WebKit-https.git
12 months agoFix assertion failure on BytecodeGenerator::recordOpcode
tzagallo@apple.com [Fri, 16 Nov 2018 15:24:04 +0000 (15:24 +0000)]
Fix assertion failure on BytecodeGenerator::recordOpcode
https://bugs.webkit.org/show_bug.cgi?id=191724
<rdar://problem/45724395>

Reviewed by Saam Barati.

JSTests:

* stress/regress-187373-2.js: Added.
(foo):

Source/JavaScriptCore:

Since https://bugs.webkit.org/show_bug.cgi?id=187373, we were not
restoring m_lastInstruction after patching the bytecode when
finalizing StructureForInContexts, only m_lastOpcodeID, which led to
the assertion failure.

* bytecompiler/BytecodeGenerator.cpp:
(JSC::StructureForInContext::finalize):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238281 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoSource/WebCore:
zalan@apple.com [Fri, 16 Nov 2018 15:07:17 +0000 (15:07 +0000)]
Source/WebCore:
[iOS] 2 subsequent taps are required to trigger certain tasks on the desktop version of YouTube.com (hover vs click).
https://bugs.webkit.org/show_bug.cgi?id=191712
<rdar://problem/45612900>

Reviewed by Simon Fraser.

In handleSyntheticClick() we use WKContentObservation to figure out whether the tap should be treated as a hover or a click.
In general, if the mouse-move event triggers a visible content change, we assume we hit a hover-like drop down menu (or something similar)
and no need to dispatch a click event.
The idea here is that if the new content (result of the mouse-move event) does not respond to mouse click, it is most likely
only for tooltip-like reasons and it's ok to proceed with the click event.

Test: fast/events/touch/ios/click-instead-of-hover-simple.html

* rendering/updating/RenderTreeUpdater.cpp:
(WebCore::CheckForVisibilityChange::~CheckForVisibilityChange):

LayoutTests:
[iOS] 2 subsequent taps are required to trigger certain tasks on the desktop version of YouTube.com
https://bugs.webkit.org/show_bug.cgi?id=191712
<rdar://problem/45612900>

Reviewed by Simon Fraser.

* fast/events/touch/ios/click-instead-of-hover-simple-expected.txt: Added.
* fast/events/touch/ios/click-instead-of-hover-simple.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238280 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[LFC][IFC] Add support for out-of-flow positioned boxes
zalan@apple.com [Fri, 16 Nov 2018 14:51:36 +0000 (14:51 +0000)]
[LFC][IFC] Add support for out-of-flow positioned boxes
https://bugs.webkit.org/show_bug.cgi?id=191726

Reviewed by Antti Koivisto.

Source/WebCore:

While laying out formatting context roots (inline-block, floats) in an inline formatting context, we need to make sure
that their out-of-flow descendants get laid out as well.

Test: fast/inline/simple-inline-with-out-of-flow-descendant.html

* layout/inlineformatting/InlineFormattingContext.cpp:
(WebCore::Layout::InlineFormattingContext::layoutFormattingContextRoot const):
* layout/layouttree/LayoutTreeBuilder.cpp: This was returning the wrong context root when the container was also a context root.
(WebCore::Layout::TreeBuilder::createSubTree):

Tools:

* LayoutReloaded/misc/LFC-passing-tests.txt:

LayoutTests:

* fast/inline/simple-inline-with-out-of-flow-descendant-expected.txt: Added.
* fast/inline/simple-inline-with-out-of-flow-descendant.html: Added.
* platform/ios/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238279 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[GStreamer][WebRTC] Add API to enable/disable device mocks
commit-queue@webkit.org [Fri, 16 Nov 2018 14:09:39 +0000 (14:09 +0000)]
[GStreamer][WebRTC] Add API to enable/disable device mocks
https://bugs.webkit.org/show_bug.cgi?id=191699

This basically us to test MediaStream/WebRTC support without
requiring cameras or microphones and is quite useful.

Also fix the GStreamerAudioMock by:
  - Stop setting `leaky-upstream` on the GStreamerCapturer queue,
    this was usefull when we were trying to bring the MediaStream
    sources inside the main pipeline, it is not the case anymore
    (and not doable with latest version of LibWebRTC).
  - Use a 'ticks' wave on the gstreamer audiotestsrc so the test
    stream is similar to what Apple port does.

Patch by Thibault Saunier <tsaunier@igalia.com> on 2018-11-16
Reviewed by Xabier Rodriguez-Calvar.

The mocks are already tested and the API is really simple.

Source/WebCore:

* platform/mediastream/gstreamer/GStreamerAudioCapturer.cpp:
(WebCore::GStreamerAudioCapturer::createSource):
* platform/mediastream/gstreamer/GStreamerAudioCapturer.h:
* platform/mediastream/gstreamer/GStreamerCapturer.cpp:
(WebCore::GStreamerCapturer::addSink):
* platform/mediastream/gstreamer/GStreamerCapturer.h:

Source/WebKit:

* UIProcess/API/glib/WebKitSettings.cpp:
(webKitSettingsSetProperty):
(webKitSettingsGetProperty):
(webkit_settings_class_init):
(webkit_settings_get_enable_mock_capture_devices):
(webkit_settings_set_enable_mock_capture_devices):
* UIProcess/API/gtk/WebKitSettings.h:
* UIProcess/API/gtk/docs/webkit2gtk-4.0-sections.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238278 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[GTK][WPE] Add "WebKitDeviceInfoPermissionRequest.h" into webkit2.h
commit-queue@webkit.org [Fri, 16 Nov 2018 14:02:41 +0000 (14:02 +0000)]
[GTK][WPE] Add "WebKitDeviceInfoPermissionRequest.h" into webkit2.h
https://bugs.webkit.org/show_bug.cgi?id=191744

It should always have been there.

Patch by Thibault Saunier <tsaunier@igalia.com> on 2018-11-16
Reviewed by Carlos Garcia Campos.

* UIProcess/API/gtk/webkit2.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238277 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[GStreamer][MediaStream] Handle track addition and removal
commit-queue@webkit.org [Fri, 16 Nov 2018 14:01:19 +0000 (14:01 +0000)]
[GStreamer][MediaStream] Handle track addition and removal
https://bugs.webkit.org/show_bug.cgi?id=191599

Patch by Thibault Saunier <tsaunier@igalia.com> on 2018-11-16
Reviewed by Xabier Rodriguez-Calvar.

Source/WebCore:

Test: fast/mediastream/MediaStream-video-element-remove-track.html

* platform/mediastream/gstreamer/GStreamerMediaStreamSource.cpp:
(WebCore::WebKitMediaStreamObserver::~WebKitMediaStreamObserver):
(WebCore::WebKitMediaStreamObserver::WebKitMediaStreamObserver):
(WebCore::webkitMediaStreamSrcFinalize):
(WebCore::webkitMediaStreamSrcChangeState):
(WebCore::webkit_media_stream_src_init):
(WebCore::webkitMediaStreamSrcSetupSrc):
(WebCore::webkitMediaStreamSrcAddTrack):
(WebCore::webkitMediaStreamSrcRemoveTrackByType):
(WebCore::webkitMediaStreamSrcSetStream):

LayoutTests:

* fast/mediastream/MediaStream-video-element-remove-track-expected.txt: Added.
* fast/mediastream/MediaStream-video-element-remove-track.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238276 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoScalableImageDecoder: don't forcefully decode image data when querying frame complete...
zandobersek@gmail.com [Fri, 16 Nov 2018 10:57:17 +0000 (10:57 +0000)]
ScalableImageDecoder: don't forcefully decode image data when querying frame completeness, duration
https://bugs.webkit.org/show_bug.cgi?id=191354

Reviewed by Michael Catanzaro.

ScalableImageDecoder::frameIsCompleteAtIndex() should only check the
index validity and, if the index is valid, check for completeness of the
corresponding frame. ScalableImageDecoder::frameDurationAtIndex() should
also only retrieve duration for already-complete frames.

Both methods avoid calling ScalableImageDecoder::frameBufferAtIndex()
as that method goes on and decodes image data to determine specific
information. The ImageSource class that's querying this information
doesn't anticipate this, and doesn't handle the increased memory
consumption of the decoded data, leaving MemoryCache in the blind about
the image resource's actual amount of consumed memory. ImageSource can
instead gracefully handle any incomplete frame by marking the decoding
status for this frame as only partial.

* platform/image-decoders/ScalableImageDecoder.cpp:
(WebCore::ScalableImageDecoder::frameIsCompleteAtIndex const):
(WebCore::ScalableImageDecoder::frameHasAlphaAtIndex const):
(WebCore::ScalableImageDecoder::frameDurationAtIndex const):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238275 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoPointerEvents should not require touch event listeners to be registered
graouts@webkit.org [Fri, 16 Nov 2018 09:09:34 +0000 (09:09 +0000)]
PointerEvents should not require touch event listeners to be registered
https://bugs.webkit.org/show_bug.cgi?id=191333
<rdar://problem/45857523>

Reviewed by Dean Jackson.

Source/WebCore:

Tests: pointerevents/ios/pointer-events-dispatch-on-touch.html
       pointerevents/ios/pointer-events-prevent-default.html

* dom/EventNames.h:
(WebCore::EventNames::isTouchEventType const):
(WebCore::EventNames::touchAndPointerEventNames const):
(WebCore::EventNames::touchEventNames const): Deleted.
* dom/Node.cpp:
(WebCore::Node::moveNodeToNewDocument):

Source/WebKit:

* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::updateTouchEventTracking):
(WebKit::WebPageProxy::touchEventTrackingType const):
(WebKit::WebPageProxy::handleTouchEventSynchronously):
(WebKit::WebPageProxy::handleTouchEventAsynchronously):
(WebKit::WebPageProxy::handleTouchEvent):
(WebKit::WebPageProxy::resetState):
* UIProcess/WebPageProxy.h:

LayoutTests:

Add two iOS tests that check that we dispatch pointer events when only those events are registered
and that we correctly account for calls to preventDefault(). To support pointer events test, we add
a new utility where we will be adding user interaction functions that should allow these tests to be
submitted to the WPT repository provided browsers implement their own versions of the UIController.

* TestExpectations:
* pointerevents/ios/pointer-events-dispatch-on-touch-expected.txt: Added.
* pointerevents/ios/pointer-events-dispatch-on-touch.html: Added.
* pointerevents/ios/pointer-events-prevent-default-expected.txt: Added.
* pointerevents/ios/pointer-events-prevent-default.html: Added.
* pointerevents/utils.js: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238274 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[iOS] Do not get stuck in indeterminate content observation state.
zalan@apple.com [Fri, 16 Nov 2018 05:46:48 +0000 (05:46 +0000)]
[iOS] Do not get stuck in indeterminate content observation state.
https://bugs.webkit.org/show_bug.cgi?id=191719

Reviewed by Simon Fraser.

Source/WebCore:

Reset the _WKContentChange flag to WKContentNoChange when the last observing timer is removed and we are in
the "can't decide yet if it's a hover or click" state.
This bug prevents us from firing click event when JS installs and removes the same set of timer(s) during mouse-move dispatch.

Test: fast/events/touch/ios/stuck-with-hover-state.html

* platform/ios/wak/WKContentObservation.cpp:
(WebThreadRemoveObservedContentModifier):

LayoutTests:

* fast/events/touch/ios/stuck-with-hover-state-expected.txt: Added.
* fast/events/touch/ios/stuck-with-hover-state.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238273 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoAdd layout test for iframe contents disappearing on view unparent/reparent
simon.fraser@apple.com [Fri, 16 Nov 2018 05:39:26 +0000 (05:39 +0000)]
Add layout test for iframe contents disappearing on view unparent/reparent
https://bugs.webkit.org/show_bug.cgi?id=191725

Reviewed by Alex Christensen.

This test tests the changes made in r238229.

* compositing/iframes/remove-reinsert-webview-with-iframe-expected.txt: Added.
* compositing/iframes/remove-reinsert-webview-with-iframe.html: Added.
* platform/ios/compositing/iframes/remove-reinsert-webview-with-iframe-expected.txt: Added.
* platform/mac-wk1/compositing/iframes/remove-reinsert-webview-with-iframe-expected.txt: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238271 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoRegExpObject's collectMatches should not be using JSArray::push to fill in its match...
mark.lam@apple.com [Fri, 16 Nov 2018 05:32:47 +0000 (05:32 +0000)]
RegExpObject's collectMatches should not be using JSArray::push to fill in its match results.
https://bugs.webkit.org/show_bug.cgi?id=191730
<rdar://problem/46048517>

Reviewed by Saam Barati.

JSTests:

* stress/regress-187006.js: Removed.
  - this test is invalid because its sole purpose is to test for the non-spec
    compliant behavior that we just fixed.

* stress/regress-191730.js: Added.

Source/JavaScriptCore:

According to the spec https://www.ecma-international.org/ecma-262/9.0/index.html#sec-regexp.prototype-@@match,
the RegExp match results are filled in using the spec's CreateDataProperty()
function which does not consult the prototype for setters.  JSArray:push()
consults the prototype for setters.  We should be using putDirectIndex() instead.

* runtime/RegExpObjectInlines.h:
(JSC::collectMatches):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238270 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoREGRESSION (r238090) Composited iframes that resize from zero don't show
simon.fraser@apple.com [Fri, 16 Nov 2018 05:29:45 +0000 (05:29 +0000)]
REGRESSION (r238090) Composited iframes that resize from zero don't show
https://bugs.webkit.org/show_bug.cgi?id=191733
rdar://problem/46107764

Reviewed by Zalan Bujtas.
Source/WebCore:

A zero-sized iframe whose contents are composited should not trigger compositing in the
parent document (see code in requiresCompositingForFrame()), but when the <iframe> element
was resized without a style change (e.g. because it's width: 100%, height: 100% and the
parent resizes), there was no code that triggered a compositing update.

Fix by having RenderLayer::updateLayerPosition() trigger an update when the size changes,
for a RenderWidget whose contents are composited.

Test: compositing/iframes/resize-from-zero-size.html

* rendering/RenderLayer.cpp:
(WebCore::RenderLayer::updateLayerPosition):

LayoutTests:

* compositing/iframes/resize-from-zero-size-expected.txt: Added.
* compositing/iframes/resize-from-zero-size.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238269 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoRegExp operations should not take fast patch if lastIndex is not numeric.
mark.lam@apple.com [Fri, 16 Nov 2018 05:12:25 +0000 (05:12 +0000)]
RegExp operations should not take fast patch if lastIndex is not numeric.
https://bugs.webkit.org/show_bug.cgi?id=191731
<rdar://problem/46017305>

Reviewed by Saam Barati.

JSTests:

* stress/regress-191731.js: Added.

Source/JavaScriptCore:

This is because if lastIndex is an object with a valueOf() method, it can execute
arbitrary code which may have side effects, and side effects are not permitted by
the RegExp fast paths.

* builtins/RegExpPrototype.js:
(globalPrivate.hasObservableSideEffectsForRegExpMatch):
(overriddenName.string_appeared_here.search):
(globalPrivate.hasObservableSideEffectsForRegExpSplit):
(intrinsic.RegExpTestIntrinsic.test):
* builtins/StringPrototype.js:
(globalPrivate.hasObservableSideEffectsForStringReplace):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238267 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoOverlay with -webkit-overflow-scrolling:touch doesn't become scrollable after added...
simon.fraser@apple.com [Fri, 16 Nov 2018 05:02:56 +0000 (05:02 +0000)]
Overlay with -webkit-overflow-scrolling:touch doesn't become scrollable after added text makes it taller
https://bugs.webkit.org/show_bug.cgi?id=158342
rdar://problem/26652811

Reviewed by Zalan Bujtas.
Source/WebCore:

Patch partly by Frédéric Wang.

This commit fixes an issue when resizing the content of a -webkit-overflow-scrolling: touch
overflow node on iOS. Indeed, the RenderLayerBacking's scrolling layer may not be properly
created and hence the UIProcess receives a null UIScrollView pointer. This triggers an
assertion in debug mode and prevents the user from scrolling the overflow node in release
mode. This was partially fixed by the refactoring of bug 90342 but this commit addresses
the remaining issues by forcing a configuration update after layout in order to ensure that
RenderLayerBacking's scrolling layer is available. For an overflow element that is not yet
composited, trigger a post-layout update that is necessary to check if we need to make it
composited when it gains scrollable overflow.

Tests: fast/scrolling/ios/change-scrollability-on-content-resize-nested.html
       fast/scrolling/ios/change-scrollability-on-content-resize.html

* rendering/RenderLayer.cpp:
(WebCore::RenderLayer::updateScrollInfoAfterLayout): Force a configuration update so that
RenderLayerCompositor::updateBackingAndHierarchy will later instantiate
RenderLayerBacking::m_scrollingLayer.

LayoutTests:

Patch partly by Frédéric Wang.

Add a test to check that scrollability of composited/non-composited overflow divs is properly
updated. This test used to fail and assert before #238090. A second test with nested divs
stills exhibit the issue and is fixed by the present commit.

* fast/scrolling/ios/change-scrollability-on-content-resize-expected.txt: Added.
* fast/scrolling/ios/change-scrollability-on-content-resize-nested-expected.txt: Added.
* fast/scrolling/ios/change-scrollability-on-content-resize-nested.html: Added.
* fast/scrolling/ios/change-scrollability-on-content-resize.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238266 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoAdditional Windows build fix after r238253.
bfulgham@apple.com [Fri, 16 Nov 2018 03:09:31 +0000 (03:09 +0000)]
Additional Windows build fix after r238253.

* pal/spi/cg/CoreGraphicsSPI.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238265 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoWKPreferencesSetFontSmoothingLevel doesn't actually do anything
mmaxfield@apple.com [Fri, 16 Nov 2018 01:55:04 +0000 (01:55 +0000)]
WKPreferencesSetFontSmoothingLevel doesn't actually do anything
https://bugs.webkit.org/show_bug.cgi?id=191708

Reviewed by Simon Fraser.

Delete the symbol because no one calls it and it doesn't do anything

Source/WebKit:

* Scripts/PreferencesTemplates/WebPreferencesStoreDefaultsMap.cpp.erb:
* Shared/FontSmoothingLevel.h: Removed.
* Shared/WebPreferences.yaml:
* UIProcess/API/C/WKAPICast.h:
(WebKit::toFontSmoothingLevel): Deleted.
* UIProcess/API/C/WKPreferences.cpp:
(WKPreferencesSetFontSmoothingLevel): Deleted.
(WKPreferencesGetFontSmoothingLevel): Deleted.
* UIProcess/API/C/WKPreferencesRefPrivate.h:
* UIProcess/WebPreferences.h:
* WebKit.xcodeproj/project.pbxproj:
* mac/WebKit2.order:

Tools:

* TestWebKitAPI/Tests/WebKit/WKPreferences.cpp:
(TestWebKitAPI::TEST):
* WebKitTestRunner/TestController.cpp:
(WTR::TestController::resetPreferencesToConsistentValues):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238264 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[curl] warning: delete called on non-final 'WebCore::CurlDownload' that has virtual...
Hironori.Fujii@sony.com [Fri, 16 Nov 2018 01:37:40 +0000 (01:37 +0000)]
[curl] warning: delete called on non-final 'WebCore::CurlDownload' that has virtual functions but non-virtual destructor [-Wdelete-non-virtual-dtor]
https://bugs.webkit.org/show_bug.cgi?id=191582

Reviewed by Alex Christensen.

No new tests because there's no behaviour change.

* platform/network/curl/CurlDownload.h: Marked CurlDownload final.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238263 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoUnreviewed, rolling out r238244.
tsavell@apple.com [Fri, 16 Nov 2018 01:07:42 +0000 (01:07 +0000)]
Unreviewed, rolling out r238244.

Caused High Sierra test runs to fail early with 50 crashes and
casued 25 API failures.

Reverted changeset:

"[css-logical] Implement flow-relative margin, padding and
border shorthands"
https://bugs.webkit.org/show_bug.cgi?id=188697
https://trac.webkit.org/changeset/238244

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238262 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoimported/w3c/web-platform-tests/encoding tests take > 30 minutes to run on debug...
ryanhaddad@apple.com [Fri, 16 Nov 2018 00:48:54 +0000 (00:48 +0000)]
imported/w3c/web-platform-tests/encoding tests take > 30 minutes to run on debug bots
https://bugs.webkit.org/show_bug.cgi?id=191720

Unreviewed test gardening.

* TestExpectations: Skip the directory on debug.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238261 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoAVKit will set videoGravity to a nil string when building against iosmac
jer.noble@apple.com [Thu, 15 Nov 2018 23:51:30 +0000 (23:51 +0000)]
AVKit will set videoGravity to a nil string when building against iosmac
https://bugs.webkit.org/show_bug.cgi?id=191573

Reviewed by Dean Jackson.

Workaround AVKit behavior by treating nil videoGravity as the default,
which is AVLayerVideoGravityResizeAspect.

* platform/ios/VideoFullscreenInterfaceAVKit.mm:
(-[WebAVPlayerLayer setVideoGravity:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238257 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoUnreviewed build fix after removing some WKSI calls on Windows.
bfulgham@apple.com [Thu, 15 Nov 2018 23:18:58 +0000 (23:18 +0000)]
Unreviewed build fix after removing some WKSI calls on Windows.

* pal/spi/cg/CoreGraphicsSPI.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238256 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoREGRESSION (r238038): [iOS] Layout test fast/forms/ios/validation-bubble-dismiss...
cdumez@apple.com [Thu, 15 Nov 2018 23:15:08 +0000 (23:15 +0000)]
REGRESSION (r238038): [iOS] Layout test fast/forms/ios/validation-bubble-dismiss-on-tap.html is a flaky failure
https://bugs.webkit.org/show_bug.cgi?id=191710
<rdar://problem/46104093>

Reviewed by Simon Fraser.

There is a small amount of time after we ask to show the popover before the popover is actually shown on screen.
During this period, clicks are eaten and not reported to the view. For this reason, we keep tapping until the
popup is dismissed.

* fast/forms/ios/validation-bubble-dismiss-on-tap-expected.txt:
* fast/forms/ios/validation-bubble-dismiss-on-tap.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238255 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoWeb Inspector: Remove unused WebInspectorProxy code
commit-queue@webkit.org [Thu, 15 Nov 2018 22:42:56 +0000 (22:42 +0000)]
Web Inspector: Remove unused WebInspectorProxy code
https://bugs.webkit.org/show_bug.cgi?id=191674

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2018-11-15
Reviewed by Dean Jackson.

* UIProcess/WebInspectorProxy.cpp:
(WebKit::WebInspectorProxy::createInspectorPage):
(WebKit::WebInspectorProxy::closeFrontendPageAndWindow):
* UIProcess/WebInspectorProxy.h:
* UIProcess/mac/WebInspectorProxyMac.mm:
(WebKit::WebInspectorProxy::setInspectorWindowFrame): Deleted.
(WebKit::WebInspectorProxy::inspectorWindowFrame): Deleted.
* WebProcess/WebPage/WebInspector.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238254 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[Win] Reduce the use of WKSI library calls: Font Handling
bfulgham@apple.com [Thu, 15 Nov 2018 22:38:39 +0000 (22:38 +0000)]
[Win] Reduce the use of WKSI library calls: Font Handling
https://bugs.webkit.org/show_bug.cgi?id=191701
<rdar://problem/46104809>

Reviewed by Myles C. Maxfield.

Source/WebCore:

Move the old Windows font handling code out of WKSI to our regular
repository. We now handle SPI differently, and don't need to keep
these implementations in a separate library. This should also help
avoid the somewhat frequent build failures caused when WKSI is not
updated in sync with WebKit.

Tested by existing Windows test cases.

* platform/graphics/FontCascade.h:
* platform/graphics/win/FontCGWin.cpp:
(WebCore::FontCascade::drawGlyphs):
(WebCore::FontCascade::setFontSmoothingLevel):
(WebCore::setCGFontSmoothingStyle):
(WebCore::FontCascade::setFontSmoothingStyle):
(WebCore::FontCascade::setFontSmoothingContrast):
(WebCore::clearTypeContrast):
(WebCore::FontCascade::systemFontSmoothingChanged):
(WebCore::FontCascade::setCGContextFontRenderingStyle):
(WebCore::renderingStyleForFont):
(WebCore::FontCascade::getGlyphAdvances):
* platform/graphics/win/GlyphPageTreeNodeCGWin.cpp:
(WebCore::GlyphPage::fill):
* platform/graphics/win/GraphicsContextCGWin.cpp:
(WebCore::GraphicsContext::drawFocusRing):
(WebCore::GraphicsContext::drawDotsForDocumentMarker):
* platform/graphics/win/SimpleFontDataCGWin.cpp:
(WebCore::Font::platformWidthForGlyph const):
* rendering/RenderMediaControls.cpp:
(WebCore::RenderMediaControls::adjustMediaSliderThumbSize):

Source/WebKitLegacy/win:

Update to use the old font handling routines moved from WKSI to FontCascade.

* WebPreferences.cpp:
(WebPreferences::setFontSmoothing):
(WebPreferences::setFontSmoothingContrast):
* WebView.cpp:
(systemParameterChanged):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238253 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoEnable process swap on cross-site navigation by default on macOS
rniwa@webkit.org [Thu, 15 Nov 2018 22:07:05 +0000 (22:07 +0000)]
Enable process swap on cross-site navigation by default on macOS
https://bugs.webkit.org/show_bug.cgi?id=191572

Reviewed by Chris Dumez.

Enabled the feature by default on macOS.

* Shared/WebPreferences.yaml:
* Shared/WebPreferencesDefaultValues.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238252 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoIncrease EWS lock timeout to 3 hours
aakash_jain@apple.com [Thu, 15 Nov 2018 22:05:27 +0000 (22:05 +0000)]
Increase EWS lock timeout to 3 hours
https://bugs.webkit.org/show_bug.cgi?id=191705

Reviewed by Alexey Proskuryakov.

* QueueStatusServer/config/queues.py: Increased EWS lock timeout to 3 hours.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238251 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoIndexedDB.IndexedDBTempFileSize API test times out with process prewarming enabled
cdumez@apple.com [Thu, 15 Nov 2018 22:01:31 +0000 (22:01 +0000)]
IndexedDB.IndexedDBTempFileSize API test times out with process prewarming enabled
https://bugs.webkit.org/show_bug.cgi?id=191671
<rdar://problem/46086062>

Reviewed by Alex Christensen.

Source/WebKit:

Enabling process prewarming caused IndexedDB.IndexedDBTempFileSize API to time out and print
the following line:
"Attempted to create a NetworkLoad with a session (id=2) that does not exist."

This actually identified a pre-existing bug with our handling of non-default data store.
Whenever a page starts using a data store, we call WebProcessPool::pageBeginUsingWebsiteDataStore()
which will call NetworkProcessProxy::addSession() if the network process was already started
to let the network process know about this non-default session. There are several issues with the
current model:
1. If the network process was not created yet when pageBeginUsingWebsiteDataStore() is called,
   then the network process will not know about the non-default session when it actually gets
   started later on. This is unlikely to happen in practice, except in case of network process
   crash because we create the network process as soon as we initialize the first WebProcessProxy.
2. Even if we successfuly managed to register the session with the network process proxy, we get
   in trouble if the network process crashes or is terminated later on as we do not re-register
   those sessions with the new network process.

To address these 2 issues, WebProcessPool::ensureNetworkProcess() now takes care of registering
all the non-default sessions (that are associated with this process pool) with the new network
process. The WebProcessPool knows about these sessions because it adds them to m_sessionToPagesMap
whenever WebProcessPool::pageBeginUsingWebsiteDataStore() is called, even if the network process
proxy was not created yet.

The reason the IndexedDB.IndexedDBTempFileSize API test was failing was because it did:
1. A load in a WebView V1 with a non-default session
2. Process prewarming kicked in after this load and would create a new WebProcessProxy.
3. Terminate the network process
4. Another load in a WebView V2 with the same non-defaut session, which would reuse the
  prewarmed process. Because the network process was terminated, constructing the new
  page would not register the session ID with the new network process when
  pageBeginUsingWebsiteDataStore() is called.
-> The load would hang because the new network process would not know about the
   non-default session when started later on.

The bug was previously hidden without process prewarming because step 4 would create a *new*
WebProcessProxy and WebProcessPool::initializeNewWebProcess() would call ensureNetworkProcess()
so that pageBeginUsingWebsiteDataStore() would successfuly register the session with the
network process later on.

I wrote a second API test (WebKit.DoLoadWithNonDefaultDataStoreAfterTerminatingNetworkProcess)
to demonstrate the pre-existing bug without process prewarming enabled.

* UIProcess/WebProcessPool.cpp:
(WebKit::WebProcessPool::ensureNetworkProcess):

Tools:

Add API test coverage.

* TestWebKitAPI/Tests/WebKitCocoa/ProcessSwapOnNavigation.mm:
* TestWebKitAPI/Tests/WebKitCocoa/WebsiteDataStoreCustomPaths.mm:
(TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238250 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoCreate feature flag (HTTPS_UPGRADE)
commit-queue@webkit.org [Thu, 15 Nov 2018 21:58:13 +0000 (21:58 +0000)]
Create feature flag (HTTPS_UPGRADE)
https://bugs.webkit.org/show_bug.cgi?id=191613
<rdar://problem/45851068>

Patch by Vivek Seth <v_seth@apple.com> on 2018-11-15
Reviewed by Dean Jackson.

* config.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238249 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoRegression(r238215) WKProcessPool.WarmInitialProcess API test is failing
cdumez@apple.com [Thu, 15 Nov 2018 21:53:42 +0000 (21:53 +0000)]
Regression(r238215) WKProcessPool.WarmInitialProcess API test is failing
https://bugs.webkit.org/show_bug.cgi?id=191693

Reviewed by Alex Christensen.

Allow *explicit* process prewarming to construct the default data store if necessary or we would
not be obeying the client's request. Only automatic process prewarming should avoid constructing
the default data store unnecessarily to avoid causing memory regressions.

* UIProcess/API/C/WKContext.cpp:
(WKContextWarmInitialProcess):
* UIProcess/API/Cocoa/WKProcessPool.mm:
(-[WKProcessPool _warmInitialProcess]):
* UIProcess/WebProcessPool.cpp:
(WebKit::WebProcessPool::prewarmProcess):
(WebKit::WebProcessPool::didReachGoodTimeToPrewarm):
* UIProcess/WebProcessPool.h:
(WebKit::WebProcessPool::sendToOneProcess):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238248 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoUnreviewed, comment improvement over r238166.
jiewen_tan@apple.com [Thu, 15 Nov 2018 21:31:34 +0000 (21:31 +0000)]
Unreviewed, comment improvement over r238166.

* UIProcess/API/C/WKWebsiteDataStoreRef.cpp:
(WKWebsiteDataStoreSetWebAuthenticationMockConfiguration):
* UIProcess/WebAuthentication/fido/CtapHidAuthenticator.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238247 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[WebAuthN] Use a real nonce for CTAPHID_INIT
jiewen_tan@apple.com [Thu, 15 Nov 2018 21:24:51 +0000 (21:24 +0000)]
[WebAuthN] Use a real nonce for CTAPHID_INIT
https://bugs.webkit.org/show_bug.cgi?id=191533
<rdar://problem/46103502>

Reviewed by Brent Fulgham.

Source/WebCore:

New tests are added into existing test files.

* Modules/webauthn/fido/FidoConstants.h:

Source/WebKit:

Use a real nonce for CTAPHID_INIT request according to:
https://fidoalliance.org/specs/fido-v2.0-ps-20170927/fido-client-to-authenticator-protocol-v2.0-ps-20170927.html#ctaphid_init-0x06.
The challenge here is the new transaction needs to start in the next runloop otherwise a dead lock will form:
wrong nonce -> new transaction -> new nonce -> write init request -> read init response from last run as it
piped in the run loop -> wrong nonce of course -> ...
To break the above dead lock, we have to start the new transaction in the next run. However, that isn't
sufficient as the arrived init response will be piped in HidConnection::m_inputReports, which is designed
on purpose to store any data packets within (initialized, terminated) time interval to prevent data loss in
the case when HidConnection::registerDataReceivedCallback is called after the first data packet's arrival.
In order to break the dead lock completely, HidConnection::invalidateCache will bnnne called prior to every
send to delete any potential init response from last run. HidConnection::invalidateCache is not necessary
for other protocols though. The above scenario is more or less a design flaw in CTAP HID.

Of course, all above scenarios are covered in our mock tests.

* UIProcess/API/C/WKWebsiteDataStoreRef.cpp:
(WKWebsiteDataStoreSetWebAuthenticationMockConfiguration):
* UIProcess/WebAuthentication/Cocoa/HidConnection.h:
(WebKit::HidConnection::invalidateCache):
* UIProcess/WebAuthentication/Mock/MockHidConnection.cpp:
(WebKit::MockHidConnection::send):
(WebKit::MockHidConnection::parseRequest):
(WebKit::MockHidConnection::feedReports):
* UIProcess/WebAuthentication/Mock/MockHidConnection.h:
* UIProcess/WebAuthentication/Mock/MockWebAuthenticationConfiguration.h:
* UIProcess/WebAuthentication/fido/CtapHidDriver.cpp:
(WebKit::CtapHidDriver::Worker::transact):
(WebKit::CtapHidDriver::CtapHidDriver):
(WebKit::CtapHidDriver::transact):
(WebKit::CtapHidDriver::continueAfterChannelAllocated):
(WebKit::CtapHidDriver::returnResponse):
* UIProcess/WebAuthentication/fido/CtapHidDriver.h:

LayoutTests:

* http/wpt/webauthn/ctap-hid-failure.https-expected.txt:
* http/wpt/webauthn/ctap-hid-failure.https.html:
* http/wpt/webauthn/ctap-hid-success.https-expected.txt:
* http/wpt/webauthn/ctap-hid-success.https.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238246 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[WebGPU] WebGPUCommandBuffer prototype
justin_fan@apple.com [Thu, 15 Nov 2018 21:20:27 +0000 (21:20 +0000)]
[WebGPU] WebGPUCommandBuffer prototype
https://bugs.webkit.org/show_bug.cgi?id=191663

Reviewed by Dean Jackson.

Begin implementation of WebGPUCommandBuffers as well as GPUQueues (MTLCommandBuffer, MTLCommandQueue).

Source/WebCore:

Test: webgpu/command-buffers.html

* CMakeLists.txt:
* DerivedSources.make:
* Modules/webgpu/GPUCommandBuffer.h: Added. Wrapper class around a MTLCommandBuffer.
* Modules/webgpu/GPUDevice.cpp:
(WebCore::GPUDevice::createCommandBuffer): Added.
(WebCore::GPUDevice::getQueue): Returns RefPtr to the device's singleton queue.
* Modules/webgpu/GPUDevice.h: Now manages the device's GPUQueue.
(WebCore::GPUDevice::platformDevice const):
* Modules/webgpu/GPUQueue.h: Added. Wrapper class around a MTLCommandQueue.
(WebCore::GPUQueue::platformQueue const):
* Modules/webgpu/GPURenderPipeline.h: Moved from Source/WebCore/Modules/webgpu/cocoa/GPURenderPipeline.h.
(WebCore::GPURenderPipeline::platformRenderPipeline const):
* Modules/webgpu/GPUShaderModule.h:
(WebCore::GPUShaderModule::platformShaderModule const):
* Modules/webgpu/GPUSwapChain.h: Moved from Source/WebCore/Modules/webgpu/cocoa/GPUSwapChain.h.
(WebCore::GPUSwapChain::platformLayer const):
* Modules/webgpu/WebGPUCommandBuffer.cpp: Added. Web interface for a GPU device's command buffer.
(WebCore::WebGPUCommandBuffer::create):
(WebCore::WebGPUCommandBuffer::WebGPUCommandBuffer):
* Modules/webgpu/WebGPUCommandBuffer.h: Added.
* Modules/webgpu/WebGPUCommandBuffer.idl: Added.
* Modules/webgpu/WebGPUDevice.cpp:
(WebCore::WebGPUDevice::createCommandBuffer const): Added.
* Modules/webgpu/WebGPUDevice.h:
* Modules/webgpu/WebGPUDevice.idl:
* Modules/webgpu/cocoa/GPUCommandBufferMetal.mm: Added. MTLCommandBuffer impl for GPUCommandBuffer.
(WebCore::GPUCommandBuffer::create): Create a MTLCommandBuffer from the MTLCommandQueue.
(WebCore::GPUCommandBuffer::GPUCommandBuffer):
* Modules/webgpu/cocoa/GPUDeviceMetal.mm:
(WebCore::GPUDevice::GPUDevice):
* Modules/webgpu/cocoa/GPUQueueMetal.mm: Added. MTLCommandQueue impl for GPUQueue.
(WebCore::GPUQueue::create):
(WebCore::GPUQueue::GPUQueue):
* Modules/webgpu/cocoa/GPURenderPipelineMetal.mm:
(WebCore::GPURenderPipeline::GPURenderPipeline):
* Modules/webgpu/cocoa/GPUShaderModuleMetal.mm:
(WebCore::GPUShaderModule::create):
(WebCore::GPUShaderModule::GPUShaderModule):
* Sources.txt:
* SourcesCocoa.txt:
* WebCore.xcodeproj/project.pbxproj:
* bindings/js/WebCoreBuiltinNames.h:

LayoutTests:

New testing for WebGPUCommandBuffers, to be fleshed out when WebGPUCommandBufferDescriptors
are added to the specification.

* webgpu/command-buffers-expected.txt: Added.
* webgpu/command-buffers.html: Basic functionality, with more stress tests to be added when the spec is updated.
* webgpu/webgpu-basics.html: Updated to create the command buffer used by later rendering operations.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238245 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[css-logical] Implement flow-relative margin, padding and border shorthands
commit-queue@webkit.org [Thu, 15 Nov 2018 21:15:26 +0000 (21:15 +0000)]
[css-logical] Implement flow-relative margin, padding and border shorthands
https://bugs.webkit.org/show_bug.cgi?id=188697

Patch by Oriol Brufau <obrufau@igalia.com> on 2018-11-15
Reviewed by Simon Fraser and Antti Koivisto.

LayoutTests/imported/w3c:

Import tests from the WPT test suite for CSS Logical Properties and Values,
and enable the CSSLogicalEnabled runtime flag for them.

They still have some failures because sideways writing modes have not been
implemented yet (https://bugs.webkit.org/show_bug.cgi?id=166941).

* web-platform-tests/css/css-logical/logical-box-border-color-expected.txt:
* web-platform-tests/css/css-logical/logical-box-border-color.html:
* web-platform-tests/css/css-logical/logical-box-border-shorthands-expected.txt:
* web-platform-tests/css/css-logical/logical-box-border-shorthands.html:
* web-platform-tests/css/css-logical/logical-box-border-style-expected.txt:
* web-platform-tests/css/css-logical/logical-box-border-style.html:
* web-platform-tests/css/css-logical/logical-box-border-width-expected.txt:
* web-platform-tests/css/css-logical/logical-box-border-width.html:
* web-platform-tests/css/css-logical/logical-box-inset-expected.txt:
* web-platform-tests/css/css-logical/logical-box-inset.html:
* web-platform-tests/css/css-logical/logical-box-margin-expected.txt:
* web-platform-tests/css/css-logical/logical-box-margin.html:
* web-platform-tests/css/css-logical/logical-box-padding-expected.txt:
* web-platform-tests/css/css-logical/logical-box-padding.html:
* web-platform-tests/css/css-logical/logical-box-size.html:
* web-platform-tests/css/css-logical/resources/test-box-properties.js:
(export.createBoxPropertyGroup):
(export.createSizingPropertyGroup):
(export.runTests.set test):
(export.runTests):
* web-platform-tests/css/css-logical/resources/test-shared.js: Added.
(export.testCSSValues):
(export.testComputedValues):
(export.makeDeclaration):
* web-platform-tests/css/css-logical/resources/w3c-import.log:

Source/WebCore:

Tests: imported/w3c/web-platform-tests/css/css-logical/logical-box-border-color.html
       imported/w3c/web-platform-tests/css/css-logical/logical-box-border-shorthands.html
       imported/w3c/web-platform-tests/css/css-logical/logical-box-border-style.html
       imported/w3c/web-platform-tests/css/css-logical/logical-box-border-width.html
       imported/w3c/web-platform-tests/css/css-logical/logical-box-margin.html
       imported/w3c/web-platform-tests/css/css-logical/logical-box-padding.html
       webexposed/css-properties-as-js-properties.html
       webexposed/css-properties-behind-flags.html
       webexposed/css-property-listing.html

* css/CSSComputedStyleDeclaration.cpp:
(WebCore::ComputedStyleExtractor::valueForPropertyinStyle):
Allow the new properties to serialize their computed value.

(WebCore::ComputedStyleExtractor::getCSSPropertyValuesFor2SidesShorthand):
(WebCore::ComputedStyleExtractor::getCSSPropertyValuesFor4SidesShorthand):
* css/CSSComputedStyleDeclaration.h:
Rename getCSSPropertyValuesForSidesShorthand to getCSSPropertyValuesFor4SidesShorthand,
and add analogous getCSSPropertyValuesFor2SidesShorthand for serializing 2-sided
shorthands.

* css/CSSProperties.json:
Add the new properties behind the CSSLogicalEnabled runtime flag.

* css/CSSStyleDeclaration.cpp:
(WebCore::CSSStyleDeclaration::supportedPropertyNames const):
Prevent CSS properties disabled behind a runtime flag from being exposed in
style declarations.

* css/StyleProperties.cpp:
(WebCore::StyleProperties::getPropertyValue const):
Allow the new properties to serialize their specified value.

(WebCore::StyleProperties::get2Values const):
Add get2Values, analogous to get4Values, for serializing 2-sided shorthands.

(WebCore::StyleProperties::borderPropertyValue const):
Allow borderPropertyValue to serialize arbitrary multi-sided border shorthands
corresponding to width, style and color.

(WebCore::MutableStyleProperties::setProperty):
Prevent CSS properties disabled behind a runtime flag from being set a value.

(WebCore::StyleProperties::asText const):
Allow the new properties to be serialized in cssText.
Prevent CSS shorthands disabled behind a runtime flag from appearing in cssText,
and serialize the longhands instead. Note that there could be another shorthand
available which is enabled, but a proper solution would require bug 190496.

* css/StyleProperties.h:
Update declarations of borderPropertyValue and get2Values.

* css/makeprop.pl:
(addProperty):
Add isEnabledCSSProperty function for checking that a CSS property is not
disabled behind a runtime flag.

* css/parser/CSSPropertyParser.cpp:
(WebCore::cssPropertyID):
Prevent CSS properties disabled behind a runtime flag from being exposed in
computed styles.

(WebCore::CSSPropertyParser::addProperty):
Prevent CSS properties disabled behind a runtime flag from being set a value.

(WebCore::CSSPropertyParser::consumeBorder):
Change consumeBorder to provide the caller with the parsed values instead of
setting properties. Then the caller can decide to which properties the values
should be set, and whether border-image should be reset or not.

(WebCore::CSSPropertyParser::consume2ValueShorthand):
(WebCore::CSSPropertyParser::consume4ValueShorthand):
Rename consume4Values to consume4ValueShorthand, and add analogous
consume2ValueShorthand for parsing shorthands with two longhands.

(WebCore::CSSPropertyParser::parseShorthand):
Allow the new properties to be parsed.

* css/parser/CSSPropertyParser.h:
Update declarations of consumeBorder, consume2ValueShorthand and
consume4ValueShorthand.

* inspector/agents/InspectorCSSAgent.cpp:
(WebCore::InspectorCSSAgent::getSupportedCSSProperties):
Prevent CSS properties disabled behind a runtime flag from being exposed in
the CSS inspector tool.

* page/RuntimeEnabledFeatures.h:
(WebCore::RuntimeEnabledFeatures::setCSSLogicalEnabled):
(WebCore::RuntimeEnabledFeatures::cssLogicalEnabled const):
Add the CSSLogicalEnabled runtime flag.

Source/WebKit:

Add a CSSLogicalEnabled runtime flag.

* Shared/WebPreferences.yaml:

Source/WebKitLegacy/mac:

Add CSSLogicalEnabled runtime flag.

* WebView/WebPreferenceKeysPrivate.h:
* WebView/WebPreferences.mm:
(+[WebPreferences initialize]):
(-[WebPreferences cssLogicalEnabled]):
(-[WebPreferences setCSSLogicalEnabled:]):
* WebView/WebPreferencesPrivate.h:
* WebView/WebView.mm:
(-[WebView _preferencesChanged:]):

Tools:

* DumpRenderTree/TestOptions.cpp:
(TestOptions::TestOptions):
* DumpRenderTree/TestOptions.h:
* DumpRenderTree/mac/DumpRenderTree.mm:
(setWebPreferencesForTestOptions):
Allow tests to enable the CSSLogicalEnabled flag in WK1.

* Scripts/webkitpy/style/checkers/jsonchecker.py:
(JSONCSSPropertiesChecker.check_codegen_properties):
Allow CSS property definitions to have a 'runtime-flag' parameter which
disables the property when the specified runtime flag is disabled.

LayoutTests:

Add tests checking that CSS properties disabled behind runtime flags
are not exposed.

* platform/ios/webexposed/css-properties-as-js-properties-expected.txt: Added.
* platform/ios/webexposed/css-property-listing-expected.txt: Added.
* platform/mac/webexposed/css-properties-as-js-properties-expected.txt: Added.
* platform/mac/webexposed/css-property-listing-expected.txt: Added.
* platform/win/webexposed/css-properties-as-js-properties-expected.txt: Added.
* platform/win/webexposed/css-property-listing-expected.txt: Added.
* webexposed/css-properties-as-js-properties-expected.txt: Added.
* webexposed/css-properties-as-js-properties.html: Added.
* webexposed/css-properties-behind-flags-expected.txt: Added.
* webexposed/css-properties-behind-flags.html: Added.
* webexposed/css-property-listing-expected.txt: Added.
* webexposed/css-property-listing.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238244 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoUnreviewed, test gardening.
jiewen_tan@apple.com [Thu, 15 Nov 2018 21:03:59 +0000 (21:03 +0000)]
Unreviewed, test gardening.

* platform/mac-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238243 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoUnreviewed, rolling out r238220.
tsavell@apple.com [Thu, 15 Nov 2018 20:40:42 +0000 (20:40 +0000)]
Unreviewed, rolling out r238220.

Introduced failing tests to iOS and is slowing down EWS

Reverted changeset:

"[css-grid] Consider scrollbars in
populateGridPositionsForDirection()"
https://bugs.webkit.org/show_bug.cgi?id=191656
https://trac.webkit.org/changeset/238220

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238242 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoREGRESSION (r238212): WebKit.BackgroundColor* API test failures.
timothy@apple.com [Thu, 15 Nov 2018 20:32:38 +0000 (20:32 +0000)]
REGRESSION (r238212): WebKit.BackgroundColor* API test failures.
https://bugs.webkit.org/show_bug.cgi?id=191697
rdar://problem/46098570

Reviewed by Dean Jackson.

* TestWebKitAPI/Tests/mac/BackgroundColor.mm:
(TestWebKitAPI::TEST): Updated for new default background color.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238241 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[WebAuthN] PublicKeyCredentialCreationOptions::AuthenticatorSelectionCriteria::Authen...
jiewen_tan@apple.com [Thu, 15 Nov 2018 20:28:07 +0000 (20:28 +0000)]
[WebAuthN] PublicKeyCredentialCreationOptions::AuthenticatorSelectionCriteria::AuthenticatorAttachment should be optional
https://bugs.webkit.org/show_bug.cgi?id=191522

Reviewed by Brent Fulgham.

Source/WebCore:

Accordign to the WebIDL, AuthenticatorSelectionCriteria::AuthenticatorAttachment should be optional.
https://www.w3.org/TR/webauthn/#dictdef-authenticatorselectioncriteria

Covered by existing tests.

* Modules/webauthn/PublicKeyCredentialCreationOptions.h:
(WebCore::PublicKeyCredentialCreationOptions::AuthenticatorSelectionCriteria::decode):

Source/WebKit:

* UIProcess/WebAuthentication/AuthenticatorManager.cpp:
(WebKit::AuthenticatorManagerInternal::collectTransports):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238240 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoDragImage should have a complete default implementation
ross.kirsling@sony.com [Thu, 15 Nov 2018 20:23:54 +0000 (20:23 +0000)]
DragImage should have a complete default implementation
https://bugs.webkit.org/show_bug.cgi?id=191666

Reviewed by Dean Jackson.

Move WPE's stub implementation down into the base implementation file.

* SourcesWPE.txt:
* platform/DragImage.cpp:
(WebCore::dragImageSize):
(WebCore::deleteDragImage):
(WebCore::scaleDragImage):
(WebCore::dissolveDragImageToFraction):
(WebCore::createDragImageFromImage):
(WebCore::createDragImageIconForCachedImageFilename):
(WebCore::createDragImageForLink):
* platform/wpe/DragImageWPE.cpp: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238239 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[WebAuthN] UserHandle can be null
jiewen_tan@apple.com [Thu, 15 Nov 2018 20:01:52 +0000 (20:01 +0000)]
[WebAuthN] UserHandle can be null
https://bugs.webkit.org/show_bug.cgi?id=191521

Reviewed by Alex Christensen.

Source/WebCore:

According to the newest spec as of 7 August, 2018: https://www.w3.org/TR/webauthn/#conforming-authenticators-u2f.
UserHandle can now be null.

Covered by existing tests.

* Modules/webauthn/AuthenticatorAssertionResponse.h:
(WebCore::AuthenticatorAssertionResponse::create):
(WebCore::AuthenticatorAssertionResponse::userHandle const):
(WebCore::AuthenticatorAssertionResponse::AuthenticatorAssertionResponse):
* Modules/webauthn/AuthenticatorAssertionResponse.idl:
* Modules/webauthn/PublicKeyCredential.cpp:
(WebCore::PublicKeyCredential::tryCreate):
* Modules/webauthn/PublicKeyCredentialData.h:
(WebCore::PublicKeyCredentialData::encode const):
(WebCore::PublicKeyCredentialData::decode):
* Modules/webauthn/fido/DeviceResponseConverter.cpp:
(fido::readCTAPGetAssertionResponse):

LayoutTests:

* http/wpt/webauthn/public-key-credential-get-success-hid.https.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238238 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agowebkitpy: Check for specific process instead of using data migrator (Follow-up fix)
jbedard@apple.com [Thu, 15 Nov 2018 19:48:45 +0000 (19:48 +0000)]
webkitpy: Check for specific process instead of using data migrator (Follow-up fix)
https://bugs.webkit.org/show_bug.cgi?id=191551
<rdar://problem/45993156>

Reviewed by Aakash Jain.

* Scripts/webkitpy/xcode/simulated_device.py:
(SimulatedDevice.is_usable): Use software_variant instead of software_version for watchOS.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238237 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[SOUP] Make Accept-Languages header look more like other browsers
mcatanzaro@igalia.com [Thu, 15 Nov 2018 19:43:59 +0000 (19:43 +0000)]
[SOUP] Make Accept-Languages header look more like other browsers
https://bugs.webkit.org/show_bug.cgi?id=191614

Reviewed by Carlos Garcia Campos.

This changes our Accept-Language header to better match other browsers, from something
like this:

Accept-Language: en-us, en;q=0.90, es-es;q=0.80, es;q=0.70

to something like this:

Accept-Language: en-US,en;q=0.90,es-ES;q=0.80,es;q=0.70

There's no particular motivation for this change. I just noticed we were different for no
clear reason.

* NetworkProcess/soup/NetworkProcessSoup.cpp:
(WebKit::buildAcceptLanguages):
* UIProcess/API/glib/WebKitWebContext.cpp:
(webkit_web_context_set_preferred_languages):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238236 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[iOS] Shift + Tab does not focus previous field
dbates@webkit.org [Thu, 15 Nov 2018 19:15:28 +0000 (19:15 +0000)]
[iOS] Shift + Tab does not focus previous field
https://bugs.webkit.org/show_bug.cgi?id=191596
<rdar://problem/45892053>

Reviewed by Wenson Hsieh.

Source/WebKit:

Wire up the the tab and shift + tab key commands to the WKWebView/WKContentView's
action forwarding mechanism. Also rename -_prevAccessoryTab to -_previousAccessoryTab.

* Platform/spi/ios/UIKitSPI.h: Add more SPI.
* UIProcess/API/Cocoa/WKWebViewInternal.h:
* UIProcess/ios/WKContentViewInteraction.h:
* UIProcess/ios/WKContentViewInteraction.mm:
(-[WKContentView canPerformActionForWebView:withSender:]):
(-[WKContentView keyCommands]):
(-[WKContentView _nextAccessoryTabForWebView:]): Added.
(-[WKContentView _previousAccessoryTabForWebView:]): Added.
(-[WKContentView _nextAccessoryTab:]): Deleted.
(-[WKContentView _prevAccessoryTab:]): Deleted.

Tools:

Add infrastructure to support testing a key down event with modifiers by creating
and dispatching a UIEvent. This infrastructure replaces the previous mechanism in
Tools/WebKitTestRunner/ios/HIDEventGenerator.mm to generate a IOHIDEvent for a
keydown as it did not support creating an event with modifier key state that would
be recognized by UIKit.

* DumpRenderTree/ios/UIScriptControllerIOS.mm:
(WTR::UIScriptController::keyDown):
(WTR::UIScriptController::keyDownUsingHardwareKeyboard): Deleted.
(WTR::UIScriptController::keyUpUsingHardwareKeyboard): Deleted.
* TestRunnerShared/UIScriptContext/Bindings/UIScriptController.idl:
Add function uiController.keyDown() that takes a character that represents a keyboard key
and an array of modifier keys. The behavior of this function is analogous to eventSender.keyDown().
Remove functions uiController.keyDownUsingHardwareKeyboard() and uiController.keyUpUsingHardwareKeyboard()
as the former is replaced by uiController.keyDown() and the latter was never used.

* TestRunnerShared/UIScriptContext/UIScriptController.cpp:
(WTR::UIScriptController::keyDown): Added.
(WTR::UIScriptController::keyUpUsingHardwareKeyboard): Deleted.
(WTR::UIScriptController::keyDownUsingHardwareKeyboard): Deleted.
* TestRunnerShared/UIScriptContext/UIScriptController.h:
* WebKitTestRunner/ios/HIDEventGenerator.h:
* WebKitTestRunner/ios/HIDEventGenerator.mm:
(createHIDKeyDownEvent): Added.
(-[HIDEventGenerator keyDown:completionBlock:]): Deleted.
(-[HIDEventGenerator keyUp:completionBlock:]): Deleted.
* WebKitTestRunner/ios/UIScriptControllerIOS.mm:
(WTR::arrayLength): Added.
(WTR::parseModifier): Added.
(WTR::parseModifierArray): Added.
(WTR::UIScriptController::keyDown):
(WTR::UIScriptController::keyDownUsingHardwareKeyboard): Deleted.
(WTR::UIScriptController::keyUpUsingHardwareKeyboard): Deleted.

LayoutTests:

Add tests to ensure that we do not regress the tab and shift + tab key commands.

* fast/events/ios/focus-tab-next-field-expected.txt: Added.
* fast/events/ios/focus-tab-next-field.html: Added.
* fast/events/ios/focus-tab-previous-field-expected.txt: Added.
* fast/events/ios/focus-tab-previous-field.html: Added.
* platform/ios-wk1/TestExpectations: Skip test focus-tab-previous-field.html in Legacy WebKit as
we do not support testing key commands with modifier keys.
* resources/ui-helper.js:
(window.UIHelper.keyDown): Updated to support taking an optional array of modifiers (defaults: [])
and use the new infrastructure to generate a key down event for the specified character.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238235 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoModernize RTCPeerConnection handling of pendingActivity
youenn@apple.com [Thu, 15 Nov 2018 18:25:49 +0000 (18:25 +0000)]
Modernize RTCPeerConnection handling of pendingActivity
https://bugs.webkit.org/show_bug.cgi?id=191661

Reviewed by Eric Carlson.

makePendingActivity is the modern way to handle set/unset of pending activity.
No change of behavior.

* Modules/mediastream/RTCPeerConnection.cpp:
(WebCore::RTCPeerConnection::create):
(WebCore::RTCPeerConnection::doStop):
* Modules/mediastream/RTCPeerConnection.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238232 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoDelete old .xcfilelist files
krollin@apple.com [Thu, 15 Nov 2018 18:20:20 +0000 (18:20 +0000)]
Delete old .xcfilelist files
https://bugs.webkit.org/show_bug.cgi?id=191669
<rdar://problem/46081994>

Reviewed by Chris Dumez.

.xcfilelist files were created and added to the Xcode project files in
https://trac.webkit.org/changeset/238008/webkit. However, they caused
build issues and they were removed from the Xcode projects in
https://trac.webkit.org/changeset/238055/webkit. This check-in removes
the files from the repository altogether. They'll ultimately be
replaced with new files with names that indicate whether the
associated files are inputs to the Run Script phase or are files
created by the Run Script phase.

Source/JavaScriptCore:

* DerivedSources.xcfilelist: Removed.
* UnifiedSources.xcfilelist: Removed.

Source/WebCore:

No new tests -- no changed functionality.

* DerivedSources.xcfilelist: Removed.
* UnifiedSources.xcfilelist: Removed.

Source/WebKit:

* DerivedSources.xcfilelist: Removed.
* UnifiedSources.xcfilelist: Removed.

Tools:

* DumpRenderTree/DerivedSources.xcfilelist: Removed.
* WebKitTestRunner/DerivedSources.xcfilelist: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238231 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoUpdate RTCPeerConnection JS built-ins to be closer to specWe
youenn@apple.com [Thu, 15 Nov 2018 16:41:26 +0000 (16:41 +0000)]
Update RTCPeerConnection JS built-ins to be closer to specWe
https://bugs.webkit.org/show_bug.cgi?id=191665

Reviewed by Eric Carlson.

LayoutTests/imported/w3c:

* web-platform-tests/webrtc/rtcpeerconnection/rtcpeerconnection-idl-expected.txt:

Source/WebCore:

Simplify JS built-ins since we no longer need to support callback versions of the API.
Make sure to have the right number of parameters in the JS built-in functions.
Make some simplification to the code.
Covered by existing tests and rebased test.

* Modules/mediastream/RTCPeerConnection.js:
(createOffer):
(createAnswer):
(setLocalDescription):
(setRemoteDescription):
(addIceCandidate):
* Modules/mediastream/RTCPeerConnectionInternals.js:
(enqueueOperation):
(callbacksAndDictionaryOverload): Deleted.

LayoutTests:

* TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238230 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoREGRESSION(r238090): Composited iframe contents disappear after switching tabs in...
simon.fraser@apple.com [Thu, 15 Nov 2018 15:57:14 +0000 (15:57 +0000)]
REGRESSION(r238090): Composited iframe contents disappear after switching tabs in Safari
https://bugs.webkit.org/show_bug.cgi?id=191673
rdar://problem/46083440

Reviewed by Antti Koivisto.

Switching tabs in Safari triggers the "setIsInWindow" code path, that detaches the layer
tree for every Frame. They get re-attached on tab show, and for subframes this involves
the triggering of a fake style recalc in the parent document via scheduleInvalidateStyleAndLayerComposition().

The style diff that's sent to RenderLayerCompositor::layerStyleChanged() as a result of that
fake style recalc is RecompositeLayer, but the code was failing to trigger the necessary
layer configuration update that gets iframe layers parented.

This stop-gap patch triggers layer config updates on every RecompositeLayer diff. A future
patch will optimize this, and add a layout test.

* rendering/RenderLayerCompositor.cpp:
(WebCore::RenderLayerCompositor::layerStyleChanged):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238229 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoUnreviewed GTK+ gardening after r238220.
magomez@igalia.com [Thu, 15 Nov 2018 15:41:14 +0000 (15:41 +0000)]
Unreviewed GTK+ gardening after r238220.

* platform/gtk/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238228 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[LFC][IFC] Skip non-inflow boxes while splitting the inline runs.
zalan@apple.com [Thu, 15 Nov 2018 15:35:21 +0000 (15:35 +0000)]
[LFC][IFC] Skip non-inflow boxes while splitting the inline runs.
https://bugs.webkit.org/show_bug.cgi?id=191690

Reviewed by Antti Koivisto.

Skip all non-inflow boxes (floats, out-of-flow positioned elements). They don't participate in the inline run context.

* layout/Verification.cpp:
(WebCore::Layout::LayoutState::verifyAndOutputMismatchingLayoutTree const):
* layout/inlineformatting/InlineFormattingContext.cpp:
(WebCore::Layout::InlineFormattingContext::splitInlineRunIfNeeded const):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238227 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[LFC] FormattingContext base class should not declare computeStaticPosition.
zalan@apple.com [Thu, 15 Nov 2018 14:58:16 +0000 (14:58 +0000)]
[LFC] FormattingContext base class should not declare computeStaticPosition.
https://bugs.webkit.org/show_bug.cgi?id=191683

Reviewed by Antti Koivisto.

Apparently only BlockFormattingContext uses it.

* layout/FormattingContext.h:
* layout/blockformatting/BlockFormattingContext.h:
* layout/inlineformatting/InlineFormattingContext.cpp:
(WebCore::Layout::InlineFormattingContext::computeStaticPosition const): Deleted.
* layout/inlineformatting/InlineFormattingContext.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238226 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[LFC][IFC] Add support for in-flow positioned inline boxes.
zalan@apple.com [Thu, 15 Nov 2018 14:42:22 +0000 (14:42 +0000)]
[LFC][IFC] Add support for in-flow positioned inline boxes.
https://bugs.webkit.org/show_bug.cgi?id=191672

Reviewed by Antti Koivisto.

Source/WebCore:

We might offset the in-flow positioned runs differently once runs are moved over to the display tree.

Test: fast/inline/simple-inline-inflow-positioned.html

* layout/Verification.cpp:
(WebCore::Layout::outputMismatchingComplexLineInformationIfNeeded):
* layout/inlineformatting/InlineFormattingContext.cpp:
(WebCore::Layout::InlineFormattingContext::splitInlineRunsIfNeeded const):
(WebCore::Layout::InlineFormattingContext::postProcessInlineRuns const):
(WebCore::Layout::InlineFormattingContext::placeInFlowPositionedChildren const):
(WebCore::Layout::InlineFormattingContext::collectInlineContentForSubtree const):
* layout/inlineformatting/InlineFormattingContext.h:
* layout/inlineformatting/InlineRun.h:
(WebCore::Layout::InlineRun::moveVertically):
* layout/layouttree/LayoutTreeBuilder.cpp:
(WebCore::Layout::outputInlineRuns):

Tools:

* LayoutReloaded/misc/LFC-passing-tests.txt:

LayoutTests:

* fast/inline/simple-inline-inflow-positioned-expected.txt: Added.
* fast/inline/simple-inline-inflow-positioned.html: Added.
* platform/ios/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238225 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[GStreamer][WebRTC] Add support for sending silence or silencing an incoming track
commit-queue@webkit.org [Thu, 15 Nov 2018 14:39:27 +0000 (14:39 +0000)]
[GStreamer][WebRTC] Add support for sending silence or silencing an incoming track
https://bugs.webkit.org/show_bug.cgi?id=191631

Patch by Thibault Saunier <tsaunier@igalia.com> on 2018-11-15
Reviewed by Xabier Rodriguez-Calvar.

This will be tested once webkit.org/b/186933 is implemented.

* platform/mediastream/gstreamer/RealtimeIncomingAudioSourceLibWebRTC.cpp:
(WebCore::RealtimeIncomingAudioSourceLibWebRTC::OnData):
* platform/mediastream/gstreamer/RealtimeOutgoingAudioSourceLibWebRTC.cpp:
(WebCore::RealtimeOutgoingAudioSourceLibWebRTC::pullAudioData):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238224 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoREGRESSION(r238178): fast/forms/access-key-mutated.html and fast/forms/access-key...
antti@apple.com [Thu, 15 Nov 2018 13:57:50 +0000 (13:57 +0000)]
REGRESSION(r238178): fast/forms/access-key-mutated.html and fast/forms/access-key-case-insensitive.html are timing out
https://bugs.webkit.org/show_bug.cgi?id=191642

Reviewed by Zalan Bujtas.

Source/WebCore:

Invalidate access key map even when thorttling style recalcs.

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

LayoutTests:

* TestExpectations: Unskip
* platform/mac-wk2/TestExpectations: Unskip
* platform/mac-wk2/accessibility/mac/selection-notification-focus-change-expected.txt:

Rebase. Focus timings depends on layout timings.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238223 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoRemove fonts from CSSFontFaceSet safely
antti@apple.com [Thu, 15 Nov 2018 13:48:52 +0000 (13:48 +0000)]
Remove fonts from CSSFontFaceSet safely
https://bugs.webkit.org/show_bug.cgi?id=191676

Reviewed by Zalan Bujtas.

Source/WebCore:

Test: fast/text/font-face-set-remove-safely.html

* css/CSSFontFaceSet.cpp:
(WebCore::CSSFontFaceSet::remove):

LayoutTests:

* fast/text/font-face-set-remove-safely-expected.txt: Added.
* fast/text/font-face-set-remove-safely.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238222 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[WebAuthN] Make AuthenticatorManager
fred.wang@free.fr [Thu, 15 Nov 2018 13:43:11 +0000 (13:43 +0000)]
[WebAuthN] Make AuthenticatorManager
https://bugs.webkit.org/show_bug.cgi?id=189279

Unreviewed build fix.

Patch by Frederic Wang <fwang@igalia.com> on 2018-11-15

* UIProcess/WebAuthentication/Cocoa/LocalAuthenticator.mm: Add missing header.
* UIProcess/WebAuthentication/Cocoa/LocalConnection.mm: Ditto.
* UIProcess/WebAuthentication/WebAuthenticatorCoordinatorProxy.cpp: Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238221 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[css-grid] Consider scrollbars in populateGridPositionsForDirection()
rego@igalia.com [Thu, 15 Nov 2018 12:19:24 +0000 (12:19 +0000)]
[css-grid] Consider scrollbars in populateGridPositionsForDirection()
https://bugs.webkit.org/show_bug.cgi?id=191656

Reviewed by Javier Fernandez.

LayoutTests/imported/w3c:

Imported tests from WPT.

* web-platform-tests/css/css-grid/grid-model/grid-container-scrollbar-001-expected.html: Added.
* web-platform-tests/css/css-grid/grid-model/grid-container-scrollbar-001.html: Added.
* web-platform-tests/css/css-grid/grid-model/grid-container-scrollbar-vertical-lr-001-expected.html: Added.
* web-platform-tests/css/css-grid/grid-model/grid-container-scrollbar-vertical-lr-001.html: Added.
* web-platform-tests/css/css-grid/grid-model/grid-container-scrollbar-vertical-rl-001-expected.html: Added.
* web-platform-tests/css/css-grid/grid-model/grid-container-scrollbar-vertical-rl-001.html: Added.
* web-platform-tests/css/css-grid/grid-model/w3c-import.log:

Source/WebCore:

We never care about scrollbars in RenderGrid::populateGridPositionsForDirection(),
that's fine if the scrollbars are at the end (e.g. on the right in horizontal writing mode and LTR direction)
but it causes problems when they're at the beginning (e.g. on the left in horizontal writing mode and RTL direction).

The patch modifies the method so it takes into account scrollbar size
in order to compute the position of the columns/rows depending on the direction and the writing mode.

Tests: imported/w3c/web-platform-tests/css/css-grid/grid-model/grid-container-scrollbar-001.html
       imported/w3c/web-platform-tests/css/css-grid/grid-model/grid-container-scrollbar-vertical-lr-001.html
       imported/w3c/web-platform-tests/css/css-grid/grid-model/grid-container-scrollbar-vertical-rl-001.html

* rendering/RenderGrid.cpp:
(WebCore::RenderGrid::populateGridPositionsForDirection):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238220 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoMove scripts for Derived and Unified Sources to external files
krollin@apple.com [Thu, 15 Nov 2018 06:01:15 +0000 (06:01 +0000)]
Move scripts for Derived and Unified Sources to external files
https://bugs.webkit.org/show_bug.cgi?id=191670
<rdar://problem/46082278>

Reviewed by Keith Miller.

Move the scripts in the Generate Derived Sources and Generate Unified
Sources Run Script phases from the Xcode projects to external shell
script files. Then invoke those scripts from the Run Script phases.
This refactoring is being performed to support later work that will
invoke these scripts in other contexts.

The scripts were maintained as-is when making the move. I did a little
reformatting and added 'set -e' to the top of each file, but that's
it.

Source/JavaScriptCore:

* JavaScriptCore.xcodeproj/project.pbxproj:
* Scripts/generate-derived-sources.sh: Added.
* Scripts/generate-unified-sources.sh: Added.

Source/WebCore:

No new tests -- no changed functionality.

* Scripts/generate-derived-sources.sh: Added.
* Scripts/generate-unified-sources.sh: Added.
* WebCore.xcodeproj/project.pbxproj:

Source/WebKit:

* Scripts/generate-derived-sources.sh: Added.
* Scripts/generate-unified-sources.sh: Added.
* WebKit.xcodeproj/project.pbxproj:

Tools:

* DumpRenderTree/DumpRenderTree.xcodeproj/project.pbxproj:
* DumpRenderTree/Scripts/generate-derived-sources.sh: Added.
* WebKitTestRunner/Scripts/generate-derived-sources.sh: Added.
* WebKitTestRunner/WebKitTestRunner.xcodeproj/project.pbxproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238219 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoWebKit.DecidePolicyForNavigationActionForHyperlinkThatRedirects API fails when PSON...
cdumez@apple.com [Thu, 15 Nov 2018 03:35:37 +0000 (03:35 +0000)]
WebKit.DecidePolicyForNavigationActionForHyperlinkThatRedirects API fails when PSON is enabled
https://bugs.webkit.org/show_bug.cgi?id=191640

Reviewed by Alex Christensen.

Source/WebKit:

The issue was that when process-swapping on a redirect, the parameters provided by the new
WebContent process for navigationActionData / originatingFrameInfoData / frameSecurityOrigin
are not correct because it does not have sufficient information (it does not know a swap
happened).

To address the issue, we now store the navigationActionData / originatingFrameInfoData /
frameSecurityOrigin on the Navigation object whenever decidePolicyForNavigationAction is
called. If the decidePolicyForNavigationAction is for a redirect, use the ones we stored
on the Navigation object instead of the ones provided by the WebContent process. In case
of redirect, those do not change anyway.

This fixes all WebKit.DecidePolicyForNavigationAction* tests with PSON enabled.

* UIProcess/API/APINavigation.h:
(API::Navigation::currentRequestIsRedirect const):
(API::Navigation::wasUserInitiated const):
(API::Navigation::shouldForceDownload const):
(API::Navigation::treatAsSameOriginNavigation const):
(API::Navigation::hasOpenedFrames const):
(API::Navigation::openedViaWindowOpenWithOpener const):
(API::Navigation::opener const):
(API::Navigation::requesterOrigin const):
(API::Navigation::lockHistory const):
(API::Navigation::lockBackForwardList const):
(API::Navigation::clientRedirectSourceForHistory const):
(API::Navigation::setLastNavigationAction):
(API::Navigation::lastNavigationAction const):
(API::Navigation::setOriginatingFrameInfo):
(API::Navigation::originatingFrameInfo const):
(API::Navigation::setDestinationFrameSecurityOrigin):
(API::Navigation::destinationFrameSecurityOrigin const):
* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::decidePolicyForNavigationActionAsync):
(WebKit::WebPageProxy::decidePolicyForNavigationAction):
(WebKit::WebPageProxy::decidePolicyForNavigationActionSync):
* UIProcess/WebPageProxy.h:

Tools:

Add API test coverage.

* TestWebKitAPI/Tests/WebKitCocoa/DecidePolicyForNavigationAction.mm:
(runDecidePolicyForNavigationActionForHyperlinkThatRedirects):
(TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238218 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoFix #end vs. #endif typo.
krollin@apple.com [Thu, 15 Nov 2018 03:33:40 +0000 (03:33 +0000)]
Fix #end vs. #endif typo.
https://bugs.webkit.org/show_bug.cgi?id=191668
<rdar://problem/46081704>

Reviewed by Alexey Proskuryakov.

Source/WebCore/SourcesCocoa.txt had a #end that should have been a
#endif. Fix this, an add a check to generate-unified-source-bundles.rb
to detect similar typos.

Source/WebCore:

No new tests -- no changed functionality.

* SourcesCocoa.txt:

Source/WTF:

* Scripts/generate-unified-source-bundles.rb:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238217 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoREGRESSION (r238115): [iOS] TestWebKitAPI.ProcessSwap.NavigateToInvalidURL is failing
cdumez@apple.com [Thu, 15 Nov 2018 03:32:32 +0000 (03:32 +0000)]
REGRESSION (r238115): [iOS] TestWebKitAPI.ProcessSwap.NavigateToInvalidURL is failing
https://bugs.webkit.org/show_bug.cgi?id=191660

Reviewed by Ryosuke Niwa.

The test tries to navigate to an invalid URL and makes sure that we do not process swap
for such load. The test was inherently racing because it was doing:
<body onload="setTimeout(() => alert('DONE'), 0); location.href = 'http://A=a%B=b'">

Its load event handler tries to navigate to 'http://A=a%B=b' and sets a 0 timer to
log an alert, which causes the API test to keep going and check that we did not swap
process. The issue is that a 0 timer may not be sufficient to trigger the load, as
the policy decision is asynchronous and relies on the UIProcess responding to it.

To address the issue, the test now waits for the next policy decision, spins the
run loop and then makes sure we did not swap process. It no longer relies on a message
from the page.

* TestWebKitAPI/Tests/WebKitCocoa/ProcessSwapOnNavigation.mm:
(-[PSONNavigationDelegate webView:decidePolicyForNavigationAction:decisionHandler:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238216 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoWebKit.WebsiteDataStoreCustomPaths API test is failing when enabling process prewarming
cdumez@apple.com [Thu, 15 Nov 2018 03:30:58 +0000 (03:30 +0000)]
WebKit.WebsiteDataStoreCustomPaths API test is failing when enabling process prewarming
https://bugs.webkit.org/show_bug.cgi?id=191638

Reviewed by Alex Christensen.

Source/WebKit:

WebProcessPool::prewarmProcess() should not create the default WebSite Data Store if it
does not exist yet. This is bad for memory consumption and it is what was causing this
API test to fail.

WebProcessPool::prewarmProcess() now tries to use the following data stores in this
order of preference:
1. WebProcessPool::m_websiteDataStore (aka this process pool's primary data store)
2. The data store of the last WebProcessProxy that was created
3. The default data store if it exists

If no suitable data store is found, we cancel the process prewarming and log a console
message instead.

* UIProcess/WebProcessPool.cpp:
(WebKit::WebProcessPool::prewarmProcess):

Tools:

Add API test coverage.

* TestWebKitAPI/Tests/WebKitCocoa/WebsiteDataStoreCustomPaths.mm:
(runWebsiteDataStoreCustomPaths):
(TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238215 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoRemove VideoFullscreenLayerManager.mm from WebCore/SourcesCocoa.txt
krollin@apple.com [Thu, 15 Nov 2018 02:40:08 +0000 (02:40 +0000)]
Remove VideoFullscreenLayerManager.mm from WebCore/SourcesCocoa.txt
https://bugs.webkit.org/show_bug.cgi?id=191667
<rdar://problem/46081286>

Reviewed by Eric Carlson.

VideoFullscreenLayerManager.mm no longer exists.

No new tests -- No changed functionality.

* SourcesCocoa.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238214 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoEnabled dark mode CSS support by default.
timothy@apple.com [Thu, 15 Nov 2018 02:27:32 +0000 (02:27 +0000)]
Enabled dark mode CSS support by default.
https://bugs.webkit.org/show_bug.cgi?id=191609
rdar://problem/46046861

Reviewed by Megan Gardner.

Source/WebCore:

* page/RuntimeEnabledFeatures.h: Set m_isDarkModeCSSEnabled to true.

Source/WebKit:

* Shared/WebPreferences.yaml: Set DarkModeCSSEnabled's defaultValue to true.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238213 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoDefault the view background color and text color to different values when in dark...
timothy@apple.com [Thu, 15 Nov 2018 01:48:20 +0000 (01:48 +0000)]
Default the view background color and text color to different values when in dark mode.
https://bugs.webkit.org/show_bug.cgi?id=191607
rdar://problem/46045854

Reviewed by Dean Jackson.

Source/WebCore:

Test: css-dark-mode/default-colors.html

* css/html.css:
(html): Set color: text on macOS.
* dom/Document.cpp:
(WebCore::Document::processSupportedColorSchemes): Call recalculateBaseBackgroundColor().
* editing/EditingStyle.cpp:
(WebCore::caretColorFromStyle): Added.
(WebCore::EditingStyle::prepareToApplyAt): Use equalIgnoringSemanticColor. Check for
caret-color directly since removeEquivalentProperties fails with semantic colors.
(WebCore::extractPropertiesNotIn): Use equalIgnoringSemanticColor. Check for caret-color
directly since removeEquivalentProperties fails with semantic colors.
* page/Frame.cpp:
(WebCore::Frame::createView): Drop backgroundColor.
* page/Frame.h:
* page/FrameView.cpp:
(WebCore::FrameView::recalculateBaseBackgroundColor): Added.
(WebCore::FrameView::updateBackgroundRecursively): Drop backgroundColor argument.
Calculate the backgroundColor based on the transparent argument only.
* page/FrameView.h:
* platform/graphics/Color.h:
(WebCore::equalIgnoringSemanticColor): Added for EditingStyle.
* rendering/RenderBox.cpp:
(WebCore::RenderBox::styleDidChange): Call recalculateBaseBackgroundColor().
* rendering/RenderLayerBacking.cpp:
(WebCore::RenderLayerBacking::shouldDumpPropertyForLayer const): Use isWhiteColor()
since it ignores the semantic color flag.
* testing/Internals.cpp:
(WebCore::Internals::setViewIsTransparent): Drop backgroundColor.
(WebCore::Internals::viewBaseBackgroundColor): Added.
* testing/Internals.h:
* testing/Internals.idl: Added viewBaseBackgroundColor.

Source/WebKit:

* UIProcess/Cocoa/WebViewImpl.mm:
(WebKit::WebViewImpl::backgroundColor const): Use controlBackgroundColor.
(WebKit::WebViewImpl::updateLayer): Simplified and use backgroundColor() getter.
* WebProcess/WebCoreSupport/WebFrameLoaderClient.cpp:
(WebKit::WebFrameLoaderClient::transitionToCommittedForNewPage): Drop backgroundColor.
* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::setDrawsBackground): Drop backgroundColor.

Source/WebKitLegacy/win:

* WebFrame.cpp:
(WebFrame::updateBackground): Drop backgroundColor.
* WebCoreSupport/WebFrameLoaderClient.cpp:
(WebFrameLoaderClient::transitionToCommittedForNewPage): Drop backgroundColor.

LayoutTests:

* css-dark-mode/default-colors-expected.txt: Added.
* css-dark-mode/default-colors.html: Added.
* inspector/css/shadow-scoped-style-expected.txt: Rebaseline.
* platform/mac/printing/width-overflow-expected.txt: Rebaseline.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238212 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoVersioning.
alancoon@apple.com [Thu, 15 Nov 2018 01:38:30 +0000 (01:38 +0000)]
Versioning.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238211 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoRevert "Versioning."
alancoon@apple.com [Thu, 15 Nov 2018 01:27:21 +0000 (01:27 +0000)]
Revert "Versioning."

This reverts commit 657b4918e31469e5d37af91eca7b4a07f0335d9c.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238210 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[WebGPU] Code quality concerns raised for 191291: [WebGPU] Experimental prototype...
justin_fan@apple.com [Thu, 15 Nov 2018 00:33:53 +0000 (00:33 +0000)]
[WebGPU] Code quality concerns raised for 191291: [WebGPU] Experimental prototype for WebGPURenderPipeline and WebGPUSwapChain
https://bugs.webkit.org/show_bug.cgi?id=191383

Reviewed by Dean Jackson.

Covered by existing WebGPU tests introduced in original patch.

* Modules/webgpu/GPUDevice.h:
* Modules/webgpu/GPUPipelineStageDescriptor.h:
* Modules/webgpu/GPURenderPipelineDescriptor.h: Now a base struct with a guaranteed vertex stage member.
(): Refactored into enum class.
(WebCore::GPURenderPipelineDescriptor::GPURenderPipelineDescriptor): Removed in favor of init-list construction.
(WebCore::GPURenderPipelineDescriptor::primitiveTopology): Now a proper enum class member.
* Modules/webgpu/GPUShaderModule.h:
* Modules/webgpu/WebGPUDevice.cpp:
(WebCore::WebGPUDevice::createRenderPipeline const):
* Modules/webgpu/WebGPUShaderModule.h:
(WebCore::WebGPUShaderModule::module const):
* Modules/webgpu/WebGPUShaderStage.h: Replaced enum with constants to better reflect IDL.
* Modules/webgpu/cocoa/GPURenderPipeline.h:
* Modules/webgpu/cocoa/GPURenderPipelineMetal.mm:
(WebCore::setFunctionsForPipelineDescriptor):
(WebCore::GPURenderPipeline::create):
* Modules/webgpu/cocoa/GPUSwapChain.h:
* WebCore.xcodeproj/project.pbxproj: Removed GPUPipelineDescriptorBase.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238208 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoLayout test fast/layers/no-clipping-overflow-hidden-added-after-transform.html is...
ryanhaddad@apple.com [Thu, 15 Nov 2018 00:32:57 +0000 (00:32 +0000)]
Layout test fast/layers/no-clipping-overflow-hidden-added-after-transform.html is a flaky failure on EWS bots
https://bugs.webkit.org/show_bug.cgi?id=191658

Unreviewed test gardening.

* platform/mac-wk2/TestExpectations: Mark this test as flaky on Sierra Release WK2 to reduce the number of retries on mac-wk2-ews.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238207 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoWeb Inspector: Pass Inspector::FrontendChannel as a reference connect/disconnect...
joepeck@webkit.org [Wed, 14 Nov 2018 23:56:07 +0000 (23:56 +0000)]
Web Inspector: Pass Inspector::FrontendChannel as a reference connect/disconnect methods
https://bugs.webkit.org/show_bug.cgi?id=191612

Reviewed by Matt Baker.

Source/JavaScriptCore:

* inspector/InspectorFrontendRouter.cpp:
(Inspector::FrontendRouter::connectFrontend):
(Inspector::FrontendRouter::disconnectFrontend):
* inspector/InspectorFrontendRouter.h:
* inspector/JSGlobalObjectInspectorController.cpp:
(Inspector::JSGlobalObjectInspectorController::connectFrontend):
(Inspector::JSGlobalObjectInspectorController::disconnectFrontend):
* inspector/JSGlobalObjectInspectorController.h:
* inspector/remote/RemoteControllableTarget.h:
* inspector/remote/cocoa/RemoteConnectionToTargetCocoa.mm:
(Inspector::RemoteConnectionToTarget::setup):
(Inspector::RemoteConnectionToTarget::close):
* inspector/remote/glib/RemoteConnectionToTargetGlib.cpp:
(Inspector::RemoteConnectionToTarget::setup):
(Inspector::RemoteConnectionToTarget::close):
* runtime/JSGlobalObjectDebuggable.cpp:
(JSC::JSGlobalObjectDebuggable::connect):
(JSC::JSGlobalObjectDebuggable::disconnect):
* runtime/JSGlobalObjectDebuggable.h:

Source/WebCore:

* inspector/InspectorController.cpp:
(WebCore::InspectorController::connectFrontend):
(WebCore::InspectorController::disconnectFrontend):
(WebCore::InspectorController::show):
* inspector/InspectorController.h:
* inspector/WorkerInspectorController.cpp:
(WebCore::WorkerInspectorController::connectFrontend):
(WebCore::WorkerInspectorController::disconnectFrontend):
* page/PageDebuggable.cpp:
(WebCore::PageDebuggable::connect):
(WebCore::PageDebuggable::disconnect):
* page/PageDebuggable.h:
* testing/Internals.cpp:
(WebCore::InspectorStubFrontend::InspectorStubFrontend):
(WebCore::InspectorStubFrontend::closeWindow):
* workers/service/context/ServiceWorkerDebuggable.cpp:
(WebCore::ServiceWorkerDebuggable::connect):
(WebCore::ServiceWorkerDebuggable::disconnect):
* workers/service/context/ServiceWorkerDebuggable.h:
* workers/service/context/ServiceWorkerInspectorProxy.cpp:
(WebCore::ServiceWorkerInspectorProxy::connectToWorker):
(WebCore::ServiceWorkerInspectorProxy::disconnectFromWorker):
* workers/service/context/ServiceWorkerInspectorProxy.h:

Source/WebKit:

* UIProcess/Automation/WebAutomationSession.cpp:
(WebKit::WebAutomationSession::connect):
(WebKit::WebAutomationSession::disconnect):
(WebKit::WebAutomationSession::terminate):
* UIProcess/Automation/WebAutomationSession.h:
* UIProcess/WebPageDebuggable.cpp:
(WebKit::WebPageDebuggable::connect):
(WebKit::WebPageDebuggable::disconnect):
* UIProcess/WebPageDebuggable.h:
* UIProcess/WebPageInspectorController.cpp:
(WebKit::WebPageInspectorController::connectFrontend):
(WebKit::WebPageInspectorController::disconnectFrontend):
* UIProcess/WebPageInspectorController.h:
* WebProcess/WebPage/WebInspector.cpp:
(WebKit::WebInspector::close):
* WebProcess/WebPage/WebPageInspectorTarget.cpp:
(WebKit::WebPageInspectorTarget::connect):
(WebKit::WebPageInspectorTarget::disconnect):

Source/WebKitLegacy/mac:

* WebCoreSupport/WebInspectorClient.mm:
(-[WebInspectorWindowController destroyInspectorView]):

Source/WebKitLegacy/win:

* WebCoreSupport/WebInspectorClient.cpp:
(WebInspectorFrontendClient::destroyInspectorView):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238206 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoVersioning.
alancoon@apple.com [Wed, 14 Nov 2018 23:40:00 +0000 (23:40 +0000)]
Versioning.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238205 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoWeb Inspector: Table should recalculate scrollable height when resized
mattbaker@apple.com [Wed, 14 Nov 2018 22:54:18 +0000 (22:54 +0000)]
Web Inspector: Table should recalculate scrollable height when resized
https://bugs.webkit.org/show_bug.cgi?id=191328
<rdar://problem/45854412>

Reviewed by Devin Rousso.

* UserInterface/Views/NetworkTableContentView.js:
(WI.NetworkTableContentView.prototype._hideDetailView):
Force table columns to layout after hiding the detail view.

* UserInterface/Views/Table.js:
(WI.Table):
(WI.Table.prototype.showColumn):
Call `_resizeColumnsAndFiller` instead of `resize`. The latter also cleared
`_cachedWidth`, which isn't unnecessary since the width hasn't changed.

(WI.Table.prototype.layout):
previously `_resizeColumnsAndFiller` always occurred after `_updateVisibleRows`,
it was just a matter of whether the cached width and height were cleared first.
That now happens in `sizeDidChange`.

(WI.Table.prototype.sizeDidChange):
(WI.Table.prototype._updateVisibleRows):
(WI.Table.prototype.resize): Deleted.
Internally this cleared the cached width and height, which moved to `sizeDidChange`.
Externally this was used by NetworkTableContentView to force columns to
layout after hiding the details view. Calling `updateLayout` on the table
accomplishes the same thing using the standard View API.

(WI.Table.prototype._resizeColumnsAndFiller):
Only update the cached width, since the height will already have been
updated. The number of rows doesn't change during this method, so that
check was removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238203 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoWeb Inspector: REGRESSION(r236540): Debugger: breakpoints are sorted alphabetically...
drousso@apple.com [Wed, 14 Nov 2018 22:39:25 +0000 (22:39 +0000)]
Web Inspector: REGRESSION(r236540): Debugger: breakpoints are sorted alphabetically instead of numerically
https://bugs.webkit.org/show_bug.cgi?id=191560

Reviewed by Joseph Pecoraro.

When sorting breakpoints, if the `WI.TreeElement`s being compared both contain instances of
`WI.Breakpoint`, sort by `WI.SourceCodeLocation` instead of title.

We only do this for `WI.Breakpoint` because they are the only debuggable item that has a
`WI.SourceCodeLocation` (e.g. `WI.EventBreakpoint` is just an event name string).

* UserInterface/Views/DebuggerSidebarPanel.js:
(WI.DebuggerSidebarPanel.prototype._addTreeElement):
(WI.DebuggerSidebarPanel.prototype._compareBreakpointTreeElements): Added.
(WI.DebuggerSidebarPanel.prototype._addIssue):
(WI.DebuggerSidebarPanel.prototype._addTreeElement.comparator): Deleted.
(WI.DebuggerSidebarPanel.prototype._compareTreeElements): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238202 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoWeb Inspector: Styles: shift-clicking on a property should extend selection
nvasilyev@apple.com [Wed, 14 Nov 2018 22:28:52 +0000 (22:28 +0000)]
Web Inspector: Styles: shift-clicking on a property should extend selection
https://bugs.webkit.org/show_bug.cgi?id=191575
<rdar://problem/46012855>

Reviewed by Devin Rousso.

When there's at least one property is selected, Shift-clicking another property should extend the selection.
Pressing Up or Down keys while holding Shift also extends the selection.

Remove unnecessary `_startedSelection`. Use `hasSelectedProperties` and `_isMousePressed` instead.

* UserInterface/Views/SpreadsheetCSSStyleDeclarationEditor.js:
`_isMousePressed` was set to `true` by mistake, causing numerous assertion fails.

(WI.SpreadsheetCSSStyleDeclarationEditor.prototype.set focusIndex):
(WI.SpreadsheetCSSStyleDeclarationEditor.prototype.hasSelectedProperties): Renamed from _hasSelectedProperties.
(WI.SpreadsheetCSSStyleDeclarationEditor.prototype.extendSelectedProperties): Added.
(WI.SpreadsheetCSSStyleDeclarationEditor.prototype.spreadsheetStylePropertyMouseLeave): Removed.
(WI.SpreadsheetCSSStyleDeclarationEditor.prototype.spreadsheetStylePropertyCopy):

(WI.SpreadsheetCSSStyleDeclarationEditor.prototype._handleKeyDown):
Implement Shift-ArrowUp/ArrowDown keys.

* UserInterface/Views/SpreadsheetCSSStyleDeclarationSection.js:
(WI.SpreadsheetCSSStyleDeclarationSection):
(WI.SpreadsheetCSSStyleDeclarationSection.prototype.spreadsheetCSSStyleDeclarationEditorPropertyBlur):
(WI.SpreadsheetCSSStyleDeclarationSection.prototype.spreadsheetCSSStyleDeclarationEditorPropertyMouseLeave): Removed.
This was only used to set _startedSelection.

(WI.SpreadsheetCSSStyleDeclarationSection.prototype._handleMouseDown):
(WI.SpreadsheetCSSStyleDeclarationSection.prototype._handleWindowClick):
(WI.SpreadsheetCSSStyleDeclarationSection.prototype._handleClick):
Remove _startedSelection and use hasSelectedProperties method instead.

* UserInterface/Views/SpreadsheetStyleProperty.js:
(WI.SpreadsheetStyleProperty):
Remove spreadsheetStylePropertyMouseLeave since it is no longer used.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238201 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoUpdate prefers-color-scheme media query matching based on GitHub issue #3278.
timothy@apple.com [Wed, 14 Nov 2018 22:24:09 +0000 (22:24 +0000)]
Update prefers-color-scheme media query matching based on GitHub issue #3278.
https://bugs.webkit.org/show_bug.cgi?id=191654
rdar://problem/46074307

Reviewed by Simon Fraser.

Source/WebCore:

Test: css-dark-mode/prefers-color-scheme.html

* css/MediaQueryEvaluator.cpp:
(WebCore::prefersColorSchemeEvaluate): Return true when there is no value. Return false
for `no-preference` since there is no macOS option for no user preference.
* css/MediaQueryExpression.cpp:
(WebCore::isFeatureValidWithoutValue): Added prefers-color-scheme.

LayoutTests:

* css-dark-mode/prefers-color-scheme-expected.txt: Rebaseline.
* css-dark-mode/prefers-color-scheme.html: Updates to better match the spec.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238200 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoWeb Inspector: Canvas: send a call stack with each action instead of an array of...
drousso@apple.com [Wed, 14 Nov 2018 22:03:14 +0000 (22:03 +0000)]
Web Inspector: Canvas: send a call stack with each action instead of an array of call frames
https://bugs.webkit.org/show_bug.cgi?id=191628

Reviewed by Dean Jackson.

Source/WebCore:

Updated existing test: inspector/model/recording.html

* inspector/InspectorCanvas.h:
* inspector/InspectorCanvas.cpp:
(WebCore::InspectorCanvas::indexForData):
(WebCore::InspectorCanvas::buildInitialState):
(WebCore::InspectorCanvas::buildAction):
Drive-by: prevent de-duplicated objects from being destroyed while recording.
Source/WebInspectorUI:

* UserInterface/Models/Recording.js:
(WI.Recording.prototype.async swizzle):

* UserInterface/Models/RecordingAction.js:
(WI.RecordingAction.fromPayload):
(WI.RecordingAction.prototype.async swizzle):

LayoutTests:

* inspector/model/recording.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238199 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoWeb Inspector: add drag+drop for importing Audits and Recordings
drousso@apple.com [Wed, 14 Nov 2018 22:01:03 +0000 (22:01 +0000)]
Web Inspector: add drag+drop for importing Audits and Recordings
https://bugs.webkit.org/show_bug.cgi?id=191566

Reviewed by Joseph Pecoraro.

Reorganize the file loading logic for the Audit and Canvas tabs/managers. Add "drop" event
handlers to the tabs that add and show the corresponding object for the imported data.

* UserInterface/Views/AuditTabContentView.js:
(WI.AuditTabContentView):
(WI.AuditTabContentView.prototype._handleDragOver): Added.
(WI.AuditTabContentView.prototype._handleDrop): Added.
* UserInterface/Views/CanvasTabContentView.js:
(WI.CanvasTabContentView):
(WI.CanvasTabContentView.prototype._addCanvas):
(WI.CanvasTabContentView.prototype._recordingImportedOrStopped):
(WI.CanvasTabContentView.prototype._handleDragOver): Added.
(WI.CanvasTabContentView.prototype._handleDrop): Added.

* UserInterface/Controllers/AuditManager.js:
(WI.AuditManager.prototype.async processJSON): Added.
(WI.AuditManager.prototype.export):
(WI.AuditManager.prototype.import): Deleted.
* UserInterface/Controllers/CanvasManager.js:
(WI.CanvasManager.prototype.processJSON): Added.
(WI.CanvasManager.prototype.importRecording): Deleted.

* UserInterface/Views/AuditNavigationSidebarPanel.js:
(WI.AuditNavigationSidebarPanel):
(WI.AuditNavigationSidebarPanel.prototype.showDefaultContentView):
(WI.AuditNavigationSidebarPanel.prototype._handleImportButtonNavigationItemClicked):
* UserInterface/Views/CanvasSidebarPanel.js:
(WI.CanvasSidebarPanel):
(WI.CanvasSidebarPanel.prototype._handleImportButtonNavigationItemClicked): Added.
(WI.CanvasSidebarPanel.prototype._recordingChanged):
(WI.CanvasSidebarPanel.prototype._updateRecordNavigationItem):

* UserInterface/Views/CanvasOverviewContentView.js:
(WI.CanvasOverviewContentView):
(WI.CanvasOverviewContentView.prototype._handleImportButtonNavigationItemClicked): Added.

* UserInterface/Base/FileUtilities.js:
(WI.FileUtilities.save): Added.
(WI.FileUtilities.importText): Added.
(WI.FileUtilities.importJSON): Added.
(WI.FileUtilities.async readText): Added.
(WI.FileUtilities.async readJSON): Added.
(WI.saveDataToFile): Deleted.
(WI.loadDataFromFile): Deleted.
* UserInterface/Base/Main.js:
(WI._contextMenuRequested):
(WI._save):
(WI._saveAs):
* UserInterface/Views/ContextMenuUtilities.js:
(WI.appendContextMenuItemsForSourceCode):
* UserInterface/Views/LogContentView.js:
(WI.LogContentView.prototype._handleContextMenuEvent):
* UserInterface/Views/NetworkTableContentView.js:
(WI.NetworkTableContentView.prototype._exportHAR):
* UserInterface/Views/RecordingContentView.js:
(WI.RecordingContentView.prototype._exportRecording):
* UserInterface/Views/SpreadsheetCSSStyleDeclarationSection.js:
(WI.SpreadsheetCSSStyleDeclarationSection.prototype._save):
Move file related functions into a static class container.

* Localizations/en.lproj/localizedStrings.js:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238198 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoWeb Inspector: Include target identifier in protocol logging
joepeck@webkit.org [Wed, 14 Nov 2018 21:59:06 +0000 (21:59 +0000)]
Web Inspector: Include target identifier in protocol logging
https://bugs.webkit.org/show_bug.cgi?id=191478

Reviewed by Devin Rousso.

* UserInterface/Protocol/Connection.js:
(InspectorBackend.Connection.prototype._dispatchResponse):
(InspectorBackend.Connection.prototype._dispatchEvent):
(InspectorBackend.Connection.prototype._sendMessageToBackend):
Call tracer log methods with the connection.

* UserInterface/Debug/CapturingProtocolTracer.js:
* UserInterface/Protocol/ProtocolTracer.js:
* UserInterface/Protocol/LoggingProtocolTracer.js:
(WI.LoggingProtocolTracer):
(WI.LoggingProtocolTracer.prototype.logFrontendException):
(WI.LoggingProtocolTracer.prototype.logFrontendRequest):
(WI.LoggingProtocolTracer.prototype.logWillHandleResponse):
(WI.LoggingProtocolTracer.prototype.logDidHandleResponse):
(WI.LoggingProtocolTracer.prototype.logDidHandleEvent):
Include the connection in log methods.

(WI.LoggingProtocolTracer.prototype.logProtocolError):
Removed, this is never called.

(WI.LoggingProtocolTracer.prototype._processEntry):
Include the target identifier in entry messages.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238197 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[Win] Compile Service Worker support
stephan.szabo@sony.com [Wed, 14 Nov 2018 21:54:26 +0000 (21:54 +0000)]
[Win] Compile Service Worker support
https://bugs.webkit.org/show_bug.cgi?id=191409

Reviewed by Youenn Fablet.

Fix compilation errors when ENABLE(SERVICE_WORKER)
on Windows with clang-cl. Clang on dllexport
platforms does not support specifying the
dllexport on both a class and members of the class
and unistd.h isn't provided but also appeared to
not be used.

No new tests, should be covered by existing tests.

* workers/service/ServiceWorkerProvider.h:
* workers/service/context/SWContextManager.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238196 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoUnreviewed attempted wincairo build fix after 238192.
joepeck@webkit.org [Wed, 14 Nov 2018 21:41:13 +0000 (21:41 +0000)]
Unreviewed attempted wincairo build fix after 238192.

* WebProcess/WebPage/WebPageInspectorTargetController.h:
* WebProcess/WebPage/WebPageInspectorTargetFrontendChannel.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238195 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoUnreviewed test gardening, move a skip expectation to a more specific file.
ryanhaddad@apple.com [Wed, 14 Nov 2018 21:37:08 +0000 (21:37 +0000)]
Unreviewed test gardening, move a skip expectation to a more specific file.

* TestExpectations:
* platform/mac-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238194 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoWeb Inspector: Follow-up: Keep Disabling Process Swap for Local Web Inspector
joepeck@webkit.org [Wed, 14 Nov 2018 21:03:53 +0000 (21:03 +0000)]
Web Inspector: Follow-up: Keep Disabling Process Swap for Local Web Inspector
https://bugs.webkit.org/show_bug.cgi?id=191494
<rdar://problem/45469854>

Reviewed by Devin Rousso.

Continue to have the Local Inspector disable PSON while open
until support is added.

* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::resetState):
* UIProcess/WebPageProxy.h:
(WebKit::WebPageProxy::hasLocalInspectorFrontend const):
(WebKit::WebPageProxy::setHasLocalInspectorFrontend):
* UIProcess/WebPageProxy.messages.in:
* UIProcess/WebProcessPool.cpp:
(WebKit::WebProcessPool::processForNavigationInternal):
* WebProcess/WebPage/WebInspector.cpp:
(WebKit::WebInspector::openFrontendConnection):
(WebKit::WebInspector::closeFrontendConnection):
* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::setHasLocalInspectorFrontend):
* WebProcess/WebPage/WebPage.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238193 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoWeb Inspector: Keep Web Inspector window alive across process swaps (PSON) (Remote...
joepeck@webkit.org [Wed, 14 Nov 2018 21:03:49 +0000 (21:03 +0000)]
Web Inspector: Keep Web Inspector window alive across process swaps (PSON) (Remote Inspector)
https://bugs.webkit.org/show_bug.cgi?id=191494
<rdar://problem/45469854>

Reviewed by Devin Rousso.

Source/JavaScriptCore:

* CMakeLists.txt:
* DerivedSources.make:
* JavaScriptCore.xcodeproj/project.pbxproj:
* Sources.txt:
New domain and resources.

* inspector/protocol/Target.json: Added.
New protocol domain, modeled after Worker.json, to allow for
multiplexing between different targets.

* inspector/InspectorTarget.h:
Each target will instantiate an InspectorTarget and must
provide an identifier, type, and means of connecting/disconnecting
to a frontend channel.

* inspector/agents/InspectorTargetAgent.cpp: Added.
(Inspector::InspectorTargetAgent::InspectorTargetAgent):
(Inspector::InspectorTargetAgent::didCreateFrontendAndBackend):
(Inspector::InspectorTargetAgent::willDestroyFrontendAndBackend):
(Inspector::InspectorTargetAgent::exists):
(Inspector::InspectorTargetAgent::initialized):
(Inspector::InspectorTargetAgent::sendMessageToTarget):
(Inspector::InspectorTargetAgent::sendMessageFromTargetToFrontend):
(Inspector::targetTypeToProtocolType):
(Inspector::buildTargetInfoObject):
(Inspector::InspectorTargetAgent::targetCreated):
(Inspector::InspectorTargetAgent::targetTerminated):
(Inspector::InspectorTargetAgent::connectToTargets):
(Inspector::InspectorTargetAgent::disconnectFromTargets):
* inspector/agents/InspectorTargetAgent.h: Added.
TargetAgent holds a list of targets, and connects/disconnects to each
of the targets when a frontend connects/disconnects.

* inspector/scripts/codegen/generator.py:
Better enum casing of ServiceWorker.

Source/WebCore:

* inspector/InspectorClient.h:
(WebCore::InspectorClient::allowRemoteInspectionToPageDirectly const):
Provide a hook so that a client may wish to allow direct remote inspection of the Page.
This is used by WebKitLegacy only.

* page/Page.cpp:
(Page::Page):
Only enable the PageDebuggable if the client wishes remote inspection of the Page directly.
This is used by WebKitLegacy only.

* inspector/InspectorController.cpp:
(WebCore::InspectorController::connectFrontend):
* inspector/InspectorController.h:
* page/PageDebuggable.cpp:
(WebCore::PageDebuggable::connect):
(WebCore::PageDebuggable::disconnect):
* page/PageDebuggable.h:
When a frontend connects, always enable the developer extras for the Page.
This is pretty much only for the remote path, which allows inspection if developer
extras were not already enabled (iOS). This simplifies the logic, and toggling
developer extras after it was already enabled is not really important.

Source/WebInspectorUI:

This starts introducing multi-target support into the Web Inspector frontend.
Specifically a backend connection that is persistent, but has the ability to
connect to and transition between Page targets received through that backend
connection.

This patch introduces the concept of a "Backend Target" which is the single
connection that the frontend contains to a backend. The old way of connecting
directly to a target is still supported. In that case the frontend constructs
a DirectBackendTarget for the debuggable type the frontend was spawns for.
However, if the frontend opens and has a TargetAgent, then it is likely
connected to a multi-target supporting backend and instead constructs a
MultiplexingBackendTarget, and will receive further information about
sub-targets to connect to. The only sub-target at the moment is a Page
sub-target.

As part of bringing up multi-target support this adds a few measures to
handle situations where the frontend is playing fast and loose with agents.
When the frontend does `FooAgent.method` it intends that to be performed
on the "main" target being debugged. Likewise when the frontend loops
over targets it expects them to be the debuggable targets. This patch
profiles a new implementation of `WI.mainTarget` and `WI.targets` to
match the assumptions being made by the frontend.

  - In a direct-target world, there is a single target which
    should be used for global agents and in WI.targets.

  - In a multi-target world, the page target is the one that
    should be used for global agents and WI.targets is the list
    of sub-targets (excluding the MultiplexingBackendTarget).

In a multi-target world, there are now commonly two Targets. The
MultiplexingBackendTarget and a PageTarget sub-target. In the future
this may include more targets, such as ServiceWorkers, DedicatedWorkers,
and perhaps even frames.

In a multi-target world, the frontend is immediately told about targets
as soon as it opens (via Target.targetCreated). In order to support
this, frontend initialization happens without a main target being available.
So this makes a few small changes to frontend initialization to perform
a bit of work once we know what the main target is.

During a page transition the frontend is told to destroy existing targets
and is soon after told about any new page targets (via Target.targetDestroyed
and Target.targetCreated). The frontend special cases this page transition.
It expects only one Page target to be alive at any time, accessible through
WI.pageTarget. When a page transition happens the WI.pageTarget changes, and
the frontend performs a bit of work to prepare the UI to handle the transition:
`<Manager>.transitionPageTarget` / WI.Notification.TransitionPageTarget.
For the most part the UI behaves fine as long once there are main frame
change and main resource change events, but those other events allow
the frontend to respond to the specific page transition cases.

* UserInterface/Base/Main.js:
(WI.loaded):
(WI.initializeBackendTarget):
(WI.initializePageTarget):
(WI.transitionPageTarget):
(WI.terminatePageTarget):
(WI.resetMainExecutionContext):
(WI.redirectGlobalAgentsToConnection):
(WI.contentLoaded):
New global functions for target initialization
and page transitioning.

* UserInterface/Test/Test.js:
(WI.loaded):
(WI.initializeBackendTarget):
(WI.initializePageTarget):
(WI.transitionPageTarget):
(WI.terminatePageTarget):
(WI.resetMainExecutionContext):
(WI.redirectGlobalAgentsToConnection):
New global functions for target initialization.
Tests continue to be a direct connection to the Page.

* UserInterface/Protocol/TargetObserver.js:
(WI.TargetObserver.prototype.targetCreated):
(WI.TargetObserver.prototype.targetDestroyed):
(WI.TargetObserver.prototype.dispatchMessageFromTarget):
New observer goes to the manager.

* UserInterface/Controllers/TargetManager.js:
(WI.TargetManager):
(WI.TargetManager.prototype.get targets):
(WI.TargetManager.prototype.get allTargets):
(WI.TargetManager.prototype.targetForIdentifier):
(WI.TargetManager.prototype.targetCreated):
(WI.TargetManager.prototype.targetDestroyed):
(WI.TargetManager.prototype.dispatchMessageFromTarget):
(WI.TargetManager.prototype.createMultiplexingBackendTarget):
(WI.TargetManager.prototype.createDirectBackendTarget):
(WI.TargetManager.prototype._addTarget):
(WI.TargetManager.prototype._removeTarget):
(WI.TargetManager.prototype._createTarget):
(WI.TargetManager.prototype._checkAndHandlePageTargetTransition):
(WI.TargetManager.prototype._checkAndHandlePageTargetTermination):
(WI.TargetManager.prototype.addTarget): Deleted.
(WI.TargetManager.prototype.removeTarget): Deleted.
(WI.TargetManager.prototype.initializeMainTarget): Deleted.
TargetManager is where we handle creating and destroying targets
and their connections. In order to simplify things a bit we make
`WI.targets`, which goes through `get targets()` an array instead
of a Set. And this includes only the sub-targets.

* UserInterface/Controllers/WorkerManager.js:
(WI.WorkerManager.prototype.workerCreated):
(WI.WorkerManager.prototype.workerTerminated):
Workers are still special-target-like things that multiplex through
WorkerAgent instead of TargetAgent. We'd like to promote these to
be full targets in the future.

* UserInterface/Protocol/DirectBackendTarget.js: Renamed from Source/WebInspectorUI/UserInterface/Protocol/MainTarget.js.
(WI.DirectBackendTarget):
(WI.DirectBackendTarget.connectionInfoForDebuggable):
(WI.DirectBackendTarget.prototype.get mainResource):
(WI.DirectBackendTarget.prototype.set mainResource):
This is the only "MainTarget" class. It is the backend target for a direct connection.

* UserInterface/Protocol/MultiplexingBackendTarget.js:
(WI.MultiplexingBackendTarget):
(WI.MultiplexingBackendTarget.prototype.initialize):
(WI.MultiplexingBackendTarget.prototype.get name):
(WI.MultiplexingBackendTarget.prototype.get executionContext):
(WI.MultiplexingBackendTarget.prototype.get mainResource):
This is the new backend target for a multi-target connection.
We don't expect it to be treated like other targets, so we don't
expect anyone to ask it for resources/executionContext/name info.

* UserInterface/Controllers/RuntimeManager.js:
(WI.TargetManager.prototype.evaluateInInspectedWindow):
This can be triggered by watch expressions before any target, and
therefore execution context, is available. Just return null, when
an execution context is available those clients will try again.

* UserInterface/Debug/Bootstrap.js:
Provide an WI.isEngineeringBuild boolean that can be used for various
debugging features.

* UserInterface/Main.html:
* UserInterface/Test.html:
New resources.

* UserInterface/Protocol/InspectorBackend.js:
(InspectorBackendClass.prototype.dispatch):
(InspectorBackendClass.prototype.runAfterPendingDispatches):
(InspectorBackend.Agent):
`InspectorBackend.mainConnection` was renamed `InspectorBackend.backendConnection`.

* UserInterface/Protocol/Connection.js:
(InspectorBackend.Connection):
(InspectorBackend.Connection.prototype._dispatchResponse):
(InspectorBackend.Connection.prototype._sendCommandToBackendWithCallback):
(InspectorBackend.Connection.prototype._sendCommandToBackendExpectingPromise):
(InspectorBackend.BackendConnection):
(InspectorBackend.BackendConnection.prototype.sendMessageToBackend):
(InspectorBackend.WorkerConnection):
(InspectorBackend.TargetConnection):
(InspectorBackend.TargetConnection.sendMessageToBackend):
Use a global sequence id to make filtering a bit easier in protocol tracing.
TargetConnection is identical to WorkerConnection except it uses TargetAgent
instead of WorkerAgent to perform multiplexing.

* UserInterface/Protocol/JavaScriptContextTarget.js:
(WI.JavaScriptContextTarget):
* UserInterface/Protocol/PageTarget.js:
(WI.PageTarget):
(WI.PageTarget.prototype.get displayName):
Specialized target types.

* UserInterface/Views/DebuggerSidebarPanel.js:
(WI.DebuggerSidebarPanel):
(WI.DebuggerSidebarPanel.prototype._targetAdded):
(WI.DebuggerSidebarPanel.prototype._targetRemoved):
(WI.DebuggerSidebarPanel.prototype._updateCallStackTreeOutline):
* UserInterface/Views/SourceCodeTextEditor.js:
(WI.SourceCodeTextEditor.prototype._targetAdded):
(WI.SourceCodeTextEditor.prototype._targetRemoved):
(WI.SourceCodeTextEditor.prototype._callFramesDidChange):
(WI.SourceCodeTextEditor.prototype._updateThreadIndicatorWidget):
(WI.SourceCodeTextEditor.prototype._reinsertAllThreadIndicators):
* UserInterface/Views/QuickConsole.js:
(WI.QuickConsole.prototype.initializeMainExecutionContextPathComponent):
(WI.QuickConsole.prototype._targetAdded):
(WI.QuickConsole.prototype._targetRemoved):
We make target added get called with all targets, including the
MultiplexingBackendTarget and PageTargets, both of which would
not have happened before. Before it was only WorkerTargets. Make
these sites a little more robust for the type of target they expect
to be able to handle.

* UserInterface/Base/Object.js:
* UserInterface/Controllers/DOMManager.js:
(WI.DOMManager):
(WI.DOMManager.prototype.transitionPageTarget):
(WI.DOMManager.prototype.requestDocument):
(WI.DOMManager.prototype._setDocument):
* UserInterface/Controllers/NetworkManager.js:
(WI.NetworkManager):
(WI.NetworkManager.prototype.transitionPageTarget):
(WI.NetworkManager.prototype.executionContextCreated):
(WI.NetworkManager.prototype._processMainFrameResourceTreePayload):
* UserInterface/Models/DefaultDashboard.js:
(WI.DefaultDashboard):
(WI.DefaultDashboard.prototype._mainResourceDidChange):
(WI.DefaultDashboard.prototype._transitionPageTarget):
* UserInterface/Views/NetworkTableContentView.js:
(WI.NetworkTableContentView):
(WI.NetworkTableContentView.prototype._mainResourceDidChange):
(WI.NetworkTableContentView.prototype._transitionPageTarget):
Special case handling when performing a page transition.

* UserInterface/Views/SettingsTabContentView.js:
* UserInterface/Debug/UncaughtExceptionReporter.js:
Document reloads are not supported right now.

Source/WebKit:

To support process swapping a slim Web Inspector backend lives in the UIProcess.
The Web Inspector frontend connects to it and is told about sub-targets, namely
pages, that it can further connect to. When performing a process swap the backend
tells the frontend to destroy existing targets and create new targets.

In the UIProcess the WebPageProxy has a WebPageInspectorController, with a single
TargetAgent holding InspectorTargetProxies to targets it knows about. Inspector
protocol messages go through this inspector controller and are routed to the
WebPage and its WebCore::Page's InspectorController. The WebPageProxy decides
when to close and expose new page targets during process swap, or basically
any time it reconnects to a WebProcess. So this patch also makes Web Inspector
stay alive and reconnect to a page when the inspected page crashes!

In the WebContentProcess the WebPage has a WebPageInspectorTarget. It also
has a WebPageInspectorTargetController in anticipation of further sub-targets
within the page (workers, frames) but none exist at the moment. The WebPage
relies on the WebPageProxy to know when to expose this target as a debuggable.

* Sources.txt:
* WebKit.xcodeproj/project.pbxproj:
New files.

* Shared/WebPageCreationParameters.cpp:
(WebKit::WebPageCreationParameters::encode const):
(WebKit::WebPageCreationParameters::decode):
* Shared/WebPageCreationParameters.h:
Remote inspector state can now stay in the UIProcess and does not need to
be passed down to the WebContentProcess.

* UIProcess/WebPageDebuggable.cpp: Copied from Source/WebCore/page/PageDebuggable.cpp.
(WebKit::WebPageDebuggable::WebPageDebuggable):
(WebKit::WebPageDebuggable::name const):
(WebKit::WebPageDebuggable::url const):
(WebKit::WebPageDebuggable::hasLocalDebugger const):
(WebKit::WebPageDebuggable::connect):
(WebKit::WebPageDebuggable::disconnect):
(WebKit::WebPageDebuggable::dispatchMessageFromRemote):
(WebKit::WebPageDebuggable::setIndicating):
(WebKit::WebPageDebuggable::setNameOverride):
* UIProcess/WebPageDebuggable.h: Copied from Source/WebCore/page/PageDebuggable.h.
Remote debuggable entry point into the UIProcess for a page.
This is pretty much identical to the PageDebuggable in WebCore.

* Scripts/webkit/messages.py:
* UIProcess/WebPageProxy.messages.in:
* UIProcess/WebPageProxy.cpp:
(WebKit::m_resetRecentCrashCountTimer):
(WebKit::WebPageProxy::finishAttachingToWebProcess):
(WebKit::WebPageProxy::close):
(WebKit::WebPageProxy::createInspectorTarget):
(WebKit::WebPageProxy::destroyInspectorTarget):
(WebKit::WebPageProxy::sendMessageToInspectorFrontend):
(WebKit::WebPageProxy::setIndicating):
(WebKit::WebPageProxy::allowsRemoteInspection const):
(WebKit::WebPageProxy::setAllowsRemoteInspection):
(WebKit::WebPageProxy::remoteInspectionNameOverride const):
(WebKit::WebPageProxy::setRemoteInspectionNameOverride):
(WebKit::WebPageProxy::remoteInspectorInformationDidChange):
(WebKit::WebPageProxy::clearInspectorTargets):
(WebKit::WebPageProxy::createInspectorTargets):
(WebKit::WebPageProxy::didCommitLoadForFrame):
(WebKit::WebPageProxy::didReceiveTitleForFrame):
(WebKit::WebPageProxy::creationParameters):
* UIProcess/WebPageProxy.h:
(WebKit::WebPageProxy::inspectorController):
(WebKit::WebPageProxy::allowsRemoteInspection const): Deleted.
(WebKit::WebPageProxy::remoteInspectionNameOverride const): Deleted.
Own more inspector state in the UIProcess including a debuggable and inspector controller.

* UIProcess/WebPageInspectorController.h: Added.
* UIProcess/WebPageInspectorController.cpp: Added.
(WebKit::WebPageInspectorController::WebPageInspectorController):
(WebKit::WebPageInspectorController::pageClosed):
(WebKit::WebPageInspectorController::hasLocalFrontend const):
(WebKit::WebPageInspectorController::hasRemoteFrontend const):
(WebKit::WebPageInspectorController::connectFrontend):
(WebKit::WebPageInspectorController::disconnectFrontend):
(WebKit::WebPageInspectorController::disconnectAllFrontends):
(WebKit::WebPageInspectorController::dispatchMessageFromFrontend):
(WebKit::WebPageInspectorController::setIndicating):
(WebKit::WebPageInspectorController::clearTargets):
(WebKit::WebPageInspectorController::createInspectorTarget):
(WebKit::WebPageInspectorController::destroyInspectorTarget):
(WebKit::WebPageInspectorController::sendMessageToInspectorFrontend):
InspectorController with a single TargetAgent in the UIProcess.

* UIProcess/WebPageInspectorTargetAgent.h:
* UIProcess/WebPageInspectorTargetAgent.cpp:
(WebKit::WebPageInspectorTargetAgent::WebPageInspectorTargetAgent):
(WebKit::WebPageInspectorTargetAgent::frontendChannel):
Target agent implementation.

* UIProcess/InspectorTargetProxy.cpp:
(WebKit::InspectorTargetProxy::create):
(WebKit::InspectorTargetProxy::InspectorTargetProxy):
(WebKit::InspectorTargetProxy::connect):
(WebKit::InspectorTargetProxy::disconnect):
(WebKit::InspectorTargetProxy::sendMessageToTargetBackend):
* UIProcess/InspectorTargetProxy.h:
UIProcess proxy for an InspectorTarget in the WebContentProcess.

* UIProcess/WebProcessPool.cpp:
(WebKit::WebProcessPool::processForNavigationInternal):

* WebProcess/WebPage/WebPage.cpp:
(WebKit::m_shouldAttachDrawingAreaOnPageTransition):
(WebKit::WebPage::connectInspector):
(WebKit::WebPage::disconnectInspector):
(WebKit::WebPage::sendMessageToTargetBackend):
(WebKit::WebPage::setIndicating):
(WebKit::WebPage::setAllowsRemoteInspection): Deleted.
(WebKit::WebPage::setRemoteInspectionNameOverride): Deleted.
* WebProcess/WebPage/WebPage.h:
* WebProcess/WebPage/WebPage.messages.in:

* WebProcess/WebPage/WebPageInspectorTarget.h:
* WebProcess/WebPage/WebPageInspectorTarget.cpp:
(WebKit::WebPageInspectorTarget::WebPageInspectorTarget):
(WebKit::WebPageInspectorTarget::identifier const):
(WebKit::WebPageInspectorTarget::connect):
(WebKit::WebPageInspectorTarget::disconnect):
(WebKit::WebPageInspectorTarget::sendMessageToTargetBackend):
InspectorTarget for this WebPage.

* WebProcess/WebPage/WebPageInspectorTargetController.cpp: Added.
(WebKit::WebPageInspectorTargetController::WebPageInspectorTargetController):
(WebKit::WebPageInspectorTargetController::~WebPageInspectorTargetController):
(WebKit::WebPageInspectorTargetController::addTarget):
(WebKit::WebPageInspectorTargetController::removeTarget):
(WebKit::WebPageInspectorTargetController::connectInspector):
(WebKit::WebPageInspectorTargetController::disconnectInspector):
(WebKit::WebPageInspectorTargetController::sendMessageToTargetBackend):
(WebKit::WebPageInspectorTargetController::sendMessageToTargetFrontend):
* WebProcess/WebPage/WebPageInspectorTargetController.h:
* WebProcess/WebPage/WebPageInspectorTargetFrontendChannel.h:
* WebProcess/WebPage/WebPageInspectorTargetFrontendChannel.cpp:
(WebKit::WebPageInspectorTargetFrontendChannel::create):
(WebKit::WebPageInspectorTargetFrontendChannel::WebPageInspectorTargetFrontendChannel):
(WebKit::WebPageInspectorTargetFrontendChannel::sendMessageToFrontend):
Preparation for more target managment in the WebContentProcess.

Source/WebKitLegacy/mac:

* WebCoreSupport/WebInspectorClient.h:
WebKitLegacy will still have remote inspection of the Page directly.

LayoutTests:

* inspector/unit-tests/target-manager.html:
WI.targets has switched to being an array instead of a set.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238192 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[ Sierra Release WK2 ] Layout Test fast/workers/worker-cloneport.html is flaky.
tsavell@apple.com [Wed, 14 Nov 2018 20:00:09 +0000 (20:00 +0000)]
[ Sierra Release WK2 ] Layout Test fast/workers/worker-cloneport.html is flaky.
https://bugs.webkit.org/show_bug.cgi?id=191644.

Unreviewed test Gardening.

* platform/mac-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238191 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoAdd SPI to show a safe browsing warning
achristensen@apple.com [Wed, 14 Nov 2018 19:47:00 +0000 (19:47 +0000)]
Add SPI to show a safe browsing warning
https://bugs.webkit.org/show_bug.cgi?id=191604

Reviewed by Andy Estes.

Source/WebKit:

Safari has a case where it shows safe browsing warnings for a reason other than being told to by the SafeBrowsing framework.
Because safe browsing is moving to WebKit, we will have Safari call this WebKit SPI in that case.
The warning looks a little different, so I had to change the structure of a SafeBrowsingResult and the location of the string processing,
but the existing behavior is the same as verified by existing unit tests.  I added a unit test to verify behavior of the new SPI.

* SourcesCocoa.txt:
* UIProcess/API/Cocoa/WKWebView.mm:
(-[WKWebView _showSafeBrowsingWarningWithTitle:warning:details:completionHandler:]):
(-[WKWebView _showSafeBrowsingWarning:completionHandler:]):
(+[WKWebView _confirmMalwareSentinel]):
(+[WKWebView _visitUnsafeWebsiteSentinel]):
* UIProcess/API/Cocoa/WKWebViewInternal.h:
* UIProcess/API/Cocoa/WKWebViewPrivate.h:
* UIProcess/Cocoa/SafeBrowsingResultCocoa.mm: Removed.
* UIProcess/Cocoa/SafeBrowsingWarningCocoa.mm: Copied from Source/WebKit/UIProcess/Cocoa/SafeBrowsingResultCocoa.mm.
(WebKit::malwareDetailsBase):
(WebKit::learnMoreURL):
(WebKit::reportAnErrorBase):
(WebKit::localizedProvider):
(WebKit::SafeBrowsingWarning::visitUnsafeWebsiteSentinel):
(WebKit::SafeBrowsingWarning::confirmMalwareSentinel):
(WebKit::replace):
(WebKit::addLinkAndReplace):
(WebKit::reportAnErrorURL):
(WebKit::malwareDetailsURL):
(WebKit::safeBrowsingTitleText):
(WebKit::safeBrowsingWarningText):
(WebKit::safeBrowsingDetailsText):
(WebKit::SafeBrowsingWarning::SafeBrowsingWarning):
(WebKit::learnMore): Deleted.
(WebKit::SafeBrowsingResult::SafeBrowsingResult): Deleted.
* UIProcess/Cocoa/WKSafeBrowsingWarning.h:
* UIProcess/Cocoa/WKSafeBrowsingWarning.mm:
(-[WKSafeBrowsingWarning initWithFrame:safeBrowsingWarning:completionHandler:]):
(-[WKSafeBrowsingWarning addContent]):
(-[WKSafeBrowsingWarning showDetailsClicked]):
(-[WKSafeBrowsingWarning clickedOnLink:]):
(-[WKSafeBrowsingTextView initWithAttributedString:forWarning:]):
(confirmMalwareSentinel): Deleted.
(visitUnsafeWebsiteSentinel): Deleted.
(replace): Deleted.
(addLinkAndReplace): Deleted.
(reportAnErrorURL): Deleted.
(malwareDetailsURL): Deleted.
(titleText): Deleted.
(warningText): Deleted.
(detailsText): Deleted.
(-[WKSafeBrowsingWarning initWithFrame:safeBrowsingResult:completionHandler:]): Deleted.
* UIProcess/Cocoa/WebPageProxyCocoa.mm:
(WebKit::WebPageProxy::beginSafeBrowsingCheck):
* UIProcess/Cocoa/WebViewImpl.h:
* UIProcess/Cocoa/WebViewImpl.mm:
(WebKit::WebViewImpl::showSafeBrowsingWarning):
* UIProcess/PageClient.h:
(WebKit::PageClient::showSafeBrowsingWarning):
* UIProcess/SafeBrowsingResult.h: Removed.
* UIProcess/SafeBrowsingWarning.h: Copied from Source/WebKit/UIProcess/SafeBrowsingResult.h.
(WebKit::SafeBrowsingWarning::create):
(WebKit::SafeBrowsingWarning::title const):
(WebKit::SafeBrowsingWarning::warning const):
(WebKit::SafeBrowsingWarning::details const):
(WebKit::SafeBrowsingResult::create): Deleted.
(WebKit::SafeBrowsingResult::isPhishing const): Deleted.
(WebKit::SafeBrowsingResult::isMalware const): Deleted.
(WebKit::SafeBrowsingResult::isUnwantedSoftware const): Deleted.
(WebKit::SafeBrowsingResult::url const): Deleted.
(WebKit::SafeBrowsingResult::provider const): Deleted.
(WebKit::SafeBrowsingResult::localizedProviderName const): Deleted.
(WebKit::SafeBrowsingResult::malwareDetailsURLBase const): Deleted.
(WebKit::SafeBrowsingResult::reportAnErrorURLBase const): Deleted.
(WebKit::SafeBrowsingResult::learnMoreURL const): Deleted.
(WebKit::SafeBrowsingResult::needsSafeBrowsingWarning const): Deleted.
(): Deleted.
* UIProcess/WebFramePolicyListenerProxy.cpp:
(WebKit::WebFramePolicyListenerProxy::didReceiveSafeBrowsingResults):
(WebKit::WebFramePolicyListenerProxy::use):
* UIProcess/WebFramePolicyListenerProxy.h:
* UIProcess/WebFrameProxy.cpp:
(WebKit::WebFrameProxy::WebFrameProxy):
(WebKit::WebFrameProxy::setUpPolicyListenerProxy):
(WebKit::WebFrameProxy::didHandleContentFilterUnblockNavigation):
* UIProcess/WebFrameProxy.h:
(WebKit::WebFrameProxy::create):
(WebKit::WebFrameProxy::page const):
* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::swapToWebProcess):
(WebKit::WebPageProxy::didCreateMainFrame):
(WebKit::WebPageProxy::didCreateSubframe):
(WebKit::WebPageProxy::decidePolicyForNavigationAction):
(WebKit::WebPageProxy::decidePolicyForNewWindowAction):
(WebKit::WebPageProxy::decidePolicyForResponse):
* UIProcess/ios/PageClientImplIOS.h:
* UIProcess/ios/PageClientImplIOS.mm:
(WebKit::PageClientImpl::showSafeBrowsingWarning):
* UIProcess/mac/PageClientImplMac.h:
* UIProcess/mac/PageClientImplMac.mm:
(WebKit::PageClientImpl::showSafeBrowsingWarning):
* UIProcess/mac/WebDataListSuggestionsDropdownMac.mm:
* WebKit.xcodeproj/project.pbxproj:

Tools:

* TestWebKitAPI/Tests/WebKitCocoa/SafeBrowsing.mm:
(TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238190 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoREGRESSION(r238178): fast/forms/access-key-mutated.html and fast/forms/access-key...
antti@apple.com [Wed, 14 Nov 2018 19:46:15 +0000 (19:46 +0000)]
REGRESSION(r238178): fast/forms/access-key-mutated.html and fast/forms/access-key-case-insensitive.html are timing out
https://bugs.webkit.org/show_bug.cgi?id=191642

* TestExpectations: Additionally skip accessibility/mac/selection-notification-focus-change.html

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238189 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[iOS] A few drag and drop tests are crashing after r238146
wenson_hsieh@apple.com [Wed, 14 Nov 2018 19:27:29 +0000 (19:27 +0000)]
[iOS] A few drag and drop tests are crashing after r238146
https://bugs.webkit.org/show_bug.cgi?id=191617

Reviewed by Dean Jackson.

Source/WebKit:

The notion of temporarily suppressing the selection assistant was introduced during iOS drag and drop
development as a way of allowing both the selection view and dropped content snapshot to fade in simultaneously
during a drop in an editable element. r238146 piggy-backs on this mechanism by changing selection suppression
state when an element is focused, when the selection changes, and when an element is blurred, depending on
whether the currently focused element is transparent.

However, in the case where the selection assistant is suppressed due to a running drop animation, if focus moves
to an element that is not fully transparent, we end up prematurely unsuppressing the text selection assistant.
This subsequently causes selection UI to immediately show up after a drop instead of animating in alongside a
snapshot of the inserted document fragment, if the drop moved focus to an editable element.

A number of drag and drop tests on iOS exercised this codepath by dragging content into editable fields and/or
moving content between editable elements in a web view. These tests began to crash due to selection views and
the accompanying callout bar appearing earlier than usual, which triggers an unrelated UIKit assertion in
<https://webkit.org/b/190401>.

This patch fixes the failing tests by refactoring our selection assistant suppression code. Instead of
maintaining a single `BOOL` flag indicating whether the selection is suppressed, we use an `OptionSet` of
`SuppressSelectionAssistantReason`s, which (at the moment) only include (1) a running drop animation, and (2)
focusing a transparent element. The text selection assistant is considered suppressed when either of the reasons
apply. This allows us to correctly handle a drop animation that occurs simultaneously as an element is focused
without unsuppressing the selection assistant early, and also allows us to handle selection assistant
suppression in more nuanced ways, depending on the suppression reason.

* UIProcess/ios/WKContentViewInteraction.h:
* UIProcess/ios/WKContentViewInteraction.mm:
(-[WKContentView cleanupInteraction]):
(-[WKContentView _displayFormNodeInputView]):

Only prevent zooming to the focused element during drop if we're suppressing the selection assistant due to
focusing a transparent element. In the case of a drop, we still want to allow scrolling and zooming.

(-[WKContentView canShowNonEmptySelectionView]):
(-[WKContentView hasSelectablePositionAtPoint:]):
(-[WKContentView pointIsNearMarkedText:]):
(-[WKContentView textInteractionGesture:shouldBeginAtPoint:]):
(-[WKContentView _startAssistingKeyboard]):
(-[WKContentView _startAssistingNode:userIsInteracting:blurPreviousNode:changingActivityState:userObject:]):
(-[WKContentView _stopAssistingNode]):
(-[WKContentView _updateChangedSelection:]):
(-[WKContentView _shouldSuppressSelectionCommands]):
(-[WKContentView _beginSuppressingSelectionAssistantForReason:]):
(-[WKContentView _stopSuppressingSelectionAssistantForReason:]):

Add helper methods for adding or removing selection assistant suppression reasons. When the last selection
assistant suppression reason is removed, we activate the selection assistant, and conversely, when the first
suppression reason is added, we deactivate the selection assistant.

(-[WKContentView _didConcludeEditDataInteraction:]):
(-[WKContentView _didPerformDragOperation:]):
(-[WKContentView dropInteraction:performDrop:]):
(-[WKContentView suppressAssistantSelectionView]): Deleted.
(-[WKContentView setSuppressAssistantSelectionView:]): Deleted.

Tools:

Augment these crashing tests to verify that selection commands are suppressed during drop over editable elements
via more robust means. These tests currently hit an assertion when revealing the callout bar too early, because
TestWebKitAPI is not a UI application (see <https://webkit.org/b/190401>).

Instead of relying on this other bug, directly ask the text input whether it is suppressing selection commands
during a drop, and remember the answer via DragAndDropSimulator.

* TestWebKitAPI/Tests/ios/DragAndDropTestsIOS.mm:
(TestWebKitAPI::TEST):

Add to the existing tests that started failing after r238146.

* TestWebKitAPI/cocoa/DragAndDropSimulator.h:
* TestWebKitAPI/cocoa/TestWKWebView.h:
* TestWebKitAPI/cocoa/TestWKWebView.mm:
(-[TestWKWebView textInputContentView]):
* TestWebKitAPI/ios/DragAndDropSimulatorIOS.mm:
(-[DragAndDropSimulator _resetSimulatedState]):
(-[DragAndDropSimulator _webView:dataInteractionOperationWasHandled:forSession:itemProviders:]):
* TestWebKitAPI/ios/UIKitSPI.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238188 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoREGRESSION(r238178): fast/forms/access-key-mutated.html and fast/forms/access-key...
antti@apple.com [Wed, 14 Nov 2018 19:18:35 +0000 (19:18 +0000)]
REGRESSION(r238178): fast/forms/access-key-mutated.html and fast/forms/access-key-case-insensitive.html are timing out
https://bugs.webkit.org/show_bug.cgi?id=191642

* TestExpectations: Skipping for now.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238187 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoSource/WebKit:
wenson_hsieh@apple.com [Wed, 14 Nov 2018 19:00:04 +0000 (19:00 +0000)]
Source/WebKit:
[Cocoa] [WebKit2] Hook up some more editing actions (-_pasteAndMatchStyle:, -makeTextWritingDirectionNatural:)
https://bugs.webkit.org/show_bug.cgi?id=191605
<rdar://problem/45813991>

Reviewed by Dean Jackson.

Hook up a couple of editing actions on iOS and macOS. The `-pasteAndMatchStyle:` selector is privately declared
in UIKit, but does not exist on macOS; as such, add `-_pasteAndMatchStyle:` as SPI on WKWebView, and
additionally implement the unprefixed version, `-pasteAndMatchStyle:`, on iOS.

Since `-makeTextWritingDirectionNatural:` is already declared on both iOS and macOS (SPI on UIResponder and API
on macOS, respectively) we simply implement the unprefixed variant on both platforms.

* Platform/spi/ios/UIKitSPI.h:
* UIProcess/API/Cocoa/WKWebView.mm:
* UIProcess/API/Cocoa/WKWebViewPrivate.h:
* UIProcess/ios/WKContentViewInteraction.h:
* UIProcess/ios/WKContentViewInteraction.mm:
(-[WKContentView makeTextWritingDirectionNaturalForWebView:]):
(-[WKContentView canPerformActionForWebView:withSender:]):

Tools:
[Cocoa] [WebKit2] Hook up some more editing actions (-pasteAndMatchStyle:, -makeTextWritingDirectionNatural:)
https://bugs.webkit.org/show_bug.cgi?id=191605
<rdar://problem/45813991>

Reviewed by Dean Jackson.

Adds two new API tests to exercise -_pasteAndMatchStyle: and -makeTextWritingDirectionNatural:.

* TestWebKitAPI/Tests/WebKitCocoa/WKAttachmentTests.mm:
(-[TestWKWebView _synchronouslyExecuteEditCommand:argument:]): Deleted.
* TestWebKitAPI/Tests/WebKitCocoa/WKWebViewEditActions.mm:
(TestWebKitAPI::TEST):
* TestWebKitAPI/cocoa/TestWKWebView.h:
* TestWebKitAPI/cocoa/TestWKWebView.mm:

Pull `-_synchronouslyExecuteEditCommand:argument:` out of WKAttachmentTests and into shared code, in a new
category on WKWebView declared in TestWKWebView. Also, fix an unrelated typo in a nearby category.

(-[WKWebView _synchronouslyExecuteEditCommand:argument:]):
* TestWebKitAPI/ios/UIKitSPI.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238186 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoUnreviewed, rolling in CodeCache in r237254
yusukesuzuki@slowstart.org [Wed, 14 Nov 2018 18:49:22 +0000 (18:49 +0000)]
Unreviewed, rolling in CodeCache in r237254
https://bugs.webkit.org/show_bug.cgi?id=190340

Land the CodeCache part without adding an additional hash value.

* bytecode/UnlinkedFunctionExecutable.cpp:
(JSC::UnlinkedFunctionExecutable::fromGlobalCode):
* bytecode/UnlinkedFunctionExecutable.h:
* parser/SourceCodeKey.h:
(JSC::SourceCodeKey::SourceCodeKey):
(JSC::SourceCodeKey::operator== const):
* runtime/CodeCache.cpp:
(JSC::CodeCache::getUnlinkedGlobalCodeBlock):
(JSC::CodeCache::getUnlinkedGlobalFunctionExecutable):
* runtime/CodeCache.h:
* runtime/FunctionConstructor.cpp:
(JSC::constructFunctionSkippingEvalEnabledCheck):
* runtime/FunctionExecutable.cpp:
(JSC::FunctionExecutable::fromGlobalCode):
* runtime/FunctionExecutable.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238185 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoREGRESSION (WEBPROCESS_WINDOWSERVER_BLOCKING): requestAnimationFrame Stops Completing
pvollan@apple.com [Wed, 14 Nov 2018 18:43:21 +0000 (18:43 +0000)]
REGRESSION (WEBPROCESS_WINDOWSERVER_BLOCKING): requestAnimationFrame Stops Completing
https://bugs.webkit.org/show_bug.cgi?id=190884

Reviewed by Dean Jackson.

Source/WebCore:

Only notify display refresh monitors with matching display ID.

Test: fast/animation/request-animation-frame-in-two-pages.html

* platform/graphics/DisplayRefreshMonitorManager.cpp:
(WebCore::DisplayRefreshMonitorManager::displayWasUpdated):
* platform/graphics/DisplayRefreshMonitorManager.h:

Source/WebKit:

Since the Web page currently owns the display link on the UI process side, the display link will be
deleted in the UI process when a Web page is closed. This is incorrect, since a display link can be
reused between Web pages in the WebContent process. This patch moves the ownership of the UI process
display links to the Web process object to address this issue. Also, the display ID of the display
being updated is included in the message to the WebContent process.

* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::close):
(WebKit::WebPageProxy::resetStateAfterProcessExited):
* UIProcess/WebPageProxy.h:
* UIProcess/WebPageProxy.messages.in:
* UIProcess/WebProcessProxy.h:
* UIProcess/WebProcessProxy.messages.in:
* UIProcess/mac/DisplayLink.cpp:
(WebKit::DisplayLink::DisplayLink):
(WebKit::DisplayLink::displayLinkCallback):
* UIProcess/mac/DisplayLink.h:
(WebKit::DisplayLink::displayID const):
* UIProcess/mac/WebPageProxyMac.mm:
(WebKit::WebPageProxy::startDisplayLink): Deleted.
(WebKit::WebPageProxy::stopDisplayLink): Deleted.
* UIProcess/mac/WebProcessProxyMac.mm:
(WebKit::WebProcessProxy::startDisplayLink):
(WebKit::WebProcessProxy::stopDisplayLink):
* WebProcess/WebPage/DrawingArea.h:
* WebProcess/WebPage/DrawingArea.messages.in:
* WebProcess/WebPage/mac/DrawingAreaMac.cpp:
(WebKit::DisplayRefreshMonitorMac::create):
(WebKit::DisplayRefreshMonitorMac::DisplayRefreshMonitorMac):
(WebKit::DisplayRefreshMonitorMac::~DisplayRefreshMonitorMac):
(WebKit::DisplayRefreshMonitorMac::requestRefreshCallback):
(WebKit::DrawingArea::createDisplayRefreshMonitor):
(WebKit::DrawingArea::displayWasRefreshed): Deleted.
* WebProcess/WebProcess.h:
* WebProcess/WebProcess.messages.in:
* WebProcess/cocoa/WebProcessCocoa.mm:
(WebKit::WebProcess::displayWasRefreshed):

LayoutTests:

* TestExpectations:
* fast/animation/request-animation-frame-in-two-pages-expected.txt: Added.
* fast/animation/request-animation-frame-in-two-pages.html: Added.
* platform/mac-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238184 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoClient should be able to disable PSON even if the experimental feature is on by default
cdumez@apple.com [Wed, 14 Nov 2018 18:13:13 +0000 (18:13 +0000)]
Client should be able to disable PSON even if the experimental feature is on by default
https://bugs.webkit.org/show_bug.cgi?id=191634

Reviewed by Dean Jackson.

Instead of overriding the APIProcessPoolConfiguration's m_processSwapsOnNavigation flag to true
when the experimental is enabled, use a separate flag to store this information. If the client
explicitly sets the APIProcessPoolConfiguration's m_processSwapsOnNavigation flag then we obey
the client's request, otherwise, we fall back to using the state from experimental features.

This allows:
- API tests to explicitely disable PSON / Process prewarming even if those are on by default
  in experimental features.
- If the client does not set those flags on the APIProcessPoolConfiguration (Safari for e.g.),
  then the experimental feature flag still fully controls the feature.

* UIProcess/API/APIProcessPoolConfiguration.cpp:
(API::ProcessPoolConfiguration::copy):
* UIProcess/API/APIProcessPoolConfiguration.h:
* UIProcess/WebProcessPool.cpp:
(WebKit::WebProcessPool::createWebPage):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238183 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoConvert libwebrtc error types to DOM exceptions
youenn@apple.com [Wed, 14 Nov 2018 17:53:21 +0000 (17:53 +0000)]
Convert libwebrtc error types to DOM exceptions
https://bugs.webkit.org/show_bug.cgi?id=191590

Reviewed by Alex Christensen.

LayoutTests/imported/w3c:

* web-platform-tests/webrtc/RTCConfiguration-rtcpMuxPolicy-expected.txt:
* web-platform-tests/webrtc/RTCPeerConnection-setLocalDescription-answer-expected.txt:
* web-platform-tests/webrtc/RTCRtpTransceiver.https-expected.txt:

Source/ThirdParty/libwebrtc:

* Configurations/libwebrtc.iOS.exp:
* Configurations/libwebrtc.iOSsim.exp:
* Configurations/libwebrtc.mac.exp:

Source/WebCore:

Make use of overloaded callback method that provides an error type.
This type is then used to create a DOM exception with the correct type.
Covered by existing tests.

* Modules/mediastream/libwebrtc/LibWebRTCMediaEndpoint.cpp:
(WebCore::LibWebRTCMediaEndpoint::doSetRemoteDescription):
(WebCore::LibWebRTCMediaEndpoint::createSessionDescriptionFailed):
(WebCore::LibWebRTCMediaEndpoint::setLocalSessionDescriptionFailed):
(WebCore::LibWebRTCMediaEndpoint::setRemoteSessionDescriptionFailed):
* Modules/mediastream/libwebrtc/LibWebRTCMediaEndpoint.h:
* Modules/mediastream/libwebrtc/LibWebRTCObservers.h:
(WebCore::toExceptionCode):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238182 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoAllow to remove MediaStreamPrivate observers when iterating over observers
youenn@apple.com [Wed, 14 Nov 2018 17:50:36 +0000 (17:50 +0000)]
Allow to remove MediaStreamPrivate observers when iterating over observers
https://bugs.webkit.org/show_bug.cgi?id=187256

Reviewed by Eric Carlson.

Migrate the observer list from a Vector to a HashSet.
This is more robust to multiple observing and keeping of order of observers is not required.
Copy the set of observers to a vector before iterating over it.
This allows to remove an observer while iterating, which is now used in UserMediaRequest.

Covered by existing tests.

* Modules/mediastream/UserMediaRequest.cpp:
(WebCore::UserMediaRequest::mediaStreamIsReady):
* platform/mediastream/MediaStreamPrivate.cpp:
(WebCore::MediaStreamPrivate::addObserver):
(WebCore::MediaStreamPrivate::removeObserver):
(WebCore::MediaStreamPrivate::forEachObserver const):
(WebCore::MediaStreamPrivate::updateActiveState):
(WebCore::MediaStreamPrivate::addTrack):
(WebCore::MediaStreamPrivate::removeTrack):
(WebCore::MediaStreamPrivate::characteristicsChanged):
* platform/mediastream/MediaStreamPrivate.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238181 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoCalling removeTrack on different RTCPeerConnection should throw InvalidAccessError
youenn@apple.com [Wed, 14 Nov 2018 17:35:02 +0000 (17:35 +0000)]
Calling removeTrack on different RTCPeerConnection should throw InvalidAccessError
https://bugs.webkit.org/show_bug.cgi?id=191603

Reviewed by Chris Dumez.

LayoutTests/imported/w3c:

* web-platform-tests/webrtc/RTCPeerConnection-removeTrack.https-expected.txt:

Source/WebCore:

Make sure to check that the sender peer connection backend is matching.
Covered by rebased WPT test.

* Modules/mediastream/RTCPeerConnection.cpp:
(WebCore::RTCPeerConnection::removeTrack):
* Modules/mediastream/RTCRtpSender.cpp:
(WebCore::RTCRtpSender::isCreatedBy const):
* Modules/mediastream/RTCRtpSender.h:

LayoutTests:

Removed obsolete test.

* fast/mediastream/RTCPeerConnection-add-removeTrack-expected.txt:
* fast/mediastream/RTCPeerConnection-add-removeTrack.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238180 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoWebKit.ApplicationManifestBasic API test is failing when enabling PSON
cdumez@apple.com [Wed, 14 Nov 2018 16:58:17 +0000 (16:58 +0000)]
WebKit.ApplicationManifestBasic API test is failing when enabling PSON
https://bugs.webkit.org/show_bug.cgi?id=191602

Reviewed by Alex Christensen.

Source/WebKit:

Add support for process swapping for a [WKWebView loadHTML:] load by storing
the necessary data on the API::Navigation and doing a loadData() instead of
a loadRequest() after process swapping when this data is present on the
navigation.

* UIProcess/API/APINavigation.cpp:
(API::Navigation::Navigation):
* UIProcess/API/APINavigation.h:
(API::Navigation::create):
(API::Navigation::substituteData const):
* UIProcess/WebNavigationState.cpp:
(WebKit::WebNavigationState::createLoadDataNavigation):
* UIProcess/WebNavigationState.h:
* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::loadData):
(WebKit::WebPageProxy::continueNavigationInNewProcess):

Tools:

Add API test coverage.

* TestWebKitAPI/Tests/WebKitCocoa/ProcessSwapOnNavigation.mm:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238179 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoAlign Mac WK2 layer flush throttling with iOS
antti@apple.com [Wed, 14 Nov 2018 16:53:25 +0000 (16:53 +0000)]
Align Mac WK2 layer flush throttling with iOS
https://bugs.webkit.org/show_bug.cgi?id=191632

Reviewed by Zalan Bujtas.

Currently Mac WK2 uses WebCore side throttling implementation in RenderLayerCompositor. This code has
throttling timer per-frame while the actual decision making and layer flushes itself are per page. These
timers generate way more flushes than expected in presence of multiple frames. There are also bugs in how
flushing state is updated when frames are created dynamically.

On iOS WK2 throttling is implemented on WebKit side and controlled by a per-page timer. Recent fixes also
make this implementation visually fast. We should align the Mac implementation and eventually unify them.

This patch implements throttling in TiledCoreAnimationDrawingArea mirroring the iOS RemoteLayerTreeDrawingArea
implementation. There are some adjustments for platform differences (local vs remote layers) and we continue
using runloop observer for the actual flushes. Timings are as in the existing Mac code.

The patch appears to be a significant performance progression.

* WebProcess/WebPage/mac/TiledCoreAnimationDrawingArea.h:
* WebProcess/WebPage/mac/TiledCoreAnimationDrawingArea.mm:
(WebKit::TiledCoreAnimationDrawingArea::TiledCoreAnimationDrawingArea):
(WebKit::TiledCoreAnimationDrawingArea::setLayerTreeStateIsFrozen):

Schedule an immediate flush when layers are unfrozen.

(WebKit::TiledCoreAnimationDrawingArea::scheduleCompositingLayerFlush):
(WebKit::TiledCoreAnimationDrawingArea::scheduleCompositingLayerFlushImmediately):

Track pending flushes with m_hasPendingFlush bit.
Delay flush if the flush throttling timer is active. Start it if we throttling but it is not active yet.

(WebKit::TiledCoreAnimationDrawingArea::flushLayers):

Clear m_hasPendingFlush if the flush succeeded.
Restart the throttling timer if we are still throttling.
Manage runloop observer invalidation here instead of the caller (and stop returning value).

(WebKit::TiledCoreAnimationDrawingArea::layerFlushRunLoopCallback):
(WebKit::TiledCoreAnimationDrawingArea::adjustLayerFlushThrottling):

Returning 'true' here disables WebCore side throttling code.

Start or stop the throttling timer on state changes.

(WebKit::TiledCoreAnimationDrawingArea::layerFlushThrottlingIsActive const):

This is used to control style and layout timers on WebCore side. Return false on frozen
state since unfreezing depends on style and layout.

(WebKit::TiledCoreAnimationDrawingArea::startLayerFlushThrottlingTimer):
(WebKit::TiledCoreAnimationDrawingArea::layerFlushThrottlingTimerFired):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238178 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agowebkitpy: Refactor port code for devices
jbedard@apple.com [Wed, 14 Nov 2018 16:24:59 +0000 (16:24 +0000)]
webkitpy: Refactor port code for devices
https://bugs.webkit.org/show_bug.cgi?id=191568
<rdar://problem/46005821>

Reviewed by Alex Christensen.

watchOS and iOS device ports share many characteristics in testing, mostly due
to the fact that both are devices.

* Scripts/webkitpy/port/device_port.py: Added.
(DevicePort): Add base class that all simulated/physical device ports share.
(DevicePort.__init__):
(DevicePort.driver_cmd_line_for_logging): Moved from iOS port.
(DevicePort._generate_all_test_configurations): Ditto.
(DevicePort.child_processes): Ditto.
(DevicePort.driver_name): Ditto.
(DevicePort.target_host): Generalize worker access to attached devices and
simulated devices based on the device manager.
(DevicePort.devices): Return a list of all devices associated with this port.
(DevicePort._create_devices):
(DevicePort._build_driver_flags): Moved from iOSSimulatorPort/iOSDevicePort.
(DevicePort._install): Moved from iOS port.
(DevicePort.setup_test_run): Ditto.
(DevicePort.clean_up_test_run): Ditto.
(DevicePort.did_spawn_worker): Ditto.
(DevicePort.setup_environ_for_server): Ditto.
* Scripts/webkitpy/port/ios.py:
(IOSPort):
(IOSPort._device_for_worker_number_map): Replaced by DevicePort.devices().
(IOSPort.driver_cmd_line_for_logging): Moved to DevicePort.
(IOSPort.driver_name): Ditto.
(IOSPort._generate_all_test_configurations): Ditto.
(IOSPort.child_processes): Ditto.
(IOSPort._testing_device): Ditto.
(IOSPort.target_host): Ditto.
(IOSPort._create_devices): Ditto.
(IOSPort.setup_test_run): Ditto.
(IOSPort.clean_up_test_run): Ditto.
(IOSPort.did_spawn_worker): Ditto.
* Scripts/webkitpy/port/ios_device.py:
(IOSDevicePort): IOSDevicePort uses a device manager from apple_additions.
(IOSDevicePort._look_for_all_crash_logs_in_log_dir): Use DevicePort.devices().
(IOSDevicePort._get_crash_log): Ditto.
(IOSDevicePort.ios_version): Ditto.
(IOSDevicePort._create_devices): Ditto.
(IOSDevicePort._device_for_worker_number_map): Replaced by DevicePort.devices().
(IOSSimulatorPort._build_driver_flags): Moved to DevicePort.
* Scripts/webkitpy/port/ios_device_unittest.py:
(IOSDeviceTest.test_get_crash_log):
* Scripts/webkitpy/port/ios_simulator.py:
(IOSSimulatorPort): IOSSimulatorPort uses the SimulatedDeviceManager.
(IOSSimulatorPort._create_devices): Clean up logging statement.
(IOSSimulatorPort.setup_environ_for_server): Some of environment set-up is shared
across all devices.
(IOSSimulatorPort._device_for_worker_number_map): Replaced by DevicePort.devices().
(IOSSimulatorPort._build_driver_flags): Moved to DevicePort.
* Scripts/webkitpy/port/ios_simulator_unittest.py:
(IOSSimulatorTest.test_32bit):
(IOSSimulatorTest.test_64bit):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238177 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[curl] Unify CookieJarCurlDatabase and the abstract class CookieJarCurl
Hironori.Fujii@sony.com [Wed, 14 Nov 2018 16:14:17 +0000 (16:14 +0000)]
[curl] Unify CookieJarCurlDatabase and the abstract class CookieJarCurl
https://bugs.webkit.org/show_bug.cgi?id=191620

Reviewed by Alex Christensen.

Remove a abstract class CookieJarCurl which is not needed anymore.
And, rename CookieJarCurlDatabase to CookieJarCurl.

No new tests because there's no behaviour change in WebCore.

* platform/Curl.cmake: Replaced CookieJarCurlDatabase.cpp with CookieJarCurl.cpp.
* platform/network/curl/CookieJarCurl.cpp: Renamed from Source/WebCore/platform/network/curl/CookieJarCurlDatabase.cpp.
* platform/network/curl/CookieJarCurl.h: Merged CookieJarCurl.h and CookieJarCurlDatabase.h.
* platform/network/curl/CookieJarCurlDatabase.h: Removed.
* platform/network/curl/NetworkStorageSessionCurl.cpp:
(WebCore::NetworkStorageSession::NetworkStorageSession): Replaced CookieJarCurlDatabase with CookieJarCurl.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238176 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[WPE] Remove glib usage in PlatformKeyboardEventWPE.cpp
chris.reid@sony.com [Wed, 14 Nov 2018 16:13:06 +0000 (16:13 +0000)]
[WPE] Remove glib usage in PlatformKeyboardEventWPE.cpp
https://bugs.webkit.org/show_bug.cgi?id=191606

Reviewed by Michael Catanzaro.

No behavior change.

Use StringBuilder::append(UChar32) as a generic way to convert a uint32_t code point to WTFString.

* platform/wpe/PlatformKeyboardEventWPE.cpp:
(WebCore::PlatformKeyboardEvent::keyValueForWPEKeyCode):
(WebCore::PlatformKeyboardEvent::singleCharacterString):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238175 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months agoWebKit.WKHTTPCookieStoreWithoutProcessPool API test is failing with process prewarmin...
cdumez@apple.com [Wed, 14 Nov 2018 15:50:05 +0000 (15:50 +0000)]
WebKit.WKHTTPCookieStoreWithoutProcessPool API test is failing with process prewarming is enabled
https://bugs.webkit.org/show_bug.cgi?id=191624

Reviewed by Alex Christensen.

Source/WebKit:

Stop setting the WebProcessPool's primary data store (m_websiteDataStore) to the default one in
WebProcessPool::prewarmProcess(). We did not really need to, we can pass the default data store
to the new WebPageProxy without having to set m_websiteDataStore. m_websiteDataStore only gets
set upon constructor if thr default data store already exists or later on when creating a WebPage
that uses the default data store.

In the case of the API test, the following was happening:
1. Create an ephemeral data store EDS
2. Create a WebView V1 using datastore EDS
3. Do a load in V1
4. Process prewarming would kick in and wrongly associated V1's process pool PP1 with the default data store
5. Create/Get the default datastore and set a few cookies on it
6. Create a WebView V2 using default datastore and a fresh new process pool PP2
7. Do a load in V2 and expect the cookies to be there

In HTTPCookieStore::setCookie() that is called at step 5, we call:
m_owningDataStore->processPoolForCookieStorageOperations()

In this case, m_owningDataStore is the default datastore and this call would previously return null because
there is no WebProcessPool yet associated with the default datastore. However, with the process prewarming
bug at step 4, the process pool PP1 would be returned since it was wrongly associated with the default
data store. As a result, we would call setCookie() on PP1's WebCookieManagerProxy and this would fail
because PP1's network process knows nothing about this session ID (it was only ever used with an ephemeral
session).

* UIProcess/WebProcessPool.cpp:
(WebKit::WebProcessPool::prewarmProcess):

Tools:

Add API test coverage.

* TestWebKitAPI/Tests/WebKitCocoa/WKHTTPCookieStore.mm:
(runWKHTTPCookieStoreWithoutProcessPool):
(TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238174 268f45cc-cd09-0410-ab3c-d52691b4dbfc

12 months ago[LFC][IFC] Construct dedicated runs when the inline element requires it (part 2)
zalan@apple.com [Wed, 14 Nov 2018 15:31:17 +0000 (15:31 +0000)]
[LFC][IFC] Construct dedicated runs when the inline element requires it (part 2)
https://bugs.webkit.org/show_bug.cgi?id=191623

Reviewed by Antti Koivisto.

This patch expands the breaking behaviour to support separate start/end breaks.

<span>parent </span><span style="padding: 10px;">start<span> middle </span>end</span><span> parent</span>

input to line breaking -> <parent start middle end parent>
output of line breaking (considering infinite constraint) -> <parent start middle end parent>
due to padding, final runs -> <parent><start middle end><parent>

"parent" -> n/a
"start" -> BreakAtStart
" middle " -> n/a
"end" -> BreakAtEnd
"parent" -> n/a

Another example:
<span>parent </span><span style="padding-right: 10px;">start<span> middle </span>end</span><span> parent</span>

line breaking -> <parent start middle end parent>
due to padding-right, final runs -> <parent start middle end><parent>

"parent" -> n/a
"start" -> n/a
" middle " -> n/a
"end" -> BreakAtEnd
"parent" -> n/a

* layout/inlineformatting/InlineFormattingContext.cpp:
(WebCore::Layout::InlineFormattingContext::splitInlineRunIfNeeded const):
(WebCore::Layout::InlineFormattingContext::collectInlineContent const): Move to a recursive algorithm (which is fine, inline contents don't tend to be too deep)
(WebCore::Layout::InlineFormattingContext::contentRequiresSeparateRun const): Deleted.
* layout/inlineformatting/InlineFormattingContext.h:
* layout/inlineformatting/InlineFormattingState.cpp:
(WebCore::Layout::InlineFormattingState::detachingRules const):
* layout/inlineformatting/InlineFormattingState.h:
(WebCore::Layout::InlineFormattingState::lastInlineItem const):
(WebCore::Layout::InlineFormattingState::addDetachingRule):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@238173 268f45cc-cd09-0410-ab3c-d52691b4dbfc