2012-09-18 Szilard Ledan EWS shouldn't sleep if there are new patches in its queue https://bugs.webkit.org/show_bug.cgi?id=83038 Reviewed by Eric Seidel. EWS tries to process a security patch. Of course it can't, because the EWS isn't the member of the security group. But the problem is that after it can't process the attachment, it says that queue is empty (but it isn't!) and it sleeps 2 minutes and push the security patch to the end of the queue. Now it stays in the loop until it finds a patch or the queue gets empty. * Scripts/webkitpy/tool/commands/queues.py: (AbstractPatchQueue._next_patch): * Scripts/webkitpy/tool/commands/queues_unittest.py: (AbstractPatchQueueTest.test_next_patch): 2012-09-18 Mikhail Pozdnyakov [WK2][WTR] InjectedBundle::booleanForKey() should handle literals effectively https://bugs.webkit.org/show_bug.cgi?id=97014 Reviewed by Kenneth Rohde Christiansen. According to http://trac.webkit.org/wiki/EfficientStrings WTF::StringBuilder::appendLiteral() shall be used for literals rather than WTF::StringBuilder::append(). * WebKitTestRunner/InjectedBundle/InjectedBundle.cpp: (WTR::InjectedBundle::booleanForKey): 2012-09-18 Simon Hausmann Update my e-mail address. * Scripts/webkitpy/common/config/committers.py: 2012-09-18 Andras Becsi Update my e-mail address. Unreviewed. * Scripts/webkitpy/common/config/committers.py: 2012-09-18 Andras Becsi [Qt] qt_webkit.pri should not be listed in Tools.pro Reviewed and rubber-stamped by Simon Hausmann and Tor Arne Vestbø. Since r128751 the module pri file is auto-generated but it was still listed in OTHER_FILES. * Tools.pro: Remove unneeded line. 2012-09-18 Simon Hausmann [Qt] Fix build with some versions of the gold linker Reviewed by Tor Arne Vestbø. Don't unconditionally pass --no-keep-memory to the linker, some versions might not support it. Instead run a compile/link test first to see if it works. * qmake/config.tests/gnuld/gnuld.pro: Added. * qmake/config.tests/gnuld/main.cpp: Added. (main): * qmake/mkspecs/features/unix/default_post.prf: 2012-09-18 Simon Hausmann [Qt] Remove forced use of gold. Reviewed by Tor Arne Vestbø. The choice of what linker to use with WebKit should be taken by Qt's build system and ideally the same for all modules of Qt. Then in turn it is usually up to the administrator of the machine. Recent Debian based systems often offer the automatic use of gold through a symlink and a dpkg-diversion when installing the gold package. * qmake/mkspecs/features/unix/default_post.prf: 2012-09-18 Simon Hausmann [Qt] Prospective Qt/Windows cross-compiling fix Reviewed by Tor Arne Vestbø. The win32 scope is not set when cross-compiling from Linux to Windows. * qmake/mkspecs/features/functions.prf: 2012-09-18 Seokju Kwon [EFL] Remove background view on EWebLauncher and MiniBrowser https://bugs.webkit.org/show_bug.cgi?id=96905 Reviewed by Gyuyoung Kim. The size of webview was changed after adding url bar. And the background view is not necessary anymore, since it was used for debugging back in the day. * EWebLauncher/main.c: (_ELauncher): (on_ecore_evas_resize): (browserCreate): * MiniBrowser/efl/main.c: (_MiniBrowser): (on_ecore_evas_resize): (browserCreate): 2012-09-18 Sergio Villar Senin [GTK] run-webkit-tests unable to find TestExpectations for WK2 https://bugs.webkit.org/show_bug.cgi?id=96998 Reviewed by Philippe Normand. We should look for TestExpectations files in all the locations where we currently look for Skipped files. This will allow run-webkit-tests to look for TestExpectations files in platform/gtk-wk2 and platform/wk2 if the "-2" flag is used. * Scripts/webkitpy/layout_tests/port/gtk.py: (GtkPort.expectations_files): 2012-09-18 Tor Arne Vestbø [Qt] Fix build without the QtQuick module Reviewed by Simon Hausmann. * qmake/mkspecs/features/features.prf: 2012-09-18 Stephanie Lewis Build fix after http://trac.webkit.org/projects/webkit/changeset/128852. Unreviewed. * DumpRenderTree/mac/DumpRenderTree.mm: (poseAsClass): 2012-09-18 Allan Sandfeld Jensen Unreviewed update of email addresses for Berlin QtWebKit office. * Scripts/webkitpy/common/config/committers.py: 2012-09-17 Zan Dobersek [GTK] fast/loader/display-image-unset-can-block-image-and-can-reload-in-place.html failing after r128645 https://bugs.webkit.org/show_bug.cgi?id=96899 Reviewed by Martin Robinson. When overriding the 'WebKitDisplayImageKey' preference, use the 'auto-load-images' property of WebKitWebSettings as the property which should be updated with the corresponding preference value. * DumpRenderTree/gtk/DumpRenderTree.cpp: (resetDefaultsToConsistentValues): * DumpRenderTree/gtk/TestRunnerGtk.cpp: (TestRunner::overridePreference): 2012-09-17 Pratik Solanki DumpRenderTree and WebKitTestRunner should compile with -Wundef on Mac https://bugs.webkit.org/show_bug.cgi?id=96973 Reviewed by Dan Bernstein. * DumpRenderTree/mac/Configurations/Base.xcconfig: * WebKitTestRunner/Configurations/Base.xcconfig: * WebKitTestRunner/PlatformWebView.h: Use #ifdef __OBJC__ and not #if. 2012-09-17 Dirk Pranke nrwt: remove "unexpected EOF" warnings https://bugs.webkit.org/show_bug.cgi?id=96970 Reviewed by Ojan Vafai. After debugging this a bit, it looks like there aren't any cases that I can reproduce where a read() of zero indicates something actually wrong; either it is a prelude to a crash, or a false negative. So, I'm removing these warnings and adding a comment. * Scripts/webkitpy/layout_tests/port/server_process.py: (ServerProcess._wait_for_data_and_update_buffers_using_select): 2012-09-17 Dirk Pranke [chromium] ASAN bot is crashing at the end of the run https://bugs.webkit.org/show_bug.cgi?id=96967 Reviewed by Abhishek Arya. The ASAN bot is crashing attempting to decode some output into UTF-8; there's no reason to do this, so let's not do this and see if something else is going on as well. * Scripts/webkitpy/layout_tests/port/chromium.py: (ChromiumPort._get_crash_log): 2012-09-17 Dirk Pranke nrwt: --results-directory isn't getting printed properly https://bugs.webkit.org/show_bug.cgi?id=96965 Reviewed by Ojan Vafai. options.results_directory isn't actually initialized with the default values until after we call print_config(), so this changes things to print the value directly. * Scripts/webkitpy/layout_tests/run_webkit_tests.py: (run): * Scripts/webkitpy/layout_tests/views/printing.py: (Printer.print_config): * Scripts/webkitpy/layout_tests/views/printing_unittest.py: (Testprinter.test_print_config): 2012-09-17 Rob Buis [BlackBerry] Enable VIDEO_TRACK https://bugs.webkit.org/show_bug.cgi?id=96949 Reviewed by Antonio Gomes. * Scripts/webkitperl/FeatureList.pm: 2012-09-17 Rick Byers Add handling of new GestureTapCancel in DRT https://bugs.webkit.org/show_bug.cgi?id=96183 Reviewed by Antonio Gomes. * DumpRenderTree/chromium/TestWebPlugin.cpp: (TestWebPlugin::handleInputEvent): * DumpRenderTree/chromium/EventSender.cpp: (EventSender::gestureTapCancel): 2012-09-17 Philip Rogers Teach style checker about preprocessor directive indentation rules https://bugs.webkit.org/show_bug.cgi?id=96874 Reviewed by Adam Barth. Preprocessor directives (#ifdef, #include, #define, etc.) should not be indented. This is not explicit in our style guide but is generally followed in our code. Searching for violations in our codebase shows these are rarely indented: #include - indented in 6 files #ifdef - indented in 0 files #ifndef - indented in 1 file #define - indented in 11 files #if - indented in 7 files * Scripts/webkitpy/style/checkers/cpp.py: (check_directive_indentation): This is the simple test where we look for spaces followed by a #. (check_style): * Scripts/webkitpy/style/checkers/cpp_unittest.py: A few tests needed to be modified because they had unintentionally indented preprocessor directives. (CppStyleTest.test_build_class.Foo): (CppStyleTest.test_build_class): (CppStyleTest.test_build_class.DERIVE_FROM_GOO): (WebKitStyleTest.test_line_breaking): (WebKitStyleTest.test_directive_indentation): 2012-09-17 Tor Arne Vestbø [Qt] Auto-generate the module pri file for QtWebKit Reviewed by Simon Hausmann. * qmake/qt_webkit.pri: Removed. 2012-09-14 Dirk Pranke nrwt: --additional-platform-dir is broken on chromium ports https://bugs.webkit.org/show_bug.cgi?id=96840 Reviewed by Ojan Vafai. Looks like I broke this when cleaning things up as part of adding ML support. Fixing and adding a test. * Scripts/webkitpy/layout_tests/port/chromium.py: (ChromiumPort.default_baseline_search_path): * Scripts/webkitpy/layout_tests/port/chromium_android.py: (ChromiumAndroidPort.default_baseline_search_path): * Scripts/webkitpy/layout_tests/port/chromium_linux.py: (ChromiumLinuxPort.default_baseline_search_path): * Scripts/webkitpy/layout_tests/port/chromium_mac.py: (ChromiumMacPort.__init__): * Scripts/webkitpy/layout_tests/port/chromium_win.py: (ChromiumWinPort.setup_environ_for_server): * Scripts/webkitpy/layout_tests/port/port_testcase.py: (test_check_build): (test_additional_platform_directory): 2012-09-13 Stephanie Lewis mac-future ignores previous platform TestExpectations. https://bugs.webkit.org/show_bug.cgi?id=96718. Reviewed by Dirk Pranke. TestExpectations match an expectation's configuraton against a list of configurations that are valid for that port. That list does not contain mac-future so none of the expectations can be applied. * Scripts/webkitpy/layout_tests/port/apple.py: (ApplePort._generate_all_test_configurations): 2012-09-14 Adam Barth Remove webkitPostMessage https://bugs.webkit.org/show_bug.cgi?id=96577 Reviewed by Ojan Vafai. Add ENABLE_LEGACY_VENDOR_PREFIXES flag. * Scripts/webkitperl/FeatureList.pm: * qmake/mkspecs/features/features.pri: 2012-09-14 Alexey Proskuryakov [WK2] webarchive/loading/javascript-url-iframe-crash.html fails https://bugs.webkit.org/show_bug.cgi?id=96817 Reviewed by Kenneth Rohde Christiansen. * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp: (WTR::InjectedBundlePage::didFailProvisionalLoadWithErrorForFrame): This callback got overlooked when adding logging elsewhere. 2012-09-14 Peter Beverloo [Chromium] Build fix for DumpRenderTree following r128628 https://bugs.webkit.org/show_bug.cgi?id=96808 Reviewed by Tony Chang. When building DumpRenderTree from the Chromium project, a compile error shows up because including the header ForwardIOStreamsAndroid.h cannot be located. TestWebKitAPI does this correctly. * DumpRenderTree/DumpRenderTree.gyp/DumpRenderTree.gyp: 2012-09-14 Dana Jansens Change cc-bugs@google.com watchlist to cc-bugs@chromium.org https://bugs.webkit.org/show_bug.cgi?id=96805 Reviewed by Dirk Pranke. * Scripts/webkitpy/common/config/committers.py: * Scripts/webkitpy/common/config/watchlist: 2012-09-14 Peter Beverloo [Chromium] Support the --{in,out,err}-fifo arguments on TestWebKitAPI and webkit_unit_tests https://bugs.webkit.org/show_bug.cgi?id=96687 Reviewed by Tony Chang. Android's DumpRenderTree currently supports these arguments, implemented as part of TestShellAndroid: http://trac.webkit.org/browser/trunk/Tools/DumpRenderTree/chromium/TestShellAndroid.cpp?rev=128496 They're used by the layout test runner to get the STDOUT and STDERR while a layout test run is in process, which is a safer alternative to parsing all the logcat output manually. The implementation can be seen here: http://trac.webkit.org/browser/trunk/Tools/Scripts/webkitpy/layout_tests/port/chromium_android.py?rev=128496#L590 This patch generalizes parsing of and applying the effects of these arguments so that they can be used for TestWebKitAPI and webkit_unit_tests as well. After this patch, this will make it possible to pull out the output-reading code from Android's layout test port and generalize it so it can be re-used in the new test-runner for the other two test suites. This has no effect when compiling and running these tests as part of Chromium code, which has a much more advanced test-runner that does parse complete log output, but also directly depends on code licensed under Apache 2. * DumpRenderTree/DumpRenderTree.gyp/DumpRenderTree.gyp: * DumpRenderTree/chromium/TestShellAndroid.cpp: (platformInit): * TestWebKitAPI/TestWebKitAPI.gyp/TestWebKitAPI.gyp: 2012-09-14 Zoltan Horvath check-webkit-style should not warn in case of NONCOPYABLE and FAST_ALLOCATED macros https://bugs.webkit.org/show_bug.cgi?id=96656 Reviewed by Adam Barth. We should allow to use WTF_MAKE_NONCOPYABLE(ClassName) and WTF_MAKE_FAST_ALLOCATED macros in 1 line. * Scripts/webkitpy/style/checkers/cpp.py: (check_style.definitions): (check_style): Add rule. * Scripts/webkitpy/style/checkers/cpp_unittest.py: (WebKitStyleTest.test_line_breaking): Add unittest. 2012-09-14 Christophe Dumez WebKitTestRunner needs layoutTestController.dumpDatabaseCallbacks https://bugs.webkit.org/show_bug.cgi?id=57570 Reviewed by Kenneth Rohde Christiansen. Implement testRunner.dumpDatabaseCallbacks in WebKitTestRunner and dump the information expected by the test cases. * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/InjectedBundle.cpp: (WTR::InjectedBundle::beginTesting): Explicitly set the default database quota. * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp: (WTR::securityOriginToStr): New convenience function to convert a security origin to the string expected in test results. (WTR): (WTR::InjectedBundlePage::InjectedBundlePage): (WTR::InjectedBundlePage::didExceedDatabaseQuota): (WTR::InjectedBundlePage::didReachApplicationCacheOriginQuota): Minor refactoring to share code with didExceedDatabaseQuota. * WebKitTestRunner/InjectedBundle/InjectedBundlePage.h: (InjectedBundlePage): * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR::TestRunner::TestRunner): * WebKitTestRunner/InjectedBundle/TestRunner.h: (WTR::TestRunner::dumpDatabaseCallbacks): (WTR::TestRunner::shouldDumpDatabaseCallbacks): (TestRunner): 2012-09-14 Tor Arne Vestbø [Qt] Set force_static_libs_as_shared early enough to be caught by configure Rubber-stamped by Simon Hausmann. * qmake/mkspecs/features/default_post.prf: 2012-09-14 Tor Arne Vestbø [Qt] Remove old cruft from the qmake build system Rubber-stamped by Simon Hausmann. * DumpRenderTree/qt/DumpRenderTree.pro: * qmake/mkspecs/features/default_post.prf: 2012-09-14 Tor Arne Vestbø [Qt] Enable force_static_libs_as_shared for development Not enabled for buildbots or production-builds (part of Qt5) Reviewed by Simon Hausmann. * qmake/mkspecs/features/default_post.prf: 2012-09-14 Tor Arne Vestbø [Qt] Make force_static_libs_as_shared work on Mac OS We had to move a few LIBS += around that were in the wrong place, and not caught when everything was just linked into the final QtWebKit library. Reviewed by Simon Hausmann. * qmake/mkspecs/features/default_post.prf: * qmake/mkspecs/features/force_static_libs_as_shared.prf: 2012-09-14 Simon Hausmann [Qt] Fix build of WTR on Windows Reviewed by Tor Arne Vestbø. WTR/qt/main.cpp includes TestController.h, which lives in just WTR/ With qmake's unix makefile generator as well as with the MingW generator, the directory where the pro file lives (WTR/) is automatically added to the include search path, but not so with MSVC. Instead of relying on qmake's implicit feature here, add WTR/ explicitly to the include search path. * WebKitTestRunner/Target.pri: 2012-09-14 Christophe Dumez WebKitTestRunner needs layoutTestController.setStopProvisionalFrameLoads https://bugs.webkit.org/show_bug.cgi?id=42691 Reviewed by Kenneth Rohde Christiansen. Add implementation for testrunner.setStopProvisionalFrameLoads() in WebKitTestRunner. Original patch by Alexey Proskuryakov. * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp: (WTR::InjectedBundlePage::didStartProvisionalLoadForFrame): * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR::TestRunner::TestRunner): * WebKitTestRunner/InjectedBundle/TestRunner.h: (TestRunner): (WTR::TestRunner::setStopProvisionalFrameLoads): (WTR::TestRunner::shouldStopProvisionalFrameLoads): 2012-09-14 Tor Arne Vestbø [Qt] Fix handling of debug/release/debug_and_release/build_all We now pick up the defaults from Qt, in default_pre (CONFIG already contains the appropriate values for debug and release). We then let the command line arguments to qmake, or the project files themselves, override the configuration, and finally we sanitize the configuration in default_post. Reviewed by Simon Hausmann. * DumpRenderTree/qt/TestNetscapePlugin/TestNetscapePlugin.pro: * Scripts/webkitdirs.pm: (buildQMakeProjects): * qmake/mkspecs/features/default_post.prf: * qmake/mkspecs/features/default_pre.prf: * qmake/mkspecs/features/mac/default_pre.prf: 2012-09-14 Sudarsana Nagineni WebKitTestRunner needs layoutTestController.setTabKeyCyclesThroughElements https://bugs.webkit.org/show_bug.cgi?id=42687 Reviewed by Kenneth Rohde Christiansen. Implement testRunner.setTabKeyCyclesThroughElements for WebKitTestRunner. * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/InjectedBundle.cpp: (WTR::InjectedBundle::beginTesting): * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR::TestRunner::setTabKeyCyclesThroughElements): (WTR): * WebKitTestRunner/InjectedBundle/TestRunner.h: (TestRunner): 2012-09-14 Sudarsana Nagineni EventSendingController::keyDown does not support non-array modifier arguments https://bugs.webkit.org/show_bug.cgi?id=96727 Reviewed by Kenneth Rohde Christiansen. Add support for handling a string as modifier argument to EventSendingController::keyDown. * WebKitTestRunner/InjectedBundle/EventSendingController.cpp: (WTR::parseModifierArray): 2012-09-14 Mikhail Pozdnyakov WebKitTestRunner needs layoutTestController.setCustomPolicyDelegate https://bugs.webkit.org/show_bug.cgi?id=42546 Reviewed by Kenneth Rohde Christiansen. Exported TestRunner::setCustomPolicyDelegate() method. Allowed Policy Delegate dumping even if we're not in 'waitUntilDone' mode. * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp: (WTR::InjectedBundlePage::decidePolicyForNavigationAction): 2012-09-14 Zan Dobersek [GTK] Clear application cache between tests in DumpRenderTree https://bugs.webkit.org/show_bug.cgi?id=96543 Reviewed by Philippe Normand. Call the new DumpRenderTreeSupportGtk method to clear application cache after each test in DumpRenderTree. Set the XDG_CACHE_HOME environment variable when running DumpRenderTree as well. It's now being set in XvfbDriver._start so that each driver is assigned a cache directory that's constructed out of the driver's name and its worker number. These directories are located in the layout tests results directory. At least on the builders this ensures a driver-specific empty cache directory with each test run. * DumpRenderTree/gtk/DumpRenderTree.cpp: (runTest): * Scripts/webkitpy/layout_tests/port/gtk.py: (GtkPort.setup_environ_for_server): * Scripts/webkitpy/layout_tests/port/xvfbdriver.py: (XvfbDriver._start): 2012-09-14 Simon Hausmann [Qt] Fix linking on Windows Reviewed by Tor Arne Vestbø. When compiling code _after_ QtWebKit.dll, we need to make sure that the QWEBKIT_EXPORT macro turns into an import macro, which means QT_MAKEDLL should only be set _up until_ the QtWebKit.dll creation in the build process and be off afterwards. This is done in win32/default_post.prf using a !contains(QT, webkit): DEFINES += QT_MAKEDLL. r128450 sanitized the values of QT and PKGCONFIG to only contain the dependencies we want for QtWebKit.dll, but unfortunately that code was applied to any binary built also after the module. Therefore QT did never contain webkit anymore and we always passed QT_MAKEDLL to DEFINES, resulting in QWEBKIT_EXPORT to _not_ turn into an import macro. This patch introduces a creating_module configuration that is used to scope the qt/pkgconfig sanitization. * qmake/mkspecs/features/default_post.prf: 2012-09-14 Christophe Dumez [WK2][WKTR] TestRunner is not dumping some frame load callbacks information https://bugs.webkit.org/show_bug.cgi?id=96740 Reviewed by Kenneth Rohde Christiansen. Dump information expected by the test cases about the following frame load callbacks: "didDisplayInsecureContentForFrame" "didRunInsecureContentForFrame" "didDetectXSSForFrame" * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp: (WTR::InjectedBundlePage::willPerformClientRedirectForFrame): Print a URL as expected by test results by calling pathSuitableForTestResult() instead of printing raw URL. (WTR::InjectedBundlePage::didDisplayInsecureContentForFrame): (WTR::InjectedBundlePage::didRunInsecureContentForFrame): (WTR::InjectedBundlePage::didDetectXSSForFrame): 2012-09-14 Peter Beverloo The runtime/unsigned category should be valid in the cpp style-checker. https://bugs.webkit.org/show_bug.cgi?id=96748 Reviewed by Csaba Osztrogonác. This is causing the webkitpy tests to fail because it's not listed in the category array. Furthermore, change an "int" to "long" because it's throwing a double error in one of the earlier asserts. * Scripts/webkitpy/style/checkers/cpp.py: (CppChecker): * Scripts/webkitpy/style/checkers/cpp_unittest.py: (WebKitStyleTest.test_names): 2012-09-13 Kenneth Rohde Christiansen Evas_Object* is a ref'ed structure, so tread it as such https://bugs.webkit.org/show_bug.cgi?id=96659 Reviewed by Adam Barth. Replace OwnPtr with RefPtr. * DumpRenderTree/efl/ImageDiff.cpp: (calculateDifference): (printImageDifferences): (readImageFromStdin): (main): 2012-09-14 Gyuyoung Kim Unreviewed, rolling out r128507. http://trac.webkit.org/changeset/128507 https://bugs.webkit.org/show_bug.cgi?id=96659 Revert. r128507 makes too many crashes in EFL layout test bots. * DumpRenderTree/efl/ImageDiff.cpp: (calculateDifference): (printImageDifferences): (readImageFromStdin): (main): 2012-09-13 Kevin Funk Make compile with both OS(WINCE) and PLATFORM(QT) support https://bugs.webkit.org/show_bug.cgi?id=95536 Reviewed by Simon Hausmann. Fix wince support in qmake files * Tools.pro: * qmake/mkspecs/features/configure.prf: * qmake/mkspecs/features/default_post.prf: * qmake/mkspecs/features/features.prf: * qmake/mkspecs/features/functions.prf: 2012-09-13 KwangYong Choi [WK2] Add color picker API support for WebKit2 https://bugs.webkit.org/show_bug.cgi?id=95058 Reviewed by Gyuyoung Kim. Initialization code for color picker callbacks in UI client. * MiniBrowser/mac/BrowserWindowController.m: (-[BrowserWindowController awakeFromNib]): * MiniBrowser/win/BrowserView.cpp: (BrowserView::create): * WebKitTestRunner/TestController.cpp: (WTR::TestController::createOtherPage): (WTR::TestController::initialize): 2012-09-13 Yuta Kitamura Upgrade yutak to reviewer https://bugs.webkit.org/show_bug.cgi?id=96721 Reviewed by Kent Tamura. * Scripts/webkitpy/common/config/committers.py: 2012-09-13 Adenilson Cavalcanti Make the style checker report usage of "unsigned int" https://bugs.webkit.org/show_bug.cgi?id=96693 Reviewed by Dirk Pranke. Add a rule to check for uses of unsigned int, as the updated coding style recomends. * Scripts/webkitpy/style/checkers/cpp.py: (check_language): 2012-09-13 Philip Rogers Add gender-neutral form of webkit-patch land-cowboy https://bugs.webkit.org/show_bug.cgi?id=96709 Reviewed by Adam Barth. Add webkit-patch land-cowhand, the confident and courageous form of webkit-patch for everyone. * Scripts/webkitpy/tool/commands/download.py: (LandCowhand): 2012-09-13 Seokju Kwon [EFL] Remove some parameters in browserCreate() https://bugs.webkit.org/show_bug.cgi?id=96499 Reviewed by Kenneth Rohde Christiansen. There are too many parameters in browserCreate(). Lots of it seems like settings from command line argument. And I have moved some parameters into User_Arguments. * EWebLauncher/main.c: (_User_Arguments): (_ELauncher): (on_key_down): (browserCreate): (findThemePath): (parseUserArguments): (main): 2012-09-13 Mark Lam Unreviewed. Adding myself to the committers list. * Scripts/webkitpy/common/config/committers.py: 2012-09-13 Ryosuke Niwa run-perf-tests output cryptic error when the config file is missing. https://bugs.webkit.org/show_bug.cgi?id=96453 Reviewed by Tony Chang. Add a special error message when a configuration file is missing. Also update the help message of --source-json-path to signify the fact it specifies the configuration file on performance tests bots. * Scripts/webkitpy/performance_tests/perftestsrunner.py: (PerfTestsRunner._parse_args): (PerfTestsRunner._generate_and_show_results): (PerfTestsRunner._merge_slave_config_json): * Scripts/webkitpy/performance_tests/perftestsrunner_unittest.py: (_test_run_with_json_output): Make upload to fail by default and assert the result in the function so that we can return logs instead. (_test_run_with_json_output.mock_upload_json): (test_run_with_json_output): (test_run_with_description): (test_run_respects_no_results): (test_run_with_slave_config_json): (test_run_with_bad_slave_config_json): (test_run_with_multiple_repositories): 2012-09-13 Kenneth Rohde Christiansen Evas_Object* is a ref'ed structure, so tread it as such https://bugs.webkit.org/show_bug.cgi?id=96659 Reviewed by Adam Barth. Replace OwnPtr with RefPtr. * DumpRenderTree/efl/ImageDiff.cpp: (calculateDifference): (printImageDifferences): (readImageFromStdin): (main): 2012-09-13 Csaba Osztrogonác One more unreviewed trivial fix after r128399. * BuildSlaveSupport/build.webkit.org-config/config.json: Fix URLs of the SVN mirror server. 2012-09-13 Csaba Osztrogonác Unreviewed trivial fix after r128399. * BuildSlaveSupport/wait-for-SVN-server.py: Renamed from Tools/BuildSlaveSupport/build.webkit.org-config/wait-for-SVN-server.py. (getLatestSVNRevision): (waitForSVNRevision): 2012-09-13 Sudarsana Nagineni [WK2][WTR] WebKitTestRunner needs testRunner.callShouldCloseOnWebView https://bugs.webkit.org/show_bug.cgi?id=96366 Reviewed by Anders Carlsson. Implement testRunner.callShouldCloseOnWebView for WebKitTestRunner by calling shouldClose() on the FrameLoader. * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR::TestRunner::callShouldCloseOnWebView): (WTR): * WebKitTestRunner/InjectedBundle/TestRunner.h: (TestRunner): 2012-09-13 Sheriff Bot Unreviewed, rolling out r128453. http://trac.webkit.org/changeset/128453 https://bugs.webkit.org/show_bug.cgi?id=96681 Having tests use the same appcache directory leads to timeouts (Requested by zdobersek on #webkit). * DumpRenderTree/gtk/DumpRenderTree.cpp: (runTest): * Scripts/webkitpy/layout_tests/port/gtk.py: (GtkPort.setup_environ_for_server): 2012-09-13 James Robinson [chromium] Temporarily remove Android Builder (dbg) from garden-o-matic view https://bugs.webkit.org/show_bug.cgi?id=96678 Reviewed by Dirk Pranke. * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/config.js: (.): 2012-09-13 Rob Buis Unreviewed changes to watchlist. - Add myself to BlackBerry bugs * Scripts/webkitpy/common/config/watchlist: 2012-09-13 Stephen Chenney Add an SVG entry to the watchlist Unreviewed config change. * Scripts/webkitpy/common/config/watchlist: - Created a new SVG watchlist label covering WebCore/svg and WebCore/rendering/svg - Added myself and pdr as watchers 2012-09-13 Simon Hausmann [Qt] Make WebKit2 work on Windows with Qt 5 https://bugs.webkit.org/show_bug.cgi?id=76776 Reviewed by Tor Arne Vestbø. WebKit2 should compile with Qt on Windows now, so enable it. This patch in fact corrects the check to disable WebKit when we don't have USE_3D_GRAPHICS available, because we need that for the GL TextureMapper (it's only built when that feature is set). * qmake/mkspecs/features/configure.prf: 2012-09-13 Tor Arne Vestbø [Qt] Ensure that QT is finalized before loading qt_module.prf As qt_module.prf does dependency tracking based on the content of the QT variable. The intermediate WebKit modules modify the variable in their .pri files, so we have to ensure we've loaded all the modules before loading qt_module. Reviewed by Tor Arne Vestbø. Patch by Oswald Buddenhagen on 2012-09-13 * qmake/mkspecs/features/default_post.prf: * qmake/mkspecs/features/webkit_modules.prf: Added. 2012-09-13 Simon Hausmann [Qt] Fix build of Qt WTR/MiniBrowser with MSVC and nmake https://bugs.webkit.org/show_bug.cgi?id=96633 Reviewed by Csaba Osztrogonác. WTR and MiniBrowser both have DRT in their include path (for the shared font stuff). Batch builds with MSVC and nmake result in nmake picking up main.cpp from the wrong directory. It's the usual problem and the usual solution is to rename one of the conflicting files. This patch chooses to rename DRT instead of WTR and MBR. Fixed include order at the same time to match style guide. * DumpRenderTree/qt/DumpRenderTree.pro: * DumpRenderTree/qt/DumpRenderTreeMain.cpp: Renamed from Tools/DumpRenderTree/qt/main.cpp. (messageHandler): (isOption): (takeOptionValue): (printUsage): (main): 2012-09-13 Kenneth Rohde Christiansen The Script run-launcher should support -2 in combination with --efl https://bugs.webkit.org/show_bug.cgi?id=96639 Reviewed by Simon Hausmann. Support running the MiniBrowser then supplied -2 as argument. * Scripts/run-launcher: 2012-09-13 Raphael Kubo da Costa Unreviewed, add my Intel email address to the list. * Scripts/webkitpy/common/config/committers.py: 2012-09-13 Tor Arne Vestbø [Qt] Don't overwrite LIBS_PRIVATE when sanitizing LIBS Reviewed by Ossy. * qmake/mkspecs/features/default_post.prf: 2012-09-13 Kenneth Rohde Christiansen [EFL][DRT] Implement LayoutTestController::layerTreeAsText https://bugs.webkit.org/show_bug.cgi?id=82294 Reviewed by Gyuyoung Kim. Implement layerTreeAsText using the DumpRenderTreeSupportEfl method. * DumpRenderTree/efl/TestRunnerEfl.cpp: (TestRunner::layerTreeAsText): 2012-09-13 Zan Dobersek [GTK] Clear application cache between tests in DumpRenderTree https://bugs.webkit.org/show_bug.cgi?id=96543 Reviewed by Philippe Normand. Call the new DumpRenderTreeSupportGtk method to clear application cache after each test. Set the XDG_CACHE_HOME environment variable when running DumpRenderTree as well. Currently it points to a subdirectory in the layout tests results directory. At least on the builders this ensures an empty cache directory with each test run. * DumpRenderTree/gtk/DumpRenderTree.cpp: (runTest): * Scripts/webkitpy/layout_tests/port/gtk.py: (GtkPort.setup_environ_for_server): 2012-09-13 Tor Arne Vestbø [Qt] Make sure LIBS, PKGCONFIG, and QT, do not affect QtWebKit's prl file Unless it's actually a dependency of the public QtWebKit API. We could replace every occurance of LIBS, PKGCONFIG, and QT, in the pri files with their _PRIVATE equivivalent, but that's likely to break when someone adds a new QT/LIBS/PKGCONFIG += foo line somewhere. Instead we clean up the variables in default_post. This means that the CONFIG -= explicitlib in linkAgainstLibrary() is no longer needed, as it was a workaround for the intermediate libraries ending up in the prl file. And, since CONFIG -= staticlib was there to support the explicitlib option, we can remove that too (as well as the exception for gprof, since it would be empty). If gprof needs tweaks to always link statically they should go in gprof.prf. Reviewed by Simon Hausmann. * qmake/mkspecs/features/default_post.prf: * qmake/mkspecs/features/functions.prf: 2012-09-13 Mario Sanchez Prada [GTK] "Infinite" loop in AccessibilityUIElementGtk.cpp https://bugs.webkit.org/show_bug.cgi?id=96632 Reviewed by Carlos Garcia Campos. Fix this by using atk_object_get_n_accessible_children instead of calling getChildren() from childCount. * WebKitTestRunner/InjectedBundle/gtk/AccessibilityUIElementGtk.cpp: (WTR::AccessibilityUIElement::childrenCount): Avoid the infinite loop by using atk_object_get_n_accessible_children(). 2012-09-13 Simon Hausmann [Qt][Win] Enable USE(3D_GRAPHICS) https://bugs.webkit.org/show_bug.cgi?id=90879 Reviewed by Kenneth Rohde Christiansen. Enable it on Windows as well as on QNX (bug fixed). * qmake/mkspecs/features/features.prf: 2012-09-13 Simon Hausmann [Qt] The ANGLE build on Windows breaks with GnuWin32's flex because it is too old https://bugs.webkit.org/show_bug.cgi?id=96359 Reviewed by Tor Arne Vestbø. GnuWin32's flex is 2.5.4 and that is too old for ANGLE's preprocessor lexer. The only viable alternative I could find on Windows that provides binaries is the winflexbison distribution (http://sourceforge.net/projects/winflexbison/), so let's use that one instead. * Scripts/webkitdirs.pm: (checkRequiredSystemConfig): * qmake/mkspecs/features/default_pre.prf: 2012-09-13 Simon Hausmann [Qt] Add configuration for production builds https://bugs.webkit.org/show_bug.cgi?id=96607 Reviewed by Kenneth Rohde Christiansen. Add configuration tweaks for production builds, to ease memory pressure with the GNU toolchain and enable all-in-one-files optimizations. * qmake/mkspecs/features/production_build.prf: Added. 2012-09-13 Simon Hausmann [Qt] GStreamer detection does not work when cross-compiling https://bugs.webkit.org/show_bug.cgi?id=96608 Reviewed by Kenneth Rohde Christiansen. When cross-compiling it is not safe to just use system(pkg-config), because there is a high probability that it will find packages in the host system instead of in the target one. Instead use qmake's new packagesExist() function that takes care of this delicate aspect of cross compilation by setting PKG_CONFIG_SYSROOT_DIR and PKG_CONFIG_LIBDIR accordingly. * qmake/mkspecs/features/features.prf: 2012-09-13 Christophe Dumez [WK2][WTR] More efficient string handling in InjectedBundlePage https://bugs.webkit.org/show_bug.cgi?id=96609 Reviewed by Kenneth Rohde Christiansen. Update StringBuilder usage in InjectedBundlePage to follow the latest recommandations for efficient string handling: http://trac.webkit.org/wiki/EfficientStrings This involves using appendLiteral() or appendNumber() instead of append() whenever possible. * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp: (WTR::dumpPath): (WTR::rangeToStr): (WTR::styleDecToStr): (WTR::frameToStr): (WTR::dumpResourceURL): (WTR): (WTR::dumpFrameDescriptionSuitableForTestResult): (WTR::dumpRequestDescriptionSuitableForTestResult): (WTR::dumpResponseDescriptionSuitableForTestResult): (WTR::dumpErrorDescriptionSuitableForTestResult): (WTR::InjectedBundlePage::didReceiveIntentForFrame): (WTR::InjectedBundlePage::registerIntentServiceForFrame): (WTR::InjectedBundlePage::didStartProvisionalLoadForFrame): (WTR::InjectedBundlePage::didReceiveServerRedirectForProvisionalLoadForFrame): (WTR::InjectedBundlePage::didCommitLoadForFrame): (WTR::InjectedBundlePage::didFinishProgress): (WTR::dumpFrameScrollPosition): (WTR::dumpFrameText): (WTR::dumpDescendantFramesText): (WTR::InjectedBundlePage::didFinishLoadForFrame): (WTR::InjectedBundlePage::didReceiveTitleForFrame): (WTR::InjectedBundlePage::didCancelClientRedirectForFrame): (WTR::InjectedBundlePage::willPerformClientRedirectForFrame): (WTR::InjectedBundlePage::didFinishDocumentLoadForFrame): (WTR::InjectedBundlePage::didHandleOnloadEventsForFrame): (WTR::InjectedBundlePage::willSendRequestForFrame): (WTR::InjectedBundlePage::didReceiveResponseForResource): (WTR::InjectedBundlePage::didFinishLoadForResource): (WTR::InjectedBundlePage::didFailLoadForResource): (WTR::InjectedBundlePage::shouldCacheResponse): (WTR::InjectedBundlePage::decidePolicyForNavigationAction): (WTR::InjectedBundlePage::willAddMessageToConsole): (WTR::InjectedBundlePage::willSetStatusbarText): (WTR::InjectedBundlePage::willRunJavaScriptAlert): (WTR::InjectedBundlePage::willRunJavaScriptConfirm): (WTR::InjectedBundlePage::willRunJavaScriptPrompt): (WTR::InjectedBundlePage::didReachApplicationCacheOriginQuota): (WTR::InjectedBundlePage::shouldBeginEditing): (WTR::InjectedBundlePage::shouldEndEditing): (WTR::InjectedBundlePage::shouldInsertNode): (WTR::InjectedBundlePage::shouldInsertText): (WTR::InjectedBundlePage::shouldDeleteRange): (WTR::InjectedBundlePage::shouldChangeSelectedRange): (WTR::InjectedBundlePage::shouldApplyStyle): (WTR::InjectedBundlePage::didBeginEditing): (WTR::InjectedBundlePage::didEndEditing): (WTR::InjectedBundlePage::didChange): (WTR::InjectedBundlePage::didChangeSelection): (WTR::InjectedBundlePage::supportsFullScreen): (WTR::InjectedBundlePage::enterFullScreenForElement): (WTR::InjectedBundlePage::exitFullScreenForElement): (WTR::InjectedBundlePage::beganEnterFullScreen): (WTR::InjectedBundlePage::beganExitFullScreen): (WTR::InjectedBundlePage::closeFullScreen): (WTR::dumpBackForwardListItem): (WTR::InjectedBundlePage::dumpBackForwardList): 2012-09-13 Sudarsana Nagineni [Qt][WK2] Memory leaks in Qt's TestRunner https://bugs.webkit.org/show_bug.cgi?id=96603 Reviewed by Csaba Osztrogonác. Fix memory leaks in Qt's TestRunner code by adopting WKString created with WKStringCreateWithUTF8CString(). * WebKitTestRunner/qt/TestControllerQt.cpp: (WTR::TestController::initializeInjectedBundlePath): (WTR::TestController::initializeTestPluginDirectory): 2012-09-13 Zan Dobersek [GTK][NRWT] Lower timeout value to 6 seconds (12 for debug configuration) https://bugs.webkit.org/show_bug.cgi?id=96506 Reviewed by Dirk Pranke. Lower the timeout value to 6 seconds for release configuration and 12 seconds for debug configuration. These are the same values the Chromium port uses. Currently the value is lowered only for the DumpRenderTree, WebKitTestRunner (and the layout tests under WebKit2 generally) requires more attention and gardening than what the current effort produces. The unit test is updated as well to test these changes. * Scripts/webkitpy/layout_tests/port/gtk.py: (GtkPort.default_timeout_ms): * Scripts/webkitpy/layout_tests/port/gtk_unittest.py: (GtkPortTest.test_default_timeout_ms): (GtkPortTest.assertLinesEqual): 2012-09-12 Csaba Osztrogonác Add SVN mirror handling feature to build.webkit.org https://bugs.webkit.org/show_bug.cgi?id=85887 Reviewed by Dirk Pranke. * BuildSlaveSupport/build.webkit.org-config/config.json: Add SVN mirrors for Qt buildslaves hosted in Szeged. * BuildSlaveSupport/build.webkit.org-config/master.cfg: (ConfigureBuild.__init__): Add SVNMirror property to be able to watch which slaves use which mirror on build.webkit.org. (ConfigureBuild.start): (CheckOutSource.__init__): Set baseURL to the SVNMirror or the default http://svn.webkit.org/repository/webkit/ (WaitForSVNServer): Add new buildstep for waiting the SVNMirror to be in sync with http://svn.webkit.org/repository/webkit/ (Factory.__init__): (BuildFactory.__init__): (TestFactory.__init__): (BuildAndTestFactory.__init__): (BuildAndPerfTestFactory.__init__): (BuildAndPerfTestWebKit2Factory.__init__): (DownloadAndPerfTestFactory.__init__): (DownloadAndPerfTestWebKit2Factory.__init__): (loadBuilderConfig): Use kwargs instead of args to be able to add the optional SVNMirror factory argument. * BuildSlaveSupport/build.webkit.org-config/mastercfg_unittest.py: (SVNMirrorTest): (SVNMirrorTest.setUp): Load config.json once at startup. (SVNMirrorTest.get_SVNMirrorFromConfig): Get the SVNMirror from config.json for a given buildslave. (SVNMirrorTest.test_CheckOutSource): Compare CheckOutSource.baseURL with SVNMirror in config.json for all builders * BuildSlaveSupport/build.webkit.org-config/wait-for-SVN-server.py: Added. (getLatestSVNRevision): Get the latest SVN revison from the given server. (waitForSVNRevision): Wait until the given SVN revision is committed to the given server. It doesn't wait if the SVN revision is empty (force build triggered without revision) or the server is unavailable. 2012-09-12 Kangil Han [WK2][WTR] Set waitUntilDone watchdog timer value equal to WK1. https://bugs.webkit.org/show_bug.cgi?id=81606 Reviewed by Tim Horton. 6 seconds is harsh for some jquery test cases. Therefore, adjust the value to match DumpRenderTree. * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR): 2012-09-12 KyungTae Kim [EFL] Support download attribute feature https://bugs.webkit.org/show_bug.cgi?id=96462 Reviewed by Gyuyoung Kim. Enable DOWNLOAD_ATTRIBUTE feature for EFL port, and add "download,request" callback to the DumpRenderTree for EFL. * DumpRenderTree/efl/DumpRenderTreeChrome.cpp: (DumpRenderTreeChrome::createView): (DumpRenderTreeChrome::onDownloadRequest): * DumpRenderTree/efl/DumpRenderTreeChrome.h: (DumpRenderTreeChrome): * Scripts/webkitperl/FeatureList.pm: 2012-09-12 Stephanie Lewis Reduce parallism on the wk2 testers. https://bugs.webkit.org/show_bug.cgi?id=95906 Reviewed by Dirk Pranke. The wk2 testers on Mountain Lion are getting stuck and timing out. The problem appears to be due to resource contention. Reducing the number of processes alleviates the issue. Starting by reducing 25%. * Scripts/webkitpy/layout_tests/port/mac.py: (MacPort.default_child_processes): 2012-09-12 Brady Eidson Assert in NetscapePlugin::destroy() with async plugin init and https://bugs.webkit.org/show_bug.cgi?id=96576 Reviewed by Anders Carlsson. Expose NPN_Invoke to plug-in tests: * DumpRenderTree/TestNetscapePlugIn/PluginTest.cpp: (PluginTest::NPN_Invoke): * DumpRenderTree/TestNetscapePlugIn/PluginTest.h: (PluginTest): Add a test that uses NPN_Invoke on the window object from inside NPP_New to remove the plug-in element: * DumpRenderTree/TestNetscapePlugIn/Tests/InvokeDestroysPluginWithinNPP_New.cpp: Added. (InvokeDestroysPluginWithinNPP_New): (InvokeDestroysPluginWithinNPP_New::InvokeDestroysPluginWithinNPP_New): (InvokeDestroysPluginWithinNPP_New::NPP_New): * DumpRenderTree/DumpRenderTree.xcodeproj/project.pbxproj: 2012-09-12 Dirk Pranke refactor TestExpectations tokenization slightly in preparation for the new syntax https://bugs.webkit.org/show_bug.cgi?id=96564 Reviewed by Ojan Vafai. This change changes how we will tokenize/lex the TestExpectations files, in preparation for supporting both the old syntax and the new one. This change by itself should be a no-op. * Scripts/webkitpy/layout_tests/models/test_expectations.py: (TestExpectationParser._tokenize_line): (TestExpectationParser): (TestExpectationParser._tokenize_line_using_old_format): (TestExpectationParser._tokenize_line_using_new_format): 2012-09-12 Dirk Pranke remove unused testexpectations editing code from webkitpy, garden-o-matic https://bugs.webkit.org/show_bug.cgi?id=96562 Reviewed by Adam Barth. This code was prototyped but never fully realized. Removing until it reanimates :). * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/checkout.js: * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/checkout_unittests.js: * Scripts/webkitpy/layout_tests/controllers/test_expectations_editor.py: Removed. * Scripts/webkitpy/layout_tests/controllers/test_expectations_editor_unittest.py: Removed. * Scripts/webkitpy/tool/servers/gardeningserver.py: (BuildCoverageExtrapolator.extrapolate_test_configurations): (GardeningHTTPRequestHandler._run_webkit_patch): (GardeningHTTPRequestHandler.ping): * Scripts/webkitpy/tool/servers/gardeningserver_unittest.py: (BuildCoverageExtrapolatorTest.test_extrapolate): (GardeningServerTest.test_rebaseline_new_port): 2012-09-12 Lucas Forschler Remove unused bot, apple-xserve-1. https://bugs.webkit.org/show_bug.cgi?id=96570 Reviewed by Stephanie Lewis. * BuildSlaveSupport/build.webkit.org-config/config.json: 2012-09-12 Xan Lopez [GTK] We attempt to rebase documentation even if it's not present https://bugs.webkit.org/show_bug.cgi?id=96553 Reviewed by Martin Robinson. Do not make the documentation rebase step fatal. This allows make install to succeed when there's no documentation generated. * gtk/generate-gtkdoc: 2012-09-12 Kenneth Rohde Christiansen Respect WEBKITOUTPUTDIR when running EFL tests https://bugs.webkit.org/show_bug.cgi?id=96528 Reviewed by Dirk Pranke. Expose user set WEBKITOUTPUTDIR to the web process. * Scripts/webkitpy/layout_tests/port/driver.py: (Driver._start): Add WEBKITOUTPUTDIR to the environment of the web process and its potential jhbuild wrapper. 2012-09-12 Dirk Pranke webkitdirs: fix uname version handling for cygwin https://bugs.webkit.org/show_bug.cgi?id=96436 Reviewed by Jon Honeycutt. Newer versions of cygwin embed an additional version string inside parentheses, so you get "1.7.16(0.249/5/3)" instead of "1.7.16". Update the code to handle that. * Scripts/webkitdirs.pm: (setupAppleWinEnv): 2012-09-12 Christophe Dumez Regression(r128338): Broke Windows build https://bugs.webkit.org/show_bug.cgi?id=96537 Unreviewed build fix. Fix Apple-Win build by allocating array dynamically since its size is not constant. * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR::stringArrayToJS): 2012-09-12 Alex Sakhartchouk [chromium] Add a virtual test suite for fast/hidpi https://bugs.webkit.org/show_bug.cgi?id=90192 Reviewed by Dirk Pranke. Add a virtual test suite to make sure the pixel tests in fast/hidpi give the same result on the hardware accelerated path as the software path. * DumpRenderTree/chromium/WebPreferences.cpp: (WebPreferences::applyTo): * Scripts/webkitpy/layout_tests/port/chromium.py: (ChromiumPort.virtual_test_suites): 2012-09-12 Adam Klein [chromium] Add content_browsertests to the flakiness dashboard https://bugs.webkit.org/show_bug.cgi?id=96535 Reviewed by Ojan Vafai. * TestResultServer/static-dashboards/dashboard_base.js: 2012-09-12 Christophe Dumez [WK2][WKTR] TestRunner needs to implement originsWithApplicationCache https://bugs.webkit.org/show_bug.cgi?id=96496 Reviewed by Kenneth Rohde Christiansen. Add implementation for originsWithApplicationCache to WebKitTestRunner. * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR::stringArrayToJS): (WTR): (WTR::TestRunner::originsWithApplicationCache): * WebKitTestRunner/InjectedBundle/TestRunner.h: (TestRunner): 2012-09-12 Simon Hausmann Fix Qt/Windows build with Python3 https://bugs.webkit.org/show_bug.cgi?id=96473 Reviewed by Csaba Osztrogonác. In Python 3 print is a real function, so we must use parentheses around the function parameters. This is backwards compatible with Python 2. * Scripts/generate-win32-export-forwards: 2012-09-12 Simon Hausmann Unreviewed trivial build fix for Qt/Windows after bug #96358. As pointed out by Kevin, we should inject the "set PATH=..." statement to add the GnuWin32 directory only if that directory actually exists. * qmake/mkspecs/features/default_post.prf: 2012-09-12 Sudarsana Nagineni [EFL] [WK2] Memory leaks in TestControllerEfl https://bugs.webkit.org/show_bug.cgi?id=96525 Reviewed by Kenneth Rohde Christiansen. Fix memory leaks in EFL's TestRunner code by adopting an allocation of WKString created with WKStringCreateWithUTF8CString(). * WebKitTestRunner/efl/TestControllerEfl.cpp: (WTR::TestController::initializeInjectedBundlePath): (WTR::TestController::initializeTestPluginDirectory): 2012-09-12 Christophe Dumez [WK2][WKTR] TestRunner needs to implement dumpApplicationCacheDelegateCallbacks https://bugs.webkit.org/show_bug.cgi?id=96374 Reviewed by Kenneth Rohde Christiansen. Implement support for dumpApplicationCacheDelegateCallbacks and disallowIncreaseForApplicationCacheQuota in WebKitTestRunner and properly dump the information expected by the tests. If the application cache quota is reached for a given security origin, WebKitTestRunner will reset the quota to its default value, unless intructed not to via disallowIncreaseForApplicationCacheQuota(). * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp: (WTR::InjectedBundlePage::InjectedBundlePage): (WTR::InjectedBundlePage::didReachApplicationCacheOriginQuota): (WTR): * WebKitTestRunner/InjectedBundle/InjectedBundlePage.h: (InjectedBundlePage): * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR::TestRunner::TestRunner): (WTR::TestRunner::disallowIncreaseForApplicationCacheQuota): (WTR): * WebKitTestRunner/InjectedBundle/TestRunner.h: (WTR::TestRunner::dumpApplicationCacheDelegateCallbacks): (TestRunner): (WTR::TestRunner::shouldDisallowIncreaseForApplicationCacheQuota): (WTR::TestRunner::shouldDumpApplicationCacheDelegateCallbacks): 2012-09-12 Jochen Eisinger [chromium] remove deprecated and unused sets import from chromium_android driver https://bugs.webkit.org/show_bug.cgi?id=96485 Reviewed by Dirk Pranke. * Scripts/webkitpy/layout_tests/port/chromium_android.py: 2012-09-12 Zan Dobersek Flakiness dashboard doesn't recognize new Chromium Android test builder https://bugs.webkit.org/show_bug.cgi?id=96523 Reviewed by Ojan Vafai. Properly return 'ANDROID' as the Chromium platform for Android builders. * TestResultServer/static-dashboards/flakiness_dashboard.js: (chromiumPlatform): * TestResultServer/static-dashboards/flakiness_dashboard_unittests.js: (test): 2012-09-12 Simon Hausmann [Qt] Build on Windows requires bison/flex in PATH https://bugs.webkit.org/show_bug.cgi?id=96358 Reviewed by Tor Arne Vestbø. The build requires flex, bison, etc. and they need to be in the PATH when building. On Mac OS X and Linux that is rarely a problem given how easily available the tools are. On Windows however a separate installation of various GNU tools is required as the operating system doesn't come with them. To make the development more convenient, Qt 5 provides a copy of the most essential tools in the gnuwin32 directory of the qt5.git top-level repository. This patch tries to detect the presence of those tools and prepends them to the PATH if found. This is required in preparation for the elimination of qt5/qtwebkit.pri, which currently expands PATH before calling build-webkit. It it also required for the upcoming introduction of win_flex as dependency over flex, which can be done with less hassle when qt5's gnuwin32 directory has been updated with the new tool. * Scripts/webkitdirs.pm: (checkRequiredSystemConfig): * qmake/mkspecs/features/default_post.prf: 2012-09-12 Tor Arne Vestbø [Qt] Update build-jsc after r128174 Reviewed by Ossy. * Scripts/build-jsc: 2012-09-12 Tor Arne Vestbø [Qt] Teach addStrictSubdirOrderBetween to handle more than two targets By hard-coding the names of the targets we defined we ended up just redefining the previous target when using addStrictSubdirOrderBetween more than once in a single project file. We now embed the two base targets into the target names. Reviewed by Simon Hausmann. * qmake/mkspecs/features/functions.prf: 2012-09-12 Philippe Liard Depend on {base,net} GYP targets rather than {base,net}_java. https://bugs.webkit.org/show_bug.cgi?id=95690 Reviewed by Adam Barth. {base,net}_java became 'private' targets on the Chromium side that clients should not depend on (see https://chromiumcodereview.appspot.com/10913083/). This removes a bunch of trailing white spaces additionally in TestWebkitAPI.gyp (which includes non-trailing white space changes). * DumpRenderTree/DumpRenderTree.gyp/DumpRenderTree.gyp: * TestWebKitAPI/TestWebKitAPI.gyp/TestWebKitAPI.gyp: 2012-09-12 Simon Hausmann [Qt] Fix the build with ENABLE_NETSCAPE_PLUGIN_API=0 https://bugs.webkit.org/show_bug.cgi?id=96494 Reviewed by Tor Arne Vestbø. WK2's ENABLE_PLUGIN_PROCESS uses the NPAPI functions unconditionally, so disable the plugin process feature if we don't have NPAPI. * qmake/mkspecs/features/features.prf: 2012-09-12 Sheriff Bot Unreviewed, rolling out r128280. http://trac.webkit.org/changeset/128280 https://bugs.webkit.org/show_bug.cgi?id=96498 it broke compilation on windows debug bot (Requested by loislo on #webkit). * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp: (WTR::InjectedBundlePage::InjectedBundlePage): * WebKitTestRunner/InjectedBundle/InjectedBundlePage.h: (InjectedBundlePage): * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR::TestRunner::TestRunner): * WebKitTestRunner/InjectedBundle/TestRunner.h: (TestRunner): 2012-09-12 Simon Hausmann [Qt] Make it possible to build with "make release" and "make debug" on Windows https://bugs.webkit.org/show_bug.cgi?id=96488 Reviewed by Tor Arne Vestbø. A "make debug" is passed through recursively and currently it aborts at Makefile.DerivedSources because there are no such targets. We want the generated sources to be independent from release or debug build configurations, so it is sufficient to provide fake debug and release targets that redirect to the same general-purpose target (first) of creating the derived sources. * qmake/mkspecs/features/default_post.prf: 2012-09-12 Kenneth Rohde Christiansen [EFL] Make DumpRenderTree smarter at finding the fonts http://webkit.org/b/96281 Reviewed by Gyuyoung Kim. Respect WEBKITOUTPUTDIR and expand the font dir from it. Use CString consistently. * DumpRenderTree/efl/FontManagement.cpp: (buildPath): (getCoreFontFiles): (addFontDirectory): (addFontFiles): (getCustomBuildDir): (getPlatformFontsPath): (addFontsToEnvironment): 2012-09-12 Christophe Dumez [WK2][WKTR] TestRunner needs to implement dumpApplicationCacheDelegateCallbacks https://bugs.webkit.org/show_bug.cgi?id=96374 Reviewed by Kenneth Rohde Christiansen. Implement support for dumpApplicationCacheDelegateCallbacks and disallowIncreaseForApplicationCacheQuota in WebKitTestRunner and properly dump the information expected by the tests. If the application cache quota is reached for a given security origin, WebKitTestRunner will reset the quota to its default value, unless intructed not to via disallowIncreaseForApplicationCacheQuota(). * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp: (WTR::InjectedBundlePage::InjectedBundlePage): (WTR::InjectedBundlePage::didReachApplicationCacheOriginQuota): (WTR): * WebKitTestRunner/InjectedBundle/InjectedBundlePage.h: (InjectedBundlePage): * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR::TestRunner::TestRunner): (WTR::TestRunner::disallowIncreaseForApplicationCacheQuota): (WTR): * WebKitTestRunner/InjectedBundle/TestRunner.h: (WTR::TestRunner::dumpApplicationCacheDelegateCallbacks): (TestRunner): (WTR::TestRunner::shouldDisallowIncreaseForApplicationCacheQuota): (WTR::TestRunner::shouldDumpApplicationCacheDelegateCallbacks): 2012-09-12 Mikhail Pozdnyakov [WK2] [WTR] WebKitTestRunner needs TestRunner.workerThreadCount https://bugs.webkit.org/show_bug.cgi?id=96388 Reviewed by Kenneth Rohde Christiansen. Exported TestRunner.workerThreadCount as readonly attribute. * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/TestRunner.cpp: Added workerThreadCount() method implementation. (WTR::TestRunner::workerThreadCount): Returns count of worker threads. (WTR): * WebKitTestRunner/InjectedBundle/TestRunner.h: Added workerThreadCount() method. (TestRunner): 2012-09-12 Jochen Eisinger [chromium] Consume a user gesture when creating a new view. https://bugs.webkit.org/show_bug.cgi?id=96373 Reviewed by Adam Barth. * DumpRenderTree/chromium/WebViewHost.cpp: (WebViewHost::createView): 2012-09-12 Mikhail Pozdnyakov [WK2][WTR] Some of TestRunner special options are not reset before testing https://bugs.webkit.org/show_bug.cgi?id=96384 Reviewed by Kenneth Rohde Christiansen. Now values of the following special options are reset: void setAcceptsEditing(in boolean value); void setCloseRemainingWindowsWhenComplete(in boolean value); void setXSSAuditorEnabled(in boolean value); void setAllowFileAccessFromFileURLs(in boolean value); void setPluginsEnabled(in boolean value); void setPopupBlockingEnabled(in boolean value); * WebKitTestRunner/InjectedBundle/InjectedBundle.cpp: (WTR::InjectedBundle::beginTesting): 2012-09-11 Ryuan Choi [CMAKE] Supply feature defines to CodeGeneratorTestRunner. https://bugs.webkit.org/show_bug.cgi?id=96273 Reviewed by Gyuyoung Kim. Passed FEATURE_DEFINES_WITH_SPACE_SEPARATOR to CodeGeneratorTestRunner in order to distinguish compile time options such as ENABLE(TOUCH_EVENTS). * WebKitTestRunner/CMakeLists.txt: 2012-09-11 Dirk Pranke [chromium] rename webkit_resources https://bugs.webkit.org/show_bug.cgi?id=95874 Reviewed by James Robinson. * DumpRenderTree/DumpRenderTree.gyp/DumpRenderTree.gyp: 2012-09-11 Jochen Eisinger [chromium] don't include content shell builders in the garden-o-matic view https://bugs.webkit.org/show_bug.cgi?id=96418 Reviewed by James Robinson. * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/config.js: 2012-09-11 Adam Klein Garden-o-matic should ignore a wider variety of warnings in buildbot json https://bugs.webkit.org/show_bug.cgi?id=96411 Reviewed by Adam Barth. Previously only the exact string "warning" was treated as a warning result. This patch treats any string with "warning" as a substring as a warning (e.g., "warnings", as seen on the cr-win buildbots). * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/builders.js: (.): * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/builders_unittests.js: 2012-09-11 Marcelo Lira Restore original value of mock scrollbars enabled in InternalSettings https://bugs.webkit.org/show_bug.cgi?id=87680 Reviewed by Adam Barth. Added backup support for mock scrollbars state in InternalSettings::Backup, and removed support for these from Qt's DumpRenderTree. * DumpRenderTree/qt/DumpRenderTreeQt.cpp: (WebCore::WebPage::resetSettings): * DumpRenderTree/qt/TestRunnerQt.cpp: (TestRunner::setFrameFlatteningEnabled): * DumpRenderTree/qt/TestRunnerQt.h: (TestRunner): 2012-09-11 Dan Bernstein Fixed a bug where kill-old-processes failed to kill clang++. Rubber-stamped by Alexey Proskuryakov. * BuildSlaveSupport/kill-old-processes: Escape the + signs in "clang++" where it is passed as a regular expression to killall. 2012-09-11 Philippe Liard Make sure that md5sum is not setup on non-Chromium Android platforms. https://bugs.webkit.org/show_bug.cgi?id=96393 Reviewed by Adam Barth. Previously md5sum was setup in ChromiumAndroidDriver's constructor which is also invoked on non-Chromium Android platforms. The following patch moves this logic to _startup_test() which is guaranteed to be invoked only on Android. This also updates the prompt unit test in chromium_android_unittest.py. * Scripts/webkitpy/layout_tests/port/chromium_android.py: (ChromiumAndroidDriver.__init__): (ChromiumAndroidDriver._setup_md5sum_and_push_data_if_needed): (ChromiumAndroidDriver._setup_test): 2012-09-11 Raphael Kubo da Costa [EFL] Rewrite the EFL-related Find modules https://bugs.webkit.org/show_bug.cgi?id=95237 Reviewed by Kenneth Rohde Christiansen. * DumpRenderTree/efl/CMakeLists.txt: Stop setting the LINK_FLAGS property and include all the Enlightenment Foundation Libraries required by the target. * EWebLauncher/CMakeLists.txt: Ditto. * MiniBrowser/efl/CMakeLists.txt: Ditto. * TestWebKitAPI/PlatformEfl.cmake: Add missing include directories now that they are not added implicitly. * WebKitTestRunner/CMakeLists.txt: Stop setting the LINK_FLAGS property. * WebKitTestRunner/PlatformEfl.cmake: Stop setting the LINK_FLAGS property and include all the Enlightenment Foundation Libraries required by the target. 2012-09-11 Christophe Dumez [WK2][WKTR] TestRunner needs to implement setApplicationCacheOriginQuota https://bugs.webkit.org/show_bug.cgi?id=96379 Reviewed by Kenneth Rohde Christiansen. Add implementation for setApplicationCacheOriginQuota to WebKitTestRunner. * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR::TestRunner::setApplicationCacheOriginQuota): (WTR): * WebKitTestRunner/InjectedBundle/TestRunner.h: (TestRunner): 2012-09-11 Luciano Wolf [Qt] [WK2] editing/inserting/typing-tab-designmode tests are failing https://bugs.webkit.org/show_bug.cgi?id=91988 Reviewed by Kenneth Rohde Christiansen. Regular key events generated by Qt for the tab key fill both keyCode and keyText, so we should do the same for our fake events. * WebKitTestRunner/qt/EventSenderProxyQt.cpp: (WTR::EventSenderProxy::keyDown): 2012-09-11 Mikhail Pozdnyakov [WK2][WTR] WebKitTestRunner needs testRunner.setSpatialNavigationEnabled https://bugs.webkit.org/show_bug.cgi?id=96269 Reviewed by Kenneth Rohde Christiansen. Exported missing testRunner.setSpatialNavigationEnabled() method. * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/InjectedBundle.cpp: (WTR::InjectedBundle::beginTesting): * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR::TestRunner::setSpatialNavigationEnabled): (WTR): * WebKitTestRunner/InjectedBundle/TestRunner.h: (TestRunner): 2012-09-11 Philippe Liard Support LayoutTests on non-rooted devices for Chromium Android. https://bugs.webkit.org/show_bug.cgi?id=95346 Reviewed by Dirk Pranke. We were requiring root permissions to sync resource files with the device by comparing host and device files' timestamp after we synchronized the device and host clocks. We now rely on md5sum provided by Chromium to handle data files synchronization. * Scripts/webkitpy/layout_tests/port/chromium.py: (ChromiumPort._check_driver_build_up_to_date): * Scripts/webkitpy/layout_tests/port/chromium_android.py: (ChromiumAndroidDriver.__init__): (ChromiumAndroidDriver._setup_test): (ChromiumAndroidDriver._push_executable): (ChromiumAndroidDriver._push_test_resources): (ChromiumAndroidDriver._read_prompt): 2012-09-11 Simon Hausmann [Qt] Fix output paths on Windows Reviewed by Tor Arne Vestbø. BUILD_ROOT_RELATIVE_TO_OUT_PWD was calculated incorrectly Windows, resulting in an incorrect DESTDIR transformation at the end of default_post.prf. The calculation should be baased no qmake paths, which can use drive letters but always use forward slashes. So the use of QMAKE_DIR_SEP is incorrect here and instead for the calculation we should use forward slashes. * qmake/mkspecs/features/default_pre.prf: 2012-09-11 Tor Arne Vestbø [Qt] Fix passing of defines from the build system We explicitly have to pass -DENABLE_FOO=0 for every single feature that's not explicitly enabled in WEBKIT_CONFIG, since Platform.h will add it's own defaults if the features are not defined. At some point we might want to run a configure-test to inspect what Platform.h will do, so we can sync up the build system's view of the feature-defines with the compile-time situation. Reviewed by Simon Hausmann. * qmake/mkspecs/features/functions.prf: 2012-09-11 Tor Arne Vestbø [Qt] Fix --qt option to build-webkit Reviewed by Ossy. * Scripts/webkitdirs.pm: (buildQMakeProjects): 2012-09-11 Tor Arne Vestbø [Qt] Add a configure step to the Qt build system This allows building the Qt port using just 'qmake WebKit.pro'. Using the build-webkit script is still supported, and will add slightly more logic to the build, such as detecting the need for clean or incremental builds. Internally, the build system now uses a new variable, WEBKIT_CONFIG, for all things related to configuring the build, such as use/have/enable flags, and these are translated to defines in default_post. Project files should no longer check contains(DEFINES,...) to detect features, but use the new enable?(), use?(), and have?() functions. The no_webkit1 and no_webkit2 options have been translated into WEBKIT_CONFIG options as well, and can be checked using build?(). Reviewed by Simon Hausmann. * DumpRenderTree/qt/DumpRenderTree.pro: * MiniBrowser/qt/MiniBrowser.pro: * QtTestBrowser/QtTestBrowser.pro: * Scripts/build-webkit: * Scripts/webkitdirs.pm: (qtFeatureDefaults): (buildQMakeProjects): * Tools.pro: * WebKitTestRunner/InjectedBundle/DerivedSources.pri: * WebKitTestRunner/InjectedBundle/Target.pri: * WebKitTestRunner/Target.pri: * qmake/.qmake.cache: Added. * qmake/configure.pri: Removed. * qmake/configure.pro: Removed. * qmake/dump-features: Added. * qmake/mkspecs/features/configure.prf: Added. * qmake/mkspecs/features/default_post.prf: * qmake/mkspecs/features/default_pre.prf: * qmake/mkspecs/features/features.prf: * qmake/mkspecs/features/features.pri: * qmake/mkspecs/features/functions.prf: * qmake/mkspecs/features/gprof.prf: * qmake/mkspecs/features/mac/default_pre.prf: * qmake/mkspecs/features/win32/default_pre.prf: 2012-09-11 Christophe Dumez [WK2][WKTR] TestRunner needs to implement clearApplicationCacheForOrigin https://bugs.webkit.org/show_bug.cgi?id=96372 Reviewed by Kenneth Rohde Christiansen. Implement clearApplicationCacheForOrigin() in WebKitTestRunner. * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR::TestRunner::clearApplicationCacheForOrigin): (WTR): * WebKitTestRunner/InjectedBundle/TestRunner.h: (TestRunner): 2012-09-11 Peter Beverloo Don't ignore .pak files when making an archive for Android https://bugs.webkit.org/show_bug.cgi?id=96375 Reviewed by Jochen Eisinger. The .pak files do need to be pushed to the device, which is currently preventing the layout test runner from correctly pushing all files. * BuildSlaveSupport/built-product-archive: (archiveBuiltProduct): 2012-09-11 Mikhail Pozdnyakov [WK2][WTR] WebKitTestRunner needs layoutTestController.setMinimumTimerInterval https://bugs.webkit.org/show_bug.cgi?id=96256 Reviewed by Kenneth Rohde Christiansen. Exported TestRunner::setMinimumTimerInterval() method. * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/InjectedBundle.cpp: (WTR::InjectedBundle::beginTesting): * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR::TestRunner::setMinimumTimerInterval): (WTR): * WebKitTestRunner/InjectedBundle/TestRunner.h: (TestRunner): 2012-09-11 Tommy Widenflycht MediaStream API: add RTCPeerConnection::onnegotiationneeded https://bugs.webkit.org/show_bug.cgi?id=96097 Reviewed by Adam Barth. * DumpRenderTree/chromium/MockWebRTCPeerConnectionHandler.cpp: (MockWebRTCPeerConnectionHandler::addStream): (MockWebRTCPeerConnectionHandler::removeStream): 2012-09-11 Peter Beverloo Recognize the "--chromium-android" argument in run-webkit-tests https://bugs.webkit.org/show_bug.cgi?id=96369 Reviewed by Jochen Eisinger. Chromium for the Android platform cannot determine the platform based on the host itself, so we'll use the --chromium-android argument as the build master passes to run-webkit-tests. Make sure the layout test runner can pick the right port when doing so. * Scripts/run-webkit-tests: * Scripts/webkitpy/layout_tests/port/factory.py: (port_options): 2012-09-11 Simon Hausmann [QT] Fix build with newer qmake from Qt 5 Rubber-stamped by Tor Arne Vestbø. Support for "qmake -query QMAKE_MKSPECS" was removed from qmake. Replace it with the use of QT_HOST_DATA/mkspecs. Based on patch by Oswald Buddenhagen . * Scripts/webkitdirs.pm: (builtDylibPathForName): (qtMakeCommand): 2012-09-11 Tor Arne Vestbø [Qt] Remove Qt 4 syncqt, no longer needed Reviewed by Simon Hausmann. * qmake/syncqt-4.8: Removed. 2012-09-11 Andras Becsi [Qt][WK2] Fix the build with recent Qt5 https://bugs.webkit.org/show_bug.cgi?id=96283 Reviewed by Simon Hausmann. We should not use deprecated Qt API because doing that results in build failure since all the deprecated API was disabled in qtbase (https://codereview.qt-project.org/#change,24890). * DumpRenderTree/qt/main.cpp: (main): * MiniBrowser/qt/MiniBrowserApplication.cpp: (MiniBrowserApplication::notify): * MiniBrowser/qt/raw/View.cpp: (View::exposeEvent): (main): 2012-09-11 Christophe Dumez WTR needs an implementation of applicationCacheDiskUsageForOrigin https://bugs.webkit.org/show_bug.cgi?id=87188 Reviewed by Kenneth Rohde Christiansen. Add implementation for applicationCacheDiskUsageForOrigin() in WebKitTestRunner. * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR::TestRunner::applicationCacheDiskUsageForOrigin): (WTR): * WebKitTestRunner/InjectedBundle/TestRunner.h: (TestRunner): 2012-09-10 Joanmarie Diggs Properly expose elements to ATs https://bugs.webkit.org/show_bug.cgi?id=84137 Reviewed by Chris Fleizach. Created a new WebCore Accessibility Role, LegendRole. Used it to map to the expected platform role, ATK_ROLE_LABEL. Also established the needed AtkRelation pair, label-for/labelled-by between the legend and fieldset. * DumpRenderTree/chromium/TestRunner/AccessibilityUIElementChromium.cpp: (roleToString): added "Legend" string for WebAccessibilityRoleLegend 2012-09-10 Jer Noble Unreviewed; rolling out r128081. http://trac.webkit.org/changeset/128081 https://bugs.webkit.org/show_bug.cgi?id=96134 New test http/tests/media/video-donottrack.html fails to run on Apple test bots. * DumpRenderTree/TestRunner.cpp: (TestRunner::TestRunner): (TestRunner::staticFunctions): * DumpRenderTree/TestRunner.h: (TestRunner): * DumpRenderTree/mac/FrameLoadDelegate.mm: * WebKitTestRunner/InjectedBundle/Bindings/TestRunner.idl: * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp: (WTR::InjectedBundlePage::InjectedBundlePage): * WebKitTestRunner/InjectedBundle/InjectedBundlePage.h: (InjectedBundlePage): * WebKitTestRunner/InjectedBundle/TestRunner.cpp: (WTR::TestRunner::TestRunner): * WebKitTestRunner/InjectedBundle/TestRunner.h: (TestRunner): 2012-09-10 Christophe Dumez TestExpectationsChecker._determine_port_from_expectations_path() does not support cascaded TestExpectations https://bugs.webkit.org/show_bug.cgi?id=96205 Reviewed by Tony Chang. _determine_port_from_expectations_path() was calling port.path_to_test_expectations_file() internally, which means that it would support only 1 TestExpectations file per port. This is an issue for ports such as EFL that support cascased TestExpectations (efl-wk2 -> efl and efl-wk1 -> efl). This patch makes _determine_port_from_expectations_path() call port.expectations_files() instead so that all the ports TestExpectations are recognized. The ports are also constructed twice, with "webkit_test_runner" option set to True and False so that we retrieve the TestExpectations paths for both WebKit1 and WebKit2. * Scripts/webkitpy/style/checkers/test_expectations.py: (TestExpectationsChecker._determine_port_from_expectations_path): * Scripts/webkitpy/style/checkers/test_expectations_unittest.py: Add corresponding unit test. (TestExpectationsTestCase.test_determine_port_from_expectations_path): 2012-09-10 Ryuan Choi [WTR] Generated source files should include config.h https://bugs.webkit.org/show_bug.cgi?id=96254 Reviewed by Benjamin Poulain. Current coding style has encouraged to include config.h at the first position, But derived sources, generated by CodeGeneratorTestRunner.pm, don't follow this rule. Therefore, this patch added it. * WebKitTestRunner/InjectedBundle/Bindings/CodeGeneratorTestRunner.pm: (_generateImplementationFile): 2012-09-10 Rick Byers Add handling of new GestureTapCancel in DRT https://bugs.webkit.org/show_bug.cgi?id=96183 Reviewed by Adam Barth. * DumpRenderTree/chromium/TestWebPlugin.cpp: (TestWebPlugin::handleInputEvent): 2012-09-10 Brady Eidson Javascript in foreground tabs should not wait synchronously for plug-ins to load and https://bugs.webkit.org/show_bug.cgi?id=96167 Reviewed by Geoff Garen. Enhance the "Slow NPP_New" plug-in to also be able to return properties to javascript. * DumpRenderTree/TestNetscapePlugIn/Tests/SlowNPPNew.cpp: (PluginObject): (SlowNPPNew::PluginObject::PluginObject): (SlowNPPNew::PluginObject::~PluginObject): (SlowNPPNew::PluginObject::hasProperty): (SlowNPPNew::PluginObject::getProperty): (SlowNPPNew::NPP_GetValue): (SlowNPPNew): 2012-09-10 Tim Horton WKTR often reports an unresponsive WebProcess on Mac bots https://bugs.webkit.org/show_bug.cgi?id=95906 Reviewed by Dan Bernstein. Attempt to work around resource contention issues by increasing the NRWT timeout when running against WKTR, as well as WKTR's own WebProcess timeout. * Scripts/webkitpy/layout_tests/port/base.py: (Port.default_timeout_ms): Default timeout when using WKTR -> 80 seconds. * WebKitTestRunner/TestController.cpp: (WTR): LongTimeout (used when waiting for WebProcess to finish) -> 60 seconds. 2012-09-07 Jer Noble