summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
...
* Fix linker errors when Qt is built statically.Sergio Martins2014-02-131-1/+4
| | | | | | | | | moc_qwebkitplatformplugin.o was both present in libWebKit1.a and libQt5WebKitWidgets.a resulting in "multiple definition of" errors when building apps against a static webkit. Change-Id: I180b6288f39d9862a1b828f1bf295889b8e99955 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@digia.com>
* Merge remote-tracking branch 'origin/stable' into devFrederik Gladhorn2014-02-1219-70/+207
|\ | | | | | | Change-Id: Iea6cf3f34fb33f2d138b72243b0e688958d9424e
| * [mips] Wrong register usage in LLInt.old/5.2Balazs Kilvady2014-02-102-52/+115
| | | | | | | | | | | | | | | | | | | | | | Fix register usage and add PIC header to all LLInt operations. This patch is taken from https://bugs.webkit.org/show_bug.cgi?id=125168. It fixes the crash of fast/js/exception-propagate-from-dfg-to-llint test on mips architecture. Change-Id: I98ad3b5766451cab48a76f7e028b210f9ebe99ed Reviewed-by: Allan Sandfeld Jensen <allan.jensen@digia.com>
| * Fix build with QT_NO_GRAPHICSVIEWSergio Martins2014-02-092-0/+6
| | | | | | | | | | Change-Id: Ia909a560db58d6cdc6ec4bbac879249eb5e286ef Reviewed-by: Allan Sandfeld Jensen <allan.jensen@digia.com>
| * Update device pixel ratio on screen changeAllan Sandfeld Jensen2014-02-075-6/+40
| | | | | | | | | | | | | | | | | | This patch makes QWebPage listen to QWindow::screenChangedEvent and update the device pixel ratio when it is received. Task-number: QTBUG-36190 Change-Id: Ia28571ef241a3a2f970876f6e474b01c1ac001e1 Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
| * Fix memory leaks in platform/image-encoders/JPEGImageEncoder.cppLiangjun Zeng2014-02-071-1/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | https://bugs.webkit.org/show_bug.cgi?id=118781 Reviewed by Brent Fulgham. We can find the function "jpeg_finish_compress" call the function "jpeg_abort" at the end. And the comments of "jpeg_abort" is "Abort processing of a JPEG compression operation, but don't destroy the object itself". (We can find these in the "jcapimin.c" of jpeg) So the compression object destroy need be called. No new tests because this doesn't change functionality. * platform/image-encoders/JPEGImageEncoder.cpp: (WebCore::compressRGBABigEndianToJPEG): Change-Id: Ic58b53c4e6bada7e92e09825be9e770f1c5445c1 git-svn-id: http://svn.webkit.org/repository/webkit/trunk@158280 268f45cc-cd09-0410-ab3c-d52691b4dbfc Reviewed-by: Michael Bruning <michael.bruning@digia.com>
| * 8 bytes memory leaks in WebProcessMainQt.cpp when proxy is initializedMinju Kim2014-02-071-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | https://bugs.webkit.org/show_bug.cgi?id=118515 Patch by Minju Kim <pmuarteo@gmail.com> on 2013-09-25 Reviewed by Christophe Dumez. * WebProcess/qt/WebProcessMainQt.cpp: (WebKit::initializeProxy): Handled EnvHttpProxyFactory by using OwnPtr to avoid memory leak when it isn't handled by NetworkProxyFactory Change-Id: I51833871d8e11441ffe60d51629814e64f77d792 git-svn-id: http://svn.webkit.org/repository/webkit/trunk@156449 268f45cc-cd09-0410-ab3c-d52691b4dbfc Reviewed-by: Michael Bruning <michael.bruning@digia.com>
| * TextureMapperGL must take hiDPI transform into accountAllan Sandfeld Jensen2014-02-061-1/+10
| | | | | | | | | | | | | | | | | | | | Since TextureMapperGL does its own native rendering it must itself apply the all transforms QPainter would normally do behind the scenes. Task-number: QTBUG-36508 Change-Id: Id4a0ee64ff7ac5c5cb51e45ffd2e4e25ed7fce74 Reviewed-by: Michael Bruning <michael.bruning@digia.com> Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
| * Fix transform of accelerated layers on hiDPI paintersAllan Sandfeld Jensen2014-02-011-3/+5
| | | | | | | | | | | | | | | | | | | | | | A mismatch of reading combinedTransform but setting worldTransform means that the pattern of only storing and restoring the transform when temporarily overriding it, causes the hiDPI transform that combinedTransform contains to be applied multiple times. Task-number: QTBUG-36508 Change-Id: I512f0752016f55935595ef259913660b4ee78ae7 Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
| * Fix assert in m_stack.isEmpty() in GraphicsContext::~GraphicsContext()Allan Sandfeld Jensen2014-01-312-1/+10
| | | | | | | | | | | | | | | | | | | | | | | | | | Save and restore is not balanced in GraphicsContext. This causes the log error "ERROR void GraphicsContext::restore() stack is empty" and assert in the GraphicsContext destructor. The cause is a mismatch of GraphicsContexts being saved and restored by TextureMapperImageBuffer beginClip and endClip. This patch fixes endClip to restore the currentContext beginClip saved. Change-Id: Ief0100dd26c7eded04734167e86712a81f1a7674 Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
| * Fix assert in BitmapTextureImageBuffer::applyFilters()Allan Sandfeld Jensen2014-01-311-2/+3
| | | | | | | | | | | | | | | | | | | | | | While WebCore should not give a BitmapTextureImageBuffer any filters applyFilters() will still be called with an empty list of filters. We should therefore not assert that applyFilters is never called by that is only ever called with an empty list of filters to apply. Change-Id: Ic24b735f1fc92e2a57c7ef068a8c7d712f0fc31f Reviewed-by: Michael Bruning <michael.bruning@digia.com>
| * Fix RenderGeometryMap.cpp ASSERT on Google PlusAllan Sandfeld Jensen2014-01-311-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | The assert roundedIntPoint(result) == roundedIntPoint(rendererMappedResult) is triggered on Google Plus. The cause appears to be premature rounding leading a wrong result. The fix is a to remove the early rounding. This follows a similar fix in upstream WebKit r142638 where the same problem was fixed for the other mapToContainer method. Change-Id: Ifae91c204f2043b3f07682dc7bed711a6cb38c91 Reviewed-by: Michael Bruning <michael.bruning@digia.com> Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
| * Don't pollute .prl & .pc files with private dependenciesJocelyn Turcotte2014-01-312-2/+3
| | | | | | | | | | | | | | | | | | | | | | | | Also change the creating_module check in needToLink to make sure that only the main module will try to link directly with the WebCore and JavaScriptCore static libraries. Task-number: QTBUG-35774 Initial-patch-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com> Change-Id: Ib5992e112803679837249400e98cb8c22636096d Reviewed-by: Oswald Buddenhagen <oswald.buddenhagen@digia.com> Reviewed-by: Allan Sandfeld Jensen <allan.jensen@digia.com>
| * Apply device pixel ratio to transparency layersAllan Sandfeld Jensen2014-01-311-2/+4
| | | | | | | | | | | | | | | | | | | | | | | | The transparency layers are currently created with taking hiDPI into account. This means that any CSS opacity causes a shift in rendering quality. This patch forwards the device pixelratio for any QPaintDevice to transparency layers created for that device. Change-Id: Ie65f76771a4cc59bc714a9b89511cab1f7617abc Reviewed-by: Michael Bruning <michael.bruning@digia.com>
| * Forward devicePixelRatio to WebCore PageAllan Sandfeld Jensen2014-01-313-0/+7
| | | | | | | | | | | | | | | | | | The device pixel ratio must be set on Page for CSS image set and CSS media queries to work correctly. Task-number: QTBUG-36190 Change-Id: I76694c1375194f35df7f62e942bc8565f065a3bc Reviewed-by: Michael Bruning <michael.bruning@digia.com>
| * Do not notifyFinished on resources never loadedAllan Sandfeld Jensen2014-01-301-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | If autoload images is disabled, the CachedImage element never starts loading but still needs loading. We therefore need to check both isLoading() and stillNeedsLoad() to determine if loading is complete. This fixes a serious memory leak when autoload images is disabled. Task-number: QTBUG-34494 Change-Id: I36811179d95e23651fabe484fbde08e2881d822b Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
* | Make building QtTestBrowser simplerAllan Sandfeld Jensen2014-02-051-1/+1
| | | | | | | | | | | | | | | | | | | | | | The QtTestBrowser needs qttestsupport and qttestsupport needs fontconfig on linux, unfortunately fontconfig is only enabled on webkit developer builds. This patch add fontconfig when qttestsupport is enabled. This makes it possible to build the QtTestBrowser in a production_build by only overriding build_testbrowser and build_qttestsupport. Change-Id: I7d49368bee82bb1eb7bc9b5d4eeaf714a3d737f8 Reviewed-by: Michael Bruning <michael.bruning@digia.com>
* | Disable CSS custom filtersAllan Sandfeld Jensen2014-02-051-3/+0
| | | | | | | | | | | | | | | | | | CSS custom filters aka CSS shaders have been removed from Chrome and Safari. We should follow the same steps and disable the feature in our builds. Change-Id: Iac63a48c1853683d764ee0401b45d15530c37e2b Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
* | Use WebKit's copy of leveldb when system version is not availableAllan Sandfeld Jensen2014-02-0511-6/+164
| | | | | | | | | | | | | | | | | | To support indexed database webkit needs leveldb, therefore webkit contains a copy of leveldb which we can use to provide support when the system doesn't ship with a suitable version. Change-Id: I231694e6a7c7cb83855bfed3f9ee24fbd2795362 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* | Fix ambiguous castAllan Sandfeld Jensen2014-02-041-1/+1
| | | | | | | | | | | | | | | | | | The conversion from uint64_t to QFlags is now ambiguous, but since we need an int QVariant in the end anyway, we can cast directly to that. Change-Id: I4478efa2b293a59b47c46d27346b56eab48b3278 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com> Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
* | Unable to upload <img src="foo.svg"> as WebGL texturedino@apple.com2014-01-292-7/+33
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | https://bugs.webkit.org/show_bug.cgi?id=123035 Reviewed by Tim Horton. Source/WebCore: If the HTMLImageElement passed to texture2D is an SVG image, paint it first into a bitmap buffer and upload that. Note that the SVG image still needs to have an intrinsic or explicit size - see how the test case must set width and height. I also renamed the cache of ImageBuffers since it is no longer only being used for video frames. * html/canvas/WebGLRenderingContext.cpp: (WebCore::WebGLRenderingContext::WebGLRenderingContext): Rename m_videoCache to m_generatedImageCache. (WebCore::WebGLRenderingContext::drawImageIntoBuffer): New method that creates an ImageBuffer of the appropriate size and renders into that. (WebCore::WebGLRenderingContext::texImage2D): If we see an SVG image, render it first. (WebCore::WebGLRenderingContext::videoFrameToImage): Renamed m_generatedImageCache. (WebCore::WebGLRenderingContext::texSubImage2D): If we see an SVG image, render it first. * html/canvas/WebGLRenderingContext.h: Renaming. Change-Id: I601eff1f3a8c11fe2a62f8b5a8a914d6eadfead4 git-svn-id: http://svn.webkit.org/repository/webkit/trunk@157647 268f45cc-cd09-0410-ab3c-d52691b4dbfc Reviewed-by: Allan Sandfeld Jensen <allan.jensen@digia.com>
* | Allow SVG images to be drawn into canvas without tainting.timothy@apple.com2014-01-297-18/+58
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | https://bugs.webkit.org/show_bug.cgi?id=119492 Reviewed by Darin Adler. Source/WebCore: * html/HTMLAnchorElement.cpp: (WebCore::HTMLAnchorElement::parseAttribute): Call shouldProhibitLinks. (WebCore::shouldProhibitLinks): Added. * html/HTMLAnchorElement.h: * html/HTMLImageElement.cpp: (WebCore::HTMLImageElement::parseAttribute): Call shouldProhibitLinks. * rendering/svg/RenderSVGRoot.cpp: (WebCore::RenderSVGRoot::isEmbeddedThroughSVGImage): Use isInSVGImage. * svg/SVGAElement.cpp: (WebCore::SVGAElement::svgAttributeChanged): Call shouldProhibitLinks. * svg/graphics/SVGImage.cpp: (WebCore::SVGImage::hasSingleSecurityOrigin): Added. (WebCore::isInSVGImage): Added. * svg/graphics/SVGImage.h: Change-Id: I294821dff7299e3c6a3e1df8f8ff3d855dc16dbd git-svn-id: http://svn.webkit.org/repository/webkit/trunk@153876 268f45cc-cd09-0410-ab3c-d52691b4dbfc Reviewed-by: Allan Sandfeld Jensen <allan.jensen@digia.com>
* | Detect AArch64 architectureAllan Sandfeld Jensen2014-01-292-1/+10
| | | | | | | | | | | | | | | | | | | | | | | | Adds WTF platform support for the AArch64 architecture. Patch is based on WebKit-gtk patch done by Riku Voipio, and was cherry-picked and tested by Marcin Juszkiewicz. Task-number: QTBUG-35442 Change-Id: I81b8c43fb3a6a65c56446f84a7b1001075998d1e Reviewed-by: Lisandro Damián Nicanor Pérez Meyer <perezmeyer@gmail.com> Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
* | Merge "Merge remote-tracking branch 'origin/stable' into dev" into ↵Frederik Gladhorn2014-01-272-9/+27
|\ \ | | | | | | | | | refs/staging/dev
| * \ Merge remote-tracking branch 'origin/stable' into devFrederik Gladhorn2014-01-272-9/+27
| |\ \ | | |/ | | | | | | Change-Id: Idc5d37689051103b5101a33ecb2fec69b30d8987
| | * Make sure the correct position for the plugin is used when showingAndy Shaw2014-01-241-3/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When the plugin was windowed then it was always being shown at 0x0 of the containing window rather than the correct position of the plugin inside the containing window. This enesures it is placed correctly in that case. Task-number: QTBUG-31730 QTBUG-33053 QTBUG-32546 Change-Id: I5ce0321924d288437d483002e9cb6dd1c98b6863 Reviewed-by: Michael Bruning <michael.bruning@digia.com>
| | * Merge remote-tracking branch 'origin/release' into stableFrederik Gladhorn2014-01-221-6/+23
| | |\ | | | | | | | | | | | | Change-Id: I82f158687d0b5da3e85c025d720111b6b1414010
| | | * Reduce memory footprint when linkingv5.2.1Allan Sandfeld Jensen2014-01-161-6/+23
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Linking QtWebKit on Linux x64 with debug symbols can use up to 11Gbyte of memory. This can make it impossible to link on many systems. This patch forces GCC 4.8+ to fully use DWARF-4 debug-types sections to eleminate duplicate symbols and makes earlier GCC versions STABS which also uses much less memory. Task-number: QTBUG-36131 Change-Id: Ib1ba3e073b5984ead3a0badb935e7b89c01b962b Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
* | | | Replace win32-g++ with mingw scopeKai Koehne2014-01-275-5/+5
|/ / / | | | | | | | | | | | | | | | | | | | | | | | | | | | Commit 773dd01 in qtbase introduced a general mingw platform scope, which is cleaner and more flexible than matching the spec name. (see 278152fff for a similar commit in qtbase). Change-Id: Id3dcd3474504aebeee16dc41cd1eeab7ac21b319 Reviewed-by: Allan Sandfeld Jensen <allan.jensen@digia.com>
* | | Simplify ImageBuffer conversion using RGBA8888 formatsAllan Sandfeld Jensen2014-01-231-106/+13
| | | | | | | | | | | | | | | | | | | | | | | | | | | Using the RGBA8888 formats introduced in Qt 5.2, we can remove the corresponding code in WebKit. This greatly simplifies the code. Change-Id: I3df43995c4df53610fc881c68998188ec08370f9 Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com> Reviewed-by: Michael Bruning <michael.bruning@digia.com>
* | | Merge remote-tracking branch 'origin/stable' into devFrederik Gladhorn2014-01-171-1/+13
|\ \ \ | |/ / | | | | | | | | | | | | | | | Conflicts: .qmake.conf Change-Id: I8721e8abb6b795242b5304ec28cf7cda56bfbce9
| * | Bump MODULE_VERSION to 5.2.2Sergio Ahumada2014-01-161-1/+1
| | | | | | | | | | | | | | | Change-Id: Ic00000a7fe9f261301096ac25aade719f2656201 Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
| * | Ensure that cursor rectangle is always valid.Luca Ottaviano2014-01-151-1/+13
| |/ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The cursor rectangle is given an invalid size when the user clicks on an input field; the reported size is 0 pixel wide. The cursor is then converted to a QRect, which is invalid if rectangle width is 0. This can interefere with input methods that check for cursorRect validity before reporting a cursorRect change. Make sure that a cursor rectangle with at least one valid dimension becomes a valid QRect. Task-number: QTBUG-35996 Change-Id: I4026f1136cfb006efc5d7915f0f10c1b5187c730 Signed-off-by: Luca Ottaviano <lottaviano@develer.com> Reviewed-by: Allan Sandfeld Jensen <allan.jensen@digia.com>
* | Make indexed database follow Qt web settings if enabledAllan Sandfeld Jensen2014-01-168-21/+65
| | | | | | | | | | | | | | | | | | | | | | | | | | Forwards the websettings for offline database storage which so far has been used for web SQL database to indexed database. Note that this does not add new API and does not make it possible to control web sql database and indexed database separately. The patch also adds a config test for leveldb which indexed db requires. Change-Id: I8808356c9910848c43c5f47e6279a1c632f17ad1 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
* | Enable HTML5 video track supportAllan Sandfeld Jensen2014-01-169-15/+148
| | | | | | | | | | | | | | | | | | | | | | | | This patch enables support for the HTML5 video track standard which defines subtitles and captions for HTML5 video. The HTML5 video controls will now have a popup button when captions are available that allows the user to select the preferred caption or turn them off. Change-Id: Id5b837d1a7b536935cd5038812594b11c48c3480 Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
* | Enable download attributeAllan Sandfeld Jensen2014-01-161-1/+1
| | | | | | | | | | | | | | | | The download attribute enables HTML links to indicate they should default to downloading and suggest a filename. Change-Id: I06e41fd9532c14657a1694c1ce69680814fd9d07 Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
* | Fix custom builds with indexeddb enabledAllan Sandfeld Jensen2014-01-156-3/+33
| | | | | | | | | | | | | | | | | | | | Our build-system needs a few updates to make indexeddb enablable again. This patch does not provide any API for indexeddb or enable indexeddb itself. It only updates the build system for default disabled feature. Change-Id: I1ca61bda0913f698bbdf716d286a50051dbe1a99 Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
* | Merge remote-tracking branch 'origin/stable' into devFrederik Gladhorn2014-01-103-1/+7
|\ \ | |/ | | | | Change-Id: Id10708a0114ed30b712426b9e78168d288691b2e
| * Check also for mingw-g++-xxx specsKai Koehne2014-01-081-1/+1
| | | | | | | | | | | | | | | | Some packagers/distributions have individual mingw-g++ mkspecs, e.g. for cross-compiling. Change-Id: I5deb2d543934b16c4df63cd8ec240e6ec9b64d9b Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
| * Reapply mouse event handling in QQuickWebViewFlickablePrivate.Michael Brüning2014-01-072-0/+6
| | | | | | | | | | | | | | | | | | | | | | | | This reapplies commit 0b789c6a785d5ef45c3f6f2fc0f8694772b3d392, which had been in Qt 5.0.x and 5.1.x, but was not upstreamed and hence overwritten when importing the new snapshot. This basically reverts http://trac.webkit.org/changeset/136119. Task-number: QTBUG-35097 Change-Id: I3e02c3feeb5193c0c06b7525469e208de1d6cafc Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
* | Detect MIPS64 architectureAllan Sandfeld Jensen2014-01-072-4/+12
| | | | | | | | | | | | | | | | | | | | Adds WTF platform support for the MIPS64 architecture. Patch is based on http://patches.openembedded.org/patch/51625/ Task-number: QTBUG-35394 Change-Id: I270d2d89cc1d93f06f5b40a6d1042ad5fbac2fe7 Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
* | Merge remote-tracking branch 'origin/stable' into devFrederik Gladhorn2014-01-0616-35/+67
|\ \ | |/ | | | | Change-Id: Ib78430b6b1cf0c9fca73466309318c43a68c70b1
| * [Qt] Re-enable plugins on Mac.Jocelyn Turcotte2014-01-039-15/+14
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | https://bugs.webkit.org/show_bug.cgi?id=116622 Reviewed by Tor Arne Vestbø. Source/WebCore: * platform/FileSystem.h: * platform/qt/FileSystemQt.cpp: (WebCore::unloadModule): - Q_WS_* aren't defined since Qt5, use Q_OS_MACX instead. * plugins/mac/PluginPackageMac.cpp: (WebCore::PluginPackage::fetchInfo): (WebCore::PluginPackage::load): - createCFString now does the adoption itself. * plugins/mac/PluginViewMac.mm: - Add missing include. Source/WebKit/qt: * WebCoreSupport/ChromeClientQt.cpp: (WebCore::ChromeClientQt::windowResizerRect): - Add missing parenthese. Tools: Plugins on Mac were only enabled if !embedded but the later was set if QPA is enabled, which is the default since Qt 5.0. Remove references to 'embedded' and fix various build issues, PluginViewMac.mm and PluginPackageMac.cpp haven't been compiled since a few months. * DumpRenderTree/qt/TestNetscapePlugin/TestNetscapePlugin.pro: * qmake/mkspecs/features/features.prf: * qmake/mkspecs/features/unix/default_pre.prf: Task-number: QTBUG-35899 git-svn-id: http://svn.webkit.org/repository/webkit/trunk@151109 268f45cc-cd09-0410-ab3c-d52691b4dbfc Change-Id: I49692a234a66c205099c5dde8eb24ffb6eadba0f Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
| * Use pkg-config to build against ICUAllan Sandfeld Jensen2014-01-031-0/+1
| | | | | | | | | | | | | | | | | | To get the correct combination of ICU includes and libraries we should use the system information available in pkg-config. Without this we might get build errors on some Linux distributions. Change-Id: I0e803e914467312c8ae586d41e08d1fda2b510b6 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
| * Remove dependency v8-privateAllan Sandfeld Jensen2014-01-021-1/+1
| | | | | | | | | | | | | | | | When building the WebKit2 tests we still try to link against v8-private, which no longer exists. Change-Id: I4848cb62ce696e941947930fb099d23a7ce0b27e Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
| * Extend disabling of whole-program-optimizations to MSVC 2013Allan Sandfeld Jensen2014-01-021-1/+1
| | | | | | | | | | | | | | | | | | Building with whole-program-optimizations remains a problem with MSVC 2013, so we must continue to disable it. Task-number: QTBUG-35835 Change-Id: Ic3131495ce0ce323ad041e51332969cc01b741c5 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
| * ASSERTION FAILED: m_repaintRect == ↵simon.fraser@apple.com2014-01-022-15/+47
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | renderer().clippedOverflowRectForRepaint(renderer().containerForRepaint()) after r135816 https://bugs.webkit.org/show_bug.cgi?id=103432 Reviewed by Dave Hyatt. RenderLayer caches repaint rects in m_repaintRect, and on updating layer positions after scrolling, asserts that the cached rect is correct. However, this assertion would sometimes fail if we were scrolling as a result of doing adjustViewSize() in the middle of layout, because we haven't updated layer positions post-layout yet. Fix by having the poorly named FrameView::repaintFixedElementsAfterScrolling() skip the layer updating if this FrameView is inside of adjusetViewSize() in layout. In order to know if we're inside view size adjusting, add a LayoutPhase member to FrameView, replacing two existing bools that track laying out state. Investigative work showed that there are many, many ways to re-enter FrameView::layout(), which makes it hard (but desirable) to more assertions about state changes, but indicated that saving and restoring the state (via TemporaryChange<LayoutPhase>) was a good idea. * page/FrameView.cpp: (WebCore::FrameView::FrameView): (WebCore::FrameView::reset): (WebCore::FrameView::updateCompositingLayersAfterStyleChange): (WebCore::FrameView::layout): (WebCore::FrameView::repaintFixedElementsAfterScrolling): * page/FrameView.h: Change-Id: If79914ee00b9250831c935bca8e7dcccf485ecf8 git-svn-id: http://svn.webkit.org/repository/webkit/trunk@159218 268f45cc-cd09-0410-ab3c-d52691b4dbfc Reviewed-by: Bartosz Brachaczek <b.brachaczek@gmail.com> Reviewed-by: Allan Sandfeld Jensen <allan.jensen@digia.com>
| * OS X Mavericks/Xcode5: Find libxslt and libxml2 headersEike Ziller2013-12-263-3/+3
| | | | | | | | | | | | | | | | | | The development headers are no longer installed in a system location, but only in the sysroot. Change-Id: Ie9b85ff638c435e7e4d0a538f6e72d2cdbae88c8 Reviewed-by: Simon Hausmann <simon.hausmann@digia.com> Reviewed-by: Allan Sandfeld Jensen <allan.jensen@digia.com>
* | Merge remote-tracking branch 'origin/stable' into devFrederik Gladhorn2013-12-237-32/+66
|\ \ | |/ | | | | Change-Id: I31a326945d624b10d5c87703d0e2b8ac723a05ab
| * Do not accept all touch eventsAllan Sandfeld Jensen2013-12-203-9/+15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | QtWebKit currently accepts all touch events to prevent the lack of accept on a TouchBegin from preventing receiving TouchUpdate and TouchEnd. This behavior prevents QGuiApplication from converting touch events to mouse events, and makes the WebView widget impossible to interact with by touch. This patch modifies the behavior slighly so that QtWebKit only accepts touch events if they are needed for the active document. Task-number: QTBUG-31938 Change-Id: Iba80ebfaf31c93f82356e3f0622039120921172a Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
View Lorry Controller Status