* heap/IncrementalSweeper.cpp:
(JSC::IncrementalSweeper::doWork): Deleted.
* heap/IncrementalSweeper.h:

Drop fullSweep() API as it no longer seems useful. garbageCollectNow()
already does a sweep after the full collection.

Source/WebCore:

* bindings/js/GCController.cpp:
(WebCore::GCController::garbageCollectNowIfNotDoneRecently):

Add new GCController::garbageCollectNowIfNotDoneRecently() API that
allows one synchronous GC per full GC timer interval. If called more
than once per interval, it becomes equivalent to garbageCollectSoon()
and merely accelerates the next collection.

* bindings/js/GCController.h:
* platform/MemoryPressureHandler.cpp:
(WebCore::MemoryPressureHandler::releaseCriticalMemory):

Call the new GCController::garbageCollectNowIfNotDoneRecently() on
critical memory pressure instead of garbageCollectionSoon() to try
as do a synchronous GC if one wasn't already done recently.
Also drop call to fullSweep() as GCController::garbageCollectNow*()
already do a sweep after the collection.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185387 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAdd code to keep track of plug-in data host names in website data records
andersca@apple.com [Tue, 9 Jun 2015 22:12:41 +0000 (22:12 +0000)]
Add code to keep track of plug-in data host names in website data records
https://bugs.webkit.org/show_bug.cgi?id=145818

Reviewed by Andreas Kling.

* Shared/WebsiteData/WebsiteData.cpp:
(WebKit::WebsiteData::encode):
(WebKit::WebsiteData::decode):
* Shared/WebsiteData/WebsiteData.h:
* UIProcess/WebsiteData/WebsiteDataRecord.cpp:
(WebKit::WebsiteDataRecord::displayNameForPluginDataHostName):
(WebKit::WebsiteDataRecord::addPluginDataHostName):
(WebKit::WebsiteDataRecord::displayNameForCookieHostName): Deleted.
* UIProcess/WebsiteData/WebsiteDataRecord.h:
* UIProcess/WebsiteData/WebsiteDataStore.cpp:
(WebKit::WebsiteDataStore::fetchData):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185386 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAdd Plug-in type declarations to WKWebsiteDataRecord and friends
andersca@apple.com [Tue, 9 Jun 2015 21:33:21 +0000 (21:33 +0000)]
Add Plug-in type declarations to WKWebsiteDataRecord and friends
https://bugs.webkit.org/show_bug.cgi?id=145816

Reviewed by Andreas Kling.

* Shared/WebsiteData/WebsiteDataTypes.h:
* UIProcess/API/Cocoa/WKWebsiteDataRecord.mm:
(dataTypesToString):
* UIProcess/API/Cocoa/WKWebsiteDataRecordInternal.h:
(WebKit::toWebsiteDataTypes):
(WebKit::toWKWebsiteDataTypes):
* UIProcess/API/Cocoa/WKWebsiteDataRecordPrivate.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185383 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[JSC] CodeBlock::m_constantRegisters should be sized-to-fit.
akling@apple.com [Tue, 9 Jun 2015 21:25:40 +0000 (21:25 +0000)]
[JSC] CodeBlock::m_constantRegisters should be sized-to-fit.
<https://webkit.org/b/145784>

Reviewed by Darin Adler.

Spotted this Vector looking chubby on cnet.com, with 1.23 MB of memory
allocated below CodeBlock::setConstantRegisters().

Use resizeToFit() instead since we know the final size up front.
Also removed some unused functions that operated on this constants vector
and the corresponding one in UnlinkedCodeBlock.

* bytecode/CodeBlock.cpp:
(JSC::CodeBlock::addOrFindConstant): Deleted.
(JSC::CodeBlock::findConstant): Deleted.
* bytecode/CodeBlock.h:
(JSC::CodeBlock::setConstantRegisters):
(JSC::CodeBlock::numberOfConstantRegisters): Deleted.
* bytecode/UnlinkedCodeBlock.cpp:
(JSC::UnlinkedCodeBlock::addOrFindConstant): Deleted.
* bytecode/UnlinkedCodeBlock.h:
(JSC::UnlinkedCodeBlock::numberOfConstantRegisters): Deleted.
(JSC::UnlinkedCodeBlock::getConstant): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185382 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[JSC] Polymorphic{Get,Put}ByIdList::addAccess() should optimize for size, not speed.
akling@apple.com [Tue, 9 Jun 2015 21:23:03 +0000 (21:23 +0000)]
[JSC] Polymorphic{Get,Put}ByIdList::addAccess() should optimize for size, not speed.
<https://webkit.org/b/145786>

Reviewed by Darin Adler.

These functions already contained comments saying they optimize for size over speed,
but they were using Vector::resize() which adds the usual slack for faster append().

Switch them over to using Vector::resizeToFit() instead, which makes the Vector
allocate a perfectly sized backing store.

Spotted 670 kB of the GetById ones, and 165 kB of PutById on cnet.com, so these
Vectors are definitely worth shrink-wrapping.

* bytecode/PolymorphicGetByIdList.cpp:
(JSC::PolymorphicGetByIdList::addAccess):
* bytecode/PolymorphicPutByIdList.cpp:
(JSC::PolymorphicPutByIdList::addAccess):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185381 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[JSC] JSPropertyNameEnumerator's property name vector should be sized-to-fit.
akling@apple.com [Tue, 9 Jun 2015 21:21:16 +0000 (21:21 +0000)]
[JSC] JSPropertyNameEnumerator's property name vector should be sized-to-fit.
<https://webkit.org/b/145787>

Reviewed by Darin Adler.

Saw 108 kB worth of JSPropertyNameEnumerator backing store Vectors on cnet.com.
Use Vector::resizeToFit() since we know the perfect size up front.

* runtime/JSPropertyNameEnumerator.cpp:
(JSC::JSPropertyNameEnumerator::finishCreation):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185380 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoFunctionExecutable::isCompiling() is weird and wrong.
akling@apple.com [Tue, 9 Jun 2015 21:20:08 +0000 (21:20 +0000)]
FunctionExecutable::isCompiling() is weird and wrong.
<https://webkit.org/b/145689>

Reviewed by Geoffrey Garen.

Remove FunctionExecutable::isCompiling() and the clearCodeIfNotCompiling() style
functions that called it before throwing away code.

isCompiling() would consider the executable to be "compiling" if it had a CodeBlock
but no JITCode. In practice, every executable gets a JITCode at the same time as it
gets a CodeBlock, by way of prepareForExecutionImpl().

* debugger/Debugger.cpp:
* heap/Heap.cpp:
(JSC::Heap::deleteAllCompiledCode):
(JSC::Heap::deleteAllUnlinkedFunctionCode):
* inspector/agents/InspectorRuntimeAgent.cpp:
(Inspector::TypeRecompiler::visit):
* runtime/Executable.cpp:
(JSC::FunctionExecutable::clearUnlinkedCodeForRecompilation):
(JSC::FunctionExecutable::clearCodeIfNotCompiling): Deleted.
(JSC::FunctionExecutable::clearUnlinkedCodeForRecompilationIfNotCompiling): Deleted.
* runtime/Executable.h:
* runtime/VM.cpp:
(JSC::StackPreservingRecompiler::visit):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185379 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agono, it was not reviewed
darin@apple.com [Tue, 9 Jun 2015 21:19:19 +0000 (21:19 +0000)]
no, it was not reviewed

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185378 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoFollow-up fix for:
darin@apple.com [Tue, 9 Jun 2015 21:18:23 +0000 (21:18 +0000)]
Follow-up fix for:
JavaScript bindings are unnecessarily checking for impossible empty JSValue arguments
https://bugs.webkit.org/show_bug.cgi?id=145811

Reviewed by Antti Koivisto.

There was one unusual case in the bindings generator that was depending on this.

* bindings/js/JSDOMBinding.h:
(WebCore::argumentOrNull): Deleted.
* bindings/scripts/CodeGeneratorJS.pm:
(GenerateParametersCheck): Stop using argumentOrNull.
* bindings/scripts/test/JS/JSTestNamedConstructor.cpp: Updated.
* bindings/scripts/test/JS/JSTestObj.cpp: Updated.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185377 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoREGRESSION(r185262): NetworkCache is in wrong location on OS X
antti@apple.com [Tue, 9 Jun 2015 20:12:15 +0000 (20:12 +0000)]
REGRESSION(r185262): NetworkCache is in wrong location on OS X
https://bugs.webkit.org/show_bug.cgi?id=145812

Reviewed by Anders Carlsson.

* UIProcess/Cocoa/WebProcessPoolCocoa.mm:
(WebKit::WebProcessPool::legacyPlatformDefaultWebSQLDatabaseDirectory):
(WebKit::WebProcessPool::legacyPlatformDefaultLocalStorageDirectory):
(WebKit::WebProcessPool::legacyPlatformDefaultMediaKeysStorageDirectory):
(WebKit::WebProcessPool::isNetworkCacheEnabled):

    Make sure static functions that use NSUserDefaults call registerUserDefaultsIfNeeded() first.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185374 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoJavaScript bindings are unnecessarily checking for impossible empty JSValue arguments
darin@apple.com [Tue, 9 Jun 2015 19:55:02 +0000 (19:55 +0000)]
JavaScript bindings are unnecessarily checking for impossible empty JSValue arguments
https://bugs.webkit.org/show_bug.cgi?id=145811

Reviewed by Antti Koivisto.

At some point, someone who didn't understand JSValue very well added checks for
isEmpty (and the function isEmpty itself) in JavaScript bindings. But the engine
never would pass an empty JSValue to a binding; that's used only inside the
engine itself. I think this was done by some Google V8 experts a while back, so
it's not entirely surprising they didn't understand this. But we don't want to
keep all that unneeded code.

* bindings/js/JSAudioTrackCustom.cpp:
(WebCore::JSAudioTrack::setKind): Removed the isEmpty clause here. I presume this
was copied and pasted from script-generated bindings. While I was at it, I used a
slightly more efficient code path that avoids doing a ref/deref on the string by
using auto&. That's trickier to do correctly for functions with more than one
argument, so I didn't do it in the bindings generator yet.
(WebCore::JSAudioTrack::setLanguage): Ditto.
* bindings/js/JSTextTrackCustom.cpp:
(WebCore::JSTextTrack::setKind): Ditto.
(WebCore::JSTextTrack::setLanguage): Ditto.
* bindings/js/JSVideoTrackCustom.cpp:
(WebCore::JSVideoTrack::setKind): Ditto.
(WebCore::JSVideoTrack::setLanguage): Ditto.

* bindings/scripts/CodeGeneratorJS.pm:
(GenerateImplementation): Use "=" syntax instead of constructor style syntax for
the argument processing. I think this is more readable.
(GenerateParametersCheck): For the enum code path, did the more efficient auto&
idiom to avoid ref/deref on a string. For the toExistingAtomicString path,
restructured so we don't have to check the flag indicating that something is atomic.
Later we could also avoid the ref/deref; added a FIXME about that.
(GetNativeType): Use String, not const String, for local variables. Sure, the
string is const, but so is every other local variable type we use! No need to be
wordy about it.
(JSValueToNative): Removed the uneeded calls to isEmpty. Since the call sites no
longer use construction syntax, used initializer syntax style for a couple classes
that are constructed with multiple arguments.

* bindings/scripts/test/JS/JSTestActiveDOMObject.cpp: Updated.
* bindings/scripts/test/JS/JSTestCustomNamedGetter.cpp: Ditto.
* bindings/scripts/test/JS/JSTestEventTarget.cpp: Ditto.
* bindings/scripts/test/JS/JSTestInterface.cpp: Ditto.
* bindings/scripts/test/JS/JSTestNamedConstructor.cpp: Ditto.
* bindings/scripts/test/JS/JSTestNondeterministic.cpp: Ditto.
* bindings/scripts/test/JS/JSTestObj.cpp: Ditto.
* bindings/scripts/test/JS/JSTestOverloadedConstructors.cpp: Ditto.
* bindings/scripts/test/JS/JSTestSerializedScriptValueInterface.cpp: Ditto.
* bindings/scripts/test/JS/JSTestTypedefs.cpp: Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185373 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoFix unused private field warning in GraphicsLayerUpdater.h
ossy@webkit.org [Tue, 9 Jun 2015 19:29:57 +0000 (19:29 +0000)]
Fix unused private field warning in GraphicsLayerUpdater.h
https://bugs.webkit.org/show_bug.cgi?id=145757

Reviewed by Darin Adler.

* platform/graphics/GraphicsLayerUpdater.cpp:
(WebCore::GraphicsLayerUpdater::GraphicsLayerUpdater):
* platform/graphics/GraphicsLayerUpdater.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185372 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agorun-benchmark should either checkout JetStream 1.0.1 or support running the latest...
rniwa@webkit.org [Tue, 9 Jun 2015 19:29:31 +0000 (19:29 +0000)]
run-benchmark should either checkout JetStream 1.0.1 or support running the latest version
https://bugs.webkit.org/show_bug.cgi?id=145764

Reviewed by Darin Adler.

Added the support for checkout a subversion directory in run-benchmark and used in the plans for
JetStream and SunSpider.

* Scripts/webkitpy/benchmark_runner/benchmark_builder/generic_benchmark_builder.py:
(GenericBenchmarkBuilder):
(GenericBenchmarkBuilder.prepare): Instead of taking each argument from benchmark plan, just accept
the whole JSON as the input.
(GenericBenchmarkBuilder._checkoutWithSubverion): Added.
(GenericBenchmarkBuilder._applyPatch): Check the existence of benchmark_patch in prepare() instead.
* Scripts/webkitpy/benchmark_runner/benchmark_runner.py:
(BenchmarkRunner.execute): Removed the check for having either 'local_copy' or 'remote_archive' in
a benchmark plan now that it's checkedi in GenericBenchmarkBuilder.prepare.
* Scripts/webkitpy/benchmark_runner/data/plans/jetstream.plan:
* Scripts/webkitpy/benchmark_runner/data/plans/sunspider.plan:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185371 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoIntroduce getter definition into static hash tables and use it for getters in RegExp...
utatane.tea@gmail.com [Tue, 9 Jun 2015 18:42:46 +0000 (18:42 +0000)]
Introduce getter definition into static hash tables and use it for getters in RegExp.prototype.
https://bugs.webkit.org/show_bug.cgi?id=145705

Reviewed by Darin Adler.

Source/JavaScriptCore:

In this patch, we introduce Accessor type into property tables.
With Accessor type, create_hash_table creates a static getter property.
This getter property is reified as the same to the static functions.

In the mean time, we only support getter because `putEntry` and `lookupPut`
only work with null setter currently. However, in the spec, there's
no need to add static setter properties. So we will add it if it becomes
necessary in the future.

And at the same time, this patch fixes the issue 145738. Before this patch,
`putEntry` in `JSObject::deleteProperty` adds `undefined` property if
`isValidOffset(...)` is false (deleted). As the result, deleting twice
revives the property with `undefined` value.

If the static functions are reified and the entry is
`BuiltinOrFunctionOrAccessor`, there's no need to execute `putEntry` with
static hash table entry. They should be handled in the normal structure's
looking up because they should be already reified. So added guard for this.

* CMakeLists.txt:
* DerivedSources.make:
* JavaScriptCore.vcxproj/JavaScriptCore.vcxproj:
* JavaScriptCore.vcxproj/JavaScriptCore.vcxproj.filters:
* JavaScriptCore.xcodeproj/project.pbxproj:
* create_hash_table:
* runtime/JSObject.cpp:
(JSC::getClassPropertyNames):
(JSC::JSObject::put):
(JSC::JSObject::deleteProperty):
(JSC::JSObject::reifyStaticFunctionsForDelete):
* runtime/Lookup.cpp:
(JSC::reifyStaticAccessor):
(JSC::setUpStaticFunctionSlot):
* runtime/Lookup.h:
(JSC::HashTableValue::propertyGetter):
(JSC::HashTableValue::propertyPutter):
(JSC::HashTableValue::accessorGetter):
(JSC::HashTableValue::accessorSetter):
(JSC::getStaticPropertySlot):
(JSC::getStaticValueSlot):
(JSC::putEntry):
(JSC::reifyStaticProperties):
* runtime/PropertySlot.h:
* runtime/RegExpObject.cpp:
(JSC::RegExpObject::getOwnPropertySlot):
(JSC::regExpObjectGlobal): Deleted.
(JSC::regExpObjectIgnoreCase): Deleted.
(JSC::regExpObjectMultiline): Deleted.
(JSC::appendLineTerminatorEscape<LChar>): Deleted.
(JSC::appendLineTerminatorEscape<UChar>): Deleted.
(JSC::regExpObjectSourceInternal): Deleted.
(JSC::regExpObjectSource): Deleted.
* runtime/RegExpPrototype.cpp:
(JSC::RegExpPrototype::getOwnPropertySlot):
(JSC::regExpProtoGetterGlobal):
(JSC::regExpProtoGetterIgnoreCase):
(JSC::regExpProtoGetterMultiline):
(JSC::appendLineTerminatorEscape<LChar>):
(JSC::appendLineTerminatorEscape<UChar>):
(JSC::regExpProtoGetterSourceInternal):
(JSC::regExpProtoGetterSource):
* tests/stress/static-function-delete.js: Added.
(shouldBe):
* tests/stress/static-function-put.js: Added.
(shouldBe):
* tests/stress/static-getter-delete.js: Added.
(shouldBe):
(shouldThrow):
* tests/stress/static-getter-descriptors.js: Added.
(shouldBe):
* tests/stress/static-getter-enumeration.js: Added.
(shouldBe):
* tests/stress/static-getter-get.js: Added.
(shouldBe):
* tests/stress/static-getter-in-names.js: Added.
(shouldBe):
* tests/stress/static-getter-names.js: Added.
(shouldBe):
* tests/stress/static-getter-put.js: Added.
(shouldBe):
(shouldThrow):

LayoutTests:

Adjust RegExp.prototype and instance's getter layout.
This is a breaking change from ES5.
So ES5 conformance tests like sputnik is needed to be fixed.

* ietestcenter/Javascript/15.2.3.3-4-212-expected.txt:
* ietestcenter/Javascript/15.2.3.3-4-213-expected.txt:
* ietestcenter/Javascript/15.2.3.3-4-214-expected.txt:
* ietestcenter/Javascript/15.2.3.3-4-215-expected.txt:
* ietestcenter/Javascript/TestCases/15.2.3.3-4-212.js:
(ES5Harness.registerTest):
* ietestcenter/Javascript/TestCases/15.2.3.3-4-213.js:
(ES5Harness.registerTest):
* ietestcenter/Javascript/TestCases/15.2.3.3-4-214.js:
(ES5Harness.registerTest):
* ietestcenter/Javascript/TestCases/15.2.3.3-4-215.js:
(ES5Harness.registerTest):
* js/Object-getOwnPropertyNames-expected.txt:
* js/dom/getOwnPropertyDescriptor-expected.txt:
* js/mozilla/strict/15.10.7-expected.txt:
* js/mozilla/strict/script-tests/15.10.7.js:
* js/pic/cached-named-property-getter.html:
* js/resources/getOwnPropertyDescriptor.js:
(.get descriptorShouldBe):
* js/script-tests/Object-getOwnPropertyNames.js:
* sputnik/Conformance/15_Native_Objects/15.10_RegExp/15.10.7/15.10.7.1_source/S15.10.7.1_A10.html:
* sputnik/Conformance/15_Native_Objects/15.10_RegExp/15.10.7/15.10.7.1_source/S15.10.7.1_A8.html:
* sputnik/Conformance/15_Native_Objects/15.10_RegExp/15.10.7/15.10.7.1_source/S15.10.7.1_A9.html:
* sputnik/Conformance/15_Native_Objects/15.10_RegExp/15.10.7/15.10.7.2_global/S15.10.7.2_A10.html:
* sputnik/Conformance/15_Native_Objects/15.10_RegExp/15.10.7/15.10.7.2_global/S15.10.7.2_A8.html:
* sputnik/Conformance/15_Native_Objects/15.10_RegExp/15.10.7/15.10.7.2_global/S15.10.7.2_A9.html:
* sputnik/Conformance/15_Native_Objects/15.10_RegExp/15.10.7/15.10.7.3_ignoreCase/S15.10.7.3_A10.html:
* sputnik/Conformance/15_Native_Objects/15.10_RegExp/15.10.7/15.10.7.3_ignoreCase/S15.10.7.3_A8.html:
* sputnik/Conformance/15_Native_Objects/15.10_RegExp/15.10.7/15.10.7.3_ignoreCase/S15.10.7.3_A9.html:
* sputnik/Conformance/15_Native_Objects/15.10_RegExp/15.10.7/15.10.7.4_multiline/S15.10.7.4_A10.html:
* sputnik/Conformance/15_Native_Objects/15.10_RegExp/15.10.7/15.10.7.4_multiline/S15.10.7.4_A8.html:
* sputnik/Conformance/15_Native_Objects/15.10_RegExp/15.10.7/15.10.7.4_multiline/S15.10.7.4_A9.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185370 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoCrash when passing Symbol to NPAPI plugin objects
utatane.tea@gmail.com [Tue, 9 Jun 2015 18:31:02 +0000 (18:31 +0000)]
Crash when passing Symbol to NPAPI plugin objects
https://bugs.webkit.org/show_bug.cgi?id=145798

Reviewed by Darin Adler.

Source/WebCore:

Test: plugins/npruntime/script-object-with-symbols.html

For C bridge APIs, we add guards for symbols.
This is the same to the existing guards in Objective-C APIs.

* bridge/c/c_class.cpp:
(JSC::Bindings::CClass::methodNamed):
(JSC::Bindings::CClass::fieldNamed):
* bridge/objc/objc_class.mm:
(JSC::Bindings::ObjcClass::methodNamed):
(JSC::Bindings::ObjcClass::fieldNamed):
(JSC::Bindings::ObjcClass::fallbackObject):

Source/WebKit2:

When the symbol is passed, `propertyName.publicName()` becomes nullptr.
So dereferencing it causes null dereference errors.
At first, this bug appears in the https://bugs.webkit.org/show_bug.cgi?id=145556,
plugin[@@toStringTag] ("string" + plugin) causes SEGV in plugins/embed-inside-object.html test.

This patch avoids it by early returning when the symbols are passed.
Methods for symbols are not implemented in the NPObject side, so it works correctly.

* WebProcess/Plugins/Netscape/JSNPObject.cpp:
(WebKit::npIdentifierFromIdentifier):
(WebKit::JSNPObject::callMethod):
(WebKit::JSNPObject::getOwnPropertySlot):
(WebKit::JSNPObject::put):
(WebKit::JSNPObject::deleteProperty):
(WebKit::JSNPObject::propertyGetter):
(WebKit::JSNPObject::methodGetter):

LayoutTests:

* plugins/npruntime/script-object-with-symbols-expected.txt: Added.
* plugins/npruntime/script-object-with-symbols.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@185369 268f45cc-cd09-0410-ab3c-d52691b4dbfc