1. 13 Aug, 2013 40 commits
    • commit-queue@webkit.org's avatar
      [BlackBerry] WebPage doesn't zoom to fit viewport after being rotated from... · ba861f8b
      commit-queue@webkit.org authored
      [BlackBerry] WebPage doesn't zoom to fit viewport after being rotated from landscape to portrait mode
      https://bugs.webkit.org/show_bug.cgi?id=119754
      
      Patch by Jacky Jiang <zhajiang@blackberry.com> on 2013-08-13
      Reviewed by Rob Buis.
      Internally reviewed by Konrad Piascik.
      
      JIRA 470951
      In this case, zoomToFitWidthScale(currentScale) = viewportWidth / contentsWidth
      which was 2.6666666666666665, initialScale = ((viewportWidth / devicePixelRatio)
      / contentsWidth) * devicePixelRatio which was 2.6666665077209473. When
      rotating the device from landscape to portrait, m_webPage->isAtInitialZoom()
      was false in setViewportSize() because of the floating-point rounding error
      above so that we had to use the scale of the previous rotation WebPage
      which was too large for the current rotation WebPage.
      Ignore the floating-point rounding error in isAtInitialZoom() so that we
      can use initialScale of the current rotation WebPage as the scale and also
      we should do it for isMaxZoomed() and isMinZoomed() as well. Please note
      that we can not use FLT_EPSILON here as it's 1E-7 on our platform.
      
      * Api/WebPage.cpp:
      (BlackBerry::WebKit::WebPagePrivate::initialScale):
      (BlackBerry::WebKit::WebPage::isMaxZoomed):
      (BlackBerry::WebKit::WebPage::isMinZoomed):
      (BlackBerry::WebKit::WebPage::isAtInitialZoom):
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154029 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      ba861f8b
    • andersca@apple.com's avatar
      <https://webkit.org/b/119771> Use NeverDestroyed for smart pointer slot static locals · 2c2dbf97
      andersca@apple.com authored
      Reviewed by Andreas Kling.
      
      This is preparation for making it a hard error to use DEFINE_STATIC_LOCAL with smart pointer types.
      
      * dom/UserTypingGestureIndicator.cpp:
      (WebCore::focusedNode):
      * page/mac/EventHandlerMac.mm:
      (WebCore::currentNSEventSlot):
      * platform/network/cf/NetworkStorageSessionCFNet.cpp:
      (WebCore::defaultNetworkStorageSession):
      (WebCore::cookieStorageOverride):
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154028 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      2c2dbf97
    • ap@apple.com's avatar
      [Mac] fast/dom/attr-style-too-lazy.html is flaky · 6f5fcda3
      ap@apple.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119763
      
      Reviewed by Darin Adler.
      
      * DumpRenderTree/mac/UIDelegate.mm:
      (-[UIDelegate webView:setStatusText:]): Don't dump anything when not running
      a test, it would just semi-randomly leak to the next one.
      (-[UIDelegate webView:didPressMissingPluginButton:]): Added same check here, just
      for good measure.
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154027 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      6f5fcda3
    • timothy_horton@apple.com's avatar
      [mac] Give MiniBrowser a reasonable bundle identifier · d40e4df9
      timothy_horton@apple.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119773
      
      Rubber-stamped by Simon Fraser.
      
      Give MiniBrowser the bundle identifier org.webkit.MiniBrowser.
      
      * MiniBrowser/mac/Info.plist:
      * MiniBrowser/mac/WebBundle/Info.plist:
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154026 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      d40e4df9
    • oliver@apple.com's avatar
      [sh4] Prepare LLINT for DFG_JIT implementation. · 22b2856d
      oliver@apple.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119755
      
      Patch by Julien Brianceau <jbrianceau@nds.com> on 2013-08-13
      Reviewed by Oliver Hunt.
      
      * LLIntOffsetsExtractor.pro: Add sh4.rb dependency.
      * offlineasm/sh4.rb:
          - Handle storeb opcode.
          - Make relative jumps when possible using braf opcode.
          - Update bmulio implementation to be consistent with baseline JIT.
          - Remove useless code from leap opcode.
          - Fix incorrect comment.
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154025 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      22b2856d
    • ap@apple.com's avatar
      Removed a bogus comment that didn't have any bugs after it. · 53cffc38
      ap@apple.com authored
              * platform/mac/TestExpectations:
      
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154024 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      53cffc38
    • andersca@apple.com's avatar
      Stop using DEFINE_STATIC_LOCAL with RefPtr · f1154b09
      andersca@apple.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119769
      
      Reviewed by Andreas Kling.
      
      No need to waste heap memory allocating RefPtr, just store the raw pointers directly.
      
      * dom/ContextFeatures.cpp:
      (WebCore::ContextFeatures::defaultSwitch):
      * dom/DocumentMarker.cpp:
      (WebCore::DocumentMarkerTextMatch::instanceFor):
      * editing/EditingStyle.cpp:
      (WebCore::StyleChange::extractTextStyles):
      * html/FTPDirectoryDocument.cpp:
      (WebCore::FTPDirectoryDocumentParser::loadDocumentTemplate):
      * html/HTMLPlugInImageElement.cpp:
      (WebCore::HTMLPlugInImageElement::didAddUserAgentShadowRoot):
      * html/track/TextTrack.cpp:
      (WebCore::TextTrack::captionMenuOffItem):
      (WebCore::TextTrack::captionMenuAutomaticItem):
      * loader/icon/IconDatabase.cpp:
      (WebCore::loadDefaultIconRecord):
      * platform/graphics/Image.cpp:
      (WebCore::Image::nullImage):
      * platform/graphics/PlatformTextTrack.h:
      (WebCore::PlatformTextTrack::captionMenuOffItem):
      (WebCore::PlatformTextTrack::captionMenuAutomaticItem):
      * rendering/style/RenderStyle.cpp:
      (WebCore::RenderStyle::initialShapeInside):
      * testing/MockCDM.cpp:
      (WebCore::initDataPrefix):
      (WebCore::keyPrefix):
      (WebCore::keyRequest):
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154023 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      f1154b09
    • dino@apple.com's avatar
      <https://webkit.org/b/119766> Make bug url more human friendly in ChangeLog · 75cfa4d4
      dino@apple.com authored
      Reviewed by Mark Rowe.
      
      Have prepare-ChangeLog produce webkit.org/b/id style
      URLs rather than exposing the inner workings of our
      cgi. I was going to use the term "Future Proofing"
      but that seems excessive :)
      
      * Scripts/prepare-ChangeLog:
      (main): Generate a more human readable form.
      (generateNewChangeLogs): Use it when printing out the ChangeLog entry.
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154022 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      75cfa4d4
    • andersca@apple.com's avatar
      Stop using DEFINE_STATIC_LOCAL with RetainPtr · 95e54306
      andersca@apple.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119765
      
      Reviewed by Jessie Berlin.
      
      No need to waste heap memory allocating RetainPtrs, just store the raw pointers directly.
      
      * platform/graphics/mac/ColorMac.mm:
      (WebCore::nsColor):
      * platform/graphics/mac/GraphicsContextMac.mm:
      (WebCore::makePatternColor):
      (WebCore::GraphicsContext::drawLineForDocumentMarker):
      * platform/graphics/mac/SimpleFontDataMac.mm:
      (WebCore::webFallbackFontFamily):
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154021 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      95e54306
    • commit-queue@webkit.org's avatar
      Unreviewed, rolling out r153925. · d1096b36
      commit-queue@webkit.org authored
      http://trac.webkit.org/changeset/153925
      https://bugs.webkit.org/show_bug.cgi?id=119768
      
      broke fast/media/media-query-list-07.html (Requested by kling
      on #webkit).
      
      * CMakeLists.txt:
      * DerivedSources.cpp:
      * DerivedSources.make:
      * DerivedSources.pri:
      * GNUmakefile.list.am:
      * Target.pri:
      * WebCore.vcxproj/WebCore.vcxproj:
      * WebCore.vcxproj/WebCore.vcxproj.filters:
      * WebCore.xcodeproj/project.pbxproj:
      * bindings/scripts/CodeGeneratorJS.pm:
      (GenerateCallbackHeader):
      (GenerateCallbackImplementation):
      (JSValueToNative):
      * bindings/scripts/IDLAttributes.txt:
      * bindings/scripts/test/JS/JSTestCallback.cpp:
      (WebCore::JSTestCallback::JSTestCallback):
      * bindings/scripts/test/JS/JSTestMediaQueryListListener.cpp:
      (WebCore::jsTestMediaQueryListListenerPrototypeFunctionMethod):
      * css/MediaAllInOne.cpp:
      * css/MediaQueryListListener.cpp: Copied from Source/WebCore/css/MediaQueryListListener.idl.
      (WebCore::MediaQueryListListener::queryChanged):
      * css/MediaQueryListListener.h:
      (WebCore::MediaQueryListListener::create):
      (WebCore::MediaQueryListListener::operator==):
      (WebCore::MediaQueryListListener::MediaQueryListListener):
      * css/MediaQueryListListener.idl:
      * css/MediaQueryMatcher.cpp:
      (WebCore::MediaQueryMatcher::Listener::evaluate):
      (WebCore::MediaQueryMatcher::styleResolverChanged):
      * css/MediaQueryMatcher.h:
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154020 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      d1096b36
    • timothy_horton@apple.com's avatar
      Remotely hosted plugins don’t work in window-server-hosted WebKit1 views · 2ce08b1d
      timothy_horton@apple.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119709
      <rdar://problem/14717572>
      
      Reviewed by Anders Carlsson.
      
      * Plugins/Hosted/NetscapePluginHostManager.h:
      Add hostLayersInWindowServer argument to instantiatePlugin.
      
      * Plugins/Hosted/NetscapePluginHostManager.mm:
      (WebKit::NetscapePluginHostManager::instantiatePlugin):
      Add hostLayersInWindowServer argument to instantiatePlugin, and pass it through to WebKitPluginHost.
      
      * Plugins/Hosted/NetscapePluginHostProxy.mm:
      (WKPCLayerHostingModeChanged):
      Forward the notification that the plugin host changed its layer hosting mode
      to the instance proxy.
      
      * Plugins/Hosted/NetscapePluginInstanceProxy.h:
      * Plugins/Hosted/NetscapePluginInstanceProxy.mm:
      (WebKit::NetscapePluginInstanceProxy::setShouldHostLayersInWindowServer):
      Ask the plugin host to re-host its layers when the layer hosting mode changes.
      
      (WebKit::NetscapePluginInstanceProxy::layerHostingModeChanged):
      Inform the WebHostedNetscapePluginView of the new layer hosting mode
      when it changes. Also, store the new CAContext ID.
      
      * Plugins/Hosted/WebHostedNetscapePluginView.h:
      * Plugins/Hosted/WebHostedNetscapePluginView.mm:
      (-[WebHostedNetscapePluginView windowHostsLayersInWindowServer]):
      Determine whether the WebView's current window uses window-server hosting for its layers.
      
      (-[WebHostedNetscapePluginView createPlugin]):
      Drive-by spelling fix.
      Factor _pluginLayer initialization out into createPluginLayer.
      Instantiate the plugin with the appropriate layer hosting mode.
      
      (-[WebHostedNetscapePluginView createPluginLayer]):
      
      (-[WebHostedNetscapePluginView setHostsLayersInWindowServer:]):
      Create a new _pluginLayer if the layer hosting mode changes.
      Attach the new _pluginLayer to our parent layer.
      
      (-[WebHostedNetscapePluginView updateAndSetWindow]):
      Inform the NetscapePluginInstanceProxy if the layer hosting mode changes,
      so it can inform the plugin host. We will asynchronously receive a reply
      with the new hosting mode and renderContextID and switch to the new CAContext,
      in setHostsLayersInWindowServer:.
      
      * Plugins/Hosted/WebKitPluginClient.defs:
      * Plugins/Hosted/WebKitPluginHost.defs:
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154019 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      2ce08b1d
    • simon.fraser@apple.com's avatar
      Every scroll causes additional layer tree work because of flatteningLayer->removeFromParent(); · 32c18dfd
      simon.fraser@apple.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119551
      
      Reviewed by Dean Jackson.
      
      Only re-parent the flattening layer if we had to update our
      layer configuration (reparenting is necessary then to get
      the correct sibling ordering), or if it was parented in some
      other layer.
      
      This avoid unnecessary flushing of state to CA.
      
      * rendering/RenderLayerBacking.cpp:
      (WebCore::RenderLayerBacking::updateGraphicsLayerConfiguration):
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154018 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      32c18dfd
    • ap@apple.com's avatar
      REGRESSION(r139282): Old caret sometimes gets "stuck" (not repainted)... · d077458a
      ap@apple.com authored
              REGRESSION(r139282): Old caret sometimes gets "stuck" (not repainted) in contenteditable elements.
              https://bugs.webkit.org/show_bug.cgi?id=119520
      
              * fast/repaint/caret-jump-between-nodes-expected.txt: Test was failing everywhere.
              Landing current Mac results to make bots greener, but Andreas should take a look at
              fixing the test.
      
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154017 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      d077458a
    • commit-queue@webkit.org's avatar
      [sh4] Prepare baseline JIT for DFG_JIT implementation. · cf80dfff
      commit-queue@webkit.org authored
      https://bugs.webkit.org/show_bug.cgi?id=119758
      
      Patch by Julien Brianceau <jbrianceau@nds.com> on 2013-08-13
      Reviewed by Oliver Hunt.
      
      * assembler/MacroAssemblerSH4.h:
          - Introduce a loadEffectiveAddress function to avoid code duplication.
          - Add ASSERTs and clean code.
      * assembler/SH4Assembler.h:
          - Prepare DFG_JIT implementation.
          - Add ASSERTs.
      * jit/JITStubs.cpp:
          - Add SH4 specific call for assertions.
      * jit/JITStubs.h:
          - Cosmetic change.
      * jit/JITStubsSH4.h:
          - Use constants to be more flexible with sh4 JIT stack frame.
      * jit/JSInterfaceJIT.h:
          - Cosmetic change.
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154016 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      cf80dfff
    • simon.fraser@apple.com's avatar
      Various media tests spew "CGContextScaleCTM: invalid context 0x0" errors · 268d5b31
      simon.fraser@apple.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119490
      
      New versions of WKSI that fix a bug that caused spew about null CGContexts,
      and update WebKitSystemInterface.h.
      
      * WebKitSystemInterface.h:
      * libWebKitSystemInterfaceLion.a:
      * libWebKitSystemInterfaceMountainLion.a:
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154015 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      268d5b31
    • berto@igalia.com's avatar
      [GTK] Closing inspector window crashes wk · 97128d1c
      berto@igalia.com authored
      https://bugs.webkit.org/show_bug.cgi?id=110865
      
      Reviewed by Carlos Garcia Campos.
      
      The previous fix for the inspector window crash breaks some unit
      tests. This one goes back to the original code and only moves the
      actual deletion of priv->corePage to the end of the function.
      
      * webkit/webkitwebview.cpp:
      (webkit_web_view_dispose):
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154014 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      97128d1c
    • robert@webkit.org's avatar
      Section's collapsed borders painted in wrong place in rtl tables · bcf251c4
      robert@webkit.org authored
      https://bugs.webkit.org/show_bug.cgi?id=119691
      
      Reviewed by Darin Adler.
      
      Source/WebCore:
      
      Table row-groups can be RTL too, so look to the section rather than the
      table when deciding the direction of a row. The section will inherit the
      table's direction if it doesn't override it.
      
      Test: fast/table/paint-collapsed-borders-rtl-section.html
      
      * rendering/RenderTableCell.cpp:
      (WebCore::RenderTableCell::hasStartBorderAdjoiningTable):
      (WebCore::RenderTableCell::hasEndBorderAdjoiningTable):
      
      LayoutTests:
      
      * fast/table/paint-collapsed-borders-rtl-section-expected.html: Added.
      * fast/table/paint-collapsed-borders-rtl-section.html: Added.
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154013 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      bcf251c4
    • ch.dumez@sisa.samsung.com's avatar
      NodeList.item() does not behave according to specification · 98e7eeb2
      ch.dumez@sisa.samsung.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119722
      
      Reviewed by Darin Adler.
      
      Source/WebCore:
      
      Make NodeList.item() indexed getter behave according to specification:
      http://dom.spec.whatwg.org/#nodelist
      
      Namely, we no longer throw if the index is negative (meaning the
      [IsIndex] extended attribute is dropped) and the argument is now
      mandatory.
      
      This behavior is consistent with both IE10 and Firefox and since
      recently Blink.
      
      Test: fast/dom/nodelist-item-parameter.html
      
      * dom/NodeList.idl:
      
      LayoutTests:
      
      Add a layout test to check that the NodeList.item() indexed getter's
      argument is properly validated. Also, update the following test case now
      that the argument is mandatory:
      fast/dom/non-numeric-values-numeric-parameters.html
      
      * fast/dom/nodelist-item-parameter-expected.txt: Added.
      * fast/dom/nodelist-item-parameter.html: Added.
      * fast/dom/non-numeric-values-numeric-parameters-expected.txt:
      * fast/dom/script-tests/non-numeric-values-numeric-parameters.js:
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154012 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      98e7eeb2
    • oliver@apple.com's avatar
      Harden executeConstruct against incorrect return types from host functions · 04bbde04
      oliver@apple.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119757
      
      Reviewed by Mark Hahnenberg.
      
      Add logic to guard against bogus return types.  There doesn't seem to be any
      class in webkit that does this wrong, but the typed array stubs in debug JSC
      do exhibit this bad behaviour.
      
      * interpreter/Interpreter.cpp:
      (JSC::Interpreter::executeConstruct):
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154011 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      04bbde04
    • andersca@apple.com's avatar
      Fix build. · 88555b5c
      andersca@apple.com authored
      * WebCore.exp.in:
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154010 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      88555b5c
    • allan.jensen@digia.com's avatar
      REGRESSION(SUBPIXEL_LAYOUT) Composited layers can cause one pixel shifts · d5cbc754
      allan.jensen@digia.com authored
      https://bugs.webkit.org/show_bug.cgi?id=115304
      
      Reviewed by David Hyatt.
      
      Source/WebCore:
      
      Accelerated layers can cause blocks at subpixel offsets to shift because
      accumulated subpixel offsets are lost between each layers.
      
      To solve this layer bounds are now calculated in LayoutUnits, and their
      subpixel offset saved so it can be used to ensure correct pixel-snapping
      during painting.
      
      Test: fast/sub-pixel/sub-pixel-composited-layers.html
      
      * WebCore.exp.in:
      * inspector/InspectorLayerTreeAgent.cpp:
      (WebCore::InspectorLayerTreeAgent::buildObjectForLayer):
      * platform/graphics/LayoutPoint.h:
      (WebCore::LayoutPoint::fraction):
      * rendering/RenderLayer.cpp:
      (WebCore::RenderLayer::setupClipPath):
      (WebCore::RenderLayer::setupFilters):
      (WebCore::RenderLayer::paintLayerContents):
      (WebCore::RenderLayer::calculateLayerBounds):
      * rendering/RenderLayer.h:
      * rendering/RenderLayerBacking.cpp:
      (WebCore::RenderLayerBacking::updateCompositedBounds):
      (WebCore::RenderLayerBacking::updateAfterWidgetResize):
      (WebCore::RenderLayerBacking::updateGraphicsLayerGeometry):
      (WebCore::RenderLayerBacking::resetContentsRect):
      (WebCore::RenderLayerBacking::contentOffsetInCompostingLayer):
      (WebCore::RenderLayerBacking::contentsBox):
      (WebCore::RenderLayerBacking::backgroundBox):
      (WebCore::RenderLayerBacking::paintIntoLayer):
      (WebCore::RenderLayerBacking::paintContents):
      (WebCore::RenderLayerBacking::compositedBounds):
      (WebCore::RenderLayerBacking::setCompositedBounds):
      * rendering/RenderLayerBacking.h:
      * rendering/RenderLayerCompositor.cpp:
      (WebCore::RenderLayerCompositor::logLayerInfo):
      (WebCore::RenderLayerCompositor::calculateCompositedBounds):
      * rendering/RenderLayerCompositor.h:
      * rendering/RenderTreeAsText.cpp:
      (WebCore::operator<<):
      * rendering/RenderTreeAsText.h:
      
      LayoutTests:
      
      * fast/sub-pixel/sub-pixel-composited-layers-expected.html: Added.
      * fast/sub-pixel/sub-pixel-composited-layers.html: Added.
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154009 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      d5cbc754
    • zandobersek@gmail.com's avatar
      [Autotools] Unicode's CFLAGS enforce -D_FORTIFY_SOURCE=2, -D_REENTRANT=1,... · f18ab334
      zandobersek@gmail.com authored
      [Autotools] Unicode's CFLAGS enforce -D_FORTIFY_SOURCE=2, -D_REENTRANT=1, causing faulty Clang builds
      https://bugs.webkit.org/show_bug.cgi?id=119685
      
      Reviewed by Gustavo Noronha Silva.
      
      icu-config includes '-D_FORTIFY_SOURCE=2 -D_REENTRANT=1' when printing out C preprocessor flags that are used
      as the C compiler flags to avoid other unwanted compiler options. This causes problems when building optimized
      builds with Clang because of a bug in that compiler:
      http://llvm.org/bugs/show_bug.cgi?id=16821
      
      To avoid that, the C preprocessor search flags, as printed by `icu-config --cppflags-searchpath` are now used
      the Unicode dependency's C compiler flags, avoiding unconditionally specifying the two macros.
      
      To adjust for that, the `-pthread` flag is added to the global CFLAGS and CXXFLAGS variables, ensuring
      the _REENTRANT define is set to 1 and declaring the flag globally instead of relying on Glib dependency's
      C compiler flags to do so for us. -D_FORTIFY_SOURCE=2 is only added to the CFLAGS and CXXFLAGS variables if
      the compiler in use is gcc or g++, preventing the Clang builds to malfunction.
      
      * Source/autotools/FindDependencies.m4:
      * Source/autotools/SetupCompilerFlags.m4:
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154008 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      f18ab334
    • zandobersek@gmail.com's avatar
      [Autotools] Don't compare $CC, $CXX to exact compiler names · c1aaaea0
      zandobersek@gmail.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119683
      
      Reviewed by Gustavo Noronha Silva.
      
      Instead of comparing $CC and $CXX to exact compiler names (like 'gcc', 'clang++' etc.),
      use the compiler version checks to also specify the broader compiler collection of which
      the used compiler is a member of. This avoids failures in some border-line cases where
      the user would still use either a GCC or a Clang compiler but provide it through a symbolic
      link that was specified via the CC/CXX environment variables.
      
      * Source/autotools/CheckSystemAndBasicDependencies.m4: Store the detected C/C++ compiler collection
      in c_compiler/cxx_compiler. Throw an error if no supported compiler was found.
      * Source/autotools/SetupCompilerFlags.m4: Test for a specific compiler by checking against
      c_compiler/cxx_compiler rather than CC/CXX values.
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154007 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      c1aaaea0
    • commit-queue@webkit.org's avatar
      [Curl] Possible infinite loop while downloading. · 85ee948c
      commit-queue@webkit.org authored
      https://bugs.webkit.org/show_bug.cgi?id=119744
      
      Patch by peavo@outlook.com <peavo@outlook.com> on 2013-08-13
      Reviewed by Darin Adler.
      
      If adding/removing curl handles fails, the download thread
      can end up in an endless loop trying to update the curl handle list.
      
      * platform/network/curl/CurlDownload.cpp:
      (WebCore::CurlDownloadManager::updateHandleList): Avoid infinite loop by ignoring return value.
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154006 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      85ee948c
    • ch.dumez@sisa.samsung.com's avatar
      HTMLSelectElement.item() does not behave according to specification · 64a2fa50
      ch.dumez@sisa.samsung.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119738
      
      Reviewed by Darin Adler.
      
      Source/WebCore:
      
      Make HTMLSelectElement.item() behave according to specification:
      http://www.w3.org/TR/html51/forms.html#dom-select-item
      http://dom.spec.whatwg.org/#dom-htmlcollection-item
      
      Namely, we no longer throw if the offset argument is negative. Also,
      the argument is now mandatory.
      
      IE10, Firefox and Blink do not throw when HTMLSelectElement.item() is
      called with a negative value (as per the specification).
      
      The argument is mandatory in Firefox and Blink, but optional in IE10
      (it returns null when called without argument). This looks like a bug
      in IE because other item() getters' argument is usually mandatory in IE
      (e.g. NodeList.item()).
      
      Test: fast/dom/HTMLSelectElement/select-element-item-argument.html
      
      * html/HTMLSelectElement.idl:
      
      LayoutTests:
      
      Add a new layout test to check that the HTMLSelectElement.item() indexed getter's
      parameter is correctly validated.
      Also update fast/dom/non-numeric-values-numeric-parameters.html now that the
      argument is mandatory.
      
      * fast/dom/HTMLSelectElement/select-element-item-argument-expected.txt: Added.
      * fast/dom/HTMLSelectElement/select-element-item-argument.html: Added.
      * fast/dom/non-numeric-values-numeric-parameters-expected.txt:
      * fast/dom/script-tests/non-numeric-values-numeric-parameters.js:
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154005 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      64a2fa50
    • ch.dumez@sisa.samsung.com's avatar
      On request error, always fire events on the XMLHttpRequestUpload before the XMLHttpRequest · 205dab32
      ch.dumez@sisa.samsung.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119714
      
      Reviewed by Alexey Proskuryakov.
      
      Source/WebCore:
      
      On request error, fire events on the XMLHttpRequestUpload object before the XMLHttpRequest
      object as per the latest specification:
      http://xhr.spec.whatwg.org/#request-error
      
      This specification change was made in Sept 2010:
      http://dev.w3.org/cvsweb/2006/webapi/XMLHttpRequest-2/Overview.src.html.diff?r1=1.138;r2=1.139;f=h
      
      It addresses the following comment:
      http://lists.w3.org/Archives/Public/public-webapps/2010JulSep/0777.html
      
      IE10 and since recently Blink behave according to specification but WebKit was firing the
      events on the XMLHttpRequest object BEFORE the XMLHttpRequestUpload object in case of
      'network error' or 'abort error'. WebKit was however behaving according to specification
      in case of 'timeout error', which was inconsistent.
      
      Test: http/tests/xmlhttprequest/upload-request-error-event-order.html
      
      * xml/XMLHttpRequest.cpp:
      (WebCore::XMLHttpRequest::networkError):
      (WebCore::XMLHttpRequest::abortError):
      
      LayoutTests:
      
      Add a layout test to validate the order in which the events are fired in case
      of a xhr request error. Also update an existing test which was relying on the
      outdated firing order.
      
      * http/tests/xmlhttprequest/simple-cross-origin-progress-events-expected.txt:
      * http/tests/xmlhttprequest/simple-cross-origin-progress-events.html:
      * http/tests/xmlhttprequest/upload-request-error-event-order-expected.txt: Added.
      * http/tests/xmlhttprequest/upload-request-error-event-order.html: Added.
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154004 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      205dab32
    • ch.dumez@sisa.samsung.com's avatar
      Simplify RadioNodeList's anonymous indexed getter · 7340ea89
      ch.dumez@sisa.samsung.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119725
      
      Reviewed by Darin Adler.
      
      Simplify RadioNodeList's anonymous indexed getter by dropping [IsIndex]
      IDL extended attribute from its parameter and making it mandatory.
      
      This change has strictly no impact on the generated bindings code
      but it simplifies the IDL.
      
      No new test, no behavior change.
      
      * html/RadioNodeList.idl:
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154003 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      7340ea89
    • allan.jensen@digia.com's avatar
      [Qt] Fix C++11 build with gcc 4.4 and 4.5 · 9081187b
      allan.jensen@digia.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119736
      
      Reviewed by Anders Carlsson.
      
      Don't force C++11 mode off anymore.
      
      * Target.pri:
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154002 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      9081187b
    • rakuco@webkit.org's avatar
      [EFL][DRT] Simplify call to reverseFind() · 80b26826
      rakuco@webkit.org authored
      https://bugs.webkit.org/show_bug.cgi?id=119746
      
      Reviewed by Christophe Dumez.
      
      Follow-up to r153977.
      
      * DumpRenderTree/efl/DumpRenderTreeChrome.cpp:
      (pathSuitableForTestResult): Instead of defining a static variable
      that's only used in a single place, pass '/' to reverseFind() directly.
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154001 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      80b26826
    • commit-queue@webkit.org's avatar
      [GTK] close the volume slider when the media control panel is hidden · 6385b615
      commit-queue@webkit.org authored
      https://bugs.webkit.org/show_bug.cgi?id=119062
      
      Patch by Danilo Cesar Lemes de Paula <danilo.cesar@collabora.co.uk> on 2013-08-13
      Reviewed by Gustavo Noronha Silva.
      
      It's not obvious how to close the volume slider on webkitGtk.
      The volume slider should close if the control panel is hidden.
      
      * html/shadow/MediaControlsGtk.cpp:
      (WebCore::MediaControlsGtk::makeTransparent):
      * html/shadow/MediaControlsGtk.h:
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@154000 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      6385b615
    • zarvai@inf.u-szeged.hu's avatar
      [Qt] Unreviewed gardening. Rebase tests after r153901 and r153915. Skip failing tests. · ba11246f
      zarvai@inf.u-szeged.hu authored
      Patch by Gabor Abraham <abrhm@inf.u-szeged.hu> on 2013-08-13
      
      * platform/qt-5.0-wk2/TestExpectations:
      * platform/qt-5.0-wk2/svg/W3C-SVG-1.1/struct-dom-03-b-expected.png: Removed.
      * platform/qt-5.0/http/tests/security/xss-DENIED-xsl-document-redirect-expected.txt: Removed.
      * platform/qt/TestExpectations:
      * platform/qt/http/tests/security/xss-DENIED-xsl-document-redirect-expected.txt: Added.
      * platform/qt/svg/W3C-SVG-1.1/struct-dom-03-b-expected.png:
      * platform/qt/svg/W3C-SVG-1.1/struct-dom-03-b-expected.txt: Added.
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@153999 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      ba11246f
    • commit-queue@webkit.org's avatar
      Unreviewed, rolling out r152563. · 642575b2
      commit-queue@webkit.org authored
      http://trac.webkit.org/changeset/152563
      https://bugs.webkit.org/show_bug.cgi?id=119740
      
      causes random crashes (Requested by carewolf on #webkit).
      
      * platform/graphics/qt/FontCacheQt.cpp:
      (WebCore::FontCache::getLastResortFallbackFont):
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@153998 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      642575b2
    • zandobersek@gmail.com's avatar
      Unreviewed GTK gardening. · 3890bb42
      zandobersek@gmail.com authored
      * platform/gtk/TestExpectations: Simplify test failure expectations to remove various lint failures,
      also removing a couple of expectations for tests that are passing in the process.
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@153997 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      3890bb42
    • zandobersek@gmail.com's avatar
      Unreviewed GTK gardening. · 5e05af5d
      zandobersek@gmail.com authored
      * platform/gtk/fast/repaint/caret-jump-between-nodes-expected.txt: Added. Rebaselining after r153815.
      * platform/gtk/svg/W3C-SVG-1.1/struct-dom-03-b-expected.png: Rebaselining after r153901.
      * platform/gtk/svg/W3C-SVG-1.1/struct-dom-03-b-expected.txt: Ditto.
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@153996 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      5e05af5d
    • zandobersek@gmail.com's avatar
      Unreviewed GTK gardening. · 7d32f466
      zandobersek@gmail.com authored
      * platform/gtk/TestExpectations: Add failure expectations for various failing tests.
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@153995 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      7d32f466
    • zarvai@inf.u-szeged.hu's avatar
      [Qt] Unreviewed gardening. Skipping new fails. · b35425a1
      zarvai@inf.u-szeged.hu authored
      Patch by Gabor Abraham <abrhm@inf.u-szeged.hu> on 2013-08-13
      
      * platform/qt-5.0-wk1/TestExpectations:
      * platform/qt/TestExpectations:
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@153994 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      b35425a1
    • abucur@adobe.com's avatar
      Unreviewed. · de330072
      abucur@adobe.com authored
      Fix the build after http://trac.webkit.org/changeset/153990.
      
      * rendering/RenderBlock.cpp:
      (WebCore::RenderBlock::updateShapesBeforeBlockLayout): Add the default return value if CSS Shapes are disabled.
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@153993 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      de330072
    • a.bah@samsung.com's avatar
      Adding testcase for verifying editing behavior for up/down caret movement between lines. · 33d62de8
      a.bah@samsung.com authored
      https://bugs.webkit.org/show_bug.cgi?id=119511
      
      Reviewed by Ryosuke Niwa.
      
      * editing/selection/verify-editing-behavior-for-line-granularity-expected.txt: Added.
      * editing/selection/verify-editing-behavior-for-line-granularity.html: Added.
      Layout testcase added for verifying that the up/down caret movement between
      lines is similar to the default text editor behavior on various platforms.
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@153992 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      33d62de8
    • berto@igalia.com's avatar
      [GTK] Closing inspector window crashes wk · c55ca377
      berto@igalia.com authored
      https://bugs.webkit.org/show_bug.cgi?id=110865
      
      Reviewed by Carlos Garcia Campos.
      
      Source/WebKit/gtk:
      
      Deleting priv->corePage during the destruction of a webView will
      trigger the deletion of InspectorFrontendClient. However that
      object is supposed to handle the webView's destroy signal first in
      order to do the necessary cleanup.
      
      The solution is to wait until webkit_web_view_dispose finishes
      before deleting priv->corePage.
      
      * webkit/webkitwebview.cpp:
      (webkit_web_view_dispose):
      
      Tools:
      
      Disconnect all signal handlers if the inspector window is
      destroyed.
      
      * GtkLauncher/LauncherInspectorWindow.c:
      (inspectorWindowDestroyed):
      (launcherInspectorWindowNew):
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@153991 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      c55ca377
    • abucur@adobe.com's avatar
      [CSS Regions] Compute correct region ranges for boxes · 33159da0
      abucur@adobe.com authored
      https://bugs.webkit.org/show_bug.cgi?id=116296
      
      Reviewed by David Hyatt.
      
      Source/WebCore:
      
      The patch extends the region ranges implementation with the following behavior:
      - the range of a box is always included in the range of its containing block (even for floats); this will simplify how overflow
      is propagated to the regions and it should later change on a case by case basis.
      - if the range of a box is not correctly estimated before the layout it will be marked for relayout; this is necessary to correctly
      position boxes that overflow naturally, like floats.
      - all the boxes have a range now, not only the blocks.
      - regionAtBlockOffset can clamp to a box.
      - an unsplittable box has a region range of a single region, the one where the box top falls into.
      
      When the layout of a child box starts it tries to give an estimate based on the maximal height of the box. After the layout,
      if the estimation was wrong, the box is relaid out. For example, if a block with a float is laid out in a region but the float
      overflows in the next region a relayout is needed for the float so it can be positioned relative to the containing block
      region (this step can be optimized by making a layout pass only for the children that don't have the region range enclosed in the
      containing block range).
      
      Tests: fast/regions/bottom-overflow-out-of-first-region-absolute.html
             fast/regions/float-pushed-width-change-2.html
             fast/regions/scrollable-paragraph-unsplittable.html
      
      * rendering/RenderBlock.cpp:
      (WebCore::RenderBlock::updateShapesBeforeBlockLayout): Doesn't do anything for regions now.
      (WebCore::RenderBlock::updateShapesAfterBlockLayout): Doesn't do anything for regions now.
      (WebCore::RenderBlock::relayoutToAvoidWidows): The layout to avoid widows. Widows use the line index to determine the break point,
      not the RootLineBox pointer as before. This is necessary to prevent stale pointers in further layouts.
      (WebCore::RenderBlock::layoutBlock):
      (WebCore::RenderBlock::layoutBlockChild):
      (WebCore::RenderBlock::layoutPositionedObjects):
      (WebCore::RenderBlock::markForPaginationRelayoutIfNeeded): Relayout for widows during layoutBlock(). Don't wait to exit the block layout.
      (WebCore::RenderBlock::positionNewFloats):
      (WebCore::RenderBlock::hasNextPage):
      (WebCore::RenderBlock::applyBeforeBreak):
      (WebCore::RenderBlock::applyAfterBreak):
      (WebCore::RenderBlock::setPageBreak):
      (WebCore::RenderBlock::updateMinimumPageHeight):
      (WebCore::RenderBlock::regionAtBlockOffset): This function can now clamp at a box.
      (WebCore::RenderBlock::computeRegionRangeForBoxChild): Calculate the box region range using the box height.
      (WebCore::RenderBlock::estimateRegionRangeForBoxChild): Estimate the box region range using a huge height for the box.
      (WebCore::RenderBlock::updateRegionRangeForBoxChild): Determine the final region range of a box.
      * rendering/RenderBlock.h:
      (WebCore::RenderBlock::logicalWidthForChild):
      (WebCore::RenderBlock::logicalHeightForChild):
      (WebCore::RenderBlock::logicalTopForChild):
      * rendering/RenderBlockLineLayout.cpp:
      (WebCore::RenderBlock::layoutInlineChildren):
      * rendering/RenderBox.cpp:
      (WebCore::RenderBox::clampToStartAndEndRegions):
      (WebCore::RenderBox::borderBoxRectInRegion):
      * rendering/RenderBox.h:
      * rendering/RenderDeprecatedFlexibleBox.cpp:
      (WebCore::RenderDeprecatedFlexibleBox::layoutBlock):
      * rendering/RenderFlexibleBox.cpp:
      (WebCore::RenderFlexibleBox::layoutBlock):
      * rendering/RenderFlowThread.cpp:
      (WebCore::RenderFlowThread::validateRegions):
      (WebCore::RenderFlowThread::regionAtBlockOffset):
      (WebCore::RenderFlowThread::adjustedPositionRelativeToOffsetParent):
      (WebCore::RenderFlowThread::pageLogicalTopForOffset):
      (WebCore::RenderFlowThread::pageLogicalWidthForOffset):
      (WebCore::RenderFlowThread::pageLogicalHeightForOffset):
      (WebCore::RenderFlowThread::pageRemainingLogicalHeightForOffset):
      (WebCore::RenderFlowThread::mapFromFlowToRegion):
      (WebCore::RenderFlowThread::logicalWidthChangedInRegionsForBlock):
      (WebCore::RenderFlowThread::clearRenderObjectCustomStyle):
      (WebCore::RenderFlowThread::clearRenderBoxRegionInfoAndCustomStyle):
      (WebCore::RenderFlowThread::setRegionRangeForBox):
      (WebCore::RenderFlowThread::applyBreakAfterContent):
      (WebCore::RenderFlowThread::addForcedRegionBreak):
      * rendering/RenderFlowThread.h:
      * rendering/RenderGrid.cpp:
      (WebCore::RenderGrid::layoutBlock):
      * rendering/RenderMultiColumnFlowThread.cpp:
      (WebCore::RenderMultiColumnFlowThread::setPageBreak):
      (WebCore::RenderMultiColumnFlowThread::updateMinimumPageHeight):
      * rendering/RenderMultiColumnFlowThread.h:
      
      LayoutTests:
      
      Tests for the changes in how region ranges are computed and applied.
      
      * fast/regions/bottom-overflow-out-of-first-region-absolute-expected.html: Added.
      * fast/regions/bottom-overflow-out-of-first-region-absolute.html: Added.
      * fast/regions/float-pushed-width-change-2-expected.html: Copied from LayoutTests/fast/regions/float-pushed-width-change-expected.html.
      * fast/regions/float-pushed-width-change-2.html: Copied from LayoutTests/fast/regions/float-pushed-width-change.html.
      * fast/regions/float-pushed-width-change-expected.html:
      * fast/regions/float-pushed-width-change.html:
      * fast/regions/scrollable-paragraph-unsplittable-expected.html: Added.
      * fast/regions/scrollable-paragraph-unsplittable.html: Added.
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@153990 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      33159da0