2012-01-30 Ojan Vafai run-webkit-tests calls out to webkit-build-directory twice https://bugs.webkit.org/show_bug.cgi?id=77248 Reviewed by Dirk Pranke. This reduces to calls to webkit-build-directory to one call and saves 500ms for run-webkit-tests over a single file. * Scripts/webkit-build-directory: If called with no arguments, print out both the top-level directory and the configuration directory. * Scripts/webkit-build-directory: * Scripts/webkitpy/layout_tests/port/config.py: (Config.build_directory): When called with no arguments, get and cache both the top-level and configuration directories. * Scripts/webkitpy/layout_tests/port/config_standalone.py: (main): * Scripts/webkitpy/layout_tests/port/config_unittest.py: (ConfigTest.assert_configuration): (ConfigTest.test_build_directory.mock_webkit_build_directory): (ConfigTest.test_build_directory): (ConfigTest.test_default_configuration__notfound): * Scripts/webkitpy/layout_tests/port/webkit.py: (WebKitPort._build_path): Cache the root directory in the options object so that other users of the options object don't need to call out to webkit-build-directory. 2012-01-30 Ojan Vafai Parsing test_expecations.txt + Skipped lists takes too long https://bugs.webkit.org/show_bug.cgi?id=77059 Reviewed by Dirk Pranke. This saves ~100ms on the Apple Mac port. -memoize a bunch of path methods. -Avoid doing multiple disk accesses per line. -Parse the skipped list directly instead of turning it into a test_expecations.txt formatting string and parsing that. * Scripts/webkitpy/layout_tests/models/test_expectations.py: (TestExpectationParser): (TestExpectationParser.expectation_for_skipped_test): (TestExpectationParser._parse_line): (TestExpectationParser._collect_matching_tests): (TestExpectations.__init__): (TestExpectations._add_skipped_tests): * Scripts/webkitpy/layout_tests/models/test_expectations_unittest.py: (test_add_skipped_tests): (test_add_skipped_tests_duplicate): * Scripts/webkitpy/layout_tests/port/base.py: (Port): (Port.test_isfile): (Port.normalize_test_name): (Port.layout_tests_dir): (Port.abspath_for_test): * Scripts/webkitpy/layout_tests/run_webkit_tests.py: 2012-01-30 Alexis Menard Unreviewed. Add myself to CSS, GStreamer, Qt related watchlists. * Scripts/webkitpy/common/config/watchlist: 2012-01-30 Gavin Barraclough Unreviewed build fix following bug#76855 * TestWebKitAPI/Tests/WTF/MetaAllocator.cpp: (): (TestWebKitAPI::TEST_F): 2012-01-26 No'am Rosenthal and Jocelyn Turcotte [Qt] WKTR: Use a software rendering pipiline when running tests. https://bugs.webkit.org/show_bug.cgi?id=76708 Reviewed by Kenneth Rohde Christiansen. * WebKitTestRunner/qt/PlatformWebViewQt.cpp: (WTR::WrapperWindow::WrapperWindow): (WTR::PlatformWebView::PlatformWebView): Use software rendering of layers since the wrapping QQuickView isn't shown. 2012-01-30 Tor Arne Vestbø [Qt] Store build config immediately instead of waiting for a successfull build Otherwise we will not pick up changes to the config if the build is aborted, for example due to build errors or if the user pressed Ctrl+Z. We now write the new config before starting a build, but still defer to write the SVN revision once the build completes successfully. Reviewed by Ossy. * Scripts/webkitdirs.pm: (buildQMakeProjects): 2012-01-30 Alexander Færøy [Qt] Add "New Window" button to the Qt MiniBrowser https://bugs.webkit.org/show_bug.cgi?id=77281 Reviewed by Tor Arne Vestbø. * MiniBrowser/qt/BrowserWindow.cpp: (BrowserWindow::BrowserWindow): * MiniBrowser/qt/MiniBrowser.qrc: * MiniBrowser/qt/icons/plus.png: Added. * MiniBrowser/qt/qml/BrowserWindow.qml: 2012-01-30 Csaba Osztrogonác Unreviewed rolling out r106222. * Scripts/webkitpy/layout_tests/controllers/single_test_runner.py: (SingleTestRunner.__init__): * Scripts/webkitpy/layout_tests/controllers/worker.py: (Worker.handle_test_list): * Scripts/webkitpy/layout_tests/models/test_input.py: (TestInput.__init__): * Scripts/webkitpy/layout_tests/run_webkit_tests.py: (_set_up_derived_options): (parse_args): * Scripts/webkitpy/layout_tests/run_webkit_tests_integrationtest.py: (MainTest.test_repeat_each): * WebKitTestRunner/TestInvocation.cpp: (WTR::TestInvocation::setIsPixelTest): 2012-01-30 Hans Wennborg Unreviewed, rolling out r106219. http://trac.webkit.org/changeset/106219 https://bugs.webkit.org/show_bug.cgi?id=77083 This broke Chromium's test_shell. * DumpRenderTree/DumpRenderTree.gypi: * DumpRenderTree/chromium/LayoutTestController.cpp: (LayoutTestController::addMockSpeechInputResult): * DumpRenderTree/chromium/LayoutTestController.h: (WebKit): (LayoutTestController): (): * DumpRenderTree/chromium/MockWebSpeechInputController.cpp: Removed. * DumpRenderTree/chromium/MockWebSpeechInputController.h: Removed. * DumpRenderTree/chromium/TestShell.cpp: * DumpRenderTree/chromium/WebViewHost.cpp: (WebViewHost::speechInputController): * DumpRenderTree/chromium/WebViewHost.h: (WebKit): (WebViewHost::speechInputControllerMock): (WebViewHost): (): 2012-01-26 Hans Wennborg Speech Input: move MockSpeechInputClient into Chromium DumpRenderTree implementation https://bugs.webkit.org/show_bug.cgi?id=77083 Reviewed by Darin Fisher. Add MockWebSpeechInputController which provides a mock implementation of the WebSpeechInputController interface, and use that in DumpRenderTree. * DumpRenderTree/DumpRenderTree.gypi: * DumpRenderTree/chromium/LayoutTestController.cpp: (LayoutTestController::addMockSpeechInputResult): * DumpRenderTree/chromium/LayoutTestController.h: (): * DumpRenderTree/chromium/MockWebSpeechInputController.cpp: Added. (MockWebSpeechInputController::create): (MockWebSpeechInputController::addMockRecognitionResult): (MockWebSpeechInputController::clearResults): (MockWebSpeechInputController::startRecognition): (MockWebSpeechInputController::cancelRecognition): (MockWebSpeechInputController::stopRecording): (MockWebSpeechInputController::MockWebSpeechInputController): (MockWebSpeechInputController::speechTaskFired): (MockWebSpeechInputController::SpeechTask::SpeechTask): (MockWebSpeechInputController::SpeechTask::stop): (MockWebSpeechInputController::SpeechTask::runIfValid): * DumpRenderTree/chromium/MockWebSpeechInputController.h: Added. (WebKit): (MockWebSpeechInputController): (MockWebSpeechInputController::taskList): (SpeechTask): * DumpRenderTree/chromium/TestShell.cpp: * DumpRenderTree/chromium/WebViewHost.cpp: (WebViewHost::speechInputController): * DumpRenderTree/chromium/WebViewHost.h: (WebKit): (WebViewHost::speechInputControllerMock): (WebViewHost): (): 2012-01-29 Sheriff Bot Unreviewed, rolling out r106109. http://trac.webkit.org/changeset/106109 https://bugs.webkit.org/show_bug.cgi?id=77302 It made tests crash (Requested by Ossy_weekend on #webkit). * WebKitTestRunner/qt/PlatformWebViewQt.cpp: (WTR::WrapperWindow::WrapperWindow): (WTR::PlatformWebView::PlatformWebView): 2012-01-28 Nico Weber [chromium] Add support for building standalone webkit/chromium checkouts with ninja. https://bugs.webkit.org/show_bug.cgi?id=77243 Reviewed by Adam Barth. * Scripts/update-webkit: * Scripts/webkitdirs.pm: (isChromiumNinja): (determineIsChromiumNinja): (buildChromiumNinja): (buildChromium): 2012-01-27 Kentaro Hara The cpp parser of prepare-ChangeLog treats if(...) {} as a method https://bugs.webkit.org/show_bug.cgi?id=77241 Reviewed by Ryosuke Niwa. The cpp parser of prepare-ChangeLog is wrong, and an inner {} block in a method in some namespace or class is treated as a method. This patch fixes the bug. class C { void func() { if (1) { // This should not be treated as a method. } } }; * Scripts/prepare-ChangeLog: (get_function_line_ranges_for_cpp): * Scripts/webkitperl/prepare-ChangeLog_unittest/resources/cpp_unittests-expected.txt: * Scripts/webkitperl/prepare-ChangeLog_unittest/resources/cpp_unittests.cpp: (Class108): (Class108::func35): 2012-01-27 Enrica Casucci Followup to http://trac.webkit.org/changeset/106171. Dan Bernstein pointed out correctly that I should have used copy instead of initWithString. Reviewed by Dan Bernstein. * DumpRenderTree/mac/DumpRenderTreePasteboard.m: (-[LocalPasteboard initWithName:]): 2012-01-27 Enrica Casucci DRT pasteboard object should return the correct pasteboard name. https://bugs.webkit.org/show_bug.cgi?id=77260 Currently the name property always returns nil. Adding a memeber variable to store the pasteboard name and return its value from the name method instead of returning nil always. Reviewed by Dan Bernstein. * DumpRenderTree/mac/DumpRenderTreePasteboard.m: (+[DumpRenderTreePasteboard _pasteboardWithName:]): (-[LocalPasteboard initWithName:]): (-[LocalPasteboard name]): 2012-01-27 Martin Robinson [GTK] [WK2] Implement the policy client https://bugs.webkit.org/show_bug.cgi?id=76343 Reviewed by Gustavo Noronha Silva. * gtk/generate-gtkdoc: (get_webkit2_options): Update the list of skipped source files to include all clients. 2012-01-27 Ada Chan Add test for WKBundleFrameGetParentFrame(). https://bugs.webkit.org/show_bug.cgi?id=77161 Reviewed by Anders Carlsson. * TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj: Add ParentFrame.cpp and ParentFrame_Bundle.cpp. * TestWebKitAPI/Tests/WebKit2/ParentFrame.cpp: Added. (TestWebKitAPI): (TestWebKitAPI::didReceiveMessageFromInjectedBundle): Get the check result from the injected bundle. (TestWebKitAPI::setInjectedBundleClient): (TestWebKitAPI::TEST): Load simple-iframe.html and then wait for the injected bundle to post result. Make sure the check is successful. * TestWebKitAPI/Tests/WebKit2/ParentFrame_Bundle.cpp: Added. (TestWebKitAPI): (ParentFrameTest): (TestWebKitAPI::ParentFrameTest::ParentFrameTest): (TestWebKitAPI::didFinishLoadForFrame): If the frame is a subframe, store it off for checking later. If the frame is the main frame, check whether it's indeed the subframe's parent frame and post the result to the test controller. (TestWebKitAPI::ParentFrameTest::didCreatePage): Set the page loader client on this page. 2012-01-27 Zan Dobersek [Gtk] DumpRenderTree lacks --no-timeout command line option https://bugs.webkit.org/show_bug.cgi?id=77196 Reviewed by Martin Robinson. Add the --no-timeout command line option and reorganize watchdog code to make running DumpRenderTree without timeout possible. * DumpRenderTree/gtk/DumpRenderTree.cpp: (setWaitToDumpWatchdog): (shouldSetWaitToDumpWatchdog): (initializeGlobalsFromCommandLineOptions): * DumpRenderTree/gtk/DumpRenderTreeGtk.h: * DumpRenderTree/gtk/LayoutTestControllerGtk.cpp: (waitToDumpWatchdogFired): (LayoutTestController::setWaitToDump): 2012-01-27 Andras Becsi [Qt][WK2] Enable the automatic touch->mouse mocking on the new Qt5 hashes https://bugs.webkit.org/show_bug.cgi?id=77122 Reviewed by Kenneth Rohde Christiansen. * MiniBrowser/qt/MiniBrowserApplication.cpp: (MiniBrowserApplication::MiniBrowserApplication): 2012-01-27 Zeno Albisser [Qt][Mac] Build fails after adding ICU support (r105997). https://bugs.webkit.org/show_bug.cgi?id=77118 Define WTF_USE_ICU_UNICODE if we are on mac. Reviewed by Tor Arne Vestbø. * qmake/mkspecs/features/features.prf: 2012-01-27 Roland Takacs Update committers.py with new contributors. https://bugs.webkit.org/show_bug.cgi?id=77207 I added six new contributors into committers.py: - Adam Kallai - Balazs Ankes - Janos Badics - Nandor Huszka - Roland Takacs - Szilard Ledan-Muntean * Scripts/webkitpy/common/config/committers.py: 2012-01-27 Vsevolod Vlasov Add Vsevolod Vlasov to reviewers list. Reviewed by Pavel Feldman. * Scripts/webkitpy/common/config/committers.py: 2012-01-26 No'am Rosenthal and Jocelyn Turcotte [Qt] WKTR: Use a software rendering pipiline when running tests. https://bugs.webkit.org/show_bug.cgi?id=76708 Reviewed by Kenneth Rohde Christiansen. * WebKitTestRunner/qt/PlatformWebViewQt.cpp: (WTR::WrapperWindow::WrapperWindow): (WTR::PlatformWebView::PlatformWebView): Use software rendering of layers since the wrapping QQuickView isn't shown. 2012-01-26 Kevin Ollivier [wx] Unreviewed. Build fix, add WebCore/testing/js to the list of source directories we build. * waf/build/settings.py: 2012-01-26 Filip Pizlo Tools/Scripts/commit-log-editor is broken due to $_ getting clobbered https://bugs.webkit.org/show_bug.cgi?id=77177 Reviewed by Jon Honeycutt. * Scripts/commit-log-editor: 2012-01-26 Ojan Vafai Decrease sleep time when killing server_process on Mac. https://bugs.webkit.org/show_bug.cgi?id=77169 Reviewed by Eric Seidel. This shaves another ~70ms off single test runtime. * Scripts/webkitpy/layout_tests/port/server_process.py: (ServerProcess.stop): 2012-01-26 Ojan Vafai Remove builder name so http://trac.webkit.org/changeset/106035 actually works. https://bugs.webkit.org/show_bug.cgi?id=77164 Reviewed by Dirk Pranke. http://trac.webkit.org/changeset/106035 depends on builder_name being falsey by default. There's other code that requires it to be iterable by default. So, make it be the empty string. * Scripts/webkitpy/layout_tests/run_webkit_tests.py: (parse_args): 2012-01-26 Ryosuke Niwa run-perf-test should include relative file paths in test names https://bugs.webkit.org/show_bug.cgi?id=77144 Reviewed by Adam Birth. Include all directory names in the test name separated by "/". e.g. Bindings/event-target-wrapper.html translates to Bindings/event-target-wrapper * Scripts/webkitpy/performance_tests/perftestsrunner.py: (PerfTestsRunner): (PerfTestsRunner._process_parser_test_result): * Scripts/webkitpy/performance_tests/perftestsrunner_unittest.py: (test_run_test_set_with_json_output): (test_run_test_set_with_json_source): 2012-01-26 Dirk Pranke webkitpy: re-land cleanup of test scaffolding https://bugs.webkit.org/show_bug.cgi?id=77153 Reviewed by Eric Seidel. This is an attempt to reland the change in bug 76973 / r105935. My earlier attempt to manipulate PYTHONPATH seems to only work some of the time, for reasons that elude me. Directly adding Tools/Scripts to sys.path from inside the test script appears to work (more?) reliably. Also, it seemed like the files didn't actually get deleted the last time; I don't know why. * Scripts/test-webkitpy: * Scripts/webkitpy/common/system/executive_unittest.py: (command_line): (ExecutiveTest): (ExecutiveTest.test_run_command_args_type): (ExecutiveTest.test_run_command_with_unicode): (ExecutiveTest.test_running_pids): (main): * Scripts/webkitpy/common/system/fileutils.py: Removed. * Scripts/webkitpy/test/cat.py: Removed. * Scripts/webkitpy/test/cat_unittest.py: Removed. * Scripts/webkitpy/test/echo.py: Removed. * Scripts/webkitpy/test/echo_unittest.py: Removed. 2012-01-26 Ojan Vafai run-webkit-tests calls "nm" when it doesn't need to https://bugs.webkit.org/show_bug.cgi?id=77143 Reviewed by Dirk Pranke. If we are not running any test that can be skipped by checking the DRT feature list or looking at the list of symbols, avoid calling out to DRT or calling "nm". This saves ~700ms on my Mac Pro. In cases where you still have to call out to "nm" it's only a couple ms slower. * Scripts/webkitpy/layout_tests/models/test_expectations.py: (TestExpectations.__init__): * Scripts/webkitpy/layout_tests/port/base.py: (Port.skipped_tests): * Scripts/webkitpy/layout_tests/port/webkit.py: (WebKitPort._should_check_for_missing_features): (WebKitPort): (WebKitPort._should_check_for_missing_symbols): (WebKitPort._skipped_tests_for_unsupported_features): (WebKitPort.skipped_layout_tests): (WebKitPort.skipped_tests): * Scripts/webkitpy/layout_tests/port/webkit_unittest.py: (WebKitPortTest.test_skipped_directories_for_symbols): (test_skipped_directories_for_features): (test_skipped_directories_for_features_no_matching_tests_in_test_list): (test_skipped_layout_tests): 2012-01-26 Benjamin Poulain Support a suffix on ChangeLog filenames based on a configuration file https://bugs.webkit.org/show_bug.cgi?id=76956 Reviewed by Kentaro Hara. Modify the tools used to generate, edit and resolve ChangeLogs to support the addition of a suffix to the ChangeLog filename. This patch makes it easier to support alternate ChangeLog files in branches. * Scripts/VCSUtils.pm: (changeLogSuffix): (changeLogFileName): * Scripts/commit-log-editor: * Scripts/prepare-ChangeLog: (getLatestChangeLogs): (generateNewChangeLogs): (generateFileList): * Scripts/resolve-ChangeLogs: (findChangeLog): 2012-01-26 Sheriff Bot Unreviewed, rolling out r106038. http://trac.webkit.org/changeset/106038 https://bugs.webkit.org/show_bug.cgi?id=77142 Caused a bunch of skipped tests to not be skipped. (Requested by ojan on #webkit). * Scripts/webkitpy/layout_tests/models/test_expectations.py: (TestExpectationParser): (TestExpectationParser.parse): (TestExpectationParser._parse_line): (TestExpectationParser._collect_matching_tests): (TestExpectations.__init__): (TestExpectations._add_skipped_tests): * Scripts/webkitpy/layout_tests/port/base.py: (Port.test_dirs): (Port.normalize_test_name): (Port.update_baseline): (Port.layout_tests_dir): (Port.relative_perf_test_filename): (Port.abspath_for_test): 2012-01-25 Ojan Vafai Parsing test_expecations.txt + Skipped lists takes too long https://bugs.webkit.org/show_bug.cgi?id=77059 Reviewed by Dirk Pranke. This saves ~100ms on the Apple Mac port. -memoize a bunch of path methods. -Avoid doing multiple disk accesses per line. -Parse the skipped list directly instead of turning it into a test_expecations.txt formatting string and parsing that. * Scripts/webkitpy/layout_tests/models/test_expectations.py: (TestExpectationParser): (TestExpectationParser.parse_skipped_tests): (TestExpectationParser._parse_line): (TestExpectationParser._collect_matching_tests): (TestExpectations.__init__): (TestExpectations._add_skipped_tests): * Scripts/webkitpy/layout_tests/port/base.py: (Port): (Port.test_isfile): (Port.normalize_test_name): (Port.layout_tests_dir): (Port.abspath_for_test): 2012-01-25 Ojan Vafai Only store the SVN revision in the summarized results if we're on a builder https://bugs.webkit.org/show_bug.cgi?id=76976 Reviewed by Dirk Pranke. This shaves another ~130ms off the single test run-webkit-tests runtime. * Scripts/webkitpy/layout_tests/controllers/manager.py: (summarize_results): * Scripts/webkitpy/layout_tests/controllers/manager_unittest.py: (ResultSummaryTest.get_result): (ResultSummaryTest): (ResultSummaryTest.get_result_summary): (ResultSummaryTest.get_unexpected_results): (ResultSummaryTest.test_no_svn_revision): (ResultSummaryTest.test_svn_revision): * Scripts/webkitpy/layout_tests/port/base.py: (Port.use_trac_links_in_results_html): 2012-01-26 Luciano Wolf [Qt][WK2] fast/loader tests failed after r94178 https://bugs.webkit.org/show_bug.cgi?id=67388 Reviewed by Simon Hausmann. Applying the same fix used for Qt-WebKit1 from bug #67254 - [Qt][DRT] Normalize file:///tmp/LayoutTests in LayoutTestController::pathToLocalResource() https://bugs.webkit.org/show_bug.cgi?id=67254 Translate file:///tmp/LayoutTests/* urls into the repository LayoutTests directory, which is derived from the running location of DumpRenderTree binary. * WebKitTestRunner/InjectedBundle/qt/LayoutTestControllerQt.cpp: (WTR::LayoutTestController::pathToLocalResource): 2012-01-26 Caio Marcelo de Oliveira Filho Implement overridePreference for boolean preferences in WebKitTestRunner https://bugs.webkit.org/show_bug.cgi?id=77033 Reviewed by Alexey Proskuryakov. For now overridePreference supports boolean values. * WebKitTestRunner/InjectedBundle/Bindings/LayoutTestController.idl: * WebKitTestRunner/InjectedBundle/LayoutTestController.cpp: (WTR::LayoutTestController::setXSSAuditorEnabled): (WTR::LayoutTestController::overridePreference): * WebKitTestRunner/InjectedBundle/LayoutTestController.h: * WebKitTestRunner/TestController.cpp: (WTR::TestController::resetStateToConsistentValues): Reset the overrides explicitly and properly disable the use of PageCache when resetting the state. Tests that use PageCache can now enable it via overridePreference. 2012-01-26 Alexander Færøy [Qt] Support opening multiple URL's from the command line in the Qt MiniBrowser https://bugs.webkit.org/show_bug.cgi?id=77100 Reviewed by Simon Hausmann. This patch makes the MiniBrowser read multiple URL's from the command line such that there will be one browser window opened for each URL. * MiniBrowser/qt/MiniBrowserApplication.cpp: (MiniBrowserApplication::handleUserOptions): 2012-01-26 Alexander Færøy [Qt] The Qt MiniBrowser crashes when given multiple URL's https://bugs.webkit.org/show_bug.cgi?id=77009 Reviewed by Simon Hausmann. This patch fixes a crash in BrowserWindow::newWindow() where a BrowserWindow-object is constructed with a 0 pointer as the WindowOptions parameter which leads to a crash. * MiniBrowser/qt/BrowserWindow.cpp: (BrowserWindow::BrowserWindow): (BrowserWindow::newWindow): * MiniBrowser/qt/BrowserWindow.h: 2012-01-26 Jesus Sanchez-Palencia [Qt] Use ICU if available https://bugs.webkit.org/show_bug.cgi?id=76821 Reviewed by Simon Hausmann. Adding unicode defines for a Qt 5 based build. * qmake/mkspecs/features/features.prf: 2012-01-26 Nándor Huszka [WK2] WebKitTestRunner needs layoutTestController.setGeolocationPermission https://bugs.webkit.org/show_bug.cgi?id=42545 Reviewed by Zoltan Herczeg. Implement the setGeolocationPermission method. * WebKitTestRunner/InjectedBundle/Bindings/LayoutTestController.idl: * WebKitTestRunner/InjectedBundle/LayoutTestController.cpp: (WTR::LayoutTestController::setGeolocationPermission): (WTR): * WebKitTestRunner/InjectedBundle/LayoutTestController.h: (LayoutTestController): 2012-01-26 Balazs Kelemen [NRWT] Support --ignore-metrics https://bugs.webkit.org/show_bug.cgi?id=76278 Reviewed by Tony Chang. Port the feature from ORWT. * Scripts/webkitpy/layout_tests/controllers/single_test_runner.py: (SingleTestRunner._run_compare_test): * Scripts/webkitpy/layout_tests/port/driver.py: (DriverOutput): (DriverOutput.strip_metrics): * Scripts/webkitpy/layout_tests/port/driver_unittest.py: (DriverOutputTest): (DriverOutputTest.test_strip_metrics): * Scripts/webkitpy/layout_tests/run_webkit_tests.py: (_set_up_derived_options): (parse_args): 2012-01-26 Sheriff Bot Unreviewed, rolling out r105935 and r105954. http://trac.webkit.org/changeset/105935 http://trac.webkit.org/changeset/105954 https://bugs.webkit.org/show_bug.cgi?id=77080 test-webkitpy is still broken (Requested by abarth on #webkit). * Scripts/test-webkitpy: * Scripts/webkitpy/common/system/executive_unittest.py: (never_ending_command): (ExecutiveTest.test_run_command_args_type): (ExecutiveTest.test_run_command_with_unicode): (ExecutiveTest.test_running_pids): * Scripts/webkitpy/common/system/fileutils.py: (make_stdout_binary): * Scripts/webkitpy/test/cat.py: (command_arguments): (main): * Scripts/webkitpy/test/cat_unittest.py: (CatTest): (CatTest.assert_cat): (CatTest.test_basic): (CatTest.test_no_newline): (CatTest.test_unicode): (CatTest.test_as_command): * Scripts/webkitpy/test/echo.py: (command_arguments): (main): * Scripts/webkitpy/test/echo_unittest.py: (EchoTest): (EchoTest.test_basic): (EchoTest.test_no_newline): (EchoTest.test_unicode): (EchoTest.test_argument_order): (EchoTest.test_empty_arguments): (EchoTest.test_no_arguments): (EchoTest.test_as_command): 2012-01-26 Gabor Rapcsanyi [GTK] Detect cross-compilation in webkitdirs.pm to not generate gtkdoc https://bugs.webkit.org/show_bug.cgi?id=76138 Reviewed by Martin Robinson. * Scripts/webkitdirs.pm: (isCrossCompilation): (buildAutotoolsProject): 2012-01-25 Adam Barth Kentaro is a reviewer now. Yay! * Scripts/webkitpy/common/config/committers.py: 2012-01-25 Dirk Pranke nrwt: running w/ no arguments fails https://bugs.webkit.org/show_bug.cgi?id=77064 Unreviewed, build fix. This change fixes a regression introduced in r105944 where running without --platform or one of the other platform switches would cause a null deref :(. * Scripts/webkitpy/layout_tests/run_webkit_tests.py: (main): 2012-01-25 Eric Seidel Automatically CC someone from Apple on any changes to Obj-C public API https://bugs.webkit.org/show_bug.cgi?id=77053 Reviewed by Timothy Hatcher. This has long been de-facto practice, seems now that we have fancy watchlist support, we should just make the tools do this automatically. * Scripts/webkitpy/common/config/watchlist: 2012-01-25 Antoine Labour Updating status in committers.py. Unreviewed. * Scripts/webkitpy/common/config/committers.py: 2012-01-25 Dirk Pranke test-webkitpy: executive_unittest fails if PYTHONPATH isn't set https://bugs.webkit.org/show_bug.cgi?id=77058 Reviewed by Eric Seidel. r105935 broke executive_unittest by changing which script we were running and not ensuring that the PYTHONPATH was set up properly. This change also fixes a regression introduced recently where test-webkitpy wasn't figuring out the path to Tools/Scripts right, either (so running ./test-webkitpy would fail). * Scripts/test-webkitpy: * Scripts/webkitpy/common/system/executive_unittest.py: (script_dir): (ExecutiveTest.setUp): (ExecutiveTest): (ExecutiveTest.tearDown): 2012-01-25 Adam Barth Disable tests on the mac-ews while we sort out hte stability issues. * Scripts/webkitpy/tool/commands/earlywarningsystem.py: (MacEWS): 2012-01-23 Dirk Pranke investigate use of 'mac' and 'win' as fully-specified port names for the apple ports https://bugs.webkit.org/show_bug.cgi?id=76475 Reviewed by Eric Seidel. This patch cleans up the internal usage of port names for the Apple mac and Apple win ports so that 'mac' and 'win are never considered "fully specified" port names: they are still legal input to the PortFactory.get() routine, but only if run on the given platform (i.e. trying to PortFactory.get('mac') on windows or linux will return an error, since we don't know which version of the port is desired. This also cleans up handling of the webkit2 variants, so that the full port names consistently follow the -[-] convention. Lastly this patch adds some assertions and removes some try/catch logic to catch more programming and usage errors; previously mac-tiger might've been allowed -- and it would've translated into using just the baselines in platform/mac -- but now it should fail. * Scripts/webkitpy/layout_tests/port/apple.py: (ApplePort.determine_full_port_name): (ApplePort.__init__): (ApplePort): (ApplePort._port_name_with_version): * Scripts/webkitpy/layout_tests/port/builders.py: * Scripts/webkitpy/layout_tests/port/factory_unittest.py: (FactoryTest.test_mac): (FactoryTest.test_win): * Scripts/webkitpy/layout_tests/port/mac.py: (MacPort.baseline_search_path): * Scripts/webkitpy/layout_tests/port/mac_unittest.py: (MacTest.test_versions): * Scripts/webkitpy/layout_tests/port/win.py: (WinPort.baseline_search_path): * Scripts/webkitpy/layout_tests/port/win_unittest.py: (WinTest.test_versions): 2012-01-25 Dirk Pranke webkitpy: print nicer errors while linting expectations files, remove redundant tests https://bugs.webkit.org/show_bug.cgi?id=76955 Reviewed by Adam Barth. This change cleans up the errors that are printed so that the are more "quickfix" compatible (path:lineno). Also, there were a bunch of redundant tests between layout_tests.models.test_expectations_unittest and style.checkers.test_expectations_unittest. The latter should just expect the former to do most of the testing. * Scripts/webkitpy/layout_tests/models/test_expectations.py: (TestExpectationParser._parse_modifiers): (TestExpectations._report_errors): (TestExpectations._add_skipped_tests): * Scripts/webkitpy/layout_tests/models/test_expectations_unittest.py: (test_parse_error_nonfatal): (SemanticTests.test_bad_bugid): * Scripts/webkitpy/style/checkers/test_expectations.py: (TestExpectationsChecker.__init__): * Scripts/webkitpy/style/checkers/test_expectations_unittest.py: (TestExpectationsTestCase.assert_lines_lint): (TestExpectationsTestCase.test_valid_expectations): (TestExpectationsTestCase.test_invalid_expectations): (TestExpectationsTestCase.test_tab): 2012-01-25 Eric Seidel webkit-patch apply-* should always continue after failures https://bugs.webkit.org/show_bug.cgi?id=77057 Reviewed by Adam Barth. As far as I can tell there is only one potential drawback to always forcing: that if you're somehow in the wrong directory it will create new directories for new files. Since webkit-patch always cd's to the root it seems that's not a drawback. The drawback of not using --force for svn-apply, is that it will stop after the first failure, which is rarely (if ever) the desired behavior. This just removes the force option (which was strangely hidden behind --non-interactive). This should make for a better user experiance. * Scripts/webkitpy/common/checkout/checkout.py: (Checkout.apply_patch): * Scripts/webkitpy/common/checkout/checkout_mock.py: (MockCheckout.apply_patch): * Scripts/webkitpy/common/checkout/checkout_unittest.py: (CheckoutTest.test_chromium_deps): (CheckoutTest): (CheckoutTest.test_apply_patch): * Scripts/webkitpy/tool/commands/download_unittest.py: (DownloadCommandsTest._default_options): * Scripts/webkitpy/tool/steps/applypatch.py: (ApplyPatch.options): (ApplyPatch.run): * Scripts/webkitpy/tool/steps/options.py: (Options): 2012-01-25 Dirk Pranke nrwt: should be able to run --platform test interactively https://bugs.webkit.org/show_bug.cgi?id=76959 Reviewed by Adam Barth. As part of refactoring the Port interfaces, in r103254 I made passing a Host object to the Port mandatory; previously TestPort objects would create their own (mock) hosts. However, new-run-webkit-tests always passed a real Host, not a MockHost, and so you couldn't run new-run-webkit-tests --platform test interactively to debug test failures. This change fixes that by creating a MockHost instead of a real host if you say --platform test*. * Scripts/webkitpy/layout_tests/run_webkit_tests.py: (main): 2012-01-25 Dirk Pranke run-webkit-tests --lint-test-files should lint all the ports by default https://bugs.webkit.org/show_bug.cgi?id=76749 Reviewed by Ojan Vafai. Currently run-webkit-tests --lint-test-files will only lint the test_expectations for the single port determined by the command line options. This is not obvious and can produce unintended results if you want to lint the chromium port (probably the normal case for using --lint-test-files) but don't specify a port name or --chromium. I think we should probably lint *all* of the ports if no port name is supplied. This change implements that, and also moves the linting logic out of the Manager class, since this really has nothing to do with Managers. It has little to do with run-webkit-tests, but splitting it into a different script would introduce compatibility issues. It might make sense to do that anyway, in a separate patch and combining that with cleaning up the style checker to share more code. * Scripts/webkitpy/layout_tests/controllers/manager.py: (Manager._strip_test_dir_prefix): * Scripts/webkitpy/layout_tests/run_webkit_tests.py: (lint): (run): * Scripts/webkitpy/layout_tests/run_webkit_tests_integrationtest.py: (LintTest.test_all_configurations): added. (LintTest.test_lint_test_files): added. (LintTest.test_lint_test_files__errors): added. (MainTest.test_lint_test_files): removed. (MainTest.test_lint_test_files__errors): removed. 2012-01-25 Dirk Pranke webkitpy: clean up a bunch of test scaffolding https://bugs.webkit.org/show_bug.cgi?id=76973 Reviewed by Eric Seidel. There were several helper modules in the webkitpy.test package that appear to only be used by executive_unittest.py. I am rolling them all into that file to make that clearer. * Scripts/webkitpy/common/system/executive_unittest.py: (ExecutiveTest.test_run_command_args_type): (ExecutiveTest.test_run_command_with_unicode): (ExecutiveTest.test_running_pids): (command_line): (main): * Scripts/webkitpy/common/system/fileutils.py: Removed. * Scripts/webkitpy/test/cat.py: Removed. * Scripts/webkitpy/test/cat_unittest.py: Removed. * Scripts/webkitpy/test/echo.py: Removed. * Scripts/webkitpy/test/echo_unittest.py: Removed. 2012-01-25 Dirk Pranke webkitpy: clean up a bunch of test scaffolding https://bugs.webkit.org/show_bug.cgi?id=76973 Reviewed by Eric Seidel. There were several helper modules in the webkitpy.test package that appear to only be used by executive_unittest.py. I am rolling them all into that file to make that clearer. * Scripts/webkitpy/common/system/executive_unittest.py: (ExecutiveTest.test_run_command_args_type): (ExecutiveTest.test_run_command_with_unicode): (ExecutiveTest.test_running_pids): (command_line): (main): * Scripts/webkitpy/common/system/fileutils.py: Removed. * Scripts/webkitpy/test/cat.py: Removed. * Scripts/webkitpy/test/cat_unittest.py: Removed. * Scripts/webkitpy/test/echo.py: Removed. * Scripts/webkitpy/test/echo_unittest.py: Removed. 2012-01-25 Eric Seidel Remove Python 2.5 support from WebKit https://bugs.webkit.org/show_bug.cgi?id=71593 Reviewed by Adam Barth. Just removing support from this one core file. I'll commit the rest of the (large) change once we've gone 24 hours and seen that no bots have exploded. We can roll out this one change if we break bots and then roll in the rest once those bots are fixed. * Scripts/webkitpy/common/system/executive.py: (Executive.cpu_count): 2012-01-25 Hajime Morita ENABLE_SHADOW_DOM should be available via build-webkit --shadow-dom https://bugs.webkit.org/show_bug.cgi?id=76863 Reviewed by Dimitri Glazkov. Added a feature flag. * Scripts/build-webkit: * qmake/mkspecs/features/features.prf: 2012-01-25 Martin Robinson [WK2] Navigation events not triggered by a mouse event claim to be triggered by the left mouse button during policy decisions https://bugs.webkit.org/show_bug.cgi?id=76344 Reviewed by Sam Weinig. Add a test verifying that the mouse button during policy decisions for events not triggered by mouse events is NoButton. * TestWebKitAPI/Tests/WebKit2/PageLoadBasic.cpp: (TestWebKitAPI::decidePolicyForNavigationAction): Added test. (TestWebKitAPI::decidePolicyForNewWindowAction): Added test. 2012-01-25 Ryosuke Niwa run-perf-tests should timeout sooner https://bugs.webkit.org/show_bug.cgi?id=77026 Reviewed by Adam Barth. Bump off the timeout from 10 minutes to 4 minutes. Also use float instead of str in test result values to compress the size of json files bots upload. * Scripts/webkitpy/performance_tests/perftestsrunner.py: (PerfTestsRunner._parse_args): (PerfTestsRunner._process_chromium_style_test_result): (PerfTestsRunner._process_parser_test_result): * Scripts/webkitpy/performance_tests/perftestsrunner_unittest.py: (test_run_test_set_for_parser_tests): (test_run_test_set_with_json_output): (test_run_test_set_with_json_source): 2012-01-25 Enrica Casucci WebView should implement typingAttributes methods to work correctly with the Inspector bar in Mail. https://bugs.webkit.org/show_bug.cgi?id=76951 Reviewed by Alexey Proskuryakov. * TestWebKitAPI/Tests/mac/InspectorBar.mm: Modified to reflect the changes to WebView and WebHTMLView. (TestWebKitAPI::TEST): 2012-01-23 MORITA Hajime [Refactoring][Internals] Should have InternalSettings https://bugs.webkit.org/show_bug.cgi?id=76424 Reviewed by Kent Tamura. * GNUmakefile.am: Added so new files. 2012-01-25 Takashi Toyoshima webkitpy: new-run-webkit-websocket is broken https://bugs.webkit.org/show_bug.cgi?id=76999 Reviewed by Dirk Pranke. r105674 broke new-run-webkit-websocket, since options.chromium doesn't exist any more. new-run-webkit-httpd was also broken and already fixed at r105835. * Scripts/new-run-webkit-websocketserver: (main): 2012-01-25 Nándor Huszka [WK2] WebKitTestRunner needs layoutTestController.setPopupBlockingEnabled https://bugs.webkit.org/show_bug.cgi?id=63458 Reviewed by Adam Roben. Implement the setPopupBlockingEnabled method. * WebKitTestRunner/InjectedBundle/Bindings/LayoutTestController.idl: * WebKitTestRunner/InjectedBundle/LayoutTestController.cpp: (WTR::LayoutTestController::setPopupBlockingEnabled): * WebKitTestRunner/InjectedBundle/LayoutTestController.h: 2012-01-24 Balazs Kelemen [Qt] Cleanup testfonts handling https://bugs.webkit.org/show_bug.cgi?id=76068 Reviewed by Simon Hausmann. Factor test font initialization into it's own file so test font users can share the implementation by adding the file to their build. Use fonts.conf from the test font directory to not require the current directory to be the WebKit root. Fix the -use-test-fonts option of QtTestBrowser by initializing the fonts earlier, before it is too late (namely before calling show on the main window). * DumpRenderTree/qt/DumpRenderTree.pro: * DumpRenderTree/qt/DumpRenderTreeQt.cpp: (WebCore::DumpRenderTree::open): * DumpRenderTree/qt/DumpRenderTreeQt.h: * DumpRenderTree/qt/QtInitializeTestFonts.cpp: Copied from Tools/WebKitTestRunner/InjectedBundle/qt/ActivateFontsQt.cpp. (WebKit::initializeTestFonts): * DumpRenderTree/qt/QtInitializeTestFonts.h: Added. * DumpRenderTree/qt/fonts.conf: Removed. No need for this since we will use the one from the test fonts directory. * DumpRenderTree/qt/main.cpp: (main): Fix the ifdefs to make it clear that the code to set up the Dpi has not been ported to Qt5 yet. * QtTestBrowser/QtTestBrowser.pro: * QtTestBrowser/launcherwindow.h: (WindowOptions::WindowOptions): * QtTestBrowser/main.cpp: (launcherMain): (LauncherApplication::handleUserOptions): * WebKitTestRunner/InjectedBundle/Target.pri: * WebKitTestRunner/InjectedBundle/qt/ActivateFontsQt.cpp: (WTR::activateFonts): Remove the code to set up the Dpi because it does not work with Qt5. 2012-01-25 Simon Hausmann [Qt] build-webkit --qt requires a git-svn or SVN checkout. Reviewed by Tor Arne Vestø. Don't require it by making svnRevisionForDirectory return "unknown" instead of bailing out with an error. * Scripts/VCSUtils.pm: (svnRevisionForDirectory): 2012-01-25 Adam Barth Temporarily disable running the unit tests on the Chromium EWS. These tests don't work properly (see https://bugs.webkit.org/show_bug.cgi?id=76979) and are currently blocking the EWS. * Scripts/webkitpy/common/config/ports.py: (ChromiumPort): 2012-01-25 Adam Barth Update results.html version of PixelZoomer https://bugs.webkit.org/show_bug.cgi?id=76965 Reviewed by Ojan Vafai. Improve readability as suggested by Ojan. * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/pixelzoomer.js: 2012-01-25 Carlos Garcia Campos [GTK] Page clients don't need to be GObjects anymore in WebKit2 GTK+ API https://bugs.webkit.org/show_bug.cgi?id=76899 Reviewed by Martin Robinson. * gtk/generate-gtkdoc: (get_webkit2_options): WebKitWebLoaderClient has been renamed to WebKitLoaderClient. 2012-01-24 Ryosuke Niwa Build fix. Trigger Chromium Mac Release (Perf) on new commits. * BuildSlaveSupport/build.webkit.org-config/config.json: 2012-01-23 Ryosuke Niwa WebKit needs public performance tests running on public bots https://bugs.webkit.org/show_bug.cgi?id=10266 Reviewed by Adam Barth. Completes the initial implementation of WebKit Perf bots. Add a build step for run-perf-test and make google-mac-4 run it, and upload the results to webkit-perf.webkit.org. * BuildSlaveSupport/build.webkit.org-config/config.json: * BuildSlaveSupport/build.webkit.org-config/master.cfg: (RunAndUploadPerfTests): (RunAndUploadPerfTests.start): (UploadPerfTestResults): (BuildAndPerfTestFactory): (BuildAndPerfTestFactory.__init__): 2012-01-24 Ojan Vafai Initializing the browser property of the Bugzilla class takes too long https://bugs.webkit.org/show_bug.cgi?id=76960 Reviewed by Adam Barth. Lazily initialize it. This saves ~150ms on a no-op run-webkit-tests call on my Mac Pro. * Scripts/webkitpy/common/net/bugzilla/bugzilla.py: (Bugzilla.__init__): (Bugzilla._get_browser): (Bugzilla): (Bugzilla._set_browser): 2012-01-24 Mark Rowe Death to JavaScriptGlue. Rubber-stamped by Geoffrey Garen. * BuildSlaveSupport/build-launcher-app: (copyNightlyLauncher): * Scripts/build-webkit: * Scripts/do-file-rename: * Scripts/do-webcore-rename: * Scripts/webkitpy/common/config/build.py: (_should_file_trigger_build): * Scripts/webkitpy/common/config/build_unittest.py: Update the test to use another Mac-specific path in conjunction with a cross-platform path. (ShouldBuildTest): 2012-01-24 Kentaro Hara webkitperl/prepare-ChangeLog_unittest/parser_unittests.pl should test stdout and stderr https://bugs.webkit.org/show_bug.cgi?id=75836 Reviewed by David Kilzer. Previously parser_unittests.pl can just test the returned value of get_function_line_ranges() in prepare-ChangeLog. This patch fixes parser_unittests.pl so that it also tests the stdout and stderr of get_function_line_ranges(). This patch adds css_unittests_warning.css, which contains test cases for warning messages of the CSS parser of prepare-ChangeLog. * Scripts/webkitperl/prepare-ChangeLog_unittest/parser_unittests.pl: (captureOutput): Captures the stdout and stderr of a given method. (convertAbsolutepathToWebKitPath): Converts absolute paths in a given string to relative paths. * Scripts/webkitperl/prepare-ChangeLog_unittest/resources/css_unittests_warning-expected.txt: Added. * Scripts/webkitperl/prepare-ChangeLog_unittest/resources/css_unittests_warning.css: Added. (element1): (element2): (element3): (element4): (): (element5): (element6): * Scripts/webkitperl/prepare-ChangeLog_unittest/resources/cpp_unittests-expected.txt: Just updated the test results, i.e. added stdout and stderr results. No essential changes in the test results. * Scripts/webkitperl/prepare-ChangeLog_unittest/resources/css_unittests-expected.txt: Ditto. * Scripts/webkitperl/prepare-ChangeLog_unittest/resources/java_unittests-expected.txt: Ditto. * Scripts/webkitperl/prepare-ChangeLog_unittest/resources/javascript_unittests-expected.txt: Ditto. * Scripts/webkitperl/prepare-ChangeLog_unittest/resources/perl_unittests-expected.txt: Ditto. * Scripts/webkitperl/prepare-ChangeLog_unittest/resources/python_unittests-expected.txt: Ditto. 2012-01-24 Dirk Pranke webkitpy: new-run-webkit-httpd is broken https://bugs.webkit.org/show_bug.cgi?id=76963 Reviewed by Adam Barth. r105674 broke new-run-webkit-httpd, since options.chromium doesn't exist any more. We didn't catch this because we don't run the integration tests by default anywhere :(. * Scripts/new-run-webkit-httpd: (run): (main): 2012-01-24 Adam Barth garden-o-matic pixel zoomer should let you look at every pixel of the image https://bugs.webkit.org/show_bug.cgi?id=76962 Reviewed by Ojan Vafai. Previously, you couldn't get up to the right and bottom pixels. * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/pixelzoomer.js: 2012-01-24 Enrica Casucci WebHTMLView should implement typingAttributes methods to work correctly with the Inspector bar in Mail. https://bugs.webkit.org/show_bug.cgi?id=76951 Added test to verify that WebHTMLView supports the typingAttributes method and that it works as expected. Reviewed by Alexey Proskuryakov. * TestWebKitAPI/Tests/mac/InspectorBar.mm: (TestWebKitAPI::TEST): 2012-01-24 James Robinson Put a space between the trac link and ellipsis in sheriffbot rollout message so it linkifies better in some IRC clients https://bugs.webkit.org/show_bug.cgi?id=76884 Reviewed by Adam Barth. * Scripts/webkitpy/tool/bot/irc_command.py: (Rollout.execute): * Scripts/webkitpy/tool/bot/sheriffircbot_unittest.py: (SheriffIRCBotTest.test_rollout): (SheriffIRCBotTest.test_revert): (SheriffIRCBotTest.test_multi_rollout): (SheriffIRCBotTest.test_rollout_with_r_in_svn_revision): (SheriffIRCBotTest.test_multi_rollout_with_r_in_svn_revision): (SheriffIRCBotTest.test_rollout_invalidate_reason): (test_multi_rollout_invalidate_reason): 2012-01-24 Kentaro Hara The cpp parser of prepare-ChangeLog cannot detect a change in classes and namespaces https://bugs.webkit.org/show_bug.cgi?id=75531 Reviewed by David Kilzer. Previously, the cpp parser of prepare-ChangeLog could not detect a change outside methods. Consider the following cpp file. namespace N { int a; // this change does not appear on ChangeLog. class C { int b; // this change does not appear on ChangeLog. void f() { int c; // this change appears on ChangeLog. } int d; // this change does not appear on ChangeLog. }; int e; // this change does not appear on ChangeLog. }; The previous prepare-ChangeLog outputs just methods in which a change is found: (N::C::f): This patch fixes prepare-ChangeLog so that it outputs namespaces, classes and methods in which a change is found: (N): (N::C): (N::C::f): * Scripts/prepare-ChangeLog: (get_function_line_ranges_for_cpp): Modified as described above. * Scripts/webkitperl/prepare-ChangeLog_unittest/resources/cpp_unittests.cpp: Added test cases. (Class104): (Class105): (Class106): (Class106::func32): (Class106::func33): (NameSpace3): (NameSpace4): (NameSpace5): (NameSpace6): (Class107): (NameSpace5::NameSpace6::Class107::func34): * Scripts/webkitperl/prepare-ChangeLog_unittest/resources/cpp_unittests-expected.txt: 2012-01-24 Kentaro Hara prepare-ChangeLog outputs warnings for top-level { ... } blocks https://bugs.webkit.org/show_bug.cgi?id=75943 Reviewed by David Kilzer. Currently, prepare-ChangeLog outputs warnings for top-level { ... } blocks, because it does not distinguish '}' at the end of a subroutine with '}' at the end of a if or for statement. foo.pl: if (1) { } for (@v) { } output: nested functions found at top-level at foo.pl:2 nested functions found at top-level at foo.pl:4 This patch fixes the bug. * Scripts/prepare-ChangeLog: (get_function_line_ranges_for_perl): * Scripts/webkitperl/prepare-ChangeLog_unittest/resources/perl_unittests.pl: Added test cases. No change in the test results. 2012-01-24 Dirk Pranke r105674 broke check-webkit-style for chromium's test_expectations. Unreviewed, build fix. We now need to pass in a platform name of 'chromium' if we want the chromium port; options.chromium is no longer recognized. I will follow this patch up with a separate patch that has a test, but this at least fixes the build. * Scripts/webkitpy/style/checkers/test_expectations.py: (TestExpectationsChecker._determine_port_from_exepectations_path): 2012-01-20 Ojan Vafai check-webkit-style of the chromium test_expectations.txt file takes too long https://bugs.webkit.org/show_bug.cgi?id=76745 Reviewed by Dimitri Glazkov. When in lint mode, have TestExpectations test all configurations instead of looping over each configuration. This also has the benefit of making the error output considerably more concise. Also, got rid of the double-printing of errors when linting through check-webkit-style. * Scripts/webkitpy/layout_tests/controllers/manager.py: (Manager.lint): * Scripts/webkitpy/layout_tests/models/test_expectations.py: (TestExpectations._report_errors): (TestExpectations._add_expectations): * Scripts/webkitpy/layout_tests/models/test_expectations_unittest.py: (test_parse_error_nonfatal): (test_error_on_different_platform): * Scripts/webkitpy/style/checkers/test_expectations.py: (TestExpectationsChecker.check_test_expectations): (TestExpectationsChecker.check): * Scripts/webkitpy/style/checkers/test_expectations_unittest.py: (TestExpectationsTestCase.test_determine_port_from_exepectations_path): (TestExpectationsTestCase.assert_lines_lint): 2012-01-24 Ryosuke Niwa Another build fix attempt after r105543. * Scripts/webkitpy/common/net/file_uploader.py: (FileUploader.upload_as_multipart_form_data): 2012-01-24 Ryosuke Niwa Build fix attempt after r105543. * Scripts/webkitpy/common/net/file_uploader.py: (FileUploader.upload_as_multipart_form_data): 2012-01-24 Vincent Scheib [Chromium] Add Pointer Lock test hooks and mock implementation to DumpRenderTree https://bugs.webkit.org/show_bug.cgi?id=76411 Reviewed by Adam Barth. Breaking up https://bugs.webkit.org/show_bug.cgi?id=75762 into a series of patches, this change adds a trivial pointer lock implementation to the Chromium port's DumpRenderTree and exposes LayoutTestController hooks to force failure conditions. * DumpRenderTree/chromium/LayoutTestController.cpp: (LayoutTestController::LayoutTestController): (LayoutTestController::didLosePointerLock): (LayoutTestController::setPointerLockWillFailAsynchronously): (LayoutTestController::setPointerLockWillFailSynchronously): * DumpRenderTree/chromium/LayoutTestController.h: * DumpRenderTree/chromium/WebViewHost.cpp: (WebViewHost::requestPointerLock): (WebViewHost::requestPointerUnlock): (WebViewHost::isPointerLocked): (WebViewHost::didCompletePointerLock): (WebViewHost::didNotCompletePointerLock): (WebViewHost::didLosePointerLock): (WebViewHost::reset): * DumpRenderTree/chromium/WebViewHost.h: (WebViewHost::setPointerLockWillFailAsynchronously): (WebViewHost::setPointerLockWillFailSynchronously): 2012-01-24 Mario Sanchez Prada Unreviewed, skip WK2 API test timing out on some GTK bots. WebKit2APITests/TestDownloads is consistently timing out on the 32bit release and 64bit debug bots. * Scripts/run-gtk-tests: (TestRunner): Skip WebKit2APITests/TestDownloads. 2012-01-24 Andras Becsi [Qt] Fix the build with the newes Qt5 hashes https://bugs.webkit.org/show_bug.cgi?id=76657 Reviewed by Simon Hausmann. * MiniBrowser/qt/MiniBrowserApplication.cpp: (MiniBrowserApplication::MiniBrowserApplication): Disable the automatic touch->mouse event synthesis for now to prevent an infinite loop where Qt always creates mouse events out of our touch mock events. 2012-01-24 Tor Arne Vestbø [Qt] Distinguish between Cygwin and cmd.exe in the way we set env variables We can't use win32-msvc* to decide whether or not to use (set FOO=bar), as building with MinGW inside cmd.exe will require the same construct. Instead we assume that when the first character in PWD is a '/' we're running under Cygwin, and use the normal unix way of setting environment variables. Reviewed by Simon Hausmann. * qmake/mkspecs/features/default_pre.prf: 2012-01-24 Simon Hausmann Some qmake build files must include Qt5's "quick" module. https://bugs.webkit.org/show_bug.cgi?id=76671 Unreviewed build fix. * MiniBrowser/qt/MiniBrowser.pro: * WebKitTestRunner/Target.pri: 2012-01-24 Andras Becsi [Qt] Fix the event timestamps for touch mocking Reviewed by Simon Hausmann. * MiniBrowser/qt/MiniBrowserApplication.cpp: (MiniBrowserApplication::sendTouchEvent): We should not set QEvent::None as the timestamp argument in the QWindowSystemInterface::handleTouchEvent call. 2012-01-24 Carlos Garcia Campos [GTK] Implement DownloadClient in WebKit2 GTK+ API https://bugs.webkit.org/show_bug.cgi?id=72952 Reviewed by Martin Robinson. * gtk/generate-gtkdoc: (get_webkit2_options): Ignore WebKitDownloadClient. 2012-01-23 Zan Dobersek [GTK] editing/deleting/5408255.html results are incorrect https://bugs.webkit.org/show_bug.cgi?id=53644 Reviewed by Martin Robinson. WEBKIT_TOP_LEVEL environment variable is now set directly in either WebKitTestRunner or DumpRenderTree through usage of a compilation-time macro. This way both tools can be run outside the test harness without the need to manually set the environment variable. * DumpRenderTree/gtk/DumpRenderTree.cpp: (getTopLevelPath): * GNUmakefile.am: * Scripts/webkitpy/layout_tests/port/gtk.py: (GtkPort.setup_environ_for_server): * WebKitTestRunner/GNUmakefile.am: * WebKitTestRunner/InjectedBundle/gtk/InjectedBundleGtk.cpp: (WTR::InjectedBundle::platformInitialize): 2012-01-23 Dmitry Lomov [Chromium] Implement layoutTestController.workerThreadCount in DRT https://bugs.webkit.org/show_bug.cgi?id=74653. Reviewed by Darin Fisher. * DumpRenderTree/chromium/LayoutTestController.cpp: (LayoutTestController::LayoutTestController): (LayoutTestController::workerThreadCount): * DumpRenderTree/chromium/LayoutTestController.h: 2012-01-23 Dirk Pranke nrwt: make --chromium work like --qt https://bugs.webkit.org/show_bug.cgi?id=76875 Reviewed by Adam Barth. --chromium used to have to be handled differently from --qt due to the way the PortFactory was implemented; there's not really a good reason for that any more so this patch makes things slightly more consistent and eliminates the options.chromium flag (--chromium is now truly a synonym for --platform chromium). * Scripts/webkitpy/layout_tests/port/factory.py: (PortFactory._default_port): (PortFactory.get): * Scripts/webkitpy/layout_tests/port/factory_unittest.py: (FactoryTest.setUp): (FactoryTest.test_chromium_mac): (FactoryTest.test_chromium_linux): (FactoryTest.test_chromium_win): * Scripts/webkitpy/layout_tests/run_webkit_tests.py: (parse_args): 2012-01-23 Dirk Pranke run-webkit-tests needs to propagate --chromium https://bugs.webkit.org/show_bug.cgi?id=76870 Reviewed by Eric Seidel. run-webkit-tests removes '--chromium' argument from @ARGV when determining which port to run, which means that that doesn't propagate to new-run-webkit-tests. That's bad (and is handled for the other ports by re-adding the flag, but apparently we're just now noticing for Chromium). * Scripts/run-webkit-tests: 2012-01-23 Brian Weinstein More build fixing after r105646. * TestWebKitAPI/Tests/WTF/RedBlackTree.cpp: (TestWebKitAPI::TestNode::key): 2012-01-23 Gavin Barraclough Unreviewed build fix - r105646 broke this. * TestWebKitAPI/Tests/WTF/RedBlackTree.cpp: (TestWebKitAPI::TestNode::TestNode): (TestWebKitAPI::TestNode::key): (TestWebKitAPI::RedBlackTreeTest::assertEqual): (TestWebKitAPI::RedBlackTreeTest::assertSameValuesForKey): (TestWebKitAPI::RedBlackTreeTest::testDriver): 2012-01-23 Sheriff Bot Unreviewed, rolling out r105658. http://trac.webkit.org/changeset/105658 https://bugs.webkit.org/show_bug.cgi?id=76883 We want this eventually, but not right at this moment (Requested by abarth on #webkit). * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/config.js: 2012-01-23 Adam Barth garden-o-matic should support Chromium Mac Lion https://bugs.webkit.org/show_bug.cgi?id=76880 Reviewed by Eric Seidel. * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/config.js: 2012-01-23 Emil A Eklund check-webkit-style whitespace/operators triggers on overloaded division operator https://bugs.webkit.org/show_bug.cgi?id=76650 Reviewed by Darin Adler. * Scripts/webkitpy/style/checkers/cpp.py: (check_spacing): * Scripts/webkitpy/style/checkers/cpp_unittest.py: (CppStyleTest.test_operator_methods): 2012-01-23 Ryosuke Niwa run-perf-tests should report server-side errors https://bugs.webkit.org/show_bug.cgi?id=76802 Reviewed by Tony Chang. Report errors whenever server's response doesn't match "OK". * Scripts/webkitpy/common/net/file_uploader.py: (FileUploader.upload_single_text_file): (FileUploader.upload_as_multipart_form_data): (FileUploader._upload_data.callback): (FileUploader): (FileUploader._upload_data): * Scripts/webkitpy/performance_tests/perftestsrunner.py: (PerfTestsRunner._upload_json): * Scripts/webkitpy/performance_tests/perftestsrunner_unittest.py: (test_upload_json): (test_upload_json.MockFileUploader.upload_single_text_file): 2012-01-22 Ryosuke Niwa run-perf-tests ignore Skipped list on chromium https://bugs.webkit.org/show_bug.cgi?id=76764 Reviewed by Dirk Pranke. Move skipped_perf_tests from WebKit port to Base port so that Chromium port can also find skipped list. Chromium port only uses test_expectations.txt for layout tests but performacne tests don't use test_expectations.txt so Chromium port also needs to use Skipped list. * Scripts/webkitpy/layout_tests/port/base.py: (Port._tests_from_skipped_file_contents): (Port): (Port._expectations_from_skipped_files): (Port.skipped_perf_tests): * Scripts/webkitpy/layout_tests/port/base_unittest.py: (PortTest.test_skipped_perf_tests): (PortTest.test_skipped_perf_tests.add_text_file): * Scripts/webkitpy/layout_tests/port/webkit.py: (WebKitPort._skipped_tests_for_unsupported_features): (WebKitPort._skipped_file_search_paths): (WebKitPort.skipped_layout_tests): 2012-01-23 Andy Estes Fix the build after r105635. * TestWebKitAPI/Tests/WTF/StringBuilder.cpp: (TestWebKitAPI::TEST): 2012-01-23 Xianzhu Wang Basic enhancements to StringBuilder https://bugs.webkit.org/show_bug.cgi?id=67081 Reviewed by Darin Adler. * TestWebKitAPI/Tests/WTF/StringBuilder.cpp: (TestWebKitAPI::TEST): 2012-01-23 Mario Sanchez Prada [GTK] run-gtk-tests randomly fails while running the xprop comand https://bugs.webkit.org/show_bug.cgi?id=76817 Reviewed by Gustavo Noronha Silva. No need to use xprop to remove the AT_SPI_BUS property since run-gtk-tests will always run new instances of Xvfb. * Scripts/run-gtk-tests: (TestRunner): Unskip TestWebKitAccessibility. (TestRunner.run): Uncomment lines for launching the accessibility bus and registry daemon, and remove lines for running xprop. 2012-01-23 Tor Arne Vestbø [Qt] Don't warn about override and final being C++11 extensions Clang will emit a warning when these extensions are used without passing --std=c++11, but we use feature checking to decide if we have the right extensions, so we can safely ignore these warnings. The XCode and Windows project files have the same workaround. The reason for adding the flag to QMAKE_OBJECTIVE_CFLAGS as well is that we only have one extra compiler for Objective-C, which is also used for Objective-C++ sources, so we have to pass the flag, even if it doesn't make sense for Objective-C. Reviewed by Simon Hausmann. * qmake/mkspecs/features/unix/default_post.prf: 2012-01-20 Tor Arne Vestbø [Qt] Change how build-webkit decides when to do full incremental builds Instead of relying on update-webkit (which isn't run on the bots) to decide when to do a full incremental build (make qmake), we let the build-webkit script itself check the current SVN revision against the previous build (by storing it in .webkit.config). If the two differ we assume a full incremental build is needed, since the new revisions might have introduced problematic things like new Q_OBJECT macros. If not, we assume the developer is doing changes locally, and revert to doing a plain 'make'. In addition, when the build fails in the latter case, we inform the developer of possible pitfalls and how to manually run 'make qmake'. Reviewed by Ossy. * Scripts/update-webkit: * Scripts/webkitdirs.pm: (buildQMakeProjects): 2012-01-23 Mario Sanchez Prada [GTK] run-gtk-tests randomly fails while running the xprop comand https://bugs.webkit.org/show_bug.cgi?id=76817 Reviewed by Philippe Normand. Temporarily comment the lines related to launching the ATSPI bus and registry daemon, which are making the bots to fail randomly. * Scripts/run-gtk-tests: (TestRunner): Skip TestWebKitAccessibility, as it won't run properly if the ATSPI infrastructure is not properly initialized. (TestRunner.run): Comment lines related to initialization of ATSPI. 2012-01-22 Ryosuke Niwa Windows python test build fix. * Scripts/webkitpy/performance_tests/perftestsrunner.py: (PerfTestsRunner.__init__): 2012-01-22 Sergio Villar Senin [GTK] DumpRenderTree converts "file:///" to a path differently https://bugs.webkit.org/show_bug.cgi?id=76631 Reviewed by Martin Robinson. DumpRenderTree should print "/" as the last path component if the path is a lone slash instead of empty output. * DumpRenderTree/gtk/DumpRenderTree.cpp: (webViewConsoleMessage): 2012-01-20 David Levin The leaky DC check should use adoptPtr as a signal instead of OwnPtr to get less false positives. https://bugs.webkit.org/show_bug.cgi?id=76752 Reviewed by Eric Seidel. * Scripts/webkitpy/style/checkers/cpp.py: (check_for_leaky_patterns): * Scripts/webkitpy/style/checkers/cpp_unittest.py: (LeakyPatternTest.test_create_dc): (LeakyPatternTest.test_create_compatible_dc): 2012-01-20 Shawn Singh Changed my status in commiters.py Unreviewed. * Scripts/webkitpy/common/config/committers.py: 2012-01-20 Ojan Vafai run-webkit-tests --lint-test-files crawls the whole LayoutTests subtree https://bugs.webkit.org/show_bug.cgi?id=76748 Reviewed by Ryosuke Niwa. It crawls the whole subtree and then doesn't use the data. Cutting this out saves 4 seconds warm and 17 seconds cold on my Mac Pro. * Scripts/webkitpy/layout_tests/run_webkit_tests.py: (run): 2012-01-20 Adam Barth Another tiny tweak to the garden-o-matic CSS. This makes things line up slightly nicer when there aren't any failures. * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/styles/onebar.css: (#summary): 2012-01-19 Ryosuke Niwa run-perf-tests should support --test-results-server option https://bugs.webkit.org/show_bug.cgi?id=76680 Reviewed by Adam Barth. Add --test-results-server, --builder-name, and --build-number options to run-perf-tests to be used by perf bots. Also refactor file_uploader as needed. * Scripts/webkitpy/common/net/file_uploader.py: (FileUploader.__init__): (FileUploader.upload_single_file): (FileUploader.upload_as_multipart_form_data): (FileUploader): (FileUploader._upload_data): (FileUploader._upload_data.callback): * Scripts/webkitpy/layout_tests/layout_package/json_results_generator.py: (JSONResultsGeneratorBase.upload_json_files): * Scripts/webkitpy/performance_tests/perftestsrunner.py: (PerfTestsRunner): (PerfTestsRunner._parse_args): (PerfTestsRunner.run): (PerfTestsRunner._generate_json): (PerfTestsRunner._upload_json): (PerfTestsRunner._run_tests_set): * Scripts/webkitpy/performance_tests/perftestsrunner_unittest.py: (create_runner): (test_run_with_upload_json): (test_run_with_upload_json.mock_upload_json): (test_upload_json): (test_upload_json.MockFileUploader): (test_upload_json.MockFileUploader.__init__): (test_upload_json.MockFileUploader.upload_single_file): (test_parse_args): 2012-01-20 Adam Barth The party time image overlaps real content! This patch makes the image centered vertically. * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/styles/onebar.css: (#onebar.partytime #summary): 2012-01-19 Ojan Vafai Refactor TestExpectationsParser in preparation for caching the results https://bugs.webkit.org/show_bug.cgi?id=76669 Reviewed by Dimitri Glazkov. Make everything private expect for the parse method. Eventually, we'll need the expectations lines to not be modified outside of TestExpectationsParser so we can cache the results. This makes check-webkit-style of the chromium test_expectations.txt file go from ~17 seconds to ~12 seconds on my Mac Pro. This patch is just a refactor in preparation, so no new tests. * Scripts/webkitpy/layout_tests/controllers/test_expectations_editor_unittest.py: (TestExpectationEditorTests.make_parsed_expectation_lines): * Scripts/webkitpy/layout_tests/models/test_expectations.py: (TestExpectationParser.parse): (TestExpectationParser): (TestExpectationParser._parse_line): (TestExpectationParser._tokenize): (TestExpectationParser._tokenize_list): (TestExpectationsModel._clear_expectations_for_test): (TestExpectations.__init__): (TestExpectations._add_expectations): (TestExpectations._add_skipped_tests): * Scripts/webkitpy/layout_tests/models/test_expectations_unittest.py: (TestExpectationParserTests.test_tokenize_blank): (TestExpectationParserTests.test_tokenize_missing_colon): (TestExpectationParserTests.test_tokenize_extra_colon): (TestExpectationParserTests.test_tokenize_empty_comment): (TestExpectationParserTests.test_tokenize_comment): (TestExpectationParserTests.test_tokenize_missing_equal): (TestExpectationParserTests.test_tokenize_extra_equal): (TestExpectationParserTests.test_tokenize_valid): (TestExpectationParserTests.test_tokenize_valid_with_comment): (TestExpectationParserTests.test_tokenize_valid_with_multiple_modifiers): (TestExpectationParserTests.test_parse_empty_string): (TestExpectationSerializerTests.assert_round_trip): (TestExpectationSerializerTests.assert_list_round_trip): * Scripts/webkitpy/tool/commands/expectations.py: (OptimizeExpectations.execute): * Scripts/webkitpy/tool/servers/gardeningserver.py: (GardeningExpectationsUpdater.update_expectations): 2012-01-20 Adam Barth Follow-up to previous patch: don't produce NaN when the revision number is missing. * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/model.js: * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/model_unittests.js: 2012-01-20 Adam Barth Garden-o-matic should tell me which revisions have been checked by all the bots https://bugs.webkit.org/show_bug.cgi?id=76722 Reviewed by Dimitri Glazkov. This information helps me know when it's safe for me to roll a candidate revision. * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/garden-o-matic.js: * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/model.js: * BuildSlaveSupport/build.webkit.org-config/public_html/TestFailures/scripts/model_unittests.js: 2012-01-20 Tor Arne Vestbø build-webkit: Don't spit out congratulations message on Ctrl+C Reviewed by Andreas Kling. * Scripts/build-webkit: 2012-01-20 Philippe Normand [GTK] Allow extra jhbuild modulesets https://bugs.webkit.org/show_bug.cgi?id=76691 Reviewed by Gustavo Noronha Silva. * gtk/jhbuildrc: Extra modulesets and modules can be added in JHBuild using the WEBKIT_EXTRA_MODULESETS and WEBKIT_EXTRA_MODULES env variables respectively. Use comma-separated values. Example: WEBKIT_EXTRA_MODULES=file:///path/to/module.set,file:///other/path. 2012-01-20 Mario Sanchez Prada [GTK] Expose accessibility hierarchy in WebKit2 to ATK/AT-SPI based ATs https://bugs.webkit.org/show_bug.cgi?id=72589 Reviewed by Martin Robinson. Ensure the accessibility infrastructure is available when running the unit tests, otherwise WK2's accessibility tests won't work. * Scripts/run-gtk-tests: (TestRunner): (TestRunner._lookup_atspi2_binary): New. Looks for the directory where at-spi2's stuff is located (like at-spi-bus-launcher and at-spi2-registryd). (TestRunner.run): Initialize the accessibility infrastructure. * gtk/generate-gtkdoc: (get_webkit2_options): Add 'WebKitWebViewBaseAccessible.*'. * gtk/jhbuild.modules: Added at-spi2-core and at-spi2-atk. 2012-01-20 Hayato Ito Add ShadowRoot.idl which is enabled by newly introduced SHADOW_DOM flag. https://bugs.webkit.org/show_bug.cgi?id=76353 Reviewed by Hajime Morita. Add ShadowRoot.idl, which is enabled only on chromium port since this is under development feature. ShadowRoot.idl contains minimum API so that we can test it. Other APIs should be added on other changes so that we can isolate issues. * DumpRenderTree/chromium/TestShell.cpp: (TestShell::TestShell): 2012-01-20 Mario Sanchez Prada [GTK] GTK's DRT not to log events for already defunct objects https://bugs.webkit.org/show_bug.cgi?id=76620 Reviewed by Martin Robinson. Do not log 'state-change:defunct' events. * DumpRenderTree/gtk/AccessibilityCallbacks.cpp: (printAccessibilityEvent): Early return if a 'state-change:defunct' signal is passed. (axObjectEventListener): Pass the signal name and value in separate parameters to printAccessibilityEvent. 2012-01-19 Gustavo Noronha Silva Unreviewed build fix. Wrong variable being used as the cwd for git remote update command. * gtk/run-with-jhbuild: (update_jhbuild): 2012-01-19 Gustavo Noronha Silva [GTK] ensure the jhbuild used by webkit is as up-to-date as needed https://bugs.webkit.org/show_bug.cgi?id=76585 Reviewed by Martin Robinson. This allows us to specify a minimum version of jhbuild required, while also providing stability against changes done to the master branch, by letting us choose what is the exact version that will be used. * gtk/run-with-jhbuild: refactored for more hackability and easier reading (jhbuild_installed): checks whether jhbuild is installed (jhbuild_cloned): checks whether jhbuild has been cloned (jhbuild_at_expected_revision): checks whether jhbuild is at the expected revision (update_jhbuild): brings jhbuild to the expected revision (clone_jhbuild): clones jhbuild (install_jhbuild): build and installs jhbuild (update_webkitgtk_libs): runs update-webkitgtk-libs (ensure_jhbuild): high-level logic to decide whether and which of the steps above are executed 2012-01-19 Alexey Proskuryakov [WK2] DumpRenderTree converts "file:///" to a path differently https://bugs.webkit.org/show_bug.cgi?id=76653 Reviewed by John Sullivan. * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp: (WTR::lastFileURLPathComponent): Do print a slash if there are no path components. 2012-01-19 Eric Seidel webkit-patch cannot rollout patches without changelog/bug number https://bugs.webkit.org/show_bug.cgi?id=75962 Reviewed by Adam Barth. * Scripts/webkitpy/tool/commands/download.py: (AbstractRolloutPrepCommand._prepare_state): * Scripts/webkitpy/tool/commands/download_unittest.py: (AbstractRolloutPrepCommandTest.test_prepare_state): 2012-01-19 Alexey Proskuryakov [WK2] fast/url/degenerate-file-base.html fails https://bugs.webkit.org/show_bug.cgi?id=76619 Reviewed by Adam Roben. A URL generated form local path for loading had an extra slash, making document.URL incorrect in all WebKit2 tests (file://localhost//path/to/test.html). * WebKitTestRunner/TestInvocation.cpp: (WTR::createWKURL): 2012-01-18 Ojan Vafai check-webkit-style of the chromium test_expectations.txt file doesn't test all chromium ports https://bugs.webkit.org/show_bug.cgi?id=76510 Reviewed by Adam Barth. Unlike other style rules, if you get the syntax of the test_expectations.txt file wrong, the layout tests won't run. Also, this check is simple and only slows down committing if you actually modify one of the test_expectations.txt files. * Scripts/webkitpy/style/checkers/test_expectations.py: (TestExpectationsChecker.check_test_expectations): (TestExpectationsChecker.check): * Scripts/webkitpy/style/checkers/test_expectations_unittest.py: (TestExpectationsTestCase.test_check_covers_all_configurations): (TestExpectationsTestCase.test_check_covers_all_configurations.mock_check_test_expectations): (TestExpectationsTestCase): (assert_lines_lint): 2012-01-19 Ryosuke Niwa Build fix. Revert an unintentional change to WebKitDriver in r105443. * Scripts/webkitpy/layout_tests/port/driver.py: (DriverProxy.__init__): 2012-01-19 Dirk Pranke remove the duplicated build rules from WebKit.gyp, Tools.gyp https://bugs.webkit.org/show_bug.cgi?id=73384 Reviewed by Tony Chang. This patch removes the no-longer-necessary conditional logic for build_webkit_exes_from_webkit_gyp; we now always build the exes from their own dedicated gyp files. * DumpRenderTree/DumpRenderTree.gyp/DumpRenderTree.gyp: 2012-01-19 Alexey Proskuryakov make-new-script-test should work with HTTP tests https://bugs.webkit.org/show_bug.cgi?id=76603 Reviewed by Adam Roben. * Scripts/make-new-script-test: (makePathToSharedSources): Renamed from makeTestRelativePathToSharedSources(), since it now also creates absolute paths for HTTP. 2012-01-19 Ryosuke Niwa Some perf tests time out when ran by run-perf-tests https://bugs.webkit.org/show_bug.cgi?id=76612 Reviewed by Dirk Pranke and Eric Seidel. Always pass --no-timeout to DumpRenderTree from run-perf-tests. Otherwise some tests such as Parser/xml-parser.html will timeout. --no-timeout option is currently supported by Chromium and Mac ports. * Scripts/webkitpy/layout_tests/port/base.py: (Port.to.create_driver): * Scripts/webkitpy/layout_tests/port/chromium.py: (ChromiumDriver.__init__): (ChromiumDriver._wrapper_options): * Scripts/webkitpy/layout_tests/port/driver.py: (Driver.__init__): (DriverProxy.__init__): * Scripts/webkitpy/layout_tests/port/webkit.py: (WebKitDriver.__init__): (WebKitDriver.cmd_line): * Scripts/webkitpy/layout_tests/port/webkit_unittest.py: (WebKitDriverTest.test_read_binary_block): (WebKitDriverTest): (WebKitDriverTest.test_no_timeout): * Scripts/webkitpy/layout_tests/run_webkit_tests_integrationtest.py: (get_tests_run.RecordingTestDriver.__init__): * Scripts/webkitpy/performance_tests/perftestsrunner.py: (PerfTestsRunner._parse_args): (PerfTestsRunner._run_tests_set): * Scripts/webkitpy/performance_tests/perftestsrunner_unittest.py: (create_runner): 2012-01-19 Ben Wells Add support for window.print to chromium DRT https://bugs.webkit.org/show_bug.cgi?id=76479 When used in a test window.print goes into print mode and then straight out again. This will allow the afterprint event to be tested by having the afterprint event happen before the dump happens. Reviewed by Mihai Parparita. * DumpRenderTree/chromium/WebViewHost.cpp: (WebViewHost::printPage): * DumpRenderTree/chromium/WebViewHost.h: 2012-01-18 Enrica Casucci editingAttributedStringFromRange in WebHTMLConverter does not handle NSUnderlineStyleAttributeName. https://bugs.webkit.org/show_bug.cgi?id=76588 Added test. Reviewed by Dan Bernstein. * TestWebKitAPI/Tests/mac/InspectorBar.mm: (TestWebKitAPI::TEST): 2012-01-19 Tor Arne Vestbø [Qt] Do full incremental builds until bots use update-webkit Rubber-stamped by Simon Hausmann. * Scripts/webkitdirs.pm: (buildQMakeProjects): 2012-01-19 Tor Arne Vestbø Fix the Qt build Unreviewed build fix. * Scripts/webkitdirs.pm: (buildQMakeProjects): 2012-01-19 Tor Arne Vestbø [Qt] Make build-webkit always do safe incremental builds after update-webkit When building against Qt5 with GCC we would just run 'make' in the build dir, and rely on the dependency files output by GCC (-MD) to handle dependency tracking, but that fails for special-cases like adding a Q_OBJECT macro to a header. To guarantee that an incrmental build will work, we have to run 'make qmake', which we now do on every build-webkit that's followed by a successful run of update-webkit. The reasoning is that update-webkit can result in such potential corner-cases being applied, and since we can't know for sure unless we inspect the diff and account for all the corner cases we assume the worst and always run 'make qmake'. After a succesful run of build-webkit we proceed to do just 'make' for any subsequent runs, since we assume that the developer knows what kind of changes he/she is doing, and when a 'make qmake' is needed. Reviewed by Simon Hausmann. * Scripts/update-webkit: * Scripts/webkitdirs.pm: (determineCurrentSVNRevision): (currentSVNRevision): (buildQMakeProjects): * qmake/mkspecs/features/default_post.prf: 2012-01-18 Ryosuke Niwa run-perf-tests should support Skipped list https://bugs.webkit.org/show_bug.cgi?id=76594 Reviewed by Adam Barth. Add a support for Skipped list in run-perf-tests; also skip files in resources directories. * Scripts/webkitpy/layout_tests/port/base.py: (Port.perf_tests_dir): (Port.skipped_perf_tests): (Port.skips_perf_test): (Port.relative_perf_test_filename): * Scripts/webkitpy/layout_tests/port/test.py: (TestPort.perf_tests_dir): * Scripts/webkitpy/layout_tests/port/webkit.py: (WebKitPort._expectations_from_skipped_files): (WebKitPort): (WebKitPort.skipped_layout_tests): (WebKitPort.skipped_perf_tests): * Scripts/webkitpy/performance_tests/perftestsrunner.py: (PerfTestsRunner._collect_tests): (PerfTestsRunner.run): * Scripts/webkitpy/performance_tests/perftestsrunner_unittest.py: (test_run_test_set_with_json_output): (test_run_test_set_with_json_source): (test_collect_tests_with_skipped_list): (test_collect_tests_with_skipped_list.add_file): 2012-01-18 Dirk Pranke Fix path to chromium_src_dir introduced in previous change. Unreviewed, build fix. * DumpRenderTree/DumpRenderTree.gyp/DumpRenderTree.gyp: 2012-01-18 Alexey Proskuryakov Need infrastructure to test Content-Disposition filename encoding support https://bugs.webkit.org/show_bug.cgi?id=76572 Reviewed by Sam Weinig. Made policy delegate log suggested file name when resource has Content-Disposition: attachment. WebKitTestRunner does not have a policy delegate yet, bug 42546. * DumpRenderTree/mac/PolicyDelegate.mm: (dispositionTypeFromContentDispositionHeader): (-[PolicyDelegate webView:decidePolicyForMIMEType:request:frame:decisionListener:]): 2012-01-18 Ojan Vafai Fix tests from http://trac.webkit.org/changeset/105332. The code assumed that if you passed an options object with a configuration attribute that it would have a valid value. A bunch of the testing mocks do things like having a configuration value of None. * Scripts/webkitpy/layout_tests/port/base.py: (Port.__init__): 2012-01-18 Ojan Vafai Shave 0.5 seconds off check-webkit-style runtime for test_expectations.txt https://bugs.webkit.org/show_bug.cgi?id=76576 Reviewed by Adam Barth. Avoid calling default_configuration in base.py. We don't need it for checking test_expectations.txt style. It takes ~0.5 seconds to run on my Mac Pro. It's the call to "perl Tools/Scripts/webkit-build-directory --top-level" from common.executive. At some point someone should probably look into why that call is so slow since it's on the critical path for run-webkit-tests and build-webkit startup. * Scripts/webkitpy/layout_tests/port/base.py: (Port.__init__): * Scripts/webkitpy/style/checkers/test_expectations.py: (TestExpectationsChecker._determine_port_from_exepectations_path): 2012-01-18 Dirk Pranke [chromium] move Tools.gyp, switch build-webkit --chromium to All.gyp https://bugs.webkit.org/show_bug.cgi?id=76505 Reviewed by Tony Chang. Update chromium build script to use All.gyp instead of WebKit.gyp * DumpRenderTree/DumpRenderTree.gyp/DumpRenderTree.gyp: Renamed from Source/WebKit/chromium/Tools.gyp. * Scripts/webkitdirs.pm: 2012-01-18 Sam Weinig Don’t hardcode the path to ‘make‘. Reviewed by Dan Bernstein. * DumpRenderTree/DumpRenderTree.xcodeproj/project.pbxproj: 2012-01-18 Joshua Bell Adding myself to committers.py https://bugs.webkit.org/show_bug.cgi?id=76569 Unreviewed. * Scripts/webkitpy/common/config/committers.py: 2012-01-18 Adam Barth Up the cap on test failures tolerated by the EWS so the mac-ews can run to completion https://bugs.webkit.org/show_bug.cgi?id=76567 Reviewed by Eric Seidel. Current the AppleMac port has slightly more failures than we allow on the EWS, causing the mac-ews to spin. This patch ups the limit a bit, which should hopefully let the mac-ews run to completion. Once AppleMac has fewer failures, we can lower the cap again. * Scripts/webkitpy/tool/bot/layouttestresultsreader_unittest.py: (LayoutTestResultsReaderTest.test_layout_test_results): * Scripts/webkitpy/tool/steps/runtests.py: (RunTests): * Scripts/webkitpy/tool/steps/steps_unittest.py: 2012-01-18 Eric Carlson Provide access to user's list of preferred languages https://bugs.webkit.org/show_bug.cgi?id=76138 Reviewed by Alexey Proskuryakov. * GNUmakefile.am: Add JSInternalsCustom.cpp. 2012-01-18 Adam Barth Add Chromium Linux GPU 32 to the list of allowed Chromium Linux configurations. * Scripts/webkitpy/layout_tests/port/chromium_linux.py: (ChromiumLinuxPort.__init__): 2012-01-18 Dominic Mazzoni Accessibility: Chromium needs methods to scroll an object into view or to a specific location. https://bugs.webkit.org/show_bug.cgi?id=73460 Reviewed by Chris Fleizach. * DumpRenderTree/AccessibilityUIElement.h: * DumpRenderTree/chromium/AccessibilityUIElement.cpp: (AccessibilityUIElement::AccessibilityUIElement): (AccessibilityUIElement::scrollToMakeVisibleCallback): (AccessibilityUIElement::scrollToMakeVisibleWithSubFocusCallback): (AccessibilityUIElement::scrollToGlobalPointCallback): * DumpRenderTree/chromium/AccessibilityUIElement.h: * DumpRenderTree/gtk/AccessibilityUIElementGtk.cpp: (AccessibilityUIElement::scrollToMakeVisible): (AccessibilityUIElement::scrollToMakeVisibleWithSubFocus): (AccessibilityUIElement::scrollToGlobalPoint): * DumpRenderTree/mac/AccessibilityUIElementMac.mm: (AccessibilityUIElement::scrollToMakeVisible): (AccessibilityUIElement::scrollToMakeVisibleWithSubFocus): (AccessibilityUIElement::scrollToGlobalPoint): * DumpRenderTree/win/AccessibilityUIElementWin.cpp: (AccessibilityUIElement::scrollToMakeVisible): (AccessibilityUIElement::scrollToMakeVisibleWithSubFocus): (AccessibilityUIElement::scrollToGlobalPoint): 2012-01-17 Enrica Casucci Missing NSForegroundColorAttributeName should be treated as black instead of transparent. https://bugs.webkit.org/show_bug.cgi?id=76490 Added TestWebKitAPI test. Reviewed by Dan Bernstein. * TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj: Added new test file. * TestWebKitAPI/Tests/mac/InspectorBar.mm: Added test that calls changeAttributes to the WebView. 2012-01-18 Balazs Kelemen [Qt] Consolidate layout test crash logging https://bugs.webkit.org/show_bug.cgi?id=75088 Reviewed by Simon Hausmann. Move backtrace generating logic into WTFReportBacktrace and add a way to deinstall signal handlers if we know that we have already printed the backtrace. * DumpRenderTree/qt/main.cpp: (crashHandler): (setupSignalHandlers): (WTFCrashHook): (main): * WebKitTestRunner/InjectedBundle/qt/InjectedBundleQt.cpp: (WTR::crashHandler): (WTR::setupSignalHandlers): (WTR::crashHook): (WTR::InjectedBundle::platformInitialize): 2012-01-18 Ilya Tikhonovsky Unreviewed build fix after r105256. There was a cyclic dependency between self._port and self._host assigments. * Scripts/webkitpy/performance_tests/perftestsrunner.py: (PerfTestsRunner.__init__): 2012-01-18 Tor Arne Vestbø Add new watchlist rule for the Qt build system Reviewed by Simon Hausmann. * Scripts/webkitpy/common/config/watchlist: 2012-01-18 Tor Arne Vestbø [Qt] Move OTHER_FILES from WebKit.pro to Tools.pro Reviewed by Simon Hausmann. * Tools.pro: 2012-01-18 Tor Arne Vestbø [Qt] Allow building only a subset of the WebKit sub-projects Fixes build-jsc, broken in r104825. We no longer need the intermediate buildQMakeQtProject(), as we're the only port using qmake, and buildQMakeProject is now highly Qt specific. Reviewed by Simon Hausmann. * Scripts/build-jsc: * Scripts/build-webkit: * Scripts/webkitdirs.pm: (buildQMakeProjects): 2012-01-18 Ryosuke Niwa run-perf-tests should generate a json file that summaries the result https://bugs.webkit.org/show_bug.cgi?id=76504 Reviewed by Adam Barth. Add an ability to generate a json file to run-perf-test in the preparation for perf bots. New option --outout-json-path specifies the json file's path, and --source-json-path specifies another json file to be merged into the generated json file. Also fixed a bug that --build wasn't set by default. * Scripts/webkitpy/layout_tests/port/test.py: (TestPort.webkit_base): * Scripts/webkitpy/performance_tests/perftestsrunner.py: (PerfTestsRunner.__init__): (PerfTestsRunner): (PerfTestsRunner._parse_args): (PerfTestsRunner.run): (PerfTestsRunner._generate_json_if_specified): (PerfTestsRunner._process_chromium_style_test_result): (PerfTestsRunner._process_parser_test_result): * Scripts/webkitpy/performance_tests/perftestsrunner_unittest.py: (MainTest): (create_runner): (test_run_test_set): (test_run_test_set_for_parser_tests): (test_run_test_set_with_summary_json): (test_collect_tests): (test_parse_args): 2012-01-17 Sheriff Bot Unreviewed, rolling out r105244. http://trac.webkit.org/changeset/105244 https://bugs.webkit.org/show_bug.cgi?id=76518 broke Chromium Mac (Requested by rolandsteiner on #webkit). * DumpRenderTree/AccessibilityUIElement.h: * DumpRenderTree/chromium/AccessibilityUIElement.cpp: (AccessibilityUIElement::AccessibilityUIElement): * DumpRenderTree/chromium/AccessibilityUIElement.h: * DumpRenderTree/gtk/AccessibilityUIElementGtk.cpp: (AccessibilityUIElement::removeSelection): * DumpRenderTree/mac/AccessibilityUIElementMac.mm: * DumpRenderTree/win/AccessibilityUIElementWin.cpp: (AccessibilityUIElement::removeSelection): 2012-01-17 Dominic Mazzoni Accessibility: Chromium needs methods to scroll an object into view or to a specific location. https://bugs.webkit.org/show_bug.cgi?id=73460 Reviewed by Chris Fleizach. * DumpRenderTree/AccessibilityUIElement.h: * DumpRenderTree/chromium/AccessibilityUIElement.cpp: (AccessibilityUIElement::AccessibilityUIElement): (AccessibilityUIElement::scrollToMakeVisibleCallback): (AccessibilityUIElement::scrollToMakeVisibleWithSubFocusCallback): (AccessibilityUIElement::scrollToGlobalPointCallback): * DumpRenderTree/chromium/AccessibilityUIElement.h: * DumpRenderTree/gtk/AccessibilityUIElementGtk.cpp: (AccessibilityUIElement::scrollToMakeVisible): (AccessibilityUIElement::scrollToMakeVisibleWithSubFocus): (AccessibilityUIElement::scrollToGlobalPoint): * DumpRenderTree/mac/AccessibilityUIElementMac.mm: (AccessibilityUIElement::scrollToMakeVisible): (AccessibilityUIElement::scrollToMakeVisibleWithSubFocus): (AccessibilityUIElement::scrollToGlobalPoint): * DumpRenderTree/win/AccessibilityUIElementWin.cpp: (AccessibilityUIElement::scrollToMakeVisible): (AccessibilityUIElement::scrollToMakeVisibleWithSubFocus): (AccessibilityUIElement::scrollToGlobalPoint): 2012-01-17 Adam Barth Enable tests on the mac-ews https://bugs.webkit.org/show_bug.cgi?id=76502 Reviewed by Simon Fraser. Apple now has five machines running the mac-ews. Lucas Forschler has asked that we try enabling tests to see if we have enough bandwidth. * Scripts/webkitpy/tool/commands/earlywarningsystem.py: (MacEWS): 2012-01-17 Sanjoy Das Add a check-webkit-style exception for GDBInterface.cpp https://bugs.webkit.org/show_bug.cgi?id=76187 Add an exception for Source/JavaScriptCore/jit/GDBInterface.cpp so that it does not fail the style check. Reviewed by Martin Robinson. * Scripts/webkitpy/style/checker.py: Tell the style checker to not run readability/naming tests on Source/JavaScriptCore/jit/GDBInterface.cpp. * Scripts/webkitpy/style/checker_unittest.py: (GlobalVariablesTest.test_path_rules_specifier): Add a test for the above exception. 2012-01-17 Hajime Morrita https://bugs.webkit.org/show_bug.cgi?id=76245 [Mac][DRT] should allow query string for the file URL. Reviewed by Tony Chang. Taught DRT about file:// as a url instead of a file path. We are now able to use a path file:///foo/bar.html?baz as a url with which DRT opens "/foo/bar.html" with "baz" as a query string. * DumpRenderTree/mac/DumpRenderTree.mm: (runTest): 2012-01-17 Ojan Vafai Check style on test_expectations.txt files before commit https://bugs.webkit.org/show_bug.cgi?id=76484 Reviewed by Adam Barth. Unlike other style rules, if you get the syntax of the test_expectations.txt file wrong, the layout tests won't run. Also, this check is simple and only slows down committing if you actually modify one of the test_expectations.txt files. * Scripts/webkitpy/tool/steps/commit.py: (Commit._check_test_expectations): (Commit.run): * Scripts/webkitpy/tool/steps/commit_unittest.py: Added. (CommitTest): (CommitTest.test_check_test_expectations): 2012-01-17 Viatcheslav Ostapenko [Qt] Debug build fails with debug qt5 https://bugs.webkit.org/show_bug.cgi?id=76463 Reviewed by Simon Hausmann. Add QNetworkCookie include in order to satisfy sizeof in QTypeInfo in debug builds. * QtTestBrowser/cookiejar.h: 2012-01-17 Dirk Pranke webkitpy: fix json import on linux https://bugs.webkit.org/show_bug.cgi?id=76481 Reviewed by Adam Barth. For some reason, the version of simplejson autoinstalled on my linux box seems to be broken. We should really only be using simplejson if json isn't available anyway, so this change looks for json first. * Scripts/webkitpy/layout_tests/layout_package/json_results_generator_unittest.py: (JSONGeneratorTest.test_test_timings_trie): 2012-01-17 Dirk Pranke Fix failures in test-webkitpy caused by r105177. Unreviewed, build fix. * Scripts/webkitpy/layout_tests/port/chromium_linux.py: (ChromiumLinuxPort._determine_driver_path_statically): (ChromiumLinuxPort._static_build_path): 2012-01-17 Dirk Pranke webkitpy: make PortFactory.get() be fully data-driven https://bugs.webkit.org/show_bug.cgi?id=76360 Reviewed by Eric Seidel. Refactor PortFactory.get() to be fully data-driven. * Scripts/webkitpy/layout_tests/port/factory.py: (PortFactory): (PortFactory._default_port): (PortFactory.get): (PortFactory.get.in): * Scripts/webkitpy/layout_tests/port/factory_unittest.py: (FactoryTest.assert_port): 2012-01-17 Dirk Pranke webkitpy: add determine_full_port_name(), clean up port.__init__() https://bugs.webkit.org/show_bug.cgi?id=76357 Reviewed by Adam Barth. Each Port class is now required to implement a 'port_name' field that will match the start of all ports constructed by this class and a 'determine_full_port_name' field that will make the port fully-qualified. * Scripts/webkitpy/layout_tests/port/apple.py: (ApplePort.__init__): * Scripts/webkitpy/layout_tests/port/base.py: (Port): (Port.determine_port_name): (Port.__init__): * Scripts/webkitpy/layout_tests/port/chromium.py: (ChromiumPort.for): (ChromiumPort.__init__): (ChromiumPort): (ChromiumPort._chromium_base_dir): (ChromiumPort.path_from_chromium_base): * Scripts/webkitpy/layout_tests/port/chromium_gpu.py: (ChromiumGpuLinuxPort): (ChromiumGpuLinuxPort.__init__): (ChromiumGpuMacPort): (ChromiumGpuMacPort.__init__): (ChromiumGpuWinPort): (ChromiumGpuWinPort.__init__): * Scripts/webkitpy/layout_tests/port/chromium_linux.py: (ChromiumLinuxPort): (ChromiumLinuxPort.determine_port_name): (ChromiumLinuxPort.__init__): * Scripts/webkitpy/layout_tests/port/chromium_mac.py: (ChromiumMacPort): (ChromiumMacPort.determine_port_name): (ChromiumMacPort.__init__): * Scripts/webkitpy/layout_tests/port/chromium_win.py: (ChromiumWinPort): (ChromiumWinPort.determine_port_name): (ChromiumWinPort.__init__): * Scripts/webkitpy/layout_tests/port/dryrun.py: (DryRunPort): (DryRunPort.determine_port_name): (DryRunPort.__init__): * Scripts/webkitpy/layout_tests/port/efl.py: (EflPort): * Scripts/webkitpy/layout_tests/port/factory.py: (PortFactory): (PortFactory._default_port): (PortFactory.get): (PortFactory.get.in): * Scripts/webkitpy/layout_tests/port/factory_unittest.py: (FactoryTest.assert_port): * Scripts/webkitpy/layout_tests/port/google_chrome.py: (GoogleChromeLinux32Port): (GoogleChromeLinux32Port.determine_port_name): (GoogleChromeLinux64Port): (GoogleChromeLinux64Port.determine_port_name): (GoogleChromeMacPort): (GoogleChromeMacPort.determine_port_name): (GoogleChromeWinPort): (GoogleChromeWinPort.determine_port_name): * Scripts/webkitpy/layout_tests/port/gtk.py: (GtkPort): * Scripts/webkitpy/layout_tests/port/mac.py: (MacPort): (MacPort.determine_port_name): (MacPort.__init__): * Scripts/webkitpy/layout_tests/port/mac_unittest.py: (test_versions): * Scripts/webkitpy/layout_tests/port/mock_drt.py: (MockDRTPort): (MockDRTPort.determine_port_name): (MockDRTPort.__init__): (MockDRT): (MockDRT.determine_port_name): * Scripts/webkitpy/layout_tests/port/port_testcase.py: (PortTestCase.make_port): * Scripts/webkitpy/layout_tests/port/qt.py: (QtPort): (QtPort.determine_port_name): (QtPort.__init__): * Scripts/webkitpy/layout_tests/port/test.py: (TestPort): (TestPort.determine_port_name): (TestPort.__init__): * Scripts/webkitpy/layout_tests/port/webkit.py: (WebKitPort.__init__): * Scripts/webkitpy/layout_tests/port/win.py: (WinPort): (WinPort.determine_port_name): * Scripts/webkitpy/tool/servers/rebaselineserver.py: (get_test_baselines.AllPlatformsPort.__init__): * Scripts/webkitpy/tool/servers/rebaselineserver_unittest.py: (get_test_config.TestMacPort): (get_test_config): 2012-01-17 Dirk Pranke webkitpy: clean up port code in preparation for static port names https://bugs.webkit.org/show_bug.cgi?id=76356 Reviewed by Adam Barth. To fix bug 76215 and be able to determine appropriate port names in webkitpy without actually constructing Port objects, we need to shuffle some logic in the chromium ports to be able to figure out whether we built DRT in 32 or 64-bit mode. * Scripts/webkitpy/layout_tests/port/chromium.py: (ChromiumPort): (ChromiumPort._chromium_base_dir): (ChromiumPort.__init__): (ChromiumPort.path_from_chromium_base): * Scripts/webkitpy/layout_tests/port/chromium_linux.py: (ChromiumLinuxPort): (ChromiumLinuxPort._determine_driver_path_statically): (ChromiumLinuxPort._static_build_path): (ChromiumLinuxPort._determine_architecture): (ChromiumLinuxPort.__init__): (ChromiumLinuxPort._build_path): 2012-01-17 Hao Zheng Do not remove pixel failures for ref tests. https://bugs.webkit.org/show_bug.cgi?id=76243 Reviewed by Tony Chang. 'NRWT --no-pixel-tests' complains when reftests are expected to be image mismatch. * Scripts/webkitpy/layout_tests/controllers/manager.py: (Manager._update_summary_with_result): * Scripts/webkitpy/layout_tests/models/test_failures.py: (is_reftest_failure): (determine_result_type): 2012-01-17 Balazs Ankes REGRESSION(r81225): ORWT should ignore reftests https://bugs.webkit.org/show_bug.cgi?id=67936 Reviewed by Csaba Osztrogonác. * Scripts/old-run-webkit-tests: (isUsedInReftest): $filename should be the first parameter (findTestsToRun): run test if it isn't reftest 2012-01-17 Gustavo Noronha Silva Unreviewed. Fix python unit test I broke with r105142 * Scripts/webkitpy/common/config/ports_unittest.py: (WebKitPortTest.test_gtk_port): 2012-01-17 Tor Arne Vestbø [Qt] Don't set the 'primary' flag when mocking touch-points in MiniBrowser Qt 5 no longer has that flag. Reviewed by Simon Hausmann. * MiniBrowser/qt/MiniBrowserApplication.cpp: (MiniBrowserApplication::notify): 2012-01-16 Gustavo Noronha Silva GTK+ EWS needs to run update-webkitgtk-libs after applying a patch https://bugs.webkit.org/show_bug.cgi?id=75857 Reviewed by Adam Barth. * Scripts/webkitdirs.pm: (buildAutotoolsProject): move cleaning the build to happen early, and handle --update-gtk right after that * Scripts/webkitpy/common/config/ports.py: (GtkPort.build_webkit_command): pass --update-gtk as an option to build-webkit, so that it is used in the EWS 2012-01-16 Tor Arne Vestbø [Qt] Don't add _debug postfix to targets when building debug only Reviewed by Simon Hausmann. * qmake/mkspecs/features/default_post.prf: * qmake/mkspecs/features/functions.prf: 2012-01-16 Tor Arne Vestbø [Qt] Add more files to project-files for easier navigation in Qt Creator Reviewed by Simon Hausmann. * Tools.pro: * qmake/config.tests/README: Added. 2012-01-17 Philippe Normand [GTK] fast/dom/constructed-objects-prototypes.html fails https://bugs.webkit.org/show_bug.cgi?id=75432 Reviewed by Martin Robinson. * Scripts/build-webkit: Enable video-track in GTK. 2012-01-16 Takashi Toyoshima Add toyoshim as a committer. https://bugs.webkit.org/show_bug.cgi?id=76390 Reviewed by Hajime Morita. * Scripts/webkitpy/common/config/committers.py: 2012-01-16 Brent Fulgham [WinCairo] Build fix after r104155. Several debug targets were not properly including the '*Common' property sheets. * DumpRenderTree/win/DumpRenderTreeLauncherDebugCairoCFLite.vsprops: * DumpRenderTree/win/ImageDiffLauncherDebugCairoCFLite.vsprops: * MiniBrowser/Configurations/MiniBrowserLauncherDebugCairoCFLite.vsprops: * WebKitTestRunner/win/WebKitTestRunnerLauncherDebugCairoCFLite.vsprops: * WinLauncher/WinLauncherLauncherDebugCairoCFLite.vsprops: 2012-01-16 Csaba Osztrogonác REGRESSION(r105072): It broke committers_unittest.py https://bugs.webkit.org/show_bug.cgi?id=76396 Reviewed by Adam Barth. * Scripts/webkitpy/common/config/committers_unittest.py: (CommittersTest._assert_fuzz_match): It should handle more names. (CommittersTest.test_contributors_by_fuzzy_match_with_legacy_names): Fix the testcase. 2012-01-16 Mario Sanchez Prada [GTK] Include gtk+3 in the jhbuild modules https://bugs.webkit.org/show_bug.cgi?id=76284 Reviewed by Philippe Normand. * gtk/jhbuild.modules: Add some gtk+3 to the jhbuild moduleset. 2012-01-16 Dan Winship Add myself as a committer Unreviewed * Scripts/webkitpy/common/config/committers.py: 2012-01-16 Raphael Kubo da Costa Unreviewed; fix the build when --as-needed and --no-copy-dt-needed-entries are passed to the linker. Some recent versions of a few Linux distributions have started passing --as-needed and --no-copy-dt-needed-entries by default to the linker, which broke the build when SHARED_CORE was turned on, as libwtf_efl.so needed a few symbols from libjavascriptcore_efl.so. * DumpRenderTree/efl/CMakeLists.txt: Explicitly pass WTF_LIBRARY_NAME before JavaScriptCore_LIBRARY_NAME in DumpRenderTree_LIBRARIES so the libraries are searched in the right order by the linker. 2012-01-16 Philippe Normand [GTK] set audio resources path in NRTW https://bugs.webkit.org/show_bug.cgi?id=76380 Reviewed by Martin Robinson. * Scripts/webkitpy/layout_tests/port/gtk.py: (GtkPort.setup_environ_for_server): Set $AUDIO_RESOURCES_PATH so the uninstalled audio resources are used during webaudio tests. 2012-01-16 Carlos Garcia Campos [GTK] ldflags and cflags should take precedence over the existing flags in gtkdoc.py https://bugs.webkit.org/show_bug.cgi?id=76369 Reviewed by Xan Lopez. If there's an installed version of webkit and there are new symbols in the build tree, gtkdoc-scanobj fails due to undefined references, because the env ldflags are given before the local ones. * gtk/gtkdoc.py: (GTKDoc._run_gtkdoc_scangobj): 2012-01-15 Pablo Flouret Fix compilation errors on build-webkit --debug --no-video on mac. https://bugs.webkit.org/show_bug.cgi?id=75867 Reviewed by Philippe Normand. Add --[no-]video-track to build-webkit, since video track feature depends on video. * Scripts/build-webkit: 2012-01-14 James Robinson [chromium] Failing webkit_unit_tests does not make the cr-linux EWS bubble red https://bugs.webkit.org/show_bug.cgi?id=76313 Reviewed by Adam Barth. The exit code of this script should be the exit code of webkit_unit_tests. Tested manually. * Scripts/run-chromium-webkit-unit-tests: 2012-01-14 Sheriff Bot Unreviewed, rolling out r105017. http://trac.webkit.org/changeset/105017 https://bugs.webkit.org/show_bug.cgi?id=76333 broke the build (Requested by philn-tp on #webkit). * gtk/jhbuild.modules: 2012-01-14 Mario Sanchez Prada [GTK] Include gtk+3 in the jhbuild modules https://bugs.webkit.org/show_bug.cgi?id=76284 Reviewed by Philippe Normand. * gtk/jhbuild.modules: Add some gtk+3 to the jhbuild moduleset. 2012-01-13 Adam Barth Remove --dry-run support from webkit-patch https://bugs.webkit.org/show_bug.cgi?id=76300 Reviewed by Dirk Pranke. This is the rest of the patch from Bug 76300. I landed it separtely because main.py had some stray conflict markers that I needed to clean up first. * Scripts/webkitpy/test/main.py: (Tester.parse_args): (Tester._run_tests): 2012-01-13 Adam Barth test-webkitpy does not run https://bugs.webkit.org/show_bug.cgi?id=76318 Unreviewed. Looks like some conflict markers got left in the file. * Scripts/webkitpy/test/main.py: (Tester.run): 2012-01-13 Adam Barth Remove --dry-run support from webkit-patch https://bugs.webkit.org/show_bug.cgi?id=76300 Reviewed by Dirk Pranke. We don't use this functionality for anything, and it doesn't really work. Someone in #webkit tried using it and and got sad/confused. We should just remove it. * Scripts/webkitpy/common/checkout/scm/git.py: (Git.push_local_commits_to_server): * Scripts/webkitpy/common/checkout/scm/scm.py: (SCM.__init__): * Scripts/webkitpy/common/checkout/scm/scm_unittest.py: (_shared_test_commit_with_message): (GitSVNTest.test_commit_text_parsing): * Scripts/webkitpy/common/checkout/scm/svn.py: (SVN.commit_with_message): * Scripts/webkitpy/common/net/bugzilla/bugzilla.py: (Bugzilla.__init__): (Bugzilla.quips): (Bugzilla.authenticate): (Bugzilla.add_attachment_to_bug): (Bugzilla.add_patch_to_bug): (Bugzilla.create_bug): (Bugzilla.clear_attachment_flags): (Bugzilla.set_flag_on_attachment): (Bugzilla.obsolete_attachment): (Bugzilla.add_cc_to_bug): (Bugzilla.post_comment_to_bug): (Bugzilla.close_bug_as_fixed): (Bugzilla.reassign_bug): (reopen_bug): * Scripts/webkitpy/test/main.py: (Tester.parse_args): (Tester.run_tests): 2012-01-13 David Levin check-webkit-style: should encourage the use of Own* classes for Windows DC. https://bugs.webkit.org/show_bug.cgi?id=76227 Reviewed by Dirk Pranke. * Scripts/webkitpy/style/checkers/cpp.py: (check_for_leaky_patterns): The new check. (process_line): Added a call to the new check. (CppChecker): Added the new error type. * Scripts/webkitpy/style/checkers/cpp_unittest.py: (CppStyleTestBase): (CppStyleTestBase.perform_leaky_pattern_check): The check for only leaky pattern errors. (LeakyPatternTest): Test cases. 2012-01-13 Dirk Pranke test-webkitpy: clean up handling of test directories, QueueStatusServer https://bugs.webkit.org/show_bug.cgi?id=76238 Reviewed by Adam Barth. This patch cleans up a bunch of hacks related to how we find all of the python unit tests and set up the sys.path variable in test-webkitpy. This patch changes how test-webkitpy works slightly in that we will now look for test files under Tools/Scripts and Source/WebKit2/Scripts rather than just Tools/Scripts/webkitpy and Source/WebKit2/Scripts/webkit. This greatly simplifies the mapping of paths to module names and appears to have no significant performance impact or other drawbacks. There are other ways that we could remove all of the hacks and still have a more flexible mapping of paths to module names, but the ones I've thought of so far add a decent amount of complexity to the implementation and don't seem to buy us much. * QueueStatusServer/__init__.py: * Scripts/test-webkitpy: * Scripts/webkitpy/test/main.py: (Tester._modules_from_paths): (Tester._is_module): 2012-01-13 Dirk Pranke test-webkitpy: should support classes and individual test names as well as modules https://bugs.webkit.org/show_bug.cgi?id=76233 Reviewed by Adam Barth. The default unittest.main() supports a list of individual TestCase classes or individual tests as well as modules. This change adds support for the same to test-webkitpy, so now you can say % test-wekbitpy webkitpy.test.cat_unittest.CatTest and % test-wekbitpy webkitpy.test.cat_unittest.CatTest.test_basic in addition to % test-wekbitpy webkitpy.test.cat_unittest * Scripts/test-webkitpy: * Scripts/webkitpy/test/cat_unittest.py: * Scripts/webkitpy/test/main.py: (Tester.parse_args): (Tester.run): (Tester): (Tester._find_modules): (Tester._exclude): (Tester._run_tests): (Tester._is_module): (Tester._log_exception): 2012-01-13 Dirk Pranke test-webkitpy: add basic code coverage support https://bugs.webkit.org/show_bug.cgi?id=76201 Reviewed by Adam Barth. This test adds basic code coverage support to test-webkitpy; if you pass the -c flag, we will generate a new coverage file. We can start simply with this and add more features as necessary. This patch requires coverage to be already installed on the machine. I am not auto-installing this for now. * Scripts/webkitpy/test/main.py: (Tester.parse_args): (Tester.run_tests): 2012-01-13 Dirk Pranke StatusBubbleTest.test_build_bubble is failing https://bugs.webkit.org/show_bug.cgi?id=76222 Reviewed by Adam Barth. * QueueStatusServer/handlers/statusbubble_unittest.py: (StatusBubbleTest.test_build_bubble): 2012-01-13 Kristóf Kosztyó nrwt should use multiple http shards https://bugs.webkit.org/show_bug.cgi?id=75958 Add --max-locked-shards option to the nrwt for manual set the maximum number of the locked shards Reviewed by Dirk Pranke. * Scripts/webkitpy/layout_tests/controllers/manager.py: (Manager._max_locked_shards): * Scripts/webkitpy/layout_tests/controllers/manager_unittest.py: (ShardingTests.get_shards): (ShardingTests.test_multiple_locked_shards): * Scripts/webkitpy/layout_tests/run_webkit_tests.py: (parse_args): 2012-01-13 Raphael Kubo da Costa prepare-ChangeLog: Look for a text editor in a few more variables. https://bugs.webkit.org/show_bug.cgi?id=76289 Reviewed by Ryosuke Niwa. A text editor was looked for only in the CHANGE_LOG_EDITOR and CHANGE_LOG_EDIT_APPLICATION environment variables before falling back to the OS X-specific open(1) command. As there does not seem to be a common way of looking for a text editor among the scripts in Tools/Scripts right now, just look at $VISUAL and $EDITOR (set much more often than the other variables in Unix systems) before falling back to open(1). * Scripts/prepare-ChangeLog: (openChangeLogs): 2012-01-13 Gabor Rapcsanyi [GTK] gtkdoc.py ignoring LDFLAGS and CFLAGS environment variables https://bugs.webkit.org/show_bug.cgi?id=76269 Reviewed by Martin Robinson. * gtk/gtkdoc.py: (GTKDoc._run_gtkdoc_scangobj): 2012-01-13 Csaba Osztrogonác [Qt][WK2]REGRESSION(r104881):It broke hundreds of tests https://bugs.webkit.org/show_bug.cgi?id=76247 Reviewed by Kenneth Rohde Christiansen. * WebKitTestRunner/TestController.cpp: (WTR::TestController::resetStateToConsistentValues): Disable mock scrollbars on Qt by default. (workaround) 2012-01-13 Hajime Morrita [Mac][DRT] Should support --no-timeout https://bugs.webkit.org/show_bug.cgi?id=76242 Reviewed by Darin Adler. * DumpRenderTree/mac/DumpRenderTree.mm: (initializeGlobalsFromCommandLineOptions): Added "--no-timeout" to the options list. (setWaitToDumpWatchdog): Added to encapsulate the waitToDumpWatchdog global variable. (shouldSetWaitToDumpWatchdog): Added to check the useTimeoutWatchdog flag set by the option. * DumpRenderTree/mac/DumpRenderTreeMac.h: * DumpRenderTree/mac/LayoutTestControllerMac.mm: (LayoutTestController::setWaitToDump): Now calls shouldSetWaitToDumpWatchdog() to check the flag. 2012-01-12 Simon Fraser Mitigate scrollbar differences when running pixel tests https://bugs.webkit.org/show_bug.cgi?id=67217 Reviewed by Dan Bernstein. Enable mock scrollbars by default for tests. * DumpRenderTree/mac/DumpRenderTree.mm: (resetDefaultsToConsistentValues): * WebKitTestRunner/TestController.cpp: (WTR::TestController::resetStateToConsistentValues): 2012-01-12 Ryosuke Niwa Need a script to run tests in PerformanceTests https://bugs.webkit.org/show_bug.cgi?id=76132 Reviewed by Adam Barth. Add run-perf-tests to run performance tests using parser/resources/runner.js. Unfortunately, there isn't an easy way of telling which test uses which format so hard-code directory that uses Chromium perf-bot style (only inspector perf. tests for now). All test outputs are re-formatted to match Chromium perf-bot style. * Scripts/run-inspector-perf-tests.py: * Scripts/run-perf-tests: Added. * Scripts/webkitpy/performance_tests/perftestsrunner.py: (PerfTestsRunner): (PerfTestsRunner.__init__): (PerfTestsRunner._collect_tests): (PerfTestsRunner.run): (PerfTestsRunner._print_status): (PerfTestsRunner._run_tests_set): (PerfTestsRunner._process_chromium_style_test_result): (PerfTestsRunner._should_ignore_line_in_parser_test_result): (PerfTestsRunner._process_parser_test_result): (PerfTestsRunner._run_single_test): * Scripts/webkitpy/performance_tests/perftestsrunner_unittest.py: (MainTest.TestDriver.run_test): (create_runner): (run_test): (test_run_test_set): (test_run_test_set_for_parser_tests): (test_collect_tests): 2012-01-12 Dirk Pranke webkitpy: clean up port unit tests in preparation for making port_name mandatory https://bugs.webkit.org/show_bug.cgi?id=76123 Reviewed by Adam Barth. This change updates the unit tests to be more consistent in how ports are created, in preparation for making port_name a mandatory parameter to the constructor and removing any version interpretation from it. There should be no functional changes in this patch, but some tests that are no longer relevant have been removed or reworked. * Scripts/webkitpy/layout_tests/port/base_unittest.py: (PortTest.test_get_option__default): * Scripts/webkitpy/layout_tests/port/chromium_linux_unittest.py: (ChromiumLinuxPortTest): (ChromiumLinuxPortTest.assert_architecture): (ChromiumLinuxPortTest.test_determine_architecture_fails): * Scripts/webkitpy/layout_tests/port/chromium_mac_unittest.py: (ChromiumMacPortTest): * Scripts/webkitpy/layout_tests/port/chromium_unittest.py: (ChromiumPortTest): (ChromiumPortTest.test_all_test_configurations): (ChromiumPortTest.TestMacPort.__init__): (ChromiumPortTest.TestLinuxPort.__init__): (ChromiumPortTest.TestWinPort.__init__): * Scripts/webkitpy/layout_tests/port/chromium_win_unittest.py: (ChromiumWinTest): * Scripts/webkitpy/layout_tests/port/efl_unittest.py: (EflPortTest): * Scripts/webkitpy/layout_tests/port/gtk_unittest.py: (GtkPortTest): * Scripts/webkitpy/layout_tests/port/mac_unittest.py: (MacTest): * Scripts/webkitpy/layout_tests/port/mock_drt_unittest.py: (MockDRTPortTest): (MockDRTPortTest.make_port): * Scripts/webkitpy/layout_tests/port/qt.py: (QtPort._port_flag_for_scripts): (QtPort.__init__): * Scripts/webkitpy/layout_tests/port/qt_unittest.py: (QtPortTest): (QtPortTest._assert_search_path): (QtPortTest.test_baseline_search_path): (QtPortTest.test_operating_system): * Scripts/webkitpy/layout_tests/port/webkit_unittest.py: (WebKitPortTest): * Scripts/webkitpy/layout_tests/port/win_unittest.py: (WinPortTest): (WinPortTest._assert_search_path): (WinPortTest.test_baseline_search_path): * Scripts/webkitpy/tool/servers/rebaselineserver_unittest.py: (get_test_config.TestMacPort.__init__): 2012-01-12 Dirk Pranke test-webkitpy: clean up logging and provide a real --help message https://bugs.webkit.org/show_bug.cgi?id=76142 Reviewed by Adam Barth. This test is mostly more refactoring - it calls objects in unittest directly instead of routing through unittest.main() -- in order to gain more control over the logging and to provide a correct --help message. This will also give us a better foundation for adding more feautures down the road. This patch adds different levels of logging controlled by the '-s', '-q', and '-v' options (see --help for details), and removes the --verbose-logging option. * Scripts/test-webkitpy: * Scripts/webkitpy/test/main.py: (Tester): (Tester.clean_packages): (Tester.__init__): (Tester.parse_args): (Tester.configure): (Tester._configure_logging): (Tester.find_modules): (Tester._exclude): (Tester._modules_from_paths): (Tester.run_tests): 2012-01-12 Simon Hausmann Make the new WTF module build on Qt https://bugs.webkit.org/show_bug.cgi?id=76163 Reviewed by Tor Arne Vestbø. * qmake/mkspecs/modules/wtf.prf: Pull in WTF from the new location. 2012-01-12 Vsevolod Vlasov Unreviewed, mac DumpRenderTree fix. * DumpRenderTree/mac/UIDelegate.mm: (-[UIDelegate webView:addMessageToConsole:withSource:]): 2012-01-12 Vsevolod Vlasov WebKitTestRunner should not print console message line number when it is equal to zero. https://bugs.webkit.org/show_bug.cgi?id=76167 Reviewed by Csaba Osztrogonác. * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp: (WTR::InjectedBundlePage::willAddMessageToConsole): 2012-01-11 Vsevolod Vlasov Make default console messages line numbers consistent. https://bugs.webkit.org/show_bug.cgi?id=74075 Reviewed by Pavel Feldman. Unset line numbers are not printed to console now. Added Source/WebCore/inspector to efl DumpRenderTree include path. * DumpRenderTree/chromium/WebViewHost.cpp: (WebViewHost::didAddMessageToConsole): * DumpRenderTree/efl/CMakeLists.txt: * DumpRenderTree/efl/DumpRenderTreeView.cpp: (onConsoleMessage): * DumpRenderTree/gtk/DumpRenderTree.cpp: (webViewConsoleMessage): * DumpRenderTree/mac/UIDelegate.mm: (-[UIDelegate webView:addMessageToConsole:withSource:]): * DumpRenderTree/qt/DumpRenderTreeQt.cpp: (WebCore::WebPage::javaScriptConsoleMessage): * DumpRenderTree/win/UIDelegate.cpp: (UIDelegate::webViewAddMessageToConsole): * DumpRenderTree/wx/DumpRenderTreeWx.cpp: (LayoutWebViewEventHandler::OnConsoleMessageEvent): 2012-01-10 Dirk Pranke test-webkitpy: push more logic into webkitpy.test.main, clean up code https://bugs.webkit.org/show_bug.cgi?id=76021 Reviewed by Eric Seidel. I plan to add more functionality to test-webkitpy, but it is difficult to hack on now; this patch is the first of two that brings test-webkitpy more inline with current coding style by pushing more logic into the webkitpy.test.main.Tester class (so that it will be testable itself). There should be no functional changes in this patch. * Scripts/test-webkitpy: (_path_from_webkit_root): * Scripts/webkitpy/test/main.py: (Tester): (Tester.init): (Tester.configure_logging): (Tester.configure_logging.filter): (Tester.clean_packages): (Tester.run_tests): 2012-01-11 Wei Jia implement layout tests for