Source/WebCore/ChangeLog

   1 2017-04-05  Chris Dumez  <cdumez@apple.com>
   2
   3         <input type="range"> changing to disabled while active breaks all pointer events
   4         https://bugs.webkit.org/show_bug.cgi?id=170447
   5         <rdar://problem/31442875>
   6
   7         Reviewed by Geoffrey Garen.
   8
   9         When a range's slider is being moved, we set SliderThumbElement's m_inDragMode flag
  10         to true and mark the range elements as the CapturingMouseEventsElement. When we get
  11         the mouseUp event, we are supposed to exit drag mode. However, when the range element
  12         gets disabled while dragging, we do not get the mouseUp event and we need to make
  13         sure we exit dragging mode anyway. r112547 tried to fix this by calling stopDragging()
  14         in SliderThumbElement::defaultEventHandler() when the input element is disabled.
  15         While this often works, this is fragile and we sometimes fail to exit dragging mode
  16         when we should.
  17
  18         This patch addressed the issue by calling stopDragging() in
  19         SliderThumbElement::disabledAttributeChanged() instead. This is much safer as we
  20         guarantee will exit dragging mode whenever the range element gets disabled, even
  21         if SliderThumbElement::defaultEventHandler() does not get called after that.
  22
  23         Test: fast/forms/range/disabled-while-dragging.html
  24
  25         * html/RangeInputType.cpp:
  26         (WebCore::RangeInputType::disabledAttributeChanged):
  27         * html/RangeInputType.h:
  28         * html/shadow/SliderThumbElement.cpp:
  29         (WebCore::SliderThumbElement::defaultEventHandler):
  30         (WebCore::SliderThumbElement::disabledAttributeChanged):
  31         * html/shadow/SliderThumbElement.h:
  32
  33 2017-04-05  Eric Carlson  <eric.carlson@apple.com>
  34
  35         [MediaStream] Video doesn't render in fullscreen on iOS
  36         https://bugs.webkit.org/show_bug.cgi?id=170404
  37
  38         Reviewed by Youenn Fablet.
  39
  40         No new tests, filed https://bugs.webkit.org/show_bug.cgi?id=170512.
  41
  42         * platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaStreamAVFObjC.h:
  43         * platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaStreamAVFObjC.mm:
  44         (WebCore::MediaPlayerPrivateMediaStreamAVFObjC::MediaPlayerPrivateMediaStreamAVFObjC): Include
  45         video fullscreen manager on iOS too.
  46         (WebCore::MediaPlayerPrivateMediaStreamAVFObjC::videoTransformationMatrix): Add paramater
  47         to force transform recalculation.
  48         (WebCore::MediaPlayerPrivateMediaStreamAVFObjC::enqueueVideoSample): Restructure code since
  49         the display layer resize happens elsewhere.
  50         (WebCore::MediaPlayerPrivateMediaStreamAVFObjC::ensureLayers): Include video fullscreen
  51         manager on iOS too.
  52         (WebCore::MediaPlayerPrivateMediaStreamAVFObjC::destroyLayers): Ditto.
  53         (WebCore::MediaPlayerPrivateMediaStreamAVFObjC::platformLayer): Ditto.
  54         (WebCore::MediaPlayerPrivateMediaStreamAVFObjC::setVideoFullscreenLayer): Ditto.
  55         (WebCore::MediaPlayerPrivateMediaStreamAVFObjC::backgroundLayerBoundsChanged): Change the
  56         display layer size and position immediately instead of waiting for the next sample buffer
  57         so the display is correct when fullscreen mode changes when paused.
  58
  59 2017-04-05  Youenn Fablet  <youenn@apple.com>
  60
  61         Deprecate and remove URL.createObjectURL(mediastream)
  62         https://bugs.webkit.org/show_bug.cgi?id=167518
  63         <rdar://problem/31149607>
  64
  65         Reviewed by Eric Carlson.
  66
  67         Covered by updated tests.
  68
  69         * CMakeLists.txt:
  70         * DerivedSources.cpp:
  71         * DerivedSources.make:
  72         * Modules/mediastream/DOMURLMediaStream.cpp: Removed.
  73         * Modules/mediastream/DOMURLMediaStream.h: Removed.
  74         * Modules/mediastream/DOMURLMediaStream.idl: Removed.
  75         * WebCore.xcodeproj/project.pbxproj:
  76         * html/DOMURL.idl:
  77
  78 2017-04-05  Alejandro G. Castro  <alex@igalia.com>
  79
  80         [Webrtc] Mock realtime sources factories should be static after r213941
  81         https://bugs.webkit.org/show_bug.cgi?id=170282
  82
  83         Reviewed by Alex Christensen.
  84
  85         If we don't make the variables static we would be returning a
  86         local variable.
  87
  88         * platform/mock/MockRealtimeAudioSource.cpp:
  89         (WebCore::MockRealtimeAudioSource::factory):
  90         * platform/mock/MockRealtimeVideoSource.cpp:
  91         (WebCore::MockRealtimeVideoSource::factory):
  92
  93 2017-04-05  Alex Christensen  <achristensen@webkit.org>
  94
  95         Fix CMake build.
  96
  97         * platform/graphics/avfoundation/objc/MediaSampleAVFObjC.mm:
  98         Some JavaScript inline functions were not being accessed from this file with different enable flags.
  99         * platform/spi/cf/CFNetworkSPI.h:
 100         * platform/spi/cocoa/NSURLConnectionSPI.h:
 101         Moved NSURLSession-specific SPI from NSURLConnectionSPI.h to CFNetworkSPI.h.
 102
 103 2017-04-05  Chris Dumez  <cdumez@apple.com>
 104
 105         _blank / _self / _parent / _top browsing context names should be case-insensitive
 106         https://bugs.webkit.org/show_bug.cgi?id=169747
 107
 108         Reviewed by Alex Christensen.
 109
 110         _blank / _self / _parent / _top browsing context names should be case-insensitive
 111         as per the HTML specification:
 112         - https://html.spec.whatwg.org/#browsing-context-names
 113
 114         This aligns our behavior with Firefox as well. See discussion at:
 115         - https://github.com/whatwg/html/issues/2443
 116
 117         Tests: imported/w3c/web-platform-tests/html/browsers/windows/browsing-context-names/browsing-context-_blank.html
 118                imported/w3c/web-platform-tests/html/browsers/windows/browsing-context-names/browsing-context-choose-parent-001.html
 119                imported/w3c/web-platform-tests/html/browsers/windows/browsing-context-names/browsing-context-choose-parent-002.html
 120                imported/w3c/web-platform-tests/html/browsers/windows/browsing-context-names/browsing-context-choose-parent-003.html
 121                imported/w3c/web-platform-tests/html/browsers/windows/browsing-context-names/browsing-context-choose-parent-004.html
 122
 123         * loader/FrameLoader.cpp:
 124         (WebCore::FrameLoader::continueLoadAfterNewWindowPolicy):
 125         (WebCore::createWindow):
 126         * page/DOMWindow.cpp:
 127         (WebCore::DOMWindow::open):
 128         * page/FrameTree.cpp:
 129         (WebCore::FrameTree::uniqueChildName):
 130         (WebCore::FrameTree::find):
 131
 132 2017-04-05  Miguel Gomez  <magomez@igalia.com>
 133
 134         [GTK+] PNG animations that should run once are not played at all
 135         https://bugs.webkit.org/show_bug.cgi?id=170499
 136
 137         Reviewed by Carlos Garcia Campos.
 138
 139         The repetition count reported bu the PNGImageDecoder is wrong. It's returning m_playCount - 1, which
 140         means 0 for the animations that need to be played once. Change it to return an appropriate value.
 141
 142         Covered by existent tests.
 143
 144         * platform/image-decoders/png/PNGImageDecoder.cpp:
 145         (WebCore::PNGImageDecoder::repetitionCount):
 146         * platform/image-decoders/png/PNGImageDecoder.h:
 147
 148 2017-04-05  Andy Estes  <aestes@apple.com>
 149
 150         REGRESSION (r202472): Data Detection overwrites existing links in detected ranges
 151         https://bugs.webkit.org/show_bug.cgi?id=170365
 152         <rdar://problem/29205721>
 153
 154         Reviewed by Tim Horton.
 155
 156         r202472 changed the node traversal in searchForLinkRemovingExistingDDLinks() to only
 157         consider nodes that are descendants of startNode, but we need to traverse all nodes between
 158         startNode and endNode to find existing non-DD links.
 159
 160         As a result, we'd add a Data Detector link to the following snippet and make the original
 161         links un-clickable:
 162
 163             <a href='#'>tomorrow</a> <a href='#'>night</a>
 164
 165         Fix this by not specifying a stayWithin node when calling NodeTraversal::next(). The loop
 166         will terminate when we reach endNode.
 167
 168         Updated WebKit2.DataDetectionReferenceDate API test.
 169
 170         * editing/cocoa/DataDetection.mm:
 171         (WebCore::searchForLinkRemovingExistingDDLinks):
 172
 173 2017-04-04  Carlos Garcia Campos  <cgarcia@igalia.com>
 174
 175         Move WebErrors from WebProcess to Shared and get rid of ErrorsGtk in WebCore
 176         https://bugs.webkit.org/show_bug.cgi?id=156974
 177
 178         Reviewed by Sam Weinig.
 179
 180         Remove ErrorsGtk.
 181
 182         * PlatformGTK.cmake:
 183         * platform/gtk/ErrorsGtk.cpp: Removed.
 184         * platform/gtk/ErrorsGtk.h: Removed.
 185
 186 2017-04-05  Andy Estes  <aestes@apple.com>
 187
 188         [ios-simulator] API test WebKit2.DataDetectionReferenceDate timing out
 189         https://bugs.webkit.org/show_bug.cgi?id=161967
 190
 191         Reviewed by Alexey Proskuryakov.
 192
 193         DataDetectorsCoreSPI.h defined DDQueryOffset as a struct of two CFIndexes, which is 16 bytes
 194         on LP64, but the struct is actually defined as two CFIndex-typed 32-bit bitfields, which is
 195         8 bytes on LP64. This breaks the ABI on Public SDK builds when calling functions that take
 196         or return DDQueryOffsets.
 197
 198         * platform/spi/cocoa/DataDetectorsCoreSPI.h: Updated the DDQueryOffset definition for
 199         Public SDK builds, and added a static_assert to detect future size changes at compile time.
 200
 201 2017-04-04  Jer Noble  <jer.noble@apple.com>
 202
 203         Move AVSampleBufferDisplayLayer declarations into AVFoundationSPI.h
 204         https://bugs.webkit.org/show_bug.cgi?id=170471
 205
 206         Reviewed by Eric Carlson.
 207
 208         Move the declaration of AVSampleBufferDisplayLayer (and related classes) into AVFoundationSPI.
 209
 210         * platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaSourceAVFObjC.mm:
 211         * platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:
 212         * platform/spi/mac/AVFoundationSPI.h:
 213
 214 2017-04-04  Youenn Fablet  <youenn@apple.com>
 215
 216         Canvas is tainted when painting a video with MediaStreamTrack
 217         https://bugs.webkit.org/show_bug.cgi?id=170486
 218
 219         Reviewed by Eric Carlson.
 220
 221         Test: http/tests/media/media-stream/getusermedia-with-canvas.html
 222
 223         Adding the notion of isolated source so that we can later on implement WebRTC isolated tracks.
 224         For now, canvas will not be tainted if painted from a MediaStreamTrack.
 225
 226         * platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaStreamAVFObjC.h:
 227         * platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaStreamAVFObjC.mm:
 228         (WebCore::MediaPlayerPrivateMediaStreamAVFObjC::didPassCORSAccessCheck):
 229         * platform/mediastream/MediaStreamTrackPrivate.h:
 230         (WebCore::MediaStreamTrackPrivate::isIsolated):
 231         * platform/mediastream/RealtimeMediaSource.h:
 232
 233 2017-04-04  Commit Queue  <commit-queue@webkit.org>
 234
 235         Unreviewed, rolling out r214894, r214895, r214907, r214912,
 236         and r214918.
 237         https://bugs.webkit.org/show_bug.cgi?id=170491
 238
 239         Caused build failures on Mac (Requested by rniwa on #webkit).
 240
 241         Reverted changesets:
 242
 243         "Build fix."
 244         http://trac.webkit.org/changeset/214894
 245
 246         "Rolling back the build fix, as it broke other builds."
 247         http://trac.webkit.org/changeset/214895
 248
 249         "Move AVSampleBufferDisplayLayer declarations into
 250         AVFoundationSPI.h"
 251         https://bugs.webkit.org/show_bug.cgi?id=170471
 252         http://trac.webkit.org/changeset/214907
 253
 254         "Unreviewed build fix: fix compilation error on Sierra."
 255         http://trac.webkit.org/changeset/214912
 256
 257         "More build fixing."
 258         http://trac.webkit.org/changeset/214918
 259
 260 2017-04-04  Youenn Fablet  <youenn@apple.com>
 261
 262         [Mac] Add back web audio support for getUserMedia MediaStreamTrack
 263         https://bugs.webkit.org/show_bug.cgi?id=170482
 264
 265         Reviewed by Eric Carlson.
 266
 267         ´Covered by reenabled test.
 268
 269         Exporting method and class used in WebKit2.
 270
 271         * WebCore.xcodeproj/project.pbxproj:
 272         * platform/audio/mac/CAAudioStreamDescription.h:
 273         * platform/mediastream/mac/WebAudioSourceProviderAVFObjC.h:
 274
 275 2017-04-04  Alexey Proskuryakov  <ap@apple.com>
 276
 277         More build fixing.
 278
 279         * platform/spi/mac/AVFoundationSPI.h:
 280
 281 2017-04-04  Simon Fraser  <simon.fraser@apple.com>
 282
 283         Do some minor FEColorMatrix code cleanup and optimization
 284         https://bugs.webkit.org/show_bug.cgi?id=170474
 285
 286         Reviewed by Dean Jackson.
 287
 288         Don't switch inside of a pixel processing loop; repeat the loop inside switch (filterType).
 289
 290         Change matrix() and saturateAndHueRotate() to dereference the source pixels once, instead
 291         of multiple times, which is faster.
 292
 293         This kind of code benefits from aligning things with spaces for readability, so do so,
 294         violating webkit style.
 295
 296         Add some off-by-default performance logging code.
 297
 298         Increases pixel processing performance from about 86ms per megapixel to 65ms per megapixel.
 299
 300         * platform/graphics/filters/FEColorMatrix.cpp:
 301         (WebCore::matrix):
 302         (WebCore::saturateAndHueRotate):
 303         (WebCore::effectType):
 304         (WebCore::FEColorMatrix::platformApplySoftware):
 305
 306 2017-04-04  Brent Fulgham  <bfulgham@apple.com>
 307
 308         Do not assert when CharacterData representing an Attr fires events
 309         https://bugs.webkit.org/show_bug.cgi?id=170454
 310         <rdar://problem/30979320>
 311
 312         Reviewed by Ryosuke Niwa.
 313
 314         Make the NoEventDispatchAssertion in CharacterData::notifyParentAfterChange conditional
 315         since Attr elements should be allowed to fire events.
 316
 317         Tests: fast/dom/no-assert-for-malformed-js-url-attribute.html
 318
 319         * dom/CharacterData.cpp:
 320         (WebCore::CharacterData::notifyParentAfterChange):
 321
 322 2017-04-04  Youenn Fablet  <youenn@apple.com>
 323
 324         LayoutTest webrtc/libwebrtc/descriptionGetters.html is a flaky failure
 325         https://bugs.webkit.org/show_bug.cgi?id=169481
 326
 327         Reviewed by Eric Carlson.
 328
 329         No need to enumerate all network interfaces in most layout tests.
 330         Adding an Internals API for that in case we want to use TCP localhost candidates.
 331
 332         * testing/Internals.cpp:
 333         (WebCore::Internals::Internals):
 334         (WebCore::Internals::setEnumeratingAllNetworkInterfacesEnabled):
 335         * testing/Internals.h:
 336         * testing/Internals.idl:
 337
 338
 339 2017-04-04  Jer Noble  <jer.noble@apple.com>
 340
 341         Unreviewed build fix: fix compilation error on Sierra.
 342
 343         * platform/spi/mac/AVFoundationSPI.h:
 344
 345 2017-04-04  Jer Noble  <jer.noble@apple.com>
 346
 347         Move AVSampleBufferDisplayLayer declarations into AVFoundationSPI.h
 348         https://bugs.webkit.org/show_bug.cgi?id=170471
 349
 350         Reviewed by Eric Carlson.
 351
 352         Move the declaration of AVSampleBufferDisplayLayer (and related classes) into AVFoundationSPI.
 353
 354         * platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaSourceAVFObjC.mm:
 355         (WebCore::MediaPlayerPrivateMediaSourceAVFObjC::addDisplayLayer): Deleted.
 356         (WebCore::MediaPlayerPrivateMediaSourceAVFObjC::removeDisplayLayer): Deleted.
 357         * platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:
 358         * platform/spi/mac/AVFoundationSPI.h:
 359
 360 2017-04-04  Saam Barati  <sbarati@apple.com>
 361
 362         WebAssembly: JSWebAssemblyCallee should not be a JSCell
 363         https://bugs.webkit.org/show_bug.cgi?id=170135
 364
 365         Reviewed by Michael Saboff.
 366
 367         * bindings/js/JSDOMWindowBase.cpp:
 368         (WebCore::callerDOMWindow):
 369
 370 2017-04-04  Simon Fraser  <simon.fraser@apple.com>
 371
 372         Change Document's lastHandledUserGestureTimestamp to be a MonotonicTime
 373         https://bugs.webkit.org/show_bug.cgi?id=170468
 374
 375         Reviewed by Zalan Bujtas.
 376
 377         Change the double to MonotonicTime. No behavior change.
 378
 379         * dom/Document.cpp:
 380         (WebCore::Document::updateLastHandledUserGestureTimestamp):
 381         * dom/Document.h:
 382         (WebCore::Document::lastHandledUserGestureTimestamp):
 383         * html/HTMLPlugInImageElement.cpp:
 384         (WebCore::documentHadRecentUserGesture):
 385         * loader/FrameLoader.cpp:
 386         (WebCore::shouldAskForNavigationConfirmation):
 387
 388 2017-04-04  Youenn Fablet  <youenn@apple.com>
 389
 390         [Debug] ASSERT(!throwScope.exception()) on imported/w3c/web-platform-tests/fetch/api/cors/cors-preflight-status-worker.html
 391         https://bugs.webkit.org/show_bug.cgi?id=170395
 392         <rdar://problem/31394017>
 393
 394         Reviewed by Mark Lam.
 395
 396         No change of behavior.
 397
 398         * bindings/js/JSDOMPromise.cpp:
 399         (WebCore::DeferredPromise::reject): Adding early ASSERT that creating an exception is happening correctly.
 400
 401 2017-04-04  Alexey Proskuryakov  <ap@apple.com>
 402
 403         Rolling back the build fix, as it broke other builds.
 404
 405         * platform/spi/mac/AVFoundationSPI.h:
 406
 407 2017-04-04  Alexey Proskuryakov  <ap@apple.com>
 408
 409         Build fix.
 410
 411         Rubber-stamped by Jer Noble.
 412
 413         * platform/spi/mac/AVFoundationSPI.h:
 414
 415 2017-04-04  Tim Horton  <timothy_horton@apple.com>
 416
 417         [Mac] -[WKWebView findMatchesForString:relativeToMatch:findOptions:maxResults:resultCollector:] invokes the resultCollector with didWrap = NO even when it wraps
 418         https://bugs.webkit.org/show_bug.cgi?id=165801
 419         <rdar://problem/29649535>
 420
 421         Reviewed by Wenson Hsieh.
 422
 423         New API tests: WebKit2.FindInPageWrapping*
 424
 425         * page/FrameTree.cpp:
 426         (WebCore::FrameTree::traverseNextWithWrap):
 427         (WebCore::FrameTree::traversePreviousWithWrap):
 428         (WebCore::FrameTree::traverseNextInPostOrderWithWrap):
 429         * page/FrameTree.h:
 430         Add CanWrap and DidWrap boolean enums, and add an optional out argument
 431         to traverse*WithWrap indicating whether a wrap actually occurred.
 432
 433         * history/CachedPage.cpp:
 434         (WebCore::firePageShowAndPopStateEvents):
 435         * history/PageCache.cpp:
 436         (WebCore::destroyRenderTree):
 437         Adjust to the new CanWrap enum.
 438
 439         * page/Page.cpp:
 440         (WebCore::incrementFrame):
 441         (WebCore::Page::findString):
 442         (WebCore::Page::findStringMatchingRanges):
 443         (WebCore::Page::rangeOfString):
 444         (WebCore::Page::findMatchesForText):
 445         (WebCore::Page::unmarkAllTextMatches):
 446         * page/Page.h:
 447         Adjust to the new CanWrap enum, and optionally plumb DidWrap through
 448         to callers of findString().
 449
 450 2017-04-04  Carlos Garcia Campos  <cgarcia@igalia.com>
 451
 452         [GTK] PLATFORM(GTK) && !USE(COORDINATED_GRAPHICS_THREADED) is no longer possible
 453         https://bugs.webkit.org/show_bug.cgi?id=170458
 454
 455         Reviewed by Carlos Alberto Lopez Perez.
 456
 457         That is not supported anymore, so we can remove dead code from VideoSinkGStreamer.cpp
 458
 459         * platform/graphics/gstreamer/VideoSinkGStreamer.cpp:
 460         (VideoRenderRequestScheduler::VideoRenderRequestScheduler):
 461
 462 2017-04-04  Per Arne Vollan  <pvollan@apple.com>
 463
 464         Unreviewed Windows build fix.
 465
 466         * bindings/js/JSDOMGuardedObject.h:
 467
 468 2017-04-03  Wenson Hsieh  <wenson_hsieh@apple.com>
 469
 470         Data interaction should register type identifiers in order of priority
 471         https://bugs.webkit.org/show_bug.cgi?id=170428
 472         <rdar://problem/30633296>
 473
 474         Reviewed by Tim Horton.
 475
 476         Currently, due to the shared logic for writing to a UIPasteboard (in the case of copy/paste) and the shared
 477         UIItemProvider-based pasteboard wrapper (in the case of data interaction), we don't enforce any particular
 478         ordering in which type identifiers are registered in the generated item provider. This is because
 479         -[UIPasteboard setItems:] only takes an unordered mapping of type identifiers to objects.
 480
 481         To fix this, we introduce a separate way to set pasteboard data that first writes a list of object
 482         representations in order of priority (which is dependent on the content being interacted with) and then register
 483         individual type-data mappings at the end.
 484
 485         Augmented existing API tests in DataInteractionTests to check for the existence and priority of type identifiers
 486         in the UIItemProviders created upon starting data interaction. Also adds a new unit test:
 487         DataInteractionTests.TextAreaToInput.
 488
 489         * WebCore.xcodeproj/project.pbxproj:
 490         * editing/cocoa/EditorCocoa.mm:
 491         (WebCore::archivedDataForAttributedString):
 492         (WebCore::Editor::writeSelectionToPasteboard):
 493         (WebCore::Editor::writeSelection):
 494         * platform/Pasteboard.h:
 495         * platform/PasteboardWriterData.h:
 496         * platform/PlatformPasteboard.h:
 497         * platform/ios/AbstractPasteboard.h:
 498         * platform/ios/AbstractPasteboard.mm: Copied from Source/WebCore/platform/ios/AbstractPasteboard.h.
 499
 500         Introduce WebPasteboardItemData, a wrapper around a list of objects representating the pasteboard data in order
 501         of priority, and a dictionary containing additional NSData blobs that contain data useful for private clients.
 502
 503         (+[WebPasteboardItemData itemWithRepresentingObjects:additionalData:]):
 504         (-[WebPasteboardItemData initWithRepresentingObjects:additionalData:]):
 505         (-[WebPasteboardItemData representingObjects]):
 506         (-[WebPasteboardItemData additionalData]):
 507         * platform/ios/PlatformPasteboardIOS.mm:
 508         (WebCore::richTextRepresentationsForPasteboardWebContent):
 509         (WebCore::PlatformPasteboard::writeObjectRepresentations):
 510         (WebCore::PlatformPasteboard::write):
 511
 512         Tweaked to check whether the pasteboard responds to -setItemsFromObjectRepresentations:. If so, uses the
 513         PlatformPasteboard::writeObjectRepresentations codepath to write data to the pasteboard, respecting type
 514         priority.
 515
 516         * platform/ios/WebItemProviderPasteboard.mm:
 517         (-[WebItemProviderPasteboard pasteboardTypes]):
 518         (-[WebItemProviderPasteboard setItemProviders:]):
 519         (-[WebItemProviderPasteboard setItemsFromObjectRepresentations:]):
 520
 521         Replaces -setItems: with -setItemsFromObjectRepresentations:, which respects the priority of each object
 522         representation of the data in the pasteboard.
 523
 524         (-[WebItemProviderPasteboard setItems:]): Deleted.
 525
 526 2017-04-03  Javier Fernandez  <jfernandez@igalia.com>
 527
 528         [css-align] Adapt place-content alignment shorthand to the new baseline syntax
 529         https://bugs.webkit.org/show_bug.cgi?id=170340
 530
 531         Reviewed by David Hyatt.
 532
 533         Now that the align-content and justify-content CSS properties are
 534         adapted to the new baseline-position CSS values syntax we can adapt the
 535         shorthand that controls such properties to the new syntax as well.
 536
 537         No new tests, just adding some additional cases to the tests we already have.
 538
 539         * css/StyleProperties.cpp:
 540         (WebCore::StyleProperties::getPropertyValue):
 541         (WebCore::StyleProperties::placeContentPropertyValue):
 542         * css/StyleProperties.h:
 543         * css/parser/CSSPropertyParser.cpp:
 544         (WebCore::isContentDistributionKeyword):
 545         (WebCore::isContentPositionKeyword):
 546         (WebCore::isOverflowKeyword):
 547         (WebCore::getBaselineKeyword):
 548         (WebCore::consumeContentDistributionOverflowPosition):
 549         (WebCore::consumeSimplifiedContentPosition):
 550
 551 2017-04-03  Nan Wang  <n_wang@apple.com>
 552
 553         AX: Expose link children when doing search predication on iOS
 554         https://bugs.webkit.org/show_bug.cgi?id=170424
 555         <rdar://problem/31413335>
 556
 557         The children of links are accessible elements on iOS instead of the link itself,
 558         so we should expose the children when doing predicated searches as well.
 559
 560         Reviewed by Chris Fleizach.
 561
 562         Test: accessibility/ios-simulator/ios-search-predicate-link-children.html
 563
 564         * accessibility/AccessibilityObject.cpp:
 565         (WebCore::AccessibilityObject::isAccessibilityObjectSearchMatchAtIndex):
 566
 567 2017-04-03  Jer Noble  <jer.noble@apple.com>
 568
 569         No audio output for MediaStream-backed audio elements on iOS
 570         https://bugs.webkit.org/show_bug.cgi?id=170427
 571
 572         Reviewed by Eric Carlson.
 573
 574         The logic for setting the output timestamp offset in AudioSampleDataSource was reversed; instead of
 575         subtracting out the timestamp of the first pull request, it effectively doubled it.
 576
 577         * platform/audio/mac/AudioSampleDataSource.mm:
 578         (WebCore::AudioSampleDataSource::pullSamplesInternal):
 579
 580 2017-04-03  Dave Hyatt  <hyatt@apple.com>
 581
 582         Japanese fonts in vertical text should support synthesized italics
 583         https://bugs.webkit.org/show_bug.cgi?id=169301
 584
 585         Reviewed by Simon Fraser.
 586
 587         Updated test in fast/text/international.
 588
 589         * platform/graphics/Font.cpp:
 590         (WebCore::Font::nonSyntheticItalicFont): Deleted.
 591         * platform/graphics/Font.h:
 592         Removed the non-synthetic italic font member, since it's not used.
 593
 594         * platform/graphics/FontCascadeFonts.cpp:
 595         (WebCore::glyphDataForNonCJKCharacterWithGlyphOrientation):
 596         Patched to ensure that a text-orientation-fallback font data is always
 597         returned in the synthetic oblique case, so that non-CJK and CJK are
 598         guaranteed to be broken up so that their slants can be applied differently.
 599
 600         (WebCore::FontCascadeFonts::glyphDataForSystemFallback):
 601         (WebCore::FontCascadeFonts::glyphDataForNormalVariant):
 602         (WebCore::glyphDataForCJKCharacterWithoutSyntheticItalic): Deleted.
 603         Patched to no longer turn off synthetic oblique for CJK.
 604
 605         * platform/graphics/cocoa/FontCascadeCocoa.mm:
 606         (WebCore::FontCascade::drawGlyphs):
 607         Apply the correct italic transform to CJK when in vertical text.
 608
 609 2017-04-03  Chris Dumez  <cdumez@apple.com>
 610
 611         REGRESSION (r206744): CSS background-image in style attribute ignored when using createHTMLDocument method of DOM parsing
 612         https://bugs.webkit.org/show_bug.cgi?id=170285
 613         <rdar://problem/31378543>
 614
 615         Reviewed by Andy Estes.
 616
 617         r206744 caused up to stop trying to resolve relative URLs when trying to load an image
 618         referred to by CSS. We already try to resolve the relative URL when parsing the CSS
 619         property so this will usually work fine. However, in the case when the CSS property
 620         is parsed in detached document and then moved to another document, we will not have
 621         the complete URL.
 622
 623         Test: fast/images/background-image-relative-url-changes-document.html
 624
 625         * css/CSSImageValue.cpp:
 626         (WebCore::CSSImageValue::loadImage):
 627
 628 2017-04-03  Jeremy Jones  <jeremyj@apple.com>
 629
 630         Do not set WebAVPlayerLayerView background to black in fullscreen.
 631         https://bugs.webkit.org/show_bug.cgi?id=170132
 632         rdar://problem/30839278
 633
 634         Reviewed by Tim Horton.
 635
 636         No new tests because no behavior change.
 637
 638         The black background on WebAVPlayerLayerView interferes with the fullscreen animation
 639         and has been removed.
 640
 641         * platform/ios/WebVideoFullscreenInterfaceAVKit.mm:
 642         (WebVideoFullscreenInterfaceAVKit::enterFullscreen):
 643         (WebVideoFullscreenInterfaceAVKit::enterFullscreenStandard):
 644
 645 2017-04-03  Antti Koivisto  <antti@apple.com>
 646
 647         REGRESSION (r207669): FileMaker Pro Help pages do not render correctly
 648         https://bugs.webkit.org/show_bug.cgi?id=170402
 649         <rdar://problem/31004344>
 650
 651         Reviewed by Simon Fraser.
 652
 653         If a new stylesheet load is started from the load event the document.styleSheets does not
 654         always reflect the already loaded stylesheets.
 655
 656         Test: fast/css/document-stylesheets-dynamic.html
 657
 658         * style/StyleScope.cpp:
 659         (WebCore::Style::Scope::updateActiveStyleSheets):
 660
 661             Remove an old optimization where we would not update active stylesheets if there were pending
 662             (head) stylesheet loads and they had not been updated already.
 663             This is probably not a valuable optimization anymore with the new lazy stylesheet update strategy.
 664
 665         * style/StyleScope.h:
 666
 667 2017-04-03  Anders Carlsson  <andersca@apple.com>
 668
 669         Tweak ApplePaySession API
 670         https://bugs.webkit.org/show_bug.cgi?id=170409
 671         rdar://problem/31405459
 672
 673         Reviewed by Tim Horton.
 674
 675         Rename "address" to "addressLines". Add "postalAddress". Reorder the ApplePayError constructor parameters.
 676
 677         * Modules/applepay/ApplePayError.cpp:
 678         (WebCore::ApplePayError::create):
 679         (WebCore::ApplePayError::ApplePayError):
 680         * Modules/applepay/ApplePayError.h:
 681         * Modules/applepay/ApplePayError.idl:
 682         * Modules/applepay/PaymentRequest.h:
 683
 684 2017-04-03  Zan Dobersek  <zdobersek@igalia.com>
 685
 686         [GCrypt] Implement CryptoKeyEC::keySizeInBits(), ::platformGeneratePair()
 687         https://bugs.webkit.org/show_bug.cgi?id=170345
 688
 689         Reviewed by Michael Catanzaro.
 690
 691         Start implementing the libgcrypt-based platform bits of CryptoKeyEC.
 692
 693         Implement keySizeInBits() by returning the appropriate size for this
 694         object's curve type. An assertion is added to ensure that this size
 695         matches the one that's returned by gcry_pk_get_nbits() for this
 696         object's EC key as represented by the m_platformKey gcry_sexp_t object.
 697
 698         Implement platformGeneratePair() by constructing a genkey s-expression
 699         that requests a generation of an EC key for the specified curve type.
 700         The s-expression is then passed to gcry_pk_genkey(), and the public
 701         and private key data is then retrieved from the returned s-expression
 702         upon success and used to create the public and private CryptoKeyEC
 703         objects.
 704
 705         The PlatformECKey type alias is changed to match gcry_sexp_t. The
 706         CryptoKeyEC destructor releases the gcry_sexp_t object through
 707         a PAL::GCrypt::HandleDeleter<gcry_sexp_t> instance.
 708
 709         The method definitions in the CryptoKeyECGCrypt.cpp file are also
 710         sorted to match the declaration order in the header.
 711
 712         No new tests -- current ones cover this sufficiently, but are not yet
 713         enabled due to other missing platform-specific SUBTLE_CRYPTO
 714         implementations.
 715
 716         * crypto/gcrypt/CryptoKeyECGCrypt.cpp:
 717         (WebCore::curveSize):
 718         (WebCore::curveName):
 719         (WebCore::CryptoKeyEC::~CryptoKeyEC):
 720         (WebCore::CryptoKeyEC::keySizeInBits):
 721         (WebCore::CryptoKeyEC::platformGeneratePair):
 722         (WebCore::CryptoKeyEC::platformImportSpki):
 723         (WebCore::CryptoKeyEC::platformImportPkcs8):
 724         (WebCore::CryptoKeyEC::platformExportRaw):
 725         (WebCore::CryptoKeyEC::platformAddFieldElements):
 726         (WebCore::CryptoKeyEC::platformExportSpki):
 727         * crypto/keys/CryptoKeyEC.h:
 728
 729 2017-04-03  Zan Dobersek  <zdobersek@igalia.com>
 730
 731         [GCrypt] Implement AES_KW support
 732         https://bugs.webkit.org/show_bug.cgi?id=170274
 733
 734         Reviewed by Michael Catanzaro.
 735
 736         Implement the CryptoAlgorithmAES_KW::platform{Wrap,Unwrap}Key()
 737         functionality for configurations that use libgcrypt. This is done
 738         by leveraging the gcry_cipher_* APIs for the AES algorithm that's
 739         deducted appropriately from the key size and the AESWRAP cipher mode.
 740
 741         No new tests -- current ones cover this sufficiently, but are not yet
 742         enabled due to other missing platform-specific SUBTLE_CRYPTO
 743         implementations.
 744
 745         * crypto/gcrypt/CryptoAlgorithmAES_KWGCrypt.cpp:
 746         (WebCore::gcryptWrapKey):
 747         (WebCore::gcryptUnwrapKey):
 748         (WebCore::CryptoAlgorithmAES_KW::platformWrapKey):
 749         (WebCore::CryptoAlgorithmAES_KW::platformUnwrapKey):
 750
 751 2017-04-03  Zan Dobersek  <zdobersek@igalia.com>
 752
 753         [GCrypt] Implement AES_GCM support
 754         https://bugs.webkit.org/show_bug.cgi?id=170271
 755
 756         Reviewed by Michael Catanzaro.
 757
 758         Implement the CryptoAlgorithmAES_GCM::platform{Encrypt,Decrypt}
 759         functionality for configurations that use libgcrypt. This is done
 760         by leveraging the gcry_cipher_* APIs for the AES algorithm that's
 761         deducted appropriately from the key size and the GCM cipher mode.
 762
 763         No new tests -- current ones cover this sufficiently, but are not yet
 764         enabled due to other missing platform-specific SUBTLE_CRYPTO
 765         implementations.
 766
 767         * crypto/gcrypt/CryptoAlgorithmAES_GCMGCrypt.cpp:
 768         (WebCore::gcryptEncrypt):
 769         (WebCore::gcryptDecrypt):
 770         (WebCore::CryptoAlgorithmAES_GCM::platformEncrypt):
 771         (WebCore::CryptoAlgorithmAES_GCM::platformDecrypt):
 772
 773 2017-04-03  Zan Dobersek  <zdobersek@igalia.com>
 774
 775         [GCrypt] Implement PBKDF2 support
 776         https://bugs.webkit.org/show_bug.cgi?id=170270
 777
 778         Reviewed by Michael Catanzaro.
 779
 780         Implement the CryptoAlgorithmPBKDF2::platformDeriveBits() functionality
 781         for configurations that use libgcrypt. This is done by leveraging the
 782         gcry_kdf_derive() API, using GCRY_KDF_PBKDF2 as the preferred KDF
 783         along with the properly deducted SHA algorithm.
 784
 785         No new tests -- current ones cover this sufficiently, but are not yet
 786         enabled due to other missing platform-specific SUBTLE_CRYPTO
 787         implementations.
 788
 789         * crypto/gcrypt/CryptoAlgorithmPBKDF2GCrypt.cpp:
 790         (WebCore::gcryptDeriveBits):
 791         (WebCore::CryptoAlgorithmPBKDF2::platformDeriveBits):
 792
 793 2017-04-01  Simon Fraser  <simon.fraser@apple.com>
 794
 795         Clean up touch event handler registration when moving nodes between documents
 796         https://bugs.webkit.org/show_bug.cgi?id=170384
 797         rdar://problem/30816694
 798
 799         Reviewed by Chris Dumez.
 800
 801         Make sure that Node::didMoveToNewDocument() does the correct unregistration on the
 802         old document, and registration on the new document for nodes with touch event listeners,
 803         and gesture event listeners. Touch "handler" nodes (those for overflow and sliders) are
 804         already correctly moved via renderer-related teardown.
 805
 806         Add assertions that fire when removal was not complete.
 807
 808         Use references in more places.
 809
 810         Tests: fast/events/touch/ios/gesture-node-move-between-documents.html
 811                fast/events/touch/ios/overflow-node-move-between-documents.html
 812                fast/events/touch/ios/slider-node-move-between-documents.html
 813                fast/events/touch/ios/touch-node-move-between-documents.html
 814
 815         * dom/EventNames.h:
 816         (WebCore::EventNames::gestureEventNames):
 817         * dom/Node.cpp:
 818         (WebCore::Node::willBeDeletedFrom):
 819         (WebCore::Node::didMoveToNewDocument):
 820         (WebCore::tryAddEventListener):
 821         (WebCore::tryRemoveEventListener):
 822         * html/shadow/SliderThumbElement.cpp:
 823         (WebCore::SliderThumbElement::registerForTouchEvents):
 824         (WebCore::SliderThumbElement::unregisterForTouchEvents):
 825         * rendering/RenderLayer.cpp:
 826         (WebCore::RenderLayer::registerAsTouchEventListenerForScrolling):
 827         (WebCore::RenderLayer::unregisterAsTouchEventListenerForScrolling):
 828
 829 2017-04-03  Youenn Fablet  <youenn@apple.com>
 830
 831         captureStream is getting black frames with webgl canvas
 832         https://bugs.webkit.org/show_bug.cgi?id=170325
 833
 834         Reviewed by Dean Jackson.
 835
 836         Test: fast/mediastream/captureStream/canvas3d.html
 837
 838         Changing the webgl context to save buffers in case the canvas is captured.
 839         Adding a canvas changed notification in case of clear.
 840         In the future, we might want to change this notification and do it when endPaint or similar is called.
 841
 842         Adding an Internals API to grab the RGBA equivalent of the next track frame.
 843         For that purpose, adding a bunch of WEBCORE_EXPORT.
 844
 845         * Modules/mediastream/CanvasCaptureMediaStreamTrack.cpp:
 846         (WebCore::CanvasCaptureMediaStreamTrack::Source::Source): Adding constraints support so that track settings
 847         getter actually transmits the width and height of the source.
 848         (WebCore::CanvasCaptureMediaStreamTrack::Source::canvasChanged): ensuring webgl canvas context keep their drawing buffer.
 849         * Modules/mediastream/MediaStreamTrack.h:
 850         * bindings/js/JSDOMGuardedObject.h:
 851         * bindings/js/JSDOMPromise.h:
 852         (WebCore::DeferredPromise::resolve):
 853         (WebCore::DeferredPromise::reject):
 854         * dom/ActiveDOMCallback.h:
 855         * html/HTMLCanvasElement.cpp:
 856         (WebCore::HTMLCanvasElement::captureStream):
 857         * html/ImageData.h:
 858         * html/ImageData.idl:
 859         * html/canvas/WebGLRenderingContext.cpp:
 860         (WebCore::WebGLRenderingContext::clear): ensuring canvas observers get notified in case of clear calls.
 861         * html/canvas/WebGLRenderingContextBase.h:
 862         (WebCore::WebGLRenderingContextBase::preserveDrawingBuffer): Added to allow canvas capture to update this property.
 863         * platform/MediaSample.h:
 864         (WebCore::MediaSample::getRGBAImageData): Added for internals API.
 865         * platform/graphics/avfoundation/MediaSampleAVFObjC.h:
 866         * platform/graphics/avfoundation/objc/MediaSampleAVFObjC.mm:
 867         (WebCore::MediaSampleAVFObjC::getRGBAImageData):
 868         * platform/graphics/cv/PixelBufferConformerCV.cpp:
 869         (WebCore::PixelBufferConformerCV::convert): Helper routine for getRGBAImageData.
 870         * platform/graphics/cv/PixelBufferConformerCV.h:
 871         * platform/mediastream/RealtimeMediaSourceSettings.h:
 872         (WebCore::RealtimeMediaSourceSettings::setSupportedConstraints):
 873         (WebCore::RealtimeMediaSourceSettings::setSupportedConstraits): Deleted.
 874         * platform/mediastream/mac/AVMediaCaptureSource.mm:
 875         (WebCore::AVMediaCaptureSource::initializeSettings):
 876         * platform/mediastream/openwebrtc/RealtimeAudioSourceOwr.h:
 877         * platform/mediastream/openwebrtc/RealtimeVideoSourceOwr.h:
 878         * platform/mock/MockRealtimeMediaSource.cpp:
 879         (WebCore::MockRealtimeMediaSource::initializeSettings):
 880         * testing/Internals.cpp:
 881         (WebCore::Internals::grabNextMediaStreamTrackFrame):
 882         (WebCore::Internals::videoSampleAvailable):
 883         * testing/Internals.h:
 884         * testing/Internals.idl:
 885
 886 2017-04-03  Per Arne Vollan  <pvollan@apple.com>
 887
 888         Implement stroke-miterlimit.
 889         https://bugs.webkit.org/show_bug.cgi?id=169078
 890
 891         Reviewed by Dean Jackson.
 892
 893         Support stroke-miterlimit for text rendering, see https://drafts.fxtf.org/paint/.
 894
 895         Tests: fast/css/stroke-miterlimit-default.html
 896                fast/css/stroke-miterlimit-large.html
 897                fast/css/stroke-miterlimit-zero.html
 898
 899         * css/CSSComputedStyleDeclaration.cpp:
 900         (WebCore::ComputedStyleExtractor::propertyValue):
 901         * css/CSSProperties.json:
 902         * css/SVGCSSComputedStyleDeclaration.cpp:
 903         (WebCore::ComputedStyleExtractor::svgPropertyValue):
 904         * rendering/TextPaintStyle.cpp:
 905         (WebCore::computeTextPaintStyle):
 906         (WebCore::updateGraphicsContext):
 907         * rendering/TextPaintStyle.h:
 908         * rendering/style/RenderStyle.cpp:
 909         (WebCore::RenderStyle::diff):
 910         * rendering/style/RenderStyle.h:
 911         (WebCore::RenderStyle::strokeMiterLimit):
 912         (WebCore::RenderStyle::setStrokeMiterLimit):
 913         (WebCore::RenderStyle::initialStrokeMiterLimit):
 914         (WebCore::RenderStyle::setStrokeDashOffset):
 915         * rendering/style/RenderStyleConstants.cpp:
 916         * rendering/style/RenderStyleConstants.h:
 917         * rendering/style/SVGRenderStyle.cpp:
 918         (WebCore::SVGRenderStyle::diff):
 919         * rendering/style/SVGRenderStyle.h:
 920         (WebCore::SVGRenderStyle::initialStrokeDashArray):
 921         (WebCore::SVGRenderStyle::strokeDashArray):
 922         (WebCore::SVGRenderStyle::initialStrokeMiterLimit): Deleted.
 923         (WebCore::SVGRenderStyle::strokeMiterLimit): Deleted.
 924         (WebCore::SVGRenderStyle::setStrokeMiterLimit): Deleted.
 925         * rendering/style/SVGRenderStyleDefs.cpp:
 926         (WebCore::StyleStrokeData::StyleStrokeData):
 927         (WebCore::StyleStrokeData::operator==):
 928         * rendering/style/SVGRenderStyleDefs.h:
 929         * rendering/style/StyleRareInheritedData.cpp:
 930         (WebCore::StyleRareInheritedData::StyleRareInheritedData):
 931         (WebCore::StyleRareInheritedData::operator==):
 932         * rendering/style/StyleRareInheritedData.h:
 933         * rendering/svg/RenderSVGShape.cpp:
 934         (WebCore::RenderSVGShape::hasSmoothStroke):
 935         * rendering/svg/SVGRenderSupport.cpp:
 936         (WebCore::SVGRenderSupport::applyStrokeStyleToContext):
 937         * rendering/svg/SVGRenderTreeAsText.cpp:
 938         (WebCore::writeStyle):
 939
 940 2017-04-03  Alejandro G. Castro  <alex@igalia.com>
 941
 942         [OWR] Fix class structure for the OWR mock classes after last modifications
 943         https://bugs.webkit.org/show_bug.cgi?id=170173
 944
 945         Reviewed by Youenn Fablet.
 946
 947         In case of OWR MockRealtimeMediaSource inherits from
 948         RealtimeMediaSourceOwr, so we have to change some of the function
 949         interfaces.
 950
 951         * platform/mock/MockRealtimeMediaSource.h:
 952
 953 2017-04-02  Alexey Proskuryakov  <ap@apple.com>
 954
 955         Build fix for
 956         Add missing text styles
 957         https://bugs.webkit.org/show_bug.cgi?id=170295
 958
 959         * rendering/RenderThemeIOS.mm:
 960         (WebCore::RenderThemeIOS::cachedSystemFontDescription):
 961         (WebCore::RenderThemeIOS::updateCachedSystemFontDescription):
 962
 963 2017-04-01  Zalan Bujtas  <zalan@apple.com>
 964
 965         Long Arabic text in ContentEditable with css white-space=pre hangs Safari
 966         https://bugs.webkit.org/show_bug.cgi?id=170245
 967
 968         Reviewed by Myles C. Maxfield.
 969
 970         While searching for mid-word break, we measure the text by codepoints in a loop until the accumulated width > available width.
 971         When we see that the accumulated width for the individual codepoints overflows, we join the codepoints and re-measure them.
 972         These 2 widths could be considerably different for number of reasons (ligatures is a prime example). When we figure that
 973         the run still fits, we go back to the main loop (since we are not supposed to wrap the line here) and take the next codepoint.
 974         However this time we start the measurement from the last whitespace, so we end up remeasuring a potentially long chuck of text
 975         until we hit the wrapping point. This is way too expensive.
 976         This patch changes the logic so that we just go back to measuring individual codepoints until we hit the constrain again.
 977
 978         Covered by existing tests.
 979
 980         * rendering/line/BreakingContext.h:
 981         (WebCore::BreakingContext::handleText): canUseSimpleFontCodePath() is just to mitigate the potential risk of regression and
 982         complex text is more likely to fall into this category.
 983
 984 2017-04-01  Jon Lee  <jonlee@apple.com>
 985
 986         Add missing text styles
 987         https://bugs.webkit.org/show_bug.cgi?id=170295
 988         rdar://problem/30219503
 989
 990         Reviewed by Dean Jackson.
 991
 992         Updated existing test to include new text styles.
 993
 994         * css/CSSValueKeywords.in: Add title0 and title4.
 995         * platform/spi/cocoa/CoreTextSPI.h:
 996         * rendering/RenderThemeIOS.mm:
 997         (WebCore::RenderThemeIOS::cachedSystemFontDescription):
 998         (WebCore::RenderThemeIOS::updateCachedSystemFontDescription):
 999
1000 2017-04-01  Dan Bernstein  <mitz@apple.com>
1001
1002         [iOS] <input type=file> label should be specified using plural rules
1003         https://bugs.webkit.org/show_bug.cgi?id=170388
1004
1005         Reviewed by Alexey Proskuryakov.
1006
1007         * English.lproj/Localizable.strings: Updated using update-webkit-localizable-strings.
1008
1009         * English.lproj/Localizable.stringsdict: Added an entry for the new key "%lu photo(s) and
1010           %lu video(s)", with plural rules covering all the different combinations in English.
1011           Other localizations may specify additional combinations as needed.
1012
1013 2017-04-01  Alexey Proskuryakov  <ap@apple.com>
1014
1015         Rolling back <https://trac.webkit.org/r214697>, as it made WebKit2.DataDetectionReferenceDate time out.
1016
1017         Was REGRESSION (r202472): Data Detection overwrites existing links in detected ranges
1018         https://bugs.webkit.org/show_bug.cgi?id=170365
1019
1020         * editing/cocoa/DataDetection.mm:
1021         (WebCore::searchForLinkRemovingExistingDDLinks):
1022
1023 2017-04-01  Chris Dumez  <cdumez@apple.com>
1024
1025         We should pause silent WebAudio rendering in background tabs
1026         https://bugs.webkit.org/show_bug.cgi?id=170299
1027         <rdar://problem/31289132>
1028
1029         Reviewed by Eric Carlson.
1030
1031         We should pause silent WebAudio rendering in background tabs since it uses CPU and is
1032         not observable by the user. Such silent WebAudio rendering seems to be used by
1033         doubleclick ads.
1034
1035         Test: webaudio/silent-audio-interrupted-in-background.html
1036
1037         * Modules/webaudio/AudioContext.cpp:
1038         (WebCore::AudioContext::lazyInitialize):
1039         (WebCore::AudioContext::uninitialize):
1040         Have AudioContext register / unregister itself with the Document to get
1041         visibility change notifications, similarly to what HTMLMediaElement was
1042         already doing.
1043
1044         (WebCore::AudioContext::visibilityStateChanged):
1045         Begin / End session interruption whenever the page visiblity changes.
1046
1047         * Modules/webaudio/AudioContext.h:
1048         * WebCore.xcodeproj/project.pbxproj:
1049
1050         * dom/Document.cpp:
1051         (WebCore::Document::registerForVisibilityStateChangedCallbacks):
1052         (WebCore::Document::unregisterForVisibilityStateChangedCallbacks):
1053         (WebCore::Document::visibilityStateChanged):
1054         * dom/Document.h:
1055         * dom/Element.h:
1056         * dom/VisibilityChangeClient.h: Added.
1057         (WebCore::VisibilityChangeClient::~VisibilityChangeClient):
1058         * html/HTMLMediaElement.h:
1059         Introduce a new VisibilityChangeClient interface and have both AudioContext
1060         and HTMLMediaElement subclass it. Previously, the visibilityStateChanged()
1061         function was on Element but this prevented AudioContext from registering
1062         itself since AudioContext is not an Element.
1063
1064 2017-04-01  Dan Bernstein  <mitz@apple.com>
1065
1066         [Cocoa] A couple of UI strings use three periods instead of an ellipsis
1067         https://bugs.webkit.org/show_bug.cgi?id=170386
1068
1069         Reviewed by Tim Horton.
1070
1071         * English.lproj/Localizable.strings: Updated using update-webkit-localizable-strings.
1072
1073         * platform/LocalizedStrings.cpp:
1074         (WebCore::mediaElementLoadingStateText): Changed "Loading..." to "Loading…".
1075
1076         * platform/cocoa/LocalizedStringsCocoa.mm:
1077         (WebCore::contextMenuItemTagStyles): Changed "Styles..." to "Styles…".
1078
1079 2017-04-01  Dan Bernstein  <mitz@apple.com>
1080
1081         Localizable strings files are out of date
1082         https://bugs.webkit.org/show_bug.cgi?id=170383
1083
1084         Reviewed by Tim Horton.
1085
1086         Ran update-webkit-localizable-strings.
1087
1088         * English.lproj/Localizable.strings:
1089
1090 2017-04-01  Dan Bernstein  <mitz@apple.com>
1091
1092         [Xcode] In engineering builds, linker warns about libwebrtc.dylib’s install name being invalid
1093         https://bugs.webkit.org/show_bug.cgi?id=170385
1094
1095         Reviewed by Tim Horton.
1096
1097         * Configurations/DebugRelease.xcconfig: Set WK_RELOCATABLE_FRAMEWORKS to YES like we do
1098           in some other projects’ DebugRelease.xcconfig. Engineering builds are always relocatable.
1099         * Configurations/WebCore.xcconfig: When WebCore is relocatable, tell the linker that it’s
1100           not going to be in the shared cache, even if its install name implies that it might be.
1101
1102 2017-04-01  Alexey Proskuryakov  <ap@apple.com>
1103
1104         Rolling back http://trac.webkit.org/r214663 - memory corruption
1105
1106         * Modules/streams/ReadableByteStreamInternals.js:
1107         (cloneArrayBuffer):
1108         * bindings/js/JSDOMGlobalObject.cpp:
1109         (WebCore::JSDOMGlobalObject::addBuiltinGlobals):
1110         * bindings/js/StructuredClone.cpp:
1111         (WebCore::structuredCloneArrayBuffer):
1112         (WebCore::cloneArrayBufferImpl): Deleted.
1113         (WebCore::cloneArrayBuffer): Deleted.
1114         * bindings/js/StructuredClone.h:
1115         * bindings/js/WebCoreBuiltinNames.h:
1116         * testing/Internals.cpp:
1117         (WebCore::markerTypeFrom):
1118         (WebCore::Internals::resetToConsistentState):
1119         (WebCore::Internals::isLoadingFromMemoryCache):
1120         (WebCore::Internals::setImageFrameDecodingDuration):
1121         (WebCore::deferredStyleRulesCountForList):
1122         (WebCore::deferredGroupRulesCountForList):
1123         (WebCore::deferredKeyframesRulesCountForList):
1124         (WebCore::Internals::eventThrottlingBehaviorOverride):
1125         (WebCore::Internals::enableMockSpeechSynthesizer):
1126         (WebCore::Internals::rangeForDictionaryLookupAtLocation):
1127         (WebCore::Internals::nodesFromRect):
1128         (WebCore::Internals::layerIDForElement):
1129         (WebCore::Internals::setElementUsesDisplayListDrawing):
1130         (WebCore::Internals::setElementTracksDisplayListReplay):
1131         (WebCore::Internals::styleRecalcCount):
1132         (WebCore::Internals::compositingUpdateCount):
1133         (WebCore::Internals::setCaptionDisplayMode):
1134         (WebCore::Internals::endMediaSessionInterruption):
1135         (WebCore::Internals::postRemoteControlCommand):
1136         (WebCore::appendOffsets):
1137         (WebCore::Internals::scrollSnapOffsets):
1138         (WebCore::Internals::setShowAllPlugins):
1139         (WebCore::Internals::cloneArrayBuffer): Deleted.
1140         * testing/Internals.h:
1141         * testing/Internals.idl:
1142
1143 2017-03-31  Zalan Bujtas  <zalan@apple.com>
1144
1145         <table>: Including <caption>, <thead> or <tbody> causes clipping across page breaks
1146         https://bugs.webkit.org/show_bug.cgi?id=170348
1147         <rdar://problem/24727151>
1148
1149         Reviewed by David Hyatt.
1150
1151         1. In RenderFlowThread::offsetFromLogicalTopOfFirstRegion() we need to take table section offset into account (they are skipped
1152         during the containing block traversal).
1153         2. Trigger paginated relayout when body is moved vertically due to caption/thead etc.
1154
1155         Test: fast/multicol/table-section-page-break.html
1156
1157         * rendering/RenderFlowThread.cpp:
1158         (WebCore::RenderFlowThread::offsetFromLogicalTopOfFirstRegion):
1159         * rendering/RenderTable.cpp:
1160         (WebCore::RenderTable::layout):
1161
1162 2017-03-31  Simon Fraser  <simon.fraser@apple.com>
1163
1164         Rename DOMWindow's m_touchEventListenerCount to m_touchAndGestureEventListenerCount
1165         https://bugs.webkit.org/show_bug.cgi?id=170371
1166
1167         Reviewed by Tim Horton.
1168
1169         This count tracks touch and gesture event listeners, so name it appropriately.
1170
1171         * page/DOMWindow.cpp:
1172         (WebCore::DOMWindow::addEventListener):
1173         (WebCore::DOMWindow::removeEventListener):
1174         (WebCore::DOMWindow::removeAllEventListeners):
1175         * page/DOMWindow.h:
1176
1177 2017-03-31  Simon Fraser  <simon.fraser@apple.com>
1178
1179         When destroying a Node, assert that it's been removed from all the touch handler maps
1180         https://bugs.webkit.org/show_bug.cgi?id=170363
1181         rdar://problem/31377469
1182
1183         Reviewed by Tim Horton.
1184
1185         Assert that the Node has been removed from the touch handler maps in all documents on destruction.
1186
1187         * dom/Document.h:
1188         (WebCore::Document::hasTouchEventHandlers):
1189         (WebCore::Document::touchEventTargetsContain):
1190         * dom/Node.cpp:
1191         (WebCore::Node::~Node):
1192
1193 2017-03-31  Alexey Proskuryakov  <ap@apple.com>
1194
1195         Rolling back https://trac.webkit.org/r214689, as it caused many crashes.
1196
1197         Was:
1198         Fix memory leak in CreateSessionDescriptionObserver::OnSuccess
1199         https://bugs.webkit.org/show_bug.cgi?id=170357
1200
1201         * Modules/mediastream/libwebrtc/LibWebRTCMediaEndpoint.cpp:
1202         (WebCore::LibWebRTCMediaEndpoint::createSessionDescriptionSucceeded):
1203
1204 2017-03-31  Youenn Fablet  <youenn@apple.com>
1205
1206         Fix memory leak in RealtimeVideoIncomingSource
1207         https://bugs.webkit.org/show_bug.cgi?id=170356
1208
1209         Reviewed by Eric Carlson.
1210
1211         No change of behavior.
1212
1213         * platform/mediastream/mac/RealtimeIncomingVideoSource.cpp:
1214         (WebCore::RealtimeIncomingVideoSource::OnFrame): Adopting the newly created pointer.
1215
1216 2017-03-31  Andy Estes  <aestes@apple.com>
1217
1218         REGRESSION (r202472): Data Detection overwrites existing links in detected ranges
1219         https://bugs.webkit.org/show_bug.cgi?id=170365
1220         <rdar://problem/29205721>
1221
1222         Reviewed by Tim Horton.
1223
1224         r202472 changed the node traversal in searchForLinkRemovingExistingDDLinks() to only
1225         consider nodes that are descendants of startNode, but we need to traverse all nodes between
1226         startNode and endNode to find existing non-DD links.
1227
1228         As a result, we'd add a Data Detector link to the following snippet and make the original
1229         links un-clickable:
1230
1231             <a href='#'>tomorrow</a> <a href='#'>night</a>
1232
1233         Fix this by not specifying a stayWithin node when calling NodeTraversal::next(). The loop
1234         will terminate when we reach endNode.
1235
1236         Updated WebKit2.DataDetectionReferenceDate API test.
1237
1238         * editing/cocoa/DataDetection.mm:
1239         (WebCore::searchForLinkRemovingExistingDDLinks):
1240
1241 2017-03-31  Eric Carlson  <eric.carlson@apple.com>
1242
1243         Incoming video source doesn't propogate frame rotation
1244         https://bugs.webkit.org/show_bug.cgi?id=170364
1245
1246         Reviewed by Youenn Fablet.
1247
1248         No new tests, the mock video source doesn't support rotation. Test will be added when
1249         this is fixed in https://bugs.webkit.org/show_bug.cgi?id=169822. The changes were
1250         verified manually.
1251
1252         * platform/mediastream/mac/RealtimeIncomingVideoSource.cpp:
1253         (WebCore::RealtimeIncomingVideoSource::OnFrame): Convert frame rotation to sample
1254         orientation and swap width and height when necessary.
1255         (WebCore::RealtimeIncomingVideoSource::processNewSample):
1256         * platform/mediastream/mac/RealtimeIncomingVideoSource.h:
1257
1258 2017-03-31  Chris Dumez  <cdumez@apple.com>
1259
1260         Possible null dereference under SourceBuffer::sourceBufferPrivateDidReceiveSample()
1261         https://bugs.webkit.org/show_bug.cgi?id=159639
1262
1263         Reviewed by Eric Carlson.
1264
1265         Add a null check for trackBuffer.description before dereferencing as it seems
1266         it can be null.
1267
1268         * Modules/mediasource/SourceBuffer.cpp:
1269         (WebCore::SourceBuffer::sourceBufferPrivateDidReceiveSample):
1270
1271 2017-03-31  Youenn Fablet  <youenn@apple.com>
1272
1273         Fix memory leak in CreateSessionDescriptionObserver::OnSuccess
1274         https://bugs.webkit.org/show_bug.cgi?id=170357
1275
1276         Reviewed by Geoffrey Garen.
1277
1278         No change of behavior.
1279
1280         * Modules/mediastream/libwebrtc/LibWebRTCMediaEndpoint.cpp:
1281         (WebCore::LibWebRTCMediaEndpoint::createSessionDescriptionSucceeded): Adopting the raw pointer parameter.
1282
1283 2017-03-31  Fujii Hironori  <Hironori.Fujii@sony.com>
1284
1285         [WinCairo] WebCore::PlatformDisplay::terminateEGLDisplay causes a crash in libGLESv2.dll while processing atexit
1286         https://bugs.webkit.org/show_bug.cgi?id=170331
1287
1288         Reviewed by Michael Catanzaro.
1289
1290         WebCore::PlatformDisplay uses atexit to destruct EGL displays
1291         while exiting process. But, when the atexit will be processed,
1292         heap of libGLESv2.dll would be already destructed and causing a
1293         crash on Windows. Do not use atexit for Windows.
1294
1295         AppleWin port does not use PlatformDisplay. Thus, it does not have
1296         this bug.
1297
1298         * platform/graphics/PlatformDisplay.cpp:
1299         (WebCore::PlatformDisplay::initializeEGLDisplay): Do not use atexit for Windows.
1300         (WebCore::PlatformDisplay::shutDownEglDisplays): Added.
1301         * platform/graphics/PlatformDisplay.h: Added a declaration of shutDownEglDisplays.
1302
1303 2017-03-31  Yoav Weiss  <yoav@yoav.ws>
1304
1305         Remove PRELOAD_DEBUG related code.
1306         https://bugs.webkit.org/show_bug.cgi?id=170352
1307
1308         Reviewed by Youenn Fablet.
1309
1310         As the PRELOAD_DEBUG related code stopped building and it seems like no one noticed,
1311         it seems safe to assume that we can remove that code. This patch removes it.
1312
1313         No new tests as there's no functional change.
1314
1315         * loader/cache/CachedResourceLoader.cpp:
1316         (WebCore::CachedResourceLoader::preload):
1317         (WebCore::CachedResourceLoader::clearPreloads):
1318         (WebCore::CachedResourceLoader::printPreloadStats): Deleted.
1319
1320 2017-03-31  Brady Eidson  <beidson@apple.com>
1321
1322         Clean up the "StorageType" enum.
1323         https://bugs.webkit.org/show_bug.cgi?id=170349
1324
1325         Reviewed by Tim Horton.
1326
1327         - Make this `enum` into an `enum class`
1328         - Add a new type specific for "transient local storage"
1329
1330         No new tests (No behavior change).
1331
1332         * WebCore.xcodeproj/project.pbxproj:
1333
1334         * inspector/InspectorDOMStorageAgent.cpp:
1335         (WebCore::InspectorDOMStorageAgent::didDispatchDOMStorageEvent):
1336
1337         * inspector/InspectorInstrumentation.h:
1338
1339         * loader/EmptyClients.cpp:
1340
1341         * storage/Storage.cpp:
1342         (WebCore::Storage::isDisabledByPrivateBrowsing):
1343
1344         * storage/StorageArea.h:
1345         (): Deleted.
1346
1347         * storage/StorageEventDispatcher.cpp:
1348         (WebCore::StorageEventDispatcher::dispatchSessionStorageEventsToFrames):
1349         (WebCore::StorageEventDispatcher::dispatchLocalStorageEventsToFrames):
1350
1351         * storage/StorageType.h:
1352         (WebCore::isLocalStorage):
1353
1354 2017-03-31  Chris Dumez  <cdumez@apple.com>
1355
1356         Unreviewed, Mark "HTML interactive form validation" as done.
1357
1358         It is shipping in Safari 10.1.
1359
1360         * features.json:
1361
1362 2017-03-31  Csaba Osztrogonác  <ossy@webkit.org>
1363
1364         Mac cmake buildfix after r214666
1365         https://bugs.webkit.org/show_bug.cgi?id=170342
1366
1367         Unreviewed buildfix.
1368
1369         * PlatformMac.cmake:
1370         * testing/MockPreviewLoaderClient.h:
1371
1372 2017-03-31  Sam Weinig  <sam@webkit.org>
1373
1374         Remove unneeded custom constructors include.
1375
1376         * WebCore.xcodeproj/project.pbxproj:
1377         * bindings/js/ios: Removed.
1378         * bindings/js/ios/TouchConstructors.cpp: Removed.
1379
1380 2017-03-31  John Wilander  <wilander@apple.com>
1381
1382         Resource Load Statistics: Check if the store exists before clearing it
1383         https://bugs.webkit.org/show_bug.cgi?id=170324
1384         <rdar://problem/31258505>
1385
1386         Reviewed by Brent Fulgham.
1387
1388         No new tests. Added a null check.
1389
1390         * loader/ResourceLoadObserver.cpp:
1391         (WebCore::ResourceLoadObserver::clearInMemoryAndPersistentStore):
1392             Added a null check.
1393
1394 2017-03-31  Romain Bellessort  <romain.bellessort@crf.canon.fr>
1395
1396         [Readable Streams API] Implement cloneArrayBuffer in WebCore
1397         https://bugs.webkit.org/show_bug.cgi?id=170008
1398
1399         Reviewed by Youenn Fablet.
1400
1401         Implemented cloneArrayBuffer based on existing structuredCloneArrayBuffer
1402         implementation. The code has been factorized so that both cloneArrayBuffer
1403         and structuredCloneArrayBuffer rely on the same code (which is basically
1404         the previous implementation of structuredCloneArrayBuffer + the ability
1405         to clone only a part of considered buffer).
1406
1407         Added test to check cloneArrayBuffer behaviour.
1408
1409         * Modules/streams/ReadableByteStreamInternals.js: Deleted cloneArrayBuffer JS implementation.
1410         * bindings/js/JSDOMGlobalObject.cpp:
1411         (WebCore::JSDOMGlobalObject::addBuiltinGlobals): Add cloneArrayBuffer private declaration.
1412         * bindings/js/StructuredClone.cpp:
1413         (WebCore::cloneArrayBufferImpl): Added (mostly based on previous structuredCloneArrayBuffer).
1414         (WebCore::cloneArrayBuffer): Added.
1415         (WebCore::structuredCloneArrayBuffer): Updated.
1416         * bindings/js/StructuredClone.h: Added cloneArrayBuffer declaration.
1417         * bindings/js/WebCoreBuiltinNames.h: Added cloneArrayBuffer declaration.
1418         * testing/Internals.cpp: Added support for testing cloneArrayBuffer.
1419         * testing/Internals.h: Added support for testing cloneArrayBuffer.
1420         * testing/Internals.idl: Added support for testing cloneArrayBuffer.
1421
1422 2017-03-31  Antoine Quint  <graouts@apple.com>
1423
1424         [mac-wk1] LayoutTest media/modern-media-controls/airplay-button/airplay-button.html is a flaky timeout
1425         https://bugs.webkit.org/show_bug.cgi?id=168409
1426         <rdar://problem/30799198>
1427
1428         Unreviewed. Yet more logging to determine under what circumstance ScriptedAnimationController gets suspended.
1429
1430         * dom/Document.cpp:
1431         (WebCore::Document::requestAnimationFrame):
1432
1433 2017-03-30  Zan Dobersek  <zdobersek@igalia.com>
1434
1435         Unreviewed GTK+ build fix. Add missing ANGLE build targets
1436         to the build.
1437
1438         * CMakeLists.txt:
1439
1440 2017-03-30  Simon Fraser  <simon.fraser@apple.com>
1441
1442         Ensure that Node::willBeDeletedFrom() always removes touch event handlers from the document
1443         https://bugs.webkit.org/show_bug.cgi?id=170323
1444         rdar://problem/23647630
1445
1446         Reviewed by Chris Dumez.
1447
1448         There are two instances where nodes are registered as touch event handlers without
1449         having normal touch event listeners: slider thumb elements, and elements with overflow scrolling,
1450         on iOS.
1451
1452         For such nodes, hasEventTargetData() will be false, but we want to ensure
1453         that they are removed from the Document's touchEventHandler set, so move the
1454         call to document.removeTouchEventHandler() outside of the conditional block.
1455
1456         This should be cheap in most cases when the touchEventHandler is empty.
1457
1458         * dom/Node.cpp:
1459         (WebCore::Node::willBeDeletedFrom):
1460
1461 2017-03-30  Simon Fraser  <simon.fraser@apple.com>
1462
1463         Minor cleanup checking for gesture event names
1464         https://bugs.webkit.org/show_bug.cgi?id=170319
1465
1466         Reviewed by Tim Horton.
1467
1468         Just use isGestureEventType() in a couple of places.
1469
1470         * dom/Node.cpp:
1471         (WebCore::tryAddEventListener):
1472         (WebCore::tryRemoveEventListener):
1473
1474 2017-03-30  Simon Fraser  <simon.fraser@apple.com>
1475
1476         Rename a touch event function, and new touch region test results
1477         https://bugs.webkit.org/show_bug.cgi?id=170309
1478         rdar://problem/31329520
1479
1480         Reviewed by Chris Dumez.
1481
1482         Adapt to a naming change in WebKitAdditions.
1483
1484         * dom/Document.cpp:
1485         (WebCore::Document::removeAllEventListeners):
1486         * page/FrameView.cpp:
1487         (WebCore::FrameView::layout):
1488         * rendering/RenderElement.cpp:
1489         (WebCore::RenderElement::styleWillChange):
1490         * rendering/RenderLayer.cpp:
1491         (WebCore::RenderLayer::scrollTo):
1492         (WebCore::RenderLayer::calculateClipRects):
1493
1494 2017-03-30  Matt Rajca  <mrajca@apple.com>
1495
1496         YouTube sometimes does not respect "user gesture" restriction for videos.
1497         https://bugs.webkit.org/show_bug.cgi?id=170297
1498
1499         I discovered a code path that does not honor the "user gesture" requirement and playback is able to begin
1500         even though we have a restriction in place. When using Media Source Extensions, which YouTube does, we transition
1501         from a "Have Metadata" to a "Future Data" state that causes playback to begin, however, we never check
1502         if we have a playback restriction in place.
1503
1504         Reviewed by Eric Carlson.
1505
1506         * html/HTMLMediaElement.cpp:
1507         (WebCore::HTMLMediaElement::setReadyState):
1508
1509 2017-03-30  Said Abou-Hallawa  <sabouhallawa@apple.com>
1510
1511         REGRESSION (r213764): Background image from sprite sheet incorrectly scaled
1512         https://bugs.webkit.org/show_bug.cgi?id=169547
1513
1514         Reviewed by Simon Fraser.
1515
1516         The bug happens when drawing only a rectangle of an image not the whole
1517         image. In BitmapImage::draw(), sizeForDrawing was calculated as the destRect
1518         scaled by the transformation which is applied to the GraphicsContext. Two
1519         problems with this approach. The first one is destRect can be only part of
1520         the image. The second problem is, the ratio destRect / srcRect is another
1521         scaling that needs to be considered.
1522
1523         To fix this issue, first the base size has to be size of the image and not
1524         destRect.size(). Secondly, we need to scale this base size with the context
1525         transformation multiplied by the ratio destRect / srcRect. This scaling is
1526         exactly the same scaling which is calculated in subsamplingScale(). Finally
1527         we use this scaled size as the sizeForDrawing to send to the ImageDecoder.
1528
1529         Test: fast/images/sprite-sheet-image-draw.html
1530
1531         * platform/graphics/BitmapImage.cpp:
1532         (WebCore::BitmapImage::draw): Fix the bug.
1533         (WebCore::BitmapImage::stopAnimation): Stops the async image decoding for animated images only.
1534         The decoding for large images will be stopped when BitmapImage::newFrameNativeImageAvailableAtIndex()
1535         is called and the decoding queue is idle.
1536         (WebCore::BitmapImage::newFrameNativeImageAvailableAtIndex): Add image logging.
1537         * platform/graphics/BitmapImage.h: Move sourceURL() to the Image class.
1538         * platform/graphics/GraphicsContext.cpp: Pass imagePaintingOptions.m_decodingMode to Image::drawTiled().
1539         (WebCore::GraphicsContext::drawTiledImage): Pass imagePaintingOptions.m_decodingMode) to Image::drawTiled().
1540         * platform/graphics/Image.cpp:
1541         (WebCore::Image::sourceURL): Moved from BitmapImage.
1542         (WebCore::Image::drawTiled): Add a DecodingMode argument instead of calling always with DecodingMode::Synchronous.
1543         * platform/graphics/Image.h:
1544         * platform/graphics/ImageFrameCache.cpp:
1545         (WebCore::ImageFrameCache::cacheAsyncFrameNativeImageAtIndex): Add image logging.
1546         (WebCore::ImageFrameCache::startAsyncDecodingQueue): Ditto,
1547         (WebCore::ImageFrameCache::requestFrameAsyncDecodingAtIndex): Ditto.
1548         (WebCore::ImageFrameCache::stopAsyncDecodingQueue): Ditto.
1549         (WebCore::ImageFrameCache::sourceURL): A helper function to avoid checking the value of m_image.
1550         * platform/graphics/ImageFrameCache.h:
1551         * platform/graphics/NativeImage.h: Rename subsamplingScale() to nativeImageDrawingScale() and return image scaling instead.
1552         * platform/graphics/cairo/NativeImageCairo.cpp:
1553         (WebCore::nativeImageDrawingScale): Ditto.
1554         (WebCore::subsamplingScale): Deleted.
1555         * platform/graphics/cg/NativeImageCG.cpp:
1556         (WebCore::nativeImageDrawingScale): Ditto.
1557         (WebCore::subsamplingScale): Deleted.
1558         * platform/graphics/win/NativeImageDirect2D.cpp:
1559         (WebCore::nativeImageDrawingScale): Ditto.
1560         (WebCore::subsamplingScale): Deleted.
1561
1562 2017-03-30  Matt Baker  <mattbaker@apple.com>
1563
1564         Web Inspector: Assertion failure in InspectorStyleProperty::setRawTextFromStyleDeclaration
1565         https://bugs.webkit.org/show_bug.cgi?id=170279
1566         <rdar://problem/30200492>
1567
1568         Reviewed by David Hyatt.
1569
1570         The SourceRange for a CSSPropertySourceData should be relative to the start
1571         of the declaration body, not the start of the StyleSheetHandler’s parsed
1572         text. This only affected the ranges of unparsed (parsedOK == false) properties
1573         lacking a trailing semi-colon.
1574
1575         This patch doesn't change the behavior of InspectorStyleSheet other than
1576         silencing an irksome assert, as String::substring does a safety check on
1577         the passed in length.
1578
1579         * inspector/InspectorStyleSheet.cpp:
1580         (WebCore::fixUnparsedProperties):
1581
1582 2017-03-30  Youenn Fablet  <youenn@apple.com> and Jon Lee <jonlee@apple.com>
1583
1584         Clean up RTCDataChannel
1585         https://bugs.webkit.org/show_bug.cgi?id=169732
1586
1587         Reviewed by Chris Dumez.
1588
1589         Test: webrtc/datachannel/datachannel-event.html
1590               webrtc/datachannel/bufferedAmountLowThreshold.html
1591
1592         Making RTCDataChannel interface closer to the spec updating implementation accordingly.
1593         See https://w3c.github.io/webrtc-pc/#rtcdatachannel.
1594         In particular adding RTCDataChannelEvent constructor, and missing bufferedAmount related attributes.
1595         Doing some additional cleaning refactoring.
1596
1597         Making bufferedAmountIsDecreasing take a bufferedAmount argument so that we get the actual value passed by
1598         libwebrtc without needing to get it from the libwebrtc network thread again.
1599         In the future, we should store the bufferedAmount value in RTCDataChannel and update its value on each libwebrtc
1600         OnBufferedAmountChange. Special treatment may be needed when the data channel is closed, in which case the bufferedAmount should just be
1601         updated to increase in the send method.
1602
1603         Added some FIXMEs as RTCDataChannel is not aligned with the spec related to send and bufferedAmount.
1604
1605         * Modules/mediastream/RTCDataChannel.cpp:
1606         (WebCore::RTCDataChannel::send):
1607         (WebCore::RTCDataChannel::close):
1608         (WebCore::RTCDataChannel::didChangeReadyState):
1609         (WebCore::RTCDataChannel::bufferedAmountIsDecreasing):
1610         * Modules/mediastream/RTCDataChannel.h:
1611         * Modules/mediastream/RTCDataChannel.idl:
1612         * Modules/mediastream/RTCDataChannelEvent.cpp:
1613         (WebCore::RTCDataChannelEvent::create):
1614         (WebCore::RTCDataChannelEvent::RTCDataChannelEvent):
1615         (WebCore::RTCDataChannelEvent::channel):
1616         * Modules/mediastream/RTCDataChannelEvent.h:
1617         * Modules/mediastream/RTCDataChannelEvent.idl:
1618         * Modules/mediastream/RTCPeerConnection.idl:
1619         * Modules/mediastream/libwebrtc/LibWebRTCDataChannelHandler.cpp:
1620         (WebCore::LibWebRTCDataChannelHandler::OnStateChange):
1621         * Modules/mediastream/libwebrtc/LibWebRTCMediaEndpoint.cpp:
1622         (WebCore::LibWebRTCMediaEndpoint::addDataChannel):
1623         * WebCore.xcodeproj/project.pbxproj:
1624         * dom/EventNames.h:
1625         * platform/mediastream/RTCDataChannelHandler.h:
1626         * platform/mediastream/RTCDataChannelHandlerClient.h:
1627         * platform/mediastream/RTCDataChannelState.h: Added.
1628         * platform/mediastream/RTCPeerConnectionHandlerClient.h:
1629         * platform/mock/RTCDataChannelHandlerMock.cpp:
1630         (WebCore::RTCDataChannelHandlerMock::setClient):
1631         (WebCore::RTCDataChannelHandlerMock::close):
1632         * platform/mock/RTCNotifiersMock.cpp:
1633         (WebCore::IceConnectionNotifier::IceConnectionNotifier):
1634         (WebCore::SignalingStateNotifier::SignalingStateNotifier):
1635         (WebCore::DataChannelStateNotifier::DataChannelStateNotifier):
1636         * platform/mock/RTCNotifiersMock.h:
1637
1638 2017-03-30  Javier Fernandez  <jfernandez@igalia.com>
1639
1640         [css-align] Adapt content-alignment properties to the new baseline syntax
1641         https://bugs.webkit.org/show_bug.cgi?id=170262
1642
1643         Reviewed by David Hyatt.
1644
1645         The baseline-position syntax has changed recently, so we need to update
1646         the CSS properties using the old syntax. This patch address only the
1647         content-alignment (align-content and justify-content).
1648
1649         I used this patch to adapt the implementation of the parsing logic for
1650         these properties to the new Blink's CSS Parsing Design.
1651
1652         The new baseline syntax is "[first | last ]? baseline" which implies
1653         modifying the parsing and computed value logic.
1654
1655         There are several layout tests affected by this change, so I'll update
1656         them accordingly.
1657
1658         No new tests, just added/modified some cases to the tests we
1659         already have using the new baseline values.
1660
1661         * css/CSSComputedStyleDeclaration.cpp:
1662         (WebCore::valueForContentPositionAndDistributionWithOverflowAlignment):
1663         * css/CSSContentDistributionValue.cpp:
1664         (WebCore::CSSContentDistributionValue::customCSSText):
1665         * css/CSSPrimitiveValueMappings.h:
1666         (WebCore::CSSPrimitiveValue::operator ItemPosition):
1667         (WebCore::CSSPrimitiveValue::operator ContentPosition):
1668         * css/CSSValueKeywords.in:
1669         * css/parser/CSSPropertyParser.cpp:
1670         (WebCore::isBaselineKeyword):
1671         (WebCore::consumeBaselineKeyword):
1672         (WebCore::consumeContentDistributionOverflowPosition):
1673         (WebCore::consumeSelfPositionOverflowPosition):
1674
1675 2017-03-30  James Craig  <jcraig@apple.com>
1676
1677         AX: Expose a new AXSubrole for explicit ARIA "group" role
1678         https://bugs.webkit.org/show_bug.cgi?id=169810
1679         <rdar://problem/31039693>
1680
1681         Reviewed by Chris Fleizach.
1682
1683         Split GroupRole into generics (GroupRole) and explicit groups
1684         (ApplicationGroupRole) so we can expose a subrole on the explicit
1685         groups. Account for the change in ARIA Tree and Menu hierachies.
1686         Update the computedRoleValue for WebKit Inspector usage, too.
1687
1688         Updated existing tests.
1689
1690         Test: accessibility/list-detection2.html:
1691         Test: accessibility/roles-computedRoleString.html:
1692         Test: inspector/dom/getAccessibilityPropertiesForNode.html:
1693         Test: inspector/dom/highlightFrame.html:
1694         Test: inspector/dom/highlightSelector.html:
1695
1696         * accessibility/AccessibilityNodeObject.cpp:
1697         (WebCore::AccessibilityNodeObject::helpText):
1698         (WebCore::AccessibilityNodeObject::hierarchicalLevel):
1699         (WebCore::AccessibilityNodeObject::remapAriaRoleDueToParent):
1700         * accessibility/AccessibilityObject.cpp:
1701         (WebCore::AccessibilityObject::accessibleNameDerivesFromContent):
1702         (WebCore::AccessibilityObject::ariaTreeItemContent):
1703         (WebCore::initializeRoleMap):
1704         (WebCore::AccessibilityObject::computedRoleString):
1705         * accessibility/AccessibilityObject.h:
1706         * accessibility/atk/WebKitAccessibleWrapperAtk.cpp:
1707         (atkRole):
1708         * accessibility/ios/WebAccessibilityObjectWrapperIOS.mm:
1709         (-[WebAccessibilityObjectWrapper determineIsAccessibilityElement]):
1710         * accessibility/mac/WebAccessibilityObjectWrapperMac.mm:
1711         (createAccessibilityRoleMap):
1712         (-[WebAccessibilityObjectWrapper subrole]):
1713
1714 2017-03-30  Chris Dumez  <cdumez@apple.com>
1715
1716         We are spending a lot of time bzero'ing AudioChannel buffers on uni-watch.com
1717         https://bugs.webkit.org/show_bug.cgi?id=170288
1718         <rdar://problem/31289132>
1719
1720         Reviewed by Eric Carlson.
1721
1722         We are spending a lot of time bzero'ing AudioChannel buffers on uni-watch.com due
1723         to doubleclick ads using WebAudio with a 0-gain GainNode. We should optimize the
1724         0-gain case as much as possible.
1725
1726         No new tests, only an optimization.
1727
1728         * Modules/webaudio/AudioNode.cpp:
1729         (WebCore::AudioNode::processIfNecessary):
1730         Drop call to unsilenceOutputs() before calling AudioNode::process(). The AudioChannel
1731         API already takes care of clearing its 'silent' flag whenever its buffer changes so
1732         there should be no need to explicitly clearing the 'silent' flag before calling
1733         process(). This was causing us to zero out buffers that were already filled with
1734         zeros because AudioChannel::zero() would no longer return early, not knowing the
1735         channel is already silent. This reduces the number of bzero() calls by ~3.5x on
1736         uni-watch.com (from 100 calls every ~20ms to 100 calls every ~70ms.
1737
1738         * Modules/webaudio/AudioNode.h:
1739         * platform/audio/mac/AudioDestinationMac.cpp:
1740         (WebCore::AudioDestinationMac::render):
1741         Avoid clamping the values in the channel buffer if the channel is silent since this
1742         will have no effect (given that the buffer only contains zeros).
1743
1744 2017-03-30  Eric Carlson  <eric.carlson@apple.com>
1745
1746         [Crash] WebCore::AudioBuffer::AudioBuffer don't checking illegal value
1747         https://bugs.webkit.org/show_bug.cgi?id=169956
1748
1749         Reviewed by Youenn Fablet.
1750
1751         Test: webaudio/audiobuffer-crash.html
1752
1753         * Modules/webaudio/AudioBuffer.cpp:
1754         (WebCore::AudioBuffer::AudioBuffer): Invalidate the object and return early if the channel
1755         array allocation fails.
1756         (WebCore::AudioBuffer::AudioBuffer): Ditto.
1757         (WebCore::AudioBuffer::invalidate): Invalidate the object.
1758         * Modules/webaudio/AudioBuffer.h:
1759
1760 2017-03-30  Antoine Quint  <graouts@apple.com>
1761
1762         [mac-wk1] LayoutTest media/modern-media-controls/airplay-button/airplay-button.html is a flaky timeout
1763         https://bugs.webkit.org/show_bug.cgi?id=168409
1764         <rdar://problem/30799198>
1765
1766         Unreview. Print the backtrace when we suspend scripted animations.
1767
1768         * page/Page.cpp:
1769         (WebCore::Page::suspendScriptedAnimations):
1770
1771 2017-03-30  Antoine Quint  <graouts@apple.com>
1772
1773         [mac-wk1] LayoutTest media/modern-media-controls/airplay-button/airplay-button.html is a flaky timeout
1774         https://bugs.webkit.org/show_bug.cgi?id=168409
1775         <rdar://problem/30799198>
1776
1777         Unreview. Pring the backtrace when we suspend ScriptedAnimationController.
1778
1779         * dom/ScriptedAnimationController.cpp:
1780         (WebCore::ScriptedAnimationController::suspend):
1781
1782 2017-03-30  Antoine Quint  <graouts@apple.com>
1783
1784         [Modern Media Controls] Invalid placard icon is not visible for short video
1785         https://bugs.webkit.org/show_bug.cgi?id=170277
1786         <rdar://problem/31327955>
1787
1788         Reviewed by Eric Carlson.
1789
1790         Make the minimum dimension needed to display a placard icon configurable and let the
1791         invalid placard use a 50pt dimension. We also make a drive-by fix to reset "text-align"
1792         for media controls as having a "text-align: center" inherited value would mess up
1793         positioning of the placard icon and make it flush to the right of the placard.
1794
1795         Test: media/modern-media-controls/invalid-placard/invalid-placard-constrained-metrics.html
1796
1797         * Modules/modern-media-controls/controls/invalid-placard.js:
1798         (InvalidPlacard):
1799         * Modules/modern-media-controls/controls/media-controls.css:
1800         (.media-controls):
1801         * Modules/modern-media-controls/controls/placard.js:
1802         (Placard.prototype.layout):
1803
1804 2017-03-30  Zan Dobersek  <zdobersek@igalia.com>
1805
1806         [GCrypt] Register missing algorithms in CryptoAlgorithmRegistry::platformRegisterAlgoritmhs()
1807         https://bugs.webkit.org/show_bug.cgi?id=170273
1808
1809         Reviewed by Michael Catanzaro.
1810
1811         * crypto/gcrypt/CryptoAlgorithmRegistryGCrypt.cpp:
1812         (WebCore::CryptoAlgorithmRegistry::platformRegisterAlgorithms):
1813         Add registerAlgorithm<> calls for AES_CFB, AES_GCM, ECDH and
1814         PBKDF2 algorithms.
1815
1816 2017-03-30  Brent Fulgham  <bfulgham@apple.com>
1817
1818         Remove unused MediaControlsApple implementation
1819         https://bugs.webkit.org/show_bug.cgi?id=170258
1820         <rdar://problem/31331056>
1821
1822         Reviewed by Eric Carlson.
1823
1824         Remove dead code related to old media controls. We switched to HTML5-based media controls
1825         several years ago.
1826
1827         No new tests. Removing this dead code should have no change in behavior.
1828
1829         * WebCore.xcodeproj/project.pbxproj: Remove unused files.
1830         * dom/EventListener.h: Remove unused event type.
1831         * html/shadow/MediaControlsApple.cpp: Removed.
1832         * html/shadow/MediaControlsApple.h: Removed.
1833
1834 2017-03-30  Fujii Hironori  <Hironori.Fujii@sony.com>
1835
1836         [WinCairo] ImageCairoWin.cpp: error C2660: 'WebCore::BitmapImage::draw': function does not take 6 arguments
1837         https://bugs.webkit.org/show_bug.cgi?id=170275
1838
1839         Reviewed by Said Abou-Hallawa.
1840
1841         Apply the similar change of ImageCGWin.cpp of r214450 to ImageCairoWin.cpp.
1842
1843         * platform/graphics/win/ImageCairoWin.cpp:
1844         (WebCore::BitmapImage::getHBITMAPOfSize): Pass DecodingMode::Synchronous to Image::draw().
1845         (WebCore::BitmapImage::drawFrameMatchingSourceSize): Ditto.
1846
1847 2017-03-27  Sergio Villar Senin  <svillar@igalia.com>
1848
1849         [css-grid] Clamp the number of autorepeat tracks
1850         https://bugs.webkit.org/show_bug.cgi?id=170120
1851
1852         Reviewed by Manuel Rego Casasnovas.
1853
1854         As suggested by the specs we do clamp the maximum number of tracks per grid in order to
1855         minimize potential OOM situations. However we were not considering the case of the recently
1856         added auto repeat syntax. Abnormally huge values for the width/height on the grid container
1857         could lead to a number of auto repeat tracks higher than the maximum.
1858
1859         A new API was added to Internals in order to test limits without having to create huge
1860         grids. This new API allows clients to set an arbitrary limit to the number of tracks. The
1861         addition of this new API forced us to add GridPosition.cpp to the project to define the
1862         global variable we use for testing. We took the chance to move part of the implementation
1863         from the header file to the source file.
1864
1865         Last but not least, several new ASSERTs were added to Grid.cpp implementation to ensure that
1866         we do not surpass the grid track limits.
1867
1868         Test: fast/css-grid-layout/grid-auto-repeat-huge-grid.html
1869
1870         * CMakeLists.txt:
1871         * WebCore.xcodeproj/project.pbxproj:
1872         * css/parser/CSSPropertyParser.cpp:
1873         (WebCore::consumeGridTrackRepeatFunction):
1874         * rendering/Grid.cpp:
1875         (WebCore::Grid::ensureGridSize): Added ASSERT.
1876         (WebCore::Grid::setSmallestTracksStart): Ditto.
1877         (WebCore::Grid::setAutoRepeatTracks): Ditto.
1878         (WebCore::Grid::setAutoRepeatEmptyColumns): Ditto.
1879         (WebCore::Grid::setAutoRepeatEmptyRows): Ditto.
1880         * rendering/RenderGrid.cpp:
1881         (WebCore::RenderGrid::clampAutoRepeatTracks): New method.
1882         (WebCore::RenderGrid::placeItemsOnGrid): Clamp the number of auto repeat tracks before
1883         passing them to the Grid.
1884         * rendering/RenderGrid.h:
1885         * rendering/style/GridArea.h:
1886         (WebCore::GridSpan::GridSpan):
1887         * rendering/style/GridPosition.cpp: Added.
1888         (WebCore::GridPosition::setExplicitPosition):
1889         (WebCore::GridPosition::setAutoPosition):
1890         (WebCore::GridPosition::setSpanPosition):
1891         (WebCore::GridPosition::setNamedGridArea):
1892         (WebCore::GridPosition::integerPosition):
1893         (WebCore::GridPosition::namedGridLine):
1894         (WebCore::GridPosition::spanPosition):
1895         (WebCore::GridPosition::operator==):
1896         * rendering/style/GridPosition.h:
1897         (WebCore::GridPosition::shouldBeResolvedAgainstOppositePosition):
1898         (WebCore::GridPosition::max):
1899         (WebCore::GridPosition::min):
1900         (WebCore::GridPosition::setMaxPositionForTesting):
1901         (WebCore::GridPosition::setExplicitPosition): Deleted.
1902         (WebCore::GridPosition::setAutoPosition): Deleted.
1903         (WebCore::GridPosition::setSpanPosition): Deleted.
1904         (WebCore::GridPosition::setNamedGridArea): Deleted.
1905         (WebCore::GridPosition::integerPosition): Deleted.
1906         (WebCore::GridPosition::namedGridLine): Deleted.
1907         (WebCore::GridPosition::spanPosition): Deleted.
1908         (WebCore::GridPosition::operator==): Deleted.
1909         * rendering/style/GridPositionsResolver.cpp:
1910         (WebCore::GridPositionsResolver::explicitGridColumnCount):
1911         (WebCore::GridPositionsResolver::explicitGridRowCount):
1912         * testing/Internals.cpp:
1913         (WebCore::Internals::setGridMaxTracksLimit):
1914         * testing/Internals.h:
1915         * testing/Internals.idl:
1916
1917 2017-03-29  Ryosuke Niwa  <rniwa@webkit.org>
1918
1919         Disconnecting a HTMLObjectElement does not always unload its content document
1920         https://bugs.webkit.org/show_bug.cgi?id=169606
1921
1922         Reviewed by Andy Estes.
1923
1924         When removing a node, we first disconnect all subframes then update the focused element as we remove each child.
1925         However, when the removed element is a focused object element with a content document, removeFocusedNodeOfSubtree
1926         can update the style tree synchronously inside Document::setFocusedElement, and reload the document.
1927
1928         Avoid this by instantiating a SubframeLoadingDisabler on the parent of the focused element.
1929
1930         Test: fast/dom/removing-focused-object-element.html
1931
1932         * dom/Document.cpp:
1933         (WebCore::Document::removeFocusedNodeOfSubtree):
1934
1935 2017-03-29  Myles C. Maxfield  <mmaxfield@apple.com>
1936
1937         Migrate to kCTFontCSSWidthAttribute
1938         https://bugs.webkit.org/show_bug.cgi?id=170265
1939
1940         Reviewed by Darin Adler.
1941
1942         Previously, we were mapping from Core Text widths to CSS widths in WebKit.
1943         However, on some OSes, Core Text can directly tell us what the CSS width
1944         value is.
1945
1946         No new tests because there is no behavior change.
1947
1948         * platform/graphics/cocoa/FontCacheCoreText.cpp:
1949         (WebCore::getCSSAttribute):
1950         (WebCore::capabilitiesForFontDescriptor):
1951         * platform/spi/cocoa/CoreTextSPI.h:
1952
1953 2017-03-28  Simon Fraser  <simon.fraser@apple.com>
1954
1955         Make it possible to dump touch event regions for testing
1956         https://bugs.webkit.org/show_bug.cgi?id=170209
1957         <rdar://problem/31309258>
1958
1959         Reviewed by Tim Horton.
1960
1961         Expose internals.touchEventRectsForEvent() and internals.passiveTouchEventListenerRects(), which
1962         fetch data via Page.
1963
1964         Because the regions are normally updated on a timer, we have to force an eager update via Document::updateTouchEventRegions().
1965
1966         Test: fast/events/touch/ios/touch-event-regions.html
1967
1968         * page/Page.cpp:
1969         (WebCore::Page::nonFastScrollableRects):
1970         (WebCore::Page::touchEventRectsForEvent):
1971         (WebCore::Page::passiveTouchEventListenerRects):
1972         * page/Page.h:
1973         * testing/Internals.cpp:
1974         (WebCore::Internals::touchEventRectsForEvent):
1975         (WebCore::Internals::passiveTouchEventListenerRects):
1976         * testing/Internals.h:
1977         * testing/Internals.idl:
1978
1979 2017-03-29  Zalan Bujtas  <zalan@apple.com>
1980
1981         RenderBlockFlow::addFloatsToNewParent should check if float is already added to the object list.
1982         https://bugs.webkit.org/show_bug.cgi?id=170259
1983         <rdar://problem/31300584>
1984
1985         Reviewed by Simon Fraser.
1986
1987         r210145 assumed that m_floatingObjects would simply ignore the floating box if it was already in the list.
1988
1989         Test: fast/block/float/placing-multiple-floats-crash.html
1990
1991         * rendering/RenderBlockFlow.cpp:
1992         (WebCore::RenderBlockFlow::addFloatsToNewParent):
1993
1994 2017-03-29  Myles C. Maxfield  <mmaxfield@apple.com>
1995
1996         Try to normalize variation ranges
1997         https://bugs.webkit.org/show_bug.cgi?id=170119
1998
1999         Unreviewed.
2000
2001         Addressing post-review comment.
2002
2003         * platform/graphics/cocoa/FontCacheCoreText.cpp:
2004         (WebCore::isGXVariableFont):
2005
2006 2017-03-29  Myles C. Maxfield  <mmaxfield@apple.com>
2007
2008         Try to normalize variation ranges
2009         https://bugs.webkit.org/show_bug.cgi?id=170119
2010
2011         Reviewed by Simon Fraser.
2012
2013         TrueType GX-style variation fonts use one particular scale for values on their
2014         weight/width/slope axes - usually the values lie between -1 and 1 on that scale.
2015         However, OpenType 1.8-style fonts use the CSS scale for values on these axes.
2016         For the purposes of font selection, these values need to lie on the same scale.
2017         However, when font selection is completed and the variation values are actually
2018         being applied to the fonts, values which lie on the font's actual scale need to
2019         be applied. This patch adds normalize*() and denormalize*() functions to perform
2020         both of these operations.
2021
2022         The conversion itself between the two scales isn't an exact mapping. Mapping
2023         slope is just a linear relationship with 0deg <=> 0 and 20deg <=> 1 (as per the
2024         CSS Fonts spec). Mapping widths is similar, it uses a 2-component piecewise
2025         linear relationship which includes the values given in the Microsoft OpenType
2026         spec for the OS/2 table's usWidthClass field. Weights are more difficult, so I
2027         plotted the CSS weights and the GX-style weights for every style of San
2028         Francisco, saw that the relationship appears to be linear, and ran a linear
2029         regression to compute the line equation.
2030
2031         As for the actual discrimination of determining whether a font is a GX-style
2032         font or not, we can use the presence of the 'STAT' table. This table didn't
2033         exist when GX fonts were being created, and OpenType 1.8 variable fonts are
2034         required to have this table.
2035
2036         Facebook uses the string ".SFNSText" in their @font-face blocks. This font is
2037         a variation font, but uses the GX-style values. Facebook asks us to create
2038         this font with a weight of 700, and because the values in the font are around
2039         1.0, we were erroneously thinking that the font wasn't bold, so we were then
2040         applying synthetic bold. This was causing text on facebook to look fuzzy and
2041         ugly.
2042
2043         Test: fast/text/variations/font-selection-properties-expected.html
2044
2045         * platform/graphics/cocoa/FontCacheCoreText.cpp:
2046         (WebCore::isGXVariableFont):
2047         (WebCore::normalizeWeight):
2048         (WebCore::normalizeSlope):
2049         (WebCore::denormalizeWeight):
2050         (WebCore::denormalizeWidth):
2051         (WebCore::denormalizeSlope):
2052         (WebCore::normalizeWidth):
2053         (WebCore::preparePlatformFont): Instead of using FontSelectionValues for the
2054         intermediate values, we should use floats instead. This is because
2055         FontSelectionValues are fixed-point numbers with the denominator having 2 bits.
2056         When using this data type to represent values on the GX scale, which are usually
2057         between 0 and 1, you lose a lot of fidelity. Instead, our intermediate
2058         calculations should be done with floats, and converted to FontSelectionValues at
2059         the end when they are representative of values on the CSS scale.
2060         (WebCore::stretchFromCoreTextTraits):
2061         (WebCore::fontWeightFromCoreText):
2062         (WebCore::extractVariationBounds):
2063         (WebCore::variationCapabilitiesForFontDescriptor):
2064         (WebCore::capabilitiesForFontDescriptor):
2065
2066 2017-03-29  Saam Barati  <sbarati@apple.com>
2067
2068         LinkBuffer and ExecutableAllocator shouldn't have anything to do with VM
2069         https://bugs.webkit.org/show_bug.cgi?id=170210
2070
2071         Reviewed by Mark Lam.
2072
2073         * cssjit/SelectorCompiler.cpp:
2074         (WebCore::SelectorCompiler::compileSelector):
2075         (WebCore::SelectorCompiler::SelectorCodeGenerator::compile):
2076
2077 2017-03-29  Javier Fernandez  <jfernandez@igalia.com>
2078
2079         [css-align] Adapt self-alignment properties to the new baseline syntax
2080         https://bugs.webkit.org/show_bug.cgi?id=170235
2081
2082         Reviewed by David Hyatt.
2083
2084         The baseline-position syntax has changed recently, so we need to update
2085         the CSS properties using the old syntax. This patch address only the
2086         self-alignment (align-self and justify-self) and default-alignment
2087         (align-items and justify-items).
2088
2089         The content-distribution properties (align-content and justify-content)
2090         will be updated in a follow up patch.
2091
2092         The new baseline syntax is "[first | last ]? baseline" which implies
2093         modifying the parsing and computed value logic.
2094
2095         There are several layout tests affected by this change, so I'll update
2096         them accordingly.
2097
2098         No new tests, just added/modified some cases to the tests we already have using the new baseline values.
2099
2100         * css/CSSComputedStyleDeclaration.cpp:
2101         (WebCore::valueForItemPositionWithOverflowAlignment):
2102         * css/CSSValueKeywords.in:
2103         * css/StyleBuilderConverter.h:
2104         (WebCore::StyleBuilderConverter::convertSelfOrDefaultAlignmentData):
2105         * css/parser/CSSPropertyParser.cpp:
2106         (WebCore::consumeBaselineKeyword):
2107         (WebCore::consumeSelfPositionOverflowPosition):
2108
2109 2017-03-29  Chris Dumez  <cdumez@apple.com>
2110
2111         Animated SVG images are not paused in pages loaded in the background
2112         https://bugs.webkit.org/show_bug.cgi?id=170043
2113         <rdar://problem/31234412>
2114
2115         Reviewed by Simon Fraser.
2116
2117         Animated SVG images are not paused in pages loaded in the background. We rely
2118         on FrameView::isOffscreen() to stop images animations in background tab (See
2119         logic in RenderElement's shouldRepaintForImageAnimation()). This works fine
2120         if a tab is visble and then becomes hidden (i.e. by switching to another
2121         tab). However, in the case where the tab gets loaded while in the background
2122         (e.g. opening link in new background tab, or session restore), then the
2123         animations would not be paused, due to FrameView::isOffscreen() erroneously
2124         returning false in this case.
2125
2126         Normally, the following chain of events happens:
2127         - Page is visible, we construct a main frame and its FrameView for loading
2128           the page. When a FrameView is constructed, we call FrameView::show() to
2129           make it visible. Then, if the page becomes non-visible, we call
2130           Page::setIsVisibleInternal(false) which calls FrameView::hide(). After
2131           that, FrameView::isOffscreen() correctly returns true because we properly
2132           called FrameView::hide().
2133
2134         However, when doing a load while the Page is hidden, the following was
2135         happening:
2136         - Page is not visible, we call Page::setIsVisibleInternal(false) which tries
2137           to call FrameView::hide() for the main frame but it does not have a FrameView
2138           yet (because the load has not started). We start the load and end up creating
2139           a FrameView. The FrameView constructor was calling FrameView::show()
2140           unconditionally, thus making the FrameView think is visible, even though its
2141           page isn't. At this point, FrameView::isOffscreen() was returning false
2142           and animations would keep running, even though the page is not visible.
2143
2144         To address the issue, we now call FrameView::show() in FrameView::create() only
2145         if the Page is actually visible, instead of calling it unconditionally. If the
2146         page ever becomes visible, Page::setIsVisibleInternal(true) will be called and
2147         it will take care of calling FrameView::show() then.
2148
2149         Tests: svg/animations/animations-paused-in-background-page-iframe.html
2150                svg/animations/animations-paused-in-background-page.html
2151
2152         * page/FrameView.cpp:
2153         (WebCore::FrameView::create):
2154
2155 2017-03-29  Wenson Hsieh  <wenson_hsieh@apple.com>
2156
2157         Links with empty hrefs should not be drag sources
2158         https://bugs.webkit.org/show_bug.cgi?id=170241
2159         <rdar://problem/31305505>
2160
2161         Reviewed by Tim Horton.
2162
2163         The m_dragSouceAction member of DragController represents the drag source actions that are available to the
2164         document, rather than the available actions given the dragging element. Thus, it is not correct to only check
2165         that (m_dragSourceAction & DragSourceActionAttachment) before proceeding down the attachment dragging codepath.
2166         This should be additionally guarded with a check that the element being dragged is, in fact, an attachment
2167         element.
2168
2169         New API test (see Tools/ChangeLog).
2170
2171         * page/DragController.cpp:
2172         (WebCore::DragController::startDrag):
2173
2174 2017-03-29  Jeremy Jones  <jeremyj@apple.com>
2175
2176         WebVideoFullscreenInterfaceAVKit needs a strong self ref before dispatching to the main thread.
2177         https://bugs.webkit.org/show_bug.cgi?id=170129
2178
2179         Reviewed by David Kilzer.
2180
2181         No new tests becuase no new behavior.
2182
2183         Hold a strong reference to WebVideoFullscreenInterfaceAVKit when dispatching to the main thread.
2184         Make WebVideoFullscreenInterfaceAVKit safe to retain from non-main thread.
2185
2186         * platform/ios/WebVideoFullscreenInterfaceAVKit.h:
2187         * platform/ios/WebVideoFullscreenInterfaceAVKit.mm:
2188         (WebVideoFullscreenInterfaceAVKit::exitFullscreen):
2189
2190 2017-03-29  Zan Dobersek  <zdobersek@igalia.com>
2191
2192         [GCrypt] Add a Handle<> class to help with GCrypt object lifetime control
2193         https://bugs.webkit.org/show_bug.cgi?id=170238
2194
2195         Reviewed by Michael Catanzaro.
2196
2197         The platform-specific CryptoAlgorithmHMAC implementation is modified
2198         to showcase the GCrypt::Handle<> use. HandleDeleter<gcry_mac_hd_t>
2199         is added accordingly.
2200
2201         * crypto/gcrypt/CryptoAlgorithmHMACGCrypt.cpp:
2202         (WebCore::calculateSignature):
2203
2204 2017-03-29  Myles C. Maxfield  <mmaxfield@apple.com>
2205
2206         Variation fonts: Make sure that feature detection and preprocessor macros are right
2207         https://bugs.webkit.org/show_bug.cgi?id=169518
2208
2209         Reviewed by Simon Fraser.
2210
2211         When I added variable fonts support, I made all OSes parse the newly accepted values,
2212         instead of just the OSes which support variable fonts.
2213
2214         Test: fast/text/font-variations-feature-detection.html
2215
2216         * css/parser/CSSPropertyParser.cpp:
2217         (WebCore::consumeFontStretch):
2218         (WebCore::consumeFontStyle):
2219         * css/parser/CSSPropertyParserHelpers.cpp:
2220         (WebCore::CSSPropertyParserHelpers::divisibleBy100):
2221         (WebCore::CSSPropertyParserHelpers::consumeFontWeightNumber):
2222
2223 2017-03-29  Antoine Quint  <graouts@apple.com>
2224
2225         [Modern Media Controls] Controls bar may disappear while captions menu is visible (redux)
2226         https://bugs.webkit.org/show_bug.cgi?id=170239
2227         <rdar://problem/31320685>
2228
2229         Reviewed by Dean Jackson.
2230
2231         We did some work in webkit.org/b/168751 to prevent the controls bar from disappearing while the
2232         captions menu is visible. But there were two cases where the behavior was not as intended:
2233
2234             1. the controls bar would hide upon exiting the video.
2235             2. clicking on the controls bar while the caption panel is up would hide the controls bar
2236                as well as the captions panel.
2237
2238         Instead of determining that the "userInteractionEnabled" property being set to "false" is indicative
2239         of secondary UI, such as the tracks panel, being attached to the controls bar, we now have an
2240         explicit property to specify this. Now, when "hasSecondaryUIAttached" is "true", we prevent the
2241         controls bar from fading when exiting the media.
2242
2243         Additionally, when dismissing the tracks panel, we check whether a mouse event lead to this and check
2244         if the mouse was positioned over the controls bar. If that is the case, we no longer hide the controls
2245         bar and only dismiss the tracks panel.
2246
2247         Test: media/modern-media-controls/tracks-panel/tracks-panel-controls-bar-remains-visible-after-clicking-over-it.html
2248
2249         * Modules/modern-media-controls/controls/controls-bar.js:
2250         (ControlsBar.prototype.set userInteractionEnabled):
2251         (ControlsBar.prototype.handleEvent):
2252         (ControlsBar.prototype._autoHideTimerFired):
2253         * Modules/modern-media-controls/controls/macos-media-controls.js:
2254         (MacOSMediaControls.prototype.showTracksPanel):
2255         (MacOSMediaControls.prototype.hideTracksPanel):
2256
2257 2017-03-29  Antoine Quint  <graouts@apple.com>
2258
2259         [Modern Media Controls] Volume slider is initially empty
2260         https://bugs.webkit.org/show_bug.cgi?id=170237
2261         <rdar://problem/31319077>
2262
2263         Reviewed by Dean Jackson.
2264
2265         Fixing fallout from https://bugs.webkit.org/show_bug.cgi?id=167935 where we changed the behavior
2266         of layout() to happen prior to a commit. In this one instance, we were overriding layout() to run
2267         after a commit, so we now override commit(). Otherwise, the <canvas> that we draw the volume slider
2268         into would have its "width" and "height" properties set after we drew, which would clear the <canvas>.
2269
2270         * Modules/modern-media-controls/controls/slider.js:
2271         (Slider.prototype.commit):
2272         (Slider.prototype.layout): Deleted.
2273
2274 2017-03-29  Zan Dobersek  <zdobersek@igalia.com>
2275
2276         [GCrypt] Move over empty GnuTLS-based SUBTLE_CRYPTO implementation files
2277         https://bugs.webkit.org/show_bug.cgi?id=170232
2278
2279         Reviewed by Michael Catanzaro.
2280
2281         Migrate the GnuTLS SUBTLE_CRYPTO implementation files to the gcrypt/
2282         directory. The implementation files themselves are no-op, so this is
2283         a simple move-and-rename operation that will enable proceeding with
2284         the libgcrypt-based implementation of SUBTLE_CRYPTO functionality.
2285
2286         No change in behavior. The SUBTLE_CRYPTO feature should build as it
2287         did before, and the implementations are empty anyway.
2288
2289         * PlatformGTK.cmake:
2290         * crypto/gcrypt/CryptoAlgorithmAES_CBCGCrypt.cpp: Renamed from Source/WebCore/crypto/gnutls/CryptoAlgorithmAES_CBCGnuTLS.cpp.
2291         * crypto/gcrypt/CryptoAlgorithmAES_CFBGCrypt.cpp: Renamed from Source/WebCore/crypto/gnutls/CryptoAlgorithmAES_CFBGnuTLS.cpp.
2292         * crypto/gcrypt/CryptoAlgorithmAES_GCMGCrypt.cpp: Renamed from Source/WebCore/crypto/gnutls/CryptoAlgorithmAES_GCMGnuTLS.cpp.
2293         * crypto/gcrypt/CryptoAlgorithmAES_KWGCrypt.cpp: Renamed from Source/WebCore/crypto/gnutls/CryptoAlgorithmAES_KWGnuTLS.cpp.
2294         * crypto/gcrypt/CryptoAlgorithmECDHGCrypt.cpp: Renamed from Source/WebCore/crypto/gnutls/CryptoAlgorithmECDHGnuTLS.cpp.
2295         * crypto/gcrypt/CryptoAlgorithmPBKDF2GCrypt.cpp: Renamed from Source/WebCore/crypto/gnutls/CryptoAlgorithmPBKDF2GnuTLS.cpp.
2296         * crypto/gcrypt/CryptoAlgorithmRSAES_PKCS1_v1_5GCrypt.cpp: Renamed from Source/WebCore/crypto/gnutls/CryptoAlgorithmRSAES_PKCS1_v1_5GnuTLS.cpp.
2297         * crypto/gcrypt/CryptoAlgorithmRSASSA_PKCS1_v1_5GCrypt.cpp: This one already exists.
2298         * crypto/gcrypt/CryptoAlgorithmRSA_OAEPGCrypt.cpp: Renamed from Source/WebCore/crypto/gnutls/CryptoAlgorithmRSA_OAEPGnuTLS.cpp.
2299         * crypto/gcrypt/CryptoAlgorithmRegistryGCrypt.cpp: Renamed from Source/WebCore/crypto/gnutls/CryptoAlgorithmRegistryGnuTLS.cpp.
2300         * crypto/gcrypt/CryptoKeyECGCrypt.cpp: Renamed from Source/WebCore/crypto/gnutls/CryptoKeyECGnuTLS.cpp.
2301         * crypto/gcrypt/CryptoKeyRSAGCrypt.cpp: Renamed from Source/WebCore/crypto/gnutls/CryptoKeyRSAGnuTLS.cpp.
2302         * crypto/gcrypt/SerializedCryptoKeyWrapGCrypt.cpp: Renamed from Source/WebCore/crypto/gnutls/SerializedCryptoKeyWrapGnuTLS.cpp.
2303         * crypto/gnutls/CryptoAlgorithmRSASSA_PKCS1_v1_5GnuTLS.cpp: Removed.
2304
2305 2017-03-29  Zan Dobersek  <zdobersek@igalia.com>
2306
2307         [GnuTLS] Remove unused CryptoDigestGnuTLS, CryptoAlgorithmHMACGnuTLS implementation files
2308         https://bugs.webkit.org/show_bug.cgi?id=170231
2309
2310         Reviewed by Michael Catanzaro.
2311
2312         * crypto/gnutls/CryptoAlgorithmHMACGnuTLS.cpp: Removed. The GCrypt counterpart
2313         is already in use, and this file wasn't even being built.
2314
2315 2017-03-29  Youenn Fablet  <youenn@apple.com>
2316
2317         Move DTMF WebRTC extension behind its own compile flag
2318         https://bugs.webkit.org/show_bug.cgi?id=170226
2319
2320         Reviewed by Eric Carlson.
2321
2322         Moving RTCDTMFSender and RTCDTMFToneChangeEvent behinf a WEB_RTC_DTMF compile flag.
2323         This compile flag is not set on any supported platform yet.
2324         Disabling related test and updated test expectations.
2325
2326         * Modules/mediastream/RTCDTMFSender.cpp:
2327         * Modules/mediastream/RTCDTMFSender.h:
2328         * Modules/mediastream/RTCDTMFSender.idl:
2329         * Modules/mediastream/RTCDTMFToneChangeEvent.cpp:
2330         * Modules/mediastream/RTCDTMFToneChangeEvent.h:
2331         * Modules/mediastream/RTCDTMFToneChangeEvent.idl:
2332         * dom/EventNames.in:
2333         * dom/EventTargetFactory.in:
2334         * platform/mediastream/RTCDTMFSenderHandler.h:
2335         * platform/mediastream/RTCDTMFSenderHandlerClient.h:
2336
2337 2017-03-29  Antoine Quint  <graouts@apple.com>
2338
2339         [mac-wk1] LayoutTest media/modern-media-controls/airplay-button/airplay-button.html is a flaky timeout
2340         https://bugs.webkit.org/show_bug.cgi?id=168409
2341         <rdar://problem/30799198>
2342
2343         Reviewed by Dean Jackson.
2344
2345         Add a new internals setting to opt into ScriptedAnimationController logging through DOM events dispatched
2346         through the document. This should allow this flaky test to get information as to why the frame isn't
2347         firing when it times out.
2348
2349         * dom/ScriptedAnimationController.cpp:
2350         (WebCore::ScriptedAnimationController::suspend):
2351         (WebCore::ScriptedAnimationController::resume):
2352         (WebCore::ScriptedAnimationController::addThrottlingReason):
2353         (WebCore::ScriptedAnimationController::removeThrottlingReason):
2354         (WebCore::ScriptedAnimationController::registerCallback):
2355         (WebCore::ScriptedAnimationController::cancelCallback):
2356         (WebCore::ScriptedAnimationController::serviceScriptedAnimations):
2357         (WebCore::ScriptedAnimationController::scheduleAnimation):
2358         (WebCore::ScriptedAnimationController::dispatchLoggingEventIfRequired):
2359         * dom/ScriptedAnimationController.h:
2360         * page/Settings.in:
2361         * testing/InternalSettings.cpp:
2362         (WebCore::InternalSettings::resetToConsistentState):
2363         (WebCore::InternalSettings::shouldDispatchRequestAnimationFrameEvents):
2364         (WebCore::InternalSettings::setShouldDispatchRequestAnimationFrameEvents):
2365         * testing/InternalSettings.h:
2366         * testing/InternalSettings.idl:
2367
2368 2017-03-28  Youenn Fablet  <youenn@apple.com>
2369
2370         [WebRTC] After r214441 addIceCandidate no longer accepts an RTCIceCandidateInit dictionary
2371         https://bugs.webkit.org/show_bug.cgi?id=170192
2372
2373         Reviewed by Jon Lee.
2374
2375         Covered by updated tests.
2376
2377         * Modules/mediastream/RTCIceCandidate.idl: Candidate is no longer a required field in RTCIceCandidateInit.
2378         * Modules/mediastream/RTCPeerConnectionInternals.js: Reworking defaultsToNull to handle both undefined and null cases.
2379
2380 2017-03-28  Youenn Fablet  <youenn@apple.com>
2381
2382         LibWebRTCProvider should allow setting encoder and decoder factories
2383         https://bugs.webkit.org/show_bug.cgi?id=170212
2384
2385         Reviewed by Eric Carlson.
2386
2387         No change of behavior.
2388         Adding the ability to set encoder/decoder libwebrtc factory getters.
2389         Setting default cocoa factory getters.
2390
2391         * platform/mediastream/libwebrtc/LibWebRTCProvider.cpp:
2392         (WebCore::staticFactoryAndThreads):
2393         (WebCore::initializePeerConnectionFactoryAndThreads):
2394         (WebCore::LibWebRTCProvider::setDecoderFactoryGetter):
2395         (WebCore::LibWebRTCProvider::setEncoderFactoryGetter):
2396         (WebCore::LibWebRTCProvider::setPeerConnectionFactory):
2397         * platform/mediastream/libwebrtc/LibWebRTCProvider.h:
2398
2399 2017-03-27  Brent Fulgham  <bfulgham@apple.com>
2400
2401         Only attach Attributes to a given element one time
2402         https://bugs.webkit.org/show_bug.cgi?id=170125
2403         <rdar://problem/31279676>
2404
2405         Reviewed by Chris Dumez.
2406
2407         Attach the attribute node to the Element before calling 'setAttributeInternal', since that method may cause
2408         arbitrary JavaScript events to fire.
2409
2410         Test: fast/dom/Attr/only-attach-attr-once.html
2411
2412         * dom/Element.cpp:
2413         (WebCore::Element::attachAttributeNodeIfNeeded): Added.
2414         (WebCore::Element::setAttributeNode): Use new method. Revise to attach attribute before calling 'setAttributeInternal'.
2415         (WebCore::Element::setAttributeNodeNS): Ditto.
2416         * dom/Element.h:
2417
2418 2017-03-28  Youenn Fablet  <youenn@apple.com>
2419
2420         Stop RTCDataChannel when closing page
2421         https://bugs.webkit.org/show_bug.cgi?id=170166
2422
2423         Reviewed by Eric Carlson.
2424
2425         Test: webrtc/datachannel/datachannel-gc.html
2426
2427         Making RTCDataChannel an ActiveDOMObject.
2428         Closing the data channel backend and freeing upon close and stop.
2429
2430         * Modules/mediastream/RTCDataChannel.cpp:
2431         (WebCore::RTCDataChannel::create):
2432         (WebCore::RTCDataChannel::RTCDataChannel):
2433         (WebCore::RTCDataChannel::close):
2434         (WebCore::RTCDataChannel::stop):
2435         * Modules/mediastream/RTCDataChannel.h:
2436         * Modules/mediastream/RTCDataChannel.idl:
2437         * Modules/mediastream/RTCPeerConnection.h:
2438
2439 2017-03-28  Myles C. Maxfield  <mmaxfield@apple.com>
2440
2441         Ranges for variation font properties are not enforced
2442         https://bugs.webkit.org/show_bug.cgi?id=169979
2443
2444         Reviewed by David Hyatt.
2445
2446         The spec specifies that:
2447         - Font weights less than or equal to 0, or greater than or equal to 1000 are parse errors
2448         - Font stretch values less than or equal to 0% are parse errors
2449         - Font style values less than or equal to -90deg or greater than or equal to 90deg are parse errors
2450
2451         Test: fast/text/variations/out-of-bounds-selection-properties.html
2452
2453         * css/parser/CSSPropertyParser.cpp:
2454         (WebCore::consumeFontWeightRange):
2455         (WebCore::fontStretchIsWithinRange):
2456         (WebCore::consumeFontStretch):
2457         (WebCore::consumeFontStretchRange):
2458         (WebCore::fontStyleIsWithinRange):
2459         (WebCore::consumeFontStyle):
2460         (WebCore::consumeFontStyleRange):
2461
2462 2017-03-28  Andy Estes  <aestes@apple.com>
2463
2464         [iOS] Crash in -[WebPreviewLoader failed] when running http/tests/multipart/policy-ignore-crash.php
2465         https://bugs.webkit.org/show_bug.cgi?id=170197
2466         <rdar://problem/30314067>
2467
2468         Reviewed by Brady Eidson.
2469
2470         If QuickLook conversion fails, we call ResourceLoader::didFail() with the NSError from
2471         QuickLook, which will call back into PreviewLoader::didFail(). We only care about network
2472         failures in PreviewLoader, not conversion failures, so check if
2473         m_finishedLoadingDataIntoConverter is set before continuing (like we do in
2474         PreviewLoader::didFinishLoading()).
2475
2476         Fixes crash in http/tests/multipart/policy-ignore-crash.php.
2477
2478         * loader/ios/PreviewLoader.mm:
2479         (WebCore::PreviewLoader::didFail):
2480
2481 2017-03-28  Chris Dumez  <cdumez@apple.com>
2482
2483         Audio indicator is visible on uni-watch.com but there is no audible audio
2484         https://bugs.webkit.org/show_bug.cgi?id=170200
2485         <rdar://problem/31289132>
2486
2487         Reviewed by Eric Carlson.
2488
2489         Cherry-pick the following patch from Blink by <rtoy@chromium.org>:
2490         - https://chromium.googlesource.com/chromium/src.git/+/439de5bb2a31384666db1a0e2dadb2b97d2f3ce4
2491
2492         When the gain of a GainNode is 0 or 1, the operation of the node can
2493         be optimized.  When gain = 1, just copy the input to the output.  When
2494         gain = 0; just zero out the output.  Currently, the input is
2495         multiplied by the gain to produce the output.  This just optimizes the
2496         multiplication away for the two special cases.
2497
2498         Also, have the GainNode set the silence hint if the gain is 0.
2499
2500         And fix a bug in processIfNecessary when unsilenceOutputs was causing the
2501         silence hint to be cleared after the node's process method was finished
2502         and may have set the silence hint.  The processing should come after
2503         unsilenceOutputs to preserve any hints from processing the node.
2504
2505         * Modules/webaudio/AudioNode.cpp:
2506         (WebCore::AudioNode::processIfNecessary):
2507         * Modules/webaudio/GainNode.cpp:
2508         (WebCore::GainNode::process):
2509         * platform/audio/AudioBus.cpp:
2510         (WebCore::AudioBus::copyWithGainFrom):
2511
2512 2017-03-28  Chris Dumez  <cdumez@apple.com>
2513
2514         Animated SVG images are not paused when outside viewport
2515         https://bugs.webkit.org/show_bug.cgi?id=170155
2516         <rdar://problem/31288893>
2517
2518         Reviewed by Antti Koivisto.
2519
2520         Make sure animated SVG images get paused when outside the viewport,
2521         similarly to what was already done for animated GIF images. Also
2522         make sure they are paused when they no longer have any renderers
2523         using them.
2524
2525         Tests: svg/animations/animated-svg-image-outside-viewport-paused.html
2526                svg/animations/animated-svg-image-removed-from-document-paused.html
2527
2528         * loader/cache/CachedImage.cpp:
2529         (WebCore::CachedImage::didAddClient):
2530         Restart the animation whenever a new CachedImage client is added. This
2531         will cause us the re-evaluate if the animation should run. The animation
2532         will pause again if the new renderer is not inside the viewport.
2533
2534         (WebCore::CachedImage::animationAdvanced):
2535         Add a flag to newImageAnimationFrameAvailable() so that the renderers can
2536         let us know if we can pause the animation. Pause the animation if all no
2537         renderer requires it (i.e. they are all outside the viewport, or there
2538         are no renderers).
2539
2540         * loader/cache/CachedImageClient.h:
2541         (WebCore::CachedImageClient::newImageAnimationFrameAvailable):
2542         By default, the CachedImageClients allow pausing. Only renderer will
2543         potentially prevent pausing if they are inside the viewport.
2544
2545         * platform/graphics/BitmapImage.cpp:
2546         (WebCore::BitmapImage::isAnimating):
2547         * platform/graphics/BitmapImage.h:
2548         * platform/graphics/Image.h:
2549         (WebCore::Image::isAnimating):
2550         Add isAnimating() flag on Image for layout testing purposes.
2551
2552         * rendering/RenderElement.cpp:
2553         (WebCore::RenderElement::newImageAnimationFrameAvailable):
2554         Set canPause flag to true if the renderer is not inside the viewport.
2555
2556         (WebCore::RenderElement::repaintForPausedImageAnimationsIfNeeded):
2557         Call startAnimation() if the renderer is now visible to resume SVG
2558         animations. Repainting is enough for GIF animations but not for SVG
2559         animations, we have to explicitly resume them.
2560
2561         * rendering/RenderElement.h:
2562         * rendering/RenderView.cpp:
2563         (WebCore::RenderView::addRendererWithPausedImageAnimations):
2564         (WebCore::RenderView::removeRendererWithPausedImageAnimations):
2565         (WebCore::RenderView::resumePausedImageAnimationsIfNeeded):
2566         * rendering/RenderView.h:
2567         Store CachedImages with the renderers that have paused animations.
2568         This is required for SVG where we need to explicitly resume the
2569         animation on the CachedImage when the renderer becomes visible
2570         again. Having access to the Image will also allow us to do smarter
2571         visibility checks in RenderElement's shouldRepaintForImageAnimation(),
2572         in the future.
2573
2574         * svg/SVGSVGElement.cpp:
2575         (WebCore::SVGSVGElement::hasActiveAnimation):
2576         * svg/SVGSVGElement.h:
2577         Add hasActiveAnimation() method.
2578
2579         * svg/graphics/SVGImage.cpp:
2580         (WebCore::SVGImage::startAnimation):
2581         Check that animations are paused before starting them. This avoid
2582         jumping due to unnecessary calls to rootElement->setCurrentTime(0).
2583
2584         (WebCore::SVGImage::isAnimating):
2585         Add isAnimating() method for layout tests purposes.
2586
2587         * svg/graphics/SVGImage.h:
2588         * svg/graphics/SVGImageClients.h:
2589         Call animationAdvanced() on the observer instead of the generic
2590         changedInRect() when the SVGImage is animating. This way, we go
2591         through the same code path as GIF animations and we end up calling
2592         CachedImage::animationAdvanced() which calls newImageAnimationFrameAvailable()
2593         on RenderElement, which determines if the animation should keep
2594         running or not.
2595
2596         * testing/Internals.cpp:
2597         (WebCore::Internals::isImageAnimating):
2598         * testing/Internals.h:
2599         * testing/Internals.idl:
2600         Add layout testing infrastructure.
2601
2602 2017-03-28  Antti Koivisto  <antti@apple.com>
2603
2604         Missing render tree position invalidation when tearing down renderers for display:contents subtree
2605         https://bugs.webkit.org/show_bug.cgi?id=170199
2606         <rdar://problem/31260856>
2607
2608         Reviewed by Zalan Bujtas.
2609
2610         Test: fast/shadow-dom/slot-renderer-teardown.html
2611
2612         * style/RenderTreeUpdater.cpp:
2613         (WebCore::RenderTreeUpdater::updateElementRenderer):
2614
2615             Invalidate the render tree position in case we do a teardown for an element without renderer.
2616
2617 2017-03-28  Ryan Haddad  <ryanhaddad@apple.com>
2618
2619         Unreviewed, rolling out r214485.
2620
2621         This change caused LayoutTest crashes.
2622
2623         Reverted changeset:
2624
2625         "Stop RTCDataChannel when closing page"
2626         https://bugs.webkit.org/show_bug.cgi?id=170166
2627         http://trac.webkit.org/changeset/214485
2628
2629 2017-03-28  Anders Carlsson  <andersca@apple.com>
2630
2631         ApplePayShippingContactUpdate.idl shouldn't have status field
2632         https://bugs.webkit.org/show_bug.cgi?id=170202
2633         rdar://problem/31307106
2634
2635         Reviewed by Beth Dakin.
2636
2637         * Modules/applepay/ApplePaySession.cpp:
2638         (WebCore::convertAndValidate):
2639         If status isn't set, infer it based on whether there are errors present or not.
2640
2641         * Modules/applepay/ApplePayShippingContactUpdate.h:
2642         Make status optional here; it's still used by the old code path.
2643
2644         * Modules/applepay/ApplePayShippingContactUpdate.idl:
2645         Remove status here.
2646
2647 2017-03-28  Brian Burg  <bburg@apple.com>
2648
2649         Web Inspector: Add "Disable Caches" option that only applies to the inspected page while Web Inspector is open
2650         https://bugs.webkit.org/show_bug.cgi?id=169865
2651         <rdar://problem/31250573>
2652
2653         Reviewed by Joseph Pecoraro.
2654
2655         Rewrite the network agent's command for disabling resource caching to use Page::setResourceCachingEnabled.
2656         The old implementation was doing weird stuff like setting no-cache headers and evicting the
2657         contents of the memory cache, neither of which is correct. The new approach has no side effects
2658         on the network, disk, or memory cache so it can be turned on temporarily without causing problems.
2659
2660         New tests:
2661         - http/tests/inspector/network/set-resource-caching-disabled-disk-cache.html
2662         - http/tests/inspector/network/set-resource-caching-disabled-memory-cache.html
2663
2664         * inspector/InspectorNetworkAgent.h:
2665         * inspector/InspectorNetworkAgent.cpp:
2666         (WebCore::InspectorNetworkAgent::setCacheDisabled): Deleted.
2667         (WebCore::InspectorNetworkAgent::setResourceCachingDisabled): Added.
2668         Implement new command.
2669
2670         (WebCore::InspectorNetworkAgent::willSendRequest):
2671         (WebCore::InspectorNetworkAgent::mainFrameNavigated):
2672         Remove crufty attempts to break caches. I believe these are intended to defeat caching
2673         proxies and similar middlemen, but this is just as likely to cause unusual loading behavior.
2674
2675         * page/Page.h:
2676         (WebCore::Page::isResourceCachingDisabled):
2677         (WebCore::Page::setResourceCachingDisabledOverride):
2678         Add an override setting so that Web Inspector's override does not mess up the value
2679         of isResourceCachingDisabled that may have been set by a WebKit API client.
2680
2681 2017-03-28  Youenn Fablet  <youenn@apple.com>
2682
2683         Fix addIceCandidate after r214441
2684         https://bugs.webkit.org/show_bug.cgi?id=170146
2685
2686         Reviewed by Chris Dumez.
2687
2688         Covered by rebased test.
2689
2690         * Modules/mediastream/RTCPeerConnection.js:
2691         (addIceCandidate): Setting function length to 1 and throwing if no parameter is passed.
2692
2693 2017-03-28  Youenn Fablet  <youenn@apple.com>
2694
2695         Stop RTCDataChannel when closing page
2696         https://bugs.webkit.org/show_bug.cgi?id=170166
2697
2698         Reviewed by Eric Carlson.
2699
2700         Test: webrtc/datachannel/datachannel-gc.html
2701
2702         Making RTCDataChannel an ActiveDOMObject.
2703         Closing the data channel backend and freeing upon close and stop.
2704
2705         * Modules/mediastream/RTCDataChannel.cpp:
2706         (WebCore::RTCDataChannel::create):
2707         (WebCore::RTCDataChannel::RTCDataChannel):
2708         (WebCore::RTCDataChannel::close):
2709         (WebCore::RTCDataChannel::stop):
2710         * Modules/mediastream/RTCDataChannel.h:
2711         * Modules/mediastream/RTCDataChannel.idl:
2712         * Modules/mediastream/RTCPeerConnection.h:
2713
2714 2017-03-27  Simon Fraser  <simon.fraser@apple.com>
2715
2716         Enhance the touch region debug overlay to show regions for the different events
2717         https://bugs.webkit.org/show_bug.cgi?id=170162
2718
2719         Reviewed by Tim Horton.
2720
2721         Have NonFastScrollableRegionOverlay use a different color for each region in EventTrackingRegions,
2722         and to draw a legend showing what the colors mean.
2723
2724         On Mac, this overlay displays the non-fast scrollable region (which we don't keep separate from the wheel event
2725         region).
2726
2727         * page/DebugPageOverlays.cpp:
2728         (WebCore::NonFastScrollableRegionOverlay::updateRegion):
2729         (WebCore::touchEventRegionColors):
2730         (WebCore::drawRightAlignedText):
2731         (WebCore::NonFastScrollableRegionOverlay::drawRect):
2732         (WebCore::RegionOverlay::drawRect):
2733         (WebCore::RegionOverlay::drawRegion):
2734
2735 2017-03-27  Simon Fraser  <simon.fraser@apple.com>
2736
2737         Make sure the non-fast scrolling debug overlay is correctly updated
2738         https://bugs.webkit.org/show_bug.cgi?id=170142
2739
2740         Reviewed by Tim Horton.
2741
2742         AsyncScrollingCoordinator::frameViewEventTrackingRegionsChanged() is called on a timer
2743         from Document code, so the existing DebugPageOverlays::didLayout() call at the end
2744         of FrameView::layout() wasn't sufficient to keep the non-fast scrollable region up-to-date
2745         on iOS.
2746
2747         * page/scrolling/AsyncScrollingCoordinator.cpp:
2748         (WebCore::AsyncScrollingCoordinator::frameViewEventTrackingRegionsChanged):
2749
2750 2017-03-28  Antoine Quint  <graouts@apple.com>
2751
2752         [Modern Media Controls] AirPlay placard text looks bad on 1x displays
2753         https://bugs.webkit.org/show_bug.cgi?id=170183
2754         <rdar://problem/30663416>
2755
2756         Reviewed by Dean Jackson.
2757
2758         Use subpixel antialiasing for all text in modern media controls.
2759
2760         * Modules/modern-media-controls/controls/media-controls.css:
2761         (.media-controls-container,):
2762
2763 2017-03-28  Antoine Quint  <graouts@apple.com>
2764
2765         [Modern Media Controls] Improve appearance of tracks panel on macOS
2766         https://bugs.webkit.org/show_bug.cgi?id=168929
2767         <rdar://problem/30741589>
2768
2769         Reviewed by Eric Carlson.
2770
2771         We use a solid color for the focus state that matches the style used on macOS
2772         and blend the titles the same way we blend other non-solid labels in the controls bar.
2773
2774         * Modules/modern-media-controls/controls/tracks-panel.css:
2775         (.tracks-panel section):
2776         (.tracks-panel section > h3):
2777         (.tracks-panel section > ul > li:focus):
2778
2779 2017-03-28  Yoav Weiss  <yoav@yoav.ws>
2780
2781         Add a warning for unused link preloads.
2782         https://bugs.webkit.org/show_bug.cgi?id=165670
2783
2784         Reviewed by Youenn Fablet.
2785
2786         Tests: http/tests/preload/single_download_preload_headers_charset.php
2787                http/tests/preload/unused_preload_warning.html
2788
2789         * dom/Document.cpp:
2790         (WebCore::Document::prepareForDestruction): Stop the timer once the document is destructed.
2791         * loader/LinkPreloadResourceClients.h: Add shouldMarkAsReferenced overides for the LinkPreloadResourceClient classes.
2792         * loader/cache/CachedResource.cpp:
2793         (WebCore::CachedResource::addClientToSet): Make sure LinkPreloadResourceClients don't set resource to be referenced.
2794         * loader/cache/CachedResourceClient.h:
2795         (WebCore::CachedResourceClient::shouldMarkAsReferenced): Make sure that ResourceClients mark preloads as referenced by default.
2796         * loader/cache/CachedResourceLoader.cpp:
2797         (WebCore::CachedResourceLoader::CachedResourceLoader): Initialize timer.
2798         (WebCore::CachedResourceLoader::~CachedResourceLoader): Stop timer.
2799         (WebCore::CachedResourceLoader::documentDidFinishLoadEvent): Trigger a timer if preloads weren't cleared at load time.
2800         (WebCore::CachedResourceLoader::stopUnusedPreloadsTimer): Stop the timer.
2801         (WebCore::CachedResourceLoader::warnUnusedPreloads): Iterate over m_preloads and issue a warning for non-referenced preloads.
2802         * loader/cache/CachedResourceLoader.h:
2803         * page/DOMWindow.cpp:
2804         (WebCore::DOMWindow::willDetachDocumentFromFrame): Clear Resource Timing buffer when document detaches, to avoid test flakiness.
2805
2806 2017-03-28  Antoine Quint  <graouts@apple.com>
2807
2808         REGRESSION: Double-clicking the captions button while the captions popover is open prevents the popover from being opened again
2809         https://bugs.webkit.org/show_bug.cgi?id=170171
2810         <rdar://problem/31095500>
2811
2812         Reviewed by Dean Jackson.
2813
2814         We used to consider that the tracks panel was presented as long as it had a parent, but since we were using for the animated
2815         transition to complete before actually removing the panel from the node hierarchy, consecutive calls to hide then present
2816         would fail to present the panel. We now use a private instance variable to track the presented state which is set immediately
2817         as presentInParent() and hide() are called.
2818
2819         Test: media/modern-media-controls/tracks-support/tracks-support-show-panel-then-double-click-on-tracks-button.html
2820
2821         * Modules/modern-media-controls/controls/tracks-panel.js:
2822         (TracksPanel.prototype.get presented):
2823         (TracksPanel.prototype.presentInParent):
2824         (TracksPanel.prototype.hide):
2825
2826 2017-03-28  Aaron Chu  <aaron_chu@apple.com>
2827
2828         AX: Media controls should be able to be re-activated after faded away
2829         https://bugs.webkit.org/show_bug.cgi?id=170048
2830         <rdar://problem/30157179>
2831
2832         Reviewed by Antoine Quint.
2833
2834         Added a "foucsin" listener for the controls bar so that when an element
2835         within fires a "focusin" event, the controls bar reappears if it is faded.
2836
2837         Test: media/modern-media-controls/media-controls/media-controls-appear-when-focus.html
2838
2839         * Modules/modern-media-controls/controls/controls-bar.js:
2840         (ControlsBar.prototype.handleEvent):
2841
2842 2017-03-28  Antoine Quint  <graouts@apple.com>
2843
2844         [Modern Media Controls] Improve media documents across macOS, iPhone and iPad
2845         https://bugs.webkit.org/show_bug.cgi?id=169145
2846         <rdar://problem/17048858>
2847
2848         Reviewed by Dean Jackson.
2849
2850         There were a variety of issues with media documents, some longstanding, and some specifically
2851         about modern media controls.
2852
2853         One issue was that fullscreen and picture-in-picture buttons would show for audio media documents,
2854         due to using a <video> element to load the audio file. We now have additional logic in MediaController
2855         to identify if the loaded media is really an audio file, and using this information to hide the
2856         fullscreen and picture-in-picture buttons.
2857
2858         Another issue was that we would inject style in MediaDocument.cpp that was specific to modern media
2859         controls when we could have the modern-media-controls module injected CSS handle this styling. We now
2860         use the injected style in the shadow root to size media documents based on the device characteristics
2861         and ensuring that page styles are overridden.
2862
2863         We also simplify how MediaDocument.cpp sets the source of the media element to simply use the "src"
2864         attribute and not a <source> element.
2865
2866         Finally, we introduce a MediaDocumentController class that is instantiated when we're dealing with
2867         a media document to hide the controls while we determine the type of media we're loading (audio vs.
2868         video) in order to apply the appropriate styling without flashes.
2869
2870         As a result of the new styles applied by the modern-media-controls module, media documents have a
2871         similar behavior on macOS and iPad, where we only enforce a min-width for video allowing them
2872         to play at their natural size otherwise, and enforcing a fixed width for audio. On iPhone however,
2873         we want to always play the media at full width, with some padding in the case of audio.
2874
2875         Tests: media/modern-media-controls/fullscreen-support/fullscreen-support-disabled-video-with-audio-tracks-only.html
2876                media/modern-media-controls/media-documents/ipad/media-document-audio-ios-sizing.html
2877                media/modern-media-controls/media-documents/ipad/media-document-video-ios-sizing.html
2878                media/modern-media-controls/media-documents/media-document-audio-ios-sizing.html
2879                media/modern-media-controls/media-documents/media-document-audio-mac-sizing.html
2880                media/modern-media-controls/media-documents/media-document-video-ios-sizing.html
2881                media/modern-media-controls/media-documents/media-document-video-mac-sizing.html
2882                media/modern-media-controls/pip-support/pip-support-disabled-video-with-audio-tracks-only.html
2883
2884         * Modules/modern-media-controls/controls/ios-inline-media-controls.css:
2885         (:host(audio) .media-controls.ios.inline > .controls-bar:before,):
2886         (:host(audio) .media-controls.ios.inline > .controls-bar:before): Deleted.
2887         * Modules/modern-media-controls/controls/macos-media-controls.css:
2888         (:host(audio) .media-controls.mac.inline > .controls-bar,):
2889         (:host(audio) .media-controls.mac.inline > .controls-bar > .background-tint,):
2890         (:host(audio) .media-controls.mac.inline > .controls-bar): Deleted.
2891         (:host(audio) .media-controls.mac.inline > .controls-bar > .background-tint): Deleted.
2892         * Modules/modern-media-controls/controls/media-document.css: Copied from Source/WebCore/Modules/modern-media-controls/controls/macos-media-controls.css.
2893         (:host(.media-document)):
2894         (:host(.media-document.ready)):
2895         (:host(.media-document.audio.mac)):
2896         (:host(.media-document.audio.ipad)):
2897         (:host(.media-document.audio.iphone)):
2898         (:host(.media-document.video.mac)):
2899         (:host(.media-document.video.ipad)):
2900         (:host(.media-document.video.iphone)):
2901         * Modules/modern-media-controls/js-files:
2902         * Modules/modern-media-controls/media/fullscreen-support.js:
2903         (FullscreenSupport.prototype.syncControl):
2904         (FullscreenSupport):
2905         * Modules/modern-media-controls/media/media-controller.js:
2906         (MediaController):
2907         (MediaController.prototype.get isAudio):
2908         * Modules/modern-media-controls/media/media-document-controller.js: Copied from Source/WebCore/Modules/modern-media-controls/media/fullscreen-support.js.
2909         (MediaDocumentController):
2910         (MediaDocumentController.prototype.handleEvent):
2911         (MediaDocumentController.prototype._mediaDocumentHasMetadata):
2912         (MediaDocumentController.prototype._mediaDocumentHasSize):
2913         * Modules/modern-media-controls/media/pip-support.js:
2914         (PiPSupport.prototype.syncControl):
2915         (PiPSupport):
2916         * html/MediaDocument.cpp:
2917         (WebCore::MediaDocumentParser::createDocumentStructure):
2918
2919 2017-03-28  Myles C. Maxfield  <mmaxfield@apple.com>
2920
2921         Follow-up patch after r214364.
2922         https://bugs.webkit.org/show_bug.cgi?id=168895
2923
2924         Unreviewed.
2925
2926         * platform/graphics/FontDescription.cpp:
2927         (WebCore::FontDescription::FontDescription):
2928
2929 2017-03-27  Said Abou-Hallawa  <sabouhallawa@apple.com>
2930
2931         REGRESSION(213764): Large images should not be decoded asynchronously when they are drawn on a canvas
2932         https://bugs.webkit.org/show_bug.cgi?id=169771
2933
2934         Reviewed by Simon Fraser.
2935
2936         Sometimes we have to draw the image immediately like when a canvas calls
2937         drawImage. In this case we have to decode the image synchronously to guarantee
2938         the drawing. Other times we need to decode with the native size of the image.
2939         The animated images have to be decoded with native size always. Otherwise
2940         the frame cache will be messed up if the same image is animated with different
2941         sizes. Currently we always decode asynchronously with sizeForDrawing. We need
2942         to decouple the decoding mode from the sizeForDrawing.
2943
2944         This patch introduce the DecodingOptions class which can store and compare the
2945         following four cases:
2946             -- Synchronous: The frame has be decoded with native size only.
2947             -- Asynchronous + anySize: This is passed from the Image::draw() callers.
2948             -- Asynchronous + fullSize: The image has to be decoded with its full size.
2949             -- Asynchronous + sizeForDrawing: The image can be decoded with sizeForDrawing unless
2950             it was decoded with either a full size or sizeForDrawing which is larger than the
2951             requested sizeForDrawing.
2952
2953         A new argument of type DecodingMode will be added to Image::draw() function.
2954         Only when the drawing comes from the render tree, it will be Asynchronous.
2955         Otherwise it will be Synchronous.
2956
2957         Tests: fast/images/animated-image-different-dest-size.html
2958                fast/images/async-image-background-image.html
2959                fast/images/async-image-canvas-draw-image.html
2960
2961         * WebCore.xcodeproj/project.pbxproj:
2962         * platform/graphics/BitmapImage.cpp:
2963         (WebCore::BitmapImage::frameImageAtIndexCacheIfNeeded): Gets the frame image, cache it synchronously if
2964         the current one is invalid. frameImageAtIndex() returns whatever stored in the cache.
2965         (WebCore::BitmapImage::nativeImage): Call frameImageAtIndexCacheIfNeeded() instead of frameImageAtIndex().
2966         (WebCore::BitmapImage::nativeImageForCurrentFrame): Ditto.
2967         (WebCore::BitmapImage::nativeImageOfSize): Ditto.
2968         (WebCore::BitmapImage::framesNativeImages): Ditto.
2969         (WebCore::BitmapImage::draw): Change the logic to do the following:
2970         -- The animated image has to be decoded with its full size.
2971         -- The animated image expects the current frame to be ready for drawing.
2972         -- The large image decoding does not need to call internalStartAnimation().
2973         -- The large image has to request async image decoding but draw the current one if it exists.
2974         (WebCore::BitmapImage::drawPattern): Draw the pattern synchronously.
2975         (WebCore::BitmapImage::shouldUseAsyncDecodingForLargeImages): Delete the call to shouldUseAsyncDecodingForTesting()
2976         since it is only applied for animated images.
2977         (WebCore::BitmapImage::shouldUseAsyncDecodingForAnimatedImages): Call shouldUseAsyncDecodingForAnimatedImageForTesting().
2978         (WebCore::BitmapImage::internalStartAnimation): Request decoding with the full size.
2979         (WebCore::BitmapImage::advanceAnimation): Call shouldUseAsyncDecodingForAnimatedImageForTesting().
2980         (WebCore::BitmapImage::internalAdvanceAnimation): Assert the current frame is not being decoding asynchronously for any size.
2981         (WebCore::BitmapImage::frameImageAtIndex): Deleted. Moved to the header file but with a new purpose: return
2982         the current frame from the frame cache as is; do not cache a new one.
2983         (WebCore::BitmapImage::shouldUseAsyncDecodingForLargeImage): Deleted. Function was renamed to shouldUseAsyncDecodingForLargeImages.
2984         (WebCore::BitmapImage::shouldUseAsyncDecodingForAnimatedImage): Deleted. Function was renamed to shouldUseAsyncDecodingForAnimatedImages.
2985         * platform/graphics/BitmapImage.h:
2986         * platform/graphics/CrossfadeGeneratedImage.cpp:
2987         (WebCore::CrossfadeGeneratedImage::draw): Add a new argument of type DecodingMode.
2988         * platform/graphics/CrossfadeGeneratedImage.h:
2989         * platform/graphics/DecodingOptions.h: Added.
2990         (WebCore::DecodingOptions::DecodingOptions): Default constructor: Synchronous mode.
2991         (WebCore::DecodingOptions::operator==): Compares two DecodingOptions for equality.
2992         (WebCore::DecodingOptions::isSynchronous): Is the frame decoded synchronously?
2993         (WebCore::DecodingOptions::isAsynchronous): Is the frame decoded asynchronously?
2994         (WebCore::DecodingOptions::isAsynchronousCompatibleWith): Is this DecodingOptions compatible with another one?
2995         (WebCore::DecodingOptions::hasFullSize): Is the decoding mode asynchronous but for the image full size?
2996         (WebCore::DecodingOptions::hasSizeForDrawing): Is this decoding mode asynchronous but for a sizeForDrawing?
2997         (WebCore::DecodingOptions::sizeForDrawing): Returns the sizeForDrawing. m_decodingModeOrSize has to hold an IntSize.
2998         (WebCore::DecodingOptions::maxDimension): Moved form ImageFrame.cpp.
2999         (WebCore::DecodingOptions::has): A helper function.
3000         (WebCore::DecodingOptions::hasDecodingMode): Does m_decodingModeOrSize a DecodingMode?
3001         (WebCore::DecodingOptions::hasSize): Does m_decodingModeOrSize an IntSize?
3002         * platform/graphics/GeneratedImage.h: Add a new argument of type DecodingMode.
3003         * platform/graphics/GradientImage.cpp:
3004         (WebCore::GradientImage::draw): Ditto.
3005         * platform/graphics/GradientImage.h: Ditto.
3006         * platform/graphics/GraphicsContext.cpp:
3007         (WebCore::GraphicsContext::drawImage): Pass the ImagePaintingOptions::m_DecodingMode to Image::draw().
3008         * platform/graphics/GraphicsContext.h:
3009         (WebCore::ImagePaintingOptions::ImagePaintingOptions): Add a new member of type DecodingMode to ImagePaintingOptions.
3010         * platform/graphics/Image.cpp:
3011         (WebCore::Image::drawTiled): Pass DecodingMode::Synchronous to Image::draw().
3012         * platform/graphics/Image.h: Add a new argument of type DecodingMode to Image::draw().
3013         * platform/graphics/ImageFrame.cpp:
3014         (WebCore::ImageFrame::operator=): Replace m_sizeForDrawing by m_decodingOptions.
3015         (WebCore::ImageFrame::hasNativeImage): Check if m_nativeImage is valid and the subsamplingLevels match.
3016         (WebCore::ImageFrame::hasFullSizeNativeImage): Checks hasNativeImage() and whether the image frame was
3017         decoded for the image full size.
3018         (WebCore::ImageFrame::hasDecodedNativeImageCompatibleWithOptions): Checks hasNativeImage() and the DecodingOptions match.
3019         (WebCore::maxDimension): Deleted. Moved to DecodingOptions.h.
3020         (WebCore::ImageFrame::isBeingDecoded): Deleted. The check for having an ImageFrame being decoded is
3021         moved to ImageFrameCache.
3022         (WebCore::ImageFrame::hasValidNativeImage): Deleted. No need to this function.
3023         * platform/graphics/ImageFrame.h:
3024         (WebCore::ImageFrame::hasNativeImage): Add an std::optional<SubsamplingLevel> argument.
3025         (WebCore::ImageFrame::hasFullSizeNativeImage): Checks whether the ImageFrame was decoded for the image full size.
3026         (WebCore::ImageFrame::enqueueSizeForDecoding): Deleted.
3027         (WebCore::ImageFrame::dequeueSizeForDecoding): Deleted.
3028         (WebCore::ImageFrame::clearSizeForDecoding): Deleted.
3029         (WebCore::ImageFrame::isBeingDecoded): Deleted.
3030         (WebCore::ImageFrame::sizeForDrawing): Deleted.
3031         (WebCore::ImageFrame::hasDecodedNativeImage): Deleted.
3032         The logic of knowing whether an ImageFrame is being decoded is moved to ImageFrameCache.
3033         * platform/graphics/ImageFrameCache.cpp:
3034         (WebCore::ImageFrameCache::cacheFrameMetadataAtIndex): Caches the metadata of an ImageFrame. If the NativeImage
3035         was decoded for a sizeForDrawing, the size of the ImageFrame will be the nativeImageSize(). Otherwise, the
3036         frameSizeAtIndex() will be called.
3037         (WebCore::ImageFrameCache::cacheFrameNativeImageAtIndex): Cache a new NativeImage which involves caching new
3038         metadata and updating the memory cache. No need to check if the existing native image is valid or not for the
3039         DecodingOptions. Actually it would be a bug if it happens. This is why cacheNativeImageForFrameRequest() asserts
3040         !frame.hasAsyncNativeImage() before calling cacheFrameNativeImageAtIndex().
3041         (WebCore::ImageFrameCache::cacheAsyncFrameNativeImageAtIndex): Cache new NativeImage which was decoded asynchronously.
3042         (WebCore::ImageFrameCache::startAsyncDecodingQueue): Call cacheAsyncFrameNativeImageAtIndex() instead of
3043         cacheNativeImageForFrameRequest() for clarity.
3044         (WebCore::ImageFrameCache::requestFrameAsyncDecodingAtIndex): Call hasAsyncNativeImage() instead of hasValidNativeImage()
3045         Call frameIsDecodingCompatibleWithOptionsAtIndex() instead of frame.isBeingDecoded(). Replace the call to enqueueSizeForDecoding()
3046         by appending the same ImageFrameRequest to m_frameCommitQueue.
3047         (WebCore::ImageFrameCache::isAsyncDecodingQueueIdle): Use m_frameCommitQueue to answer the question whether the decodingQueue is idle.
3048         (WebCore::ImageFrameCache::stopAsyncDecodingQueue): Use m_frameCommitQueue to loop through all the ImageFrames which are currently being decoded.
3049         (WebCore::ImageFrameCache::frameAtIndexCacheIfNeeded): For getting the metadata, this function needs a valid frame. If it is requested
3050         to decode the nativeImage, it has to do it synchronously.
3051         (WebCore::ImageFrameCache::singlePixelSolidColor): Don't cache the frame if it is an animated image or the size is not a single pixel.
3052         (WebCore::ImageFrameCache::frameIsBeingDecodedAndIsCompatibleWithOptionsAtIndex): Use m_frameCommitQueue to answer the question whether an ImageFrame
3053         is being decoded and is compatible with DecodingOptions.
3054         (WebCore::ImageFrameCache::frameHasFullSizeNativeImageAtIndex): Calls ImageFrame::hasFullNativeImage() for a frame.
3055         (WebCore::ImageFrameCache::frameHasDecodedNativeImageCompatibleWithOptionsAtIndex): Calls ImageFrame::hasDecodedNativeImageCompatibleWithOptions() for a frame.
3056         (WebCore::ImageFrameCache::frameImageAtIndex): Returns the current NativeImage without caching.
3057         (WebCore::ImageFrameCache::frameImageAtIndexCacheIfNeeded): Returns the current NativeImage but cache it synchronously if needed.
3058         (WebCore::ImageFrameCache::setFrameNativeImageAtIndex): Deleted.
3059         (WebCore::ImageFrameCache::setFrameMetadataAtIndex): Deleted.
3060         (WebCore::ImageFrameCache::replaceFrameNativeImageAtIndex): Deleted.
3061         (WebCore::ImageFrameCache::frameIsBeingDecodedAtIndex): Deleted.
3062         (WebCore::ImageFrameCache::frameHasImageAtIndex): Deleted.
3063         (WebCore::ImageFrameCache::frameHasValidNativeImageAtIndex): Deleted.
3064         (WebCore::ImageFrameCache::frameHasDecodedNativeImage): Deleted.
3065         * platform/graphics/ImageFrameCache.h: Two ImageFrameRequest queues will be used.
3066         -- The existing one m_frameRequestQueue which is shared between the main thread and decoding thread. The requests will be
3067         dequeued from it before starting the decoding. The decoded NativeImage will be cached only on the main thread. The decoding
3068         thread is not blocked by the callOnMainThread(). This means there might be multiple ImageFrameRequests which were dequeued
3069         while their NativeImages have not been cached yet.
3070         -- A new one m_frameCommitQueue which is track all the ImageFrameRequests whose NativeImages have not been cached yet.
3071         (WebCore::ImageFrameCache::frameAtIndexCacheIfNeeded): Be explicit about caching the image frame. frameImageAtIndex()
3072         returns the current image frame without caching. frameAtIndexCacheIfNeeded(). returns the current image frame but cache
3073         it if needed.
3074         (WebCore::ImageFrameCache::ImageFrameRequest::operator==): Compares two ImageFrameRequests for equality.
3075         * platform/graphics/ImageSource.cpp:
3076         (WebCore::ImageSource::frameImageAtIndexCacheIfNeeded):
3077         (WebCore::ImageSource::frameImageAtIndex): Deleted.
3078         * platform/graphics/ImageSource.h:
3079         (WebCore::ImageSource::requestFrameAsyncDecodingAtIndex): Change the type of the argument from IntSize to be const std::optional<IntSize>.
3080         (WebCore::ImageSource::frameIsBeingDecodedAndIsCompatibleWithOptionsAtIndex): Rename of frameIsBeingDecodedAtIndex(). Replace the argument of type
3081         std::optional<IntSize> by an argument of type DecodingOptions.
3082         (WebCore::ImageSource::frameHasFullSizeNativeImageAtIndex): A wrapper around the ImageFrameCache function.
3083         (WebCore::ImageSource::frameHasDecodedNativeImageCompatibleWithOptionsAtIndex): Ditto.
3084         (WebCore::ImageSource::frameImageAtIndex): Ditto.
3085         (WebCore::ImageSource::frameIsBeingDecodedAtIndex): Deleted.
3086         (WebCore::ImageSource::frameHasValidNativeImageAtIndex): Deleted.
3087         (WebCore::ImageSource::frameHasDecodedNativeImage): Deleted.
3088         * platform/graphics/NamedImageGeneratedImage.cpp:
3089         (WebCore::NamedImageGeneratedImage::draw): Add a new argument of type DecodingMode.
3090         * platform/graphics/NamedImageGeneratedImage.h: Ditto.
3091         * platform/graphics/cairo/ImageBufferCairo.cpp:
3092         (WebCore::ImageBuffer::draw): Add a new argument of type DecodingMode.
3093         * platform/graphics/cg/ImageDecoderCG.cpp:
3094         (WebCore::ImageDecoder::createFrameImageAtIndex): Replace the sizeForDrawing argument by a DecodingMode argument. Add a new handling
3095         for decoding asynchronously for the image full size.
3096         * platform/graphics/cg/ImageDecoderCG.h: Change the prototype of the function.
3097         * platform/graphics/cg/PDFDocumentImage.cpp:
3098         (WebCore::PDFDocumentImage::draw): Add a new argument of type DecodingMode.
3099         * platform/graphics/cg/PDFDocumentImage.h:
3100         * platform/graphics/win/ImageCGWin.cpp:
3101         (WebCore::BitmapImage::getHBITMAPOfSize): Pass DecodingMode::Synchronous to Image::draw().
3102         (WebCore::BitmapImage::drawFrameMatchingSourceSize): Ditto.
3103         * platform/graphics/win/ImageDecoderDirect2D.cpp:
3104         (WebCore::ImageDecoder::createFrameImageAtIndex): Replace the sizeForDrawing argument by a DecodingMode argument.
3105         * platform/graphics/win/ImageDecoderDirect2D.h: Change the prototype of the function.
3106         * platform/image-decoders/ImageDecoder.cpp:
3107         (WebCore::ImageDecoder::createFrameImageAtIndex): Replace the sizeForDrawing argument by a DecodingMode argument.
3108         * platform/image-decoders/ImageDecoder.h: Change the prototype of the function.
3109         * rendering/RenderBoxModelObject.cpp:
3110         (WebCore::RenderBoxModelObject::paintFillLayerExtended): Draw the background image asynchronously if the image size is large.
3111         * rendering/RenderImage.cpp:
3112         (WebCore::RenderImage::paintIntoRect): Draw the background image element asynchronously if the image size is large.
3113         * svg/graphics/SVGImage.cpp:
3114         (WebCore::SVGImage::drawForContainer): Pass DecodingMode::Synchronous to draw().
3115         (WebCore::SVGImage::nativeImageForCurrentFrame): Ditto.
3116         (WebCore::SVGImage::nativeImage): Ditto.
3117         (WebCore::SVGImage::draw): Add a new argument of type DecodingMode.
3118         * svg/graphics/SVGImage.h: Change the prototype of the function.
3119         * svg/graphics/SVGImageForContainer.cpp:
3120         (WebCore::SVGImageForContainer::draw): Add a new argument of type DecodingMode.
3121         * svg/graphics/SVGImageForContainer.h: Change the prototype of the function.
3122
3123 2017-03-27  Youenn Fablet  <youenn@apple.com>
3124
3125         Activate release libwebrtc logging when WebRTC log channel is on
3126         https://bugs.webkit.org/show_bug.cgi?id=169659
3127
3128         Reviewed by Alex Christensen.
3129
3130         * platform/mediastream/libwebrtc/LibWebRTCProvider.cpp:
3131         (WebCore::initializePeerConnectionFactoryAndThreads):
3132
3133 2017-03-27  Antti Koivisto  <antti@apple.com>
3134
3135         Move visibleInViewportStateChanged callback from Element to render tree
3136         https://bugs.webkit.org/show_bug.cgi?id=170039
3137
3138         Reviewed by Zalan Bujtas.
3139
3140         Make it easier to use from the render tree.
3141
3142         Also for simplicity move the bits from RenderObject rare data to RenderElement.
3143         There is plenty of space there.
3144
3145         * dom/Element.h:
3146         (WebCore::Element::isVisibleInViewportChanged): Deleted.
3147         * html/HTMLMediaElement.h:
3148         * rendering/RenderElement.cpp:
3149         (WebCore::RenderElement::RenderElement):
3150         (WebCore::RenderElement::willBeDestroyed):
3151         (WebCore::RenderElement::registerForVisibleInViewportCallback):
3152         (WebCore::RenderElement::unregisterForVisibleInViewportCallback):
3153         (WebCore::RenderElement::setVisibleInViewportState):
3154         (WebCore::RenderElement::visibleInViewportStateChanged):
3155         * rendering/RenderElement.h:
3156         (WebCore::RenderElement::visibleInViewportState):
3157         * rendering/RenderObject.cpp:
3158         (WebCore::RenderObject::setIsRegisteredForVisibleInViewportCallback): Deleted.
3159         (WebCore::RenderObject::setVisibleInViewportState): Deleted.
3160         * rendering/RenderObject.h:
3161         (WebCore::RenderObject::hasOutlineAutoAncestor):
3162         (WebCore::RenderObject::RenderObjectRareData::RenderObjectRareData):
3163         (WebCore::RenderObject::isRegisteredForVisibleInViewportCallback): Deleted.
3164         (WebCore::RenderObject::visibleInViewportState): Deleted.
3165         * rendering/RenderVideo.cpp:
3166         (WebCore::RenderVideo::visibleInViewportStateChanged):
3167         * rendering/RenderVideo.h:
3168         * rendering/RenderView.cpp:
3169         (WebCore::RenderView::updateVisibleViewportRect):
3170
3171 2017-03-27  Youenn Fablet  <youenn@apple.com>
3172
3173         addIceCandidate should not throw if passed null or undefined
3174         https://bugs.webkit.org/show_bug.cgi?id=170118
3175
3176         Reviewed by Eric Carlson.
3177
3178         Covered by updated test.
3179
3180         A null/undefined candidate passed to addIceCandidate means end of Ice candidate..
3181
3182         * Modules/mediastream/PeerConnectionBackend.cpp:
3183         (WebCore::PeerConnectionBackend::addIceCandidate):
3184         * Modules/mediastream/PeerConnectionBackend.h:
3185         (WebCore::PeerConnectionBackend::endOfIceCandidates):
3186         * Modules/mediastream/RTCPeerConnection.cpp:
3187         (WebCore::RTCPeerConnection::queuedAddIceCandidate):
3188         * Modules/mediastream/RTCPeerConnection.h:
3189         * Modules/mediastream/RTCPeerConnection.idl:
3190         * Modules/mediastream/RTCPeerConnection.js:
3191         (addIceCandidate):
3192
3193 2017-03-27  Antti Koivisto  <antti@apple.com>
3194
3195         Allow the page to render before <link> stylesheet tags in body
3196         https://bugs.webkit.org/show_bug.cgi?id=149157
3197         <rdar://problem/24658830>
3198
3199         Reviewed by Simon Fraser.
3200
3201         Currently we block style and renderer building completely if document has any loading
3202         stylesheets. In case a script queries something layout dependent we construct the render
3203         tree with whatever style we have but block painting in it.
3204
3205         This patch changes behavior so that a loading stylesheet in body only blocks rendering for elements
3206         that are after it. The expectation is that such stylesheets rarely affect elements before them
3207         and the elements can be rendered without causing ugly visible styling changes.
3208
3209         The patch replaces the old flash-of-unstyled-content (FOUC) preventation mechanism with a more
3210         fine-grained one. Paint blocking is now done on per-renderer basis with based on isNonFinal flag in
3211         RenderStyle.
3212
3213         For stylesheets in head the behavior should be largely unchanged.
3214
3215         Test: http/tests/incremental/stylesheet-body-incremental-rendering.html
3216
3217         * css/StyleResolver.cpp:
3218         (WebCore::StyleResolver::pseudoStyleRulesForElement):
3219         * dom/Document.cpp:
3220         (WebCore::Document::Document):
3221         (WebCore::Document::resolveStyle):
3222         (WebCore::Document::updateLayoutIgnorePendingStylesheets):
3223
3224             Remove the old FOUC preventation state tracking.
3225
3226         (WebCore::Document::shouldScheduleLayout):
3227         (WebCore::Document::didRemoveAllPendingStylesheet):
3228
3229             Repaints will now get triggered by the normal style mechanism.
3230
3231         * dom/Document.h:
3232         (WebCore::Document::hasNodesWithNonFinalStyle):
3233         (WebCore::Document::setHasNodesWithNonFinalStyle):
3234
3235             Track if we need to recompute the style later because non-final or unstyled elements.
3236
3237         (WebCore::Document::didLayoutWithPendingStylesheets): Deleted.
3238         (WebCore::Document::hasNodesWithPlaceholderStyle): Deleted.
3239         (WebCore::Document::setHasNodesWithPlaceholderStyle): Deleted.
3240         * html/HTMLFrameSetElement.cpp:
3241         (WebCore::HTMLFrameSetElement::rendererIsNeeded):
3242         * page/FrameView.cpp:
3243         (WebCore::FrameView::qualifiesAsVisuallyNonEmpty):
3244
3245             Don't qualify as visually non-empty if we have loading stylesheets in head (even if there is
3246             a fouc-prevented render tree).
3247
3248         (WebCore::FrameView::fireLayoutRelatedMilestonesIfNeeded):
3249         * rendering/RenderBlock.cpp:
3250         (WebCore::RenderBlock::paintContents):
3251
3252             Instead of a global test, block painting if isNonFinal is set in the renderer's style.
3253
3254         * rendering/RenderLayer.cpp:
3255         (WebCore::shouldSuppressPaintingLayer):
3256         * rendering/style/RenderStyle.cpp:
3257         (WebCore::RenderStyle::changeRequiresRepaint):
3258
3259             The isNonFinal flag prevents painting so we need to trigger repaint when it gets cleared.
3260
3261         * rendering/style/RenderStyle.h:
3262         (WebCore::RenderStyle::isNotFinal):
3263         (WebCore::RenderStyle::setIsNotFinal):
3264         (WebCore::RenderStyle::isPlaceholderStyle): Deleted.
3265         (WebCore::RenderStyle::setIsPlaceholderStyle): Deleted.
3266
3267             There is no need for placeholder styles anymore. Reuse the bit for isNotFinal.
3268
3269         * rendering/style/StyleRareNonInheritedData.cpp:
3270         (WebCore::StyleRareNonInheritedData::StyleRareNonInheritedData):
3271         (WebCore::StyleRareNonInheritedData::operator==):
3272         * rendering/style/StyleRareNonInheritedData.h:
3273         * style/StyleScope.cpp:
3274         (WebCore::Style::Scope::analyzeStyleSheetChange):
3275         (WebCore::Style::Scope::updateActiveStyleSheets):
3276         * style/StyleTreeResolver.cpp:
3277         (WebCore::Style::TreeResolver::styleForElement):
3278         (WebCore::Style::TreeResolver::resolveElement):
3279
3280             If we have seens a loading stylesheet and don't have a renderer yet don't style the element.
3281             In case there is a renderer or we are ignoring pending sheets, resolve the style normally
3282             but mark it as non-final.
3283
3284         (WebCore::Style::makePlaceholderStyle): Deleted.
3285
3286 2017-03-27  Myles C. Maxfield  <mmaxfield@apple.com>
3287
3288         Test variation font ranges in the CSS Font Loading API
3289         https://bugs.webkit.org/show_bug.cgi?id=170022
3290
3291         Reviewed by Dean Jackson.
3292
3293         Test: fast/text/variations/font-loading-api-parse-ranges.html
3294
3295         Don't use keywords when reporting font variation range values.
3296
3297         * css/CSSComputedStyleDeclaration.cpp:
3298         (WebCore::ComputedStyleExtractor::fontNonKeywordWeightFromStyleValue):
3299         (WebCore::ComputedStyleExtractor::fontWeightFromStyleValue):
3300         (WebCore::ComputedStyleExtractor::fontNonKeywordStretchFromStyleValue):
3301         (WebCore::ComputedStyleExtractor::fontStretchFromStyleValue):
3302         (WebCore::ComputedStyleExtractor::fontNonKeywordStyleFromStyleValue):
3303         (WebCore::ComputedStyleExtractor::fontStyleFromStyleValue):
3304         * css/CSSComputedStyleDeclaration.h:
3305         * css/FontFace.cpp:
3306         (WebCore::FontFace::style):
3307         (WebCore::FontFace::weight):
3308         (WebCore::FontFace::stretch):
3309
3310 2017-03-27  Anders Carlsson  <andersca@apple.com>
3311
3312         Propagate shipping contact update errors
3313         https://bugs.webkit.org/show_bug.cgi?id=170141
3314         rdar://problem/31276576
3315
3316         Reviewed by Tim Horton.
3317
3318         * Modules/applepay/ApplePaySession.cpp:
3319         (WebCore::convertAndValidate):
3320
3321 2017-03-27  Ryan Haddad  <ryanhaddad@apple.com>
3322
3323         Unreviewed, rolling out r214411.
3324
3325         Two of the LayoutTests for this change time out on ios-
3326         simulator.
3327
3328         Reverted changeset:
3329
3330         "[Modern Media Controls] Improve media documents across macOS,
3331         iPhone and iPad"
3332         https://bugs.webkit.org/show_bug.cgi?id=169145
3333         http://trac.webkit.org/changeset/214411
3334
3335 2017-03-27  Antoine Quint  <graouts@apple.com>
3336
3337         [Modern Media Controls] Clicking on the tracks button when the tracks panel is up in a media document pauses the video
3338         https://bugs.webkit.org/show_bug.cgi?id=168517
3339         <rdar://problem/30577636>
3340
3341         Reviewed by Dean Jackson.
3342
3343         We completely turn off default event handling in MediaDocument.cpp since we're implementing the
3344         behavior we expect to pause and resume the video in the modern-media-controls module already. This
3345         gets rid of this odd case where the content would not see the "click" event while the C++ side would
3346         handle it and pause the video.
3347
3348         * Modules/modern-media-controls/media/media-controller.js:
3349         (MediaController):
3350         (MediaController.prototype.handleEvent):
3351         (MediaController.prototype._containerWasClicked): Deleted.
3352         * html/MediaDocument.cpp:
3353         (WebCore::MediaDocument::defaultEventHandler):
3354
3355 2017-03-27  Youenn Fablet  <youenn@apple.com>
3356
3357         Tighten RTCDatachannel creation and parameter getters
3358         https://bugs.webkit.org/show_bug.cgi?id=170081
3359
3360         Reviewed by Eric Carlson.
3361
3362         Covered by updated tests.
3363
3364         Adding some parameter checks when creating data channels.
3365         Making some getters nullable as per the spec.
3366
3367         * Modules/mediastream/RTCDataChannel.h:
3368         * Modules/mediastream/RTCDataChannel.idl:
3369         * Modules/mediastream/RTCPeerConnection.cpp:
3370         (WebCore::RTCPeerConnection::createDataChannel):
3371         * Modules/mediastream/RTCPeerConnection.idl:
3372         * Modules/mediastream/libwebrtc/LibWebRTCMediaEndpoint.cpp:
3373         (WebCore::LibWebRTCMediaEndpoint::createDataChannel):
3374         (WebCore::LibWebRTCMediaEndpoint::addDataChannel):
3375         * platform/mediastream/RTCDataChannelHandler.h:
3376         (): Deleted.
3377
3378 2017-03-27  Youenn Fablet  <youenn@apple.com>
3379
3380         Add support for RTCRtpReceiver/RTCRtpSender getParameters
3381         https://bugs.webkit.org/show_bug.cgi?id=170057
3382
3383         Reviewed by Alex Christensen.
3384
3385         Test: webrtc/video-getParameters.html
3386
3387         getParameters returns a dictionary of values taken from libwebrtc RtpReceiverInterface/RtpSenderInrterface objects.
3388         Added a direct link between WebCore RTCRtpReceiver and libwebrtc RtpReceiverInterface object.
3389         Making the link between WebCore RTCRtpSender and libwebrtc RtpSenderInrterface object through
3390         PeerConnectionBackend to keep the current architecture shared with OpenWebRTC.
3391         In the future, we should try to make the link more direct.
3392
3393         Added routines to support the conversion from libwebrtc to WebCore.
3394         Ensured that RTCRtpReceiver is cleaning its backend when being stopped.
3395
3396         * CMakeLists.txt:
3397         * DerivedSources.make:
3398         * Modules/mediastream/MediaEndpointPeerConnection.cpp:
3399         (WebCore::MediaEndpointPeerConnection::setRemoteDescriptionTask):
3400         * Modules/mediastream/PeerConnectionBackend.h:
3401         (WebCore::PeerConnectionBackend::getParameters):
3402         * Modules/mediastream/RTCPeerConnection.cpp:
3403         (WebCore::RTCPeerConnection::doClose):
3404         (WebCore::RTCPeerConnection::getParameters):
3405         * Modules/mediastream/RTCPeerConnection.h:
3406         * Modules/mediastream/RTCRtpParameters.h: Added.
3407         * Modules/mediastream/RTCRtpParameters.idl: Added.
3408         * Modules/mediastream/RTCRtpReceiver.cpp:
3409         (WebCore::RTCRtpReceiver::RTCRtpReceiver):