cb4017562f639e670555d523787f5099afcd18ad
[WebKit-https.git] / Tools / ChangeLog
1 2011-03-02  Adam Roben  <aroben@apple.com>
2
3         Look in the new-to-SnowLeopard location for crash logs on SnowLeopard and newer
4
5         ReportCrash saves logs to ~/Library/Logs/CrashReporter on Leopard, but to
6         ~/Library/Logs/DiagnosticReports on SnowLeopard. old-run-webkit-tests was only looking in
7         the former location, but was getting lucky on some SnowLeopard machines because symlinks
8         were being created there pointing into DiagnosticReports. For machines without these
9         symlinks, crash logs were not getting captured.
10
11         Fixes <http://webkit.org/b/55607> old-run-webkit-tests doesn't capture crash logs on some
12         SnowLeopard machines
13
14         Reviewed by Eric Seidel.
15
16         * Scripts/old-run-webkit-tests:
17         (captureSavedCrashLog): Look in ~/Library/Logs/DiagnosticReports on SnowLeopard and newer.
18
19 2011-03-02  Adam Roben  <aroben@apple.com>
20
21         Don't try to capture crash logs on platforms that haven't implemented that feature
22
23         Fixes <http://webkit.org/b/55504> Use of uninitialized value in numeric lt (<) at
24         Tools/Scripts/old-run-webkit-tests line 1778 seen on Qt bots after a crash
25
26         Reviewed by Joseph Pecoraro.
27
28         * Scripts/old-run-webkit-tests:
29         (captureSavedCrashLog): Bail if we don't have a crash log glob.
30         (findNewestFileMatchingGlob): Fix a buggy test for when no paths match the glob. This isn't
31         strictly required to fix bug 55504, but seems worthwhile.
32
33 2011-01-26  John Knottenbelt  <jknotten@chromium.org>
34
35         Reviewed by Dmitry Titov.
36
37         Detach Geolocation from Frame when Page destroyed.
38         https://bugs.webkit.org/show_bug.cgi?id=52877
39
40         Extend the layout test controller to expose the number of pending
41         geolocation requests, so that we can test that the requests have
42         been cancelled on page close.
43
44         * DumpRenderTree/LayoutTestController.cpp:
45         (numberOfPendingGeolocationPermissionRequestsCallback):
46         (LayoutTestController::staticFunctions):
47         * DumpRenderTree/LayoutTestController.h:
48         * DumpRenderTree/chromium/LayoutTestController.cpp:
49         (LayoutTestController::LayoutTestController):
50         (LayoutTestController::numberOfPendingGeolocationPermissionRequests):
51         * DumpRenderTree/chromium/LayoutTestController.h:
52         * DumpRenderTree/gtk/LayoutTestControllerGtk.cpp:
53         (LayoutTestController::numberOfPendingGeolocationPermissionRequests):
54         * DumpRenderTree/mac/LayoutTestControllerMac.mm:
55         (LayoutTestController::numberOfPendingGeolocationPermissionRequests):
56         * DumpRenderTree/mac/UIDelegate.h:
57         * DumpRenderTree/mac/UIDelegate.mm:
58         (-[UIDelegate numberOfPendingGeolocationPermissionRequests]):
59         * DumpRenderTree/qt/LayoutTestControllerQt.cpp:
60         (LayoutTestController::numberOfPendingGeolocationPermissionRequests):
61         * DumpRenderTree/qt/LayoutTestControllerQt.h:
62         * DumpRenderTree/win/LayoutTestControllerWin.cpp:
63         (LayoutTestController::numberOfPendingGeolocationPermissionRequests):
64         * DumpRenderTree/wx/LayoutTestControllerWx.cpp:
65         (LayoutTestController::numberOfPendingGeolocationPermissionRequests):
66
67
68 2011-03-01  Kent Tamura  <tkent@chromium.org>
69
70         Reviewed by Dimitri Glazkov.
71
72         Assertion fails when a form validation bubble appears
73         https://bugs.webkit.org/show_bug.cgi?id=55550
74
75         Test: fast/forms/interactive-validation-attach-assertion.html
76
77         * DumpRenderTree/chromium/WebPreferences.cpp:
78         (WebPreferences::applyTo): Enable the interactive validation feature by
79           default. This matches to Chromium browser.
80
81 2011-03-01  Gyuyoung Kim  <gyuyoung.kim@samsung.com>
82
83         Reviewed by Kent Tamura.
84
85         Remove efl port style exception of Webcore/JavaScriptCore
86         https://bugs.webkit.org/show_bug.cgi?id=55549
87
88         WebCore and JavaScriptCore need to adhere WebKit coding style.
89
90         * Scripts/webkitpy/style/checker.py:
91
92 2011-03-01  Tony Chang  <tony@chromium.org>
93
94         Reviewed by Ojan Vafai.
95
96         [chromium] clean up temp files on chromium-mac builder too
97         https://bugs.webkit.org/show_bug.cgi?id=55525
98
99         * BuildSlaveSupport/chromium/remove-crash-logs:
100
101 2011-02-28  Ojan Vafai  <ojan@chromium.org>
102
103         Reviewed by Tony Chang.
104
105         wrap json in a function call to afford cross-domain loading
106         https://bugs.webkit.org/show_bug.cgi?id=55353
107
108         Also delete the code that compacts the JSON. It turns out this
109         breaks the rebaseline tool and is probably a premature optimization anyways.
110
111         * Scripts/webkitpy/layout_tests/layout_package/json_results_generator.py:
112         * Scripts/webkitpy/layout_tests/layout_package/test_runner.py:
113         * Scripts/webkitpy/layout_tests/run_webkit_tests.py:
114         * Scripts/webkitpy/tool/commands/rebaselineserver.py:
115
116 2011-03-01  Sheriff Bot  <webkit.review.bot@gmail.com>
117
118         Unreviewed, rolling out r80079.
119         http://trac.webkit.org/changeset/80079
120         https://bugs.webkit.org/show_bug.cgi?id=55547
121
122         "Broke the Win debug build?" (Requested by dcheng on #webkit).
123
124         * Scripts/build-webkit:
125
126 2011-03-01  Mark Rowe  <mrowe@apple.com>
127
128         Reviewed by Sam Weinig.
129
130         Update verification scripts to handle WebKit2.
131
132         * Scripts/check-for-global-initializers: Skip object files that use RefCountedLeakChecker.
133         * Scripts/check-for-inappropriate-files-in-framework: Add some inappropriate file types for
134         WebKit2.
135
136 2011-03-01  Daniel Cheng  <dcheng@chromium.org>
137
138         Reviewed by David Levin.
139
140         Add feature define for data transfer items
141         https://bugs.webkit.org/show_bug.cgi?id=55510
142
143         * Scripts/build-webkit:
144
145 2011-02-23  Joseph Pecoraro  <joepeck@webkit.org>
146
147         Reviewed by Timothy Hatcher.
148
149         All Console Messages should be passed to ChromeClients.
150         https://bugs.webkit.org/show_bug.cgi?id=54926
151
152         Update DumpRenderTree's UIDelegate to use the new console message delegate.
153
154         * DumpRenderTree/mac/UIDelegate.mm:
155         (-[UIDelegate webView:addMessageToConsole:withSource:]):
156
157 2011-03-01  Dirk Pranke  <dpranke@chromium.org>
158
159         Reviewed by Adam Barth.
160
161         Tragically, we stopped running the python unit tests
162         on Chromium Win just as I was getting them to be green.
163         This patch turns them back on.
164
165         https://bugs.webkit.org/show_bug.cgi?id=55521
166
167         * BuildSlaveSupport/build.webkit.org-config/master.cfg:
168
169 2011-03-01  Dirk Pranke  <dpranke@chromium.org>
170
171         Reviewed by Adam Roben.
172
173         Disable
174         webkitpy.common.system.executive_unittest.ExecutiveTest.\
175            test_kill_all fails on win until it can be properly triaged.
176
177         https://bugs.webkit.org/show_bug.cgi?id=54790
178
179         * Scripts/webkitpy/common/system/executive_unittest.py:
180
181     2011-03-01  Helder Correia  <helder@sencha.com>
182
183             Unreviewed. Add myself to committers.py.
184
185             * Scripts/webkitpy/common/config/committers.py:
186
187 2011-03-01  Adam Roben  <aroben@apple.com>
188
189         Give up if a crash log for the web process is taking too long to be saved on Windows
190
191         This should keep the bots from getting stuck while waiting for a crash log, if writing a
192         crash log ever hangs or the UI process for some reason doesn't notice it's finished.
193
194         Fixes <http://webkit.org/b/55499> WebKitTestRunner can hang forever waiting for a crash log
195         to be saved for the web process
196
197         Reviewed by Steve Falkenburg.
198
199         * WebKitTestRunner/win/TestControllerWin.cpp:
200         (WTR::runRunLoopUntil): Moved code to run the run loop here here from platformRunUntil.
201         Generalized the code slightly to handle the optional object to wait on.
202         (WTR::TestController::platformRunUntil): Use the new runRunLoopUntil function, and also use
203         it (with a timeout) when waiting for a crash log for the web process to be saved. This will
204         prevent us from waiting forever.
205
206 2011-03-01  Adam Roben  <aroben@apple.com>
207
208         Give WebKitLauncherWin's PDB file a name different from WebKit's
209
210         WebKitLauncherWin (which builds WebKit.exe) and WebKit (which builds WebKit.dll) were
211         fighting over the same PDB file, meaning that one of them was going to lose and not have
212         symbols. WebKitLauncherWin's PDB file is now named WebKitLauncherWin.pdb.
213
214         Fixes <http://webkit.org/b/55495> Crashes in WebKit.dll have bad backtraces for some test
215         runs (due to missing symbols)
216
217         Reviewed by David Kilzer.
218
219         * WebKitLauncherWin/WebKitLauncherWinCommon.vsprops: Use the project name,
220         "WebKitLauncherWin", instead of the target name, "WebKit", when naming the PDB file.
221
222 2011-03-01  Adam Roben  <aroben@apple.com>
223
224         Count web process crashes toward the --exit-after-n-crashes-or-timeouts limit
225
226         Fixes <http://webkit.org/b/55492> Web process crashes aren't counted toward the
227         --exit-after-n-crashes-or-timeouts limit
228
229         Reviewed by David Kilzer.
230
231         * Scripts/old-run-webkit-tests:
232         (stopRunningTestsEarlyIfNeeded): Include web process crashes in the computation, and print
233         the number of tests that caused the web process to crash when exiting early.
234
235 2011-03-01  Adam Roben  <aroben@apple.com>
236
237         Save a crash log when the web process crashes
238
239         On Windows, WebKitTestRunner now detects when the web process is crashing and waits to exit
240         until it has finished crashing, which guarantees that the crash log will have had time to be
241         saved, too. On Mac, we always wait until ReportCrash has exited before capturing the crash
242         log, so all we have to do is choose the right crash log out of the CrashReporter directory.
243
244         Fixes <http://webkit.org/b/44121> <rdar://problem/8320759> When the web process crashes and
245         a crash log is being saved, WebKitTestRunner thinks the web process has become unresponsive
246
247         Reviewed by Sam Weinig.
248
249         * Scripts/old-run-webkit-tests:
250         (testCrashedOrTimedOut): Don't kill WebKitTestRunner when the web process crashes. It will
251         kill itself. On Windows, this will cause us to wait until the crash log has been saved. On
252         Mac, it should have no effect. Capture saved crash logs for web process crashes, too.
253         (captureSavedCrashLog): Added $webProcessCrashed argument. On Mac, look for
254         WebProces_*.crash files when the web process crashes.
255
256         * WebKitTestRunner/InjectedBundle/InjectedBundle.cpp:
257         (WTR::InjectedBundle::initialize): Added an initializationUserData argument. Updated for
258         initializePlatformDefaults -> platformInitialize rename. Pass the initializationUserData
259         along to platformInitialize.
260
261         * WebKitTestRunner/InjectedBundle/InjectedBundle.h: See above.
262
263         * WebKitTestRunner/InjectedBundle/InjectedBundleMain.cpp:
264         (WKBundleInitialize): Pass along the initializationUserData to the InjectedBundle.
265
266         * WebKitTestRunner/InjectedBundle/mac/InjectedBundleMac.mm:
267         (WTR::InjectedBundle::platformInitialize):
268         * WebKitTestRunner/InjectedBundle/qt/InjectedBundleQt.cpp:
269         (WTR::InjectedBundle::platformInitialize):
270         Updated function signature.
271
272         * WebKitTestRunner/InjectedBundle/win/InjectedBundleWin.cpp:
273         (WTR::exceptionFilter): Added. Tells the UI process we're crashing by signaling the
274         webProcessCrashingEvent, then lets the crash continue as normal.
275
276         (WTR::InjectedBundle::platformInitialize): Hook up exceptionFilter. Retrieve the name of the
277         event we should use to tell the UI process we're crashing from the initializationUserData,
278         and get a handle to that event.
279
280         * WebKitTestRunner/TestController.cpp:
281         (WTR::TestController::TestController): Initialize new members.
282         (WTR::TestController::processDidCrash): Removed unnecessary WKPageRef argument. Changed to
283         only print the "#CRASHED - WebProcess" message once, since this can be called more than once
284         when a crash log is being saved on Windows. Exit right away if specified. (This is the
285         default.)
286
287         * WebKitTestRunner/TestController.h: Added new members.
288
289         * WebKitTestRunner/win/TestControllerWin.cpp:
290         (WTR::TestController::platformInitialize): Set up the event the web process will use to tell
291         us it's crashing.
292         (WTR::TestController::platformRunUntil): Pass MWMO_INPUTAVAILABLE to
293         ::MsgWaitForMultipleObjectsEx so we'll process messages that have already been seen by
294         ::PeekMessage. (This is unrelated to the bug fix.) Notice when the webProcessCrashingEvent
295         has been signaled. When this happens, print the "#CRASHED - WebProcess" message right away
296         so the test harness will know the web process has crashed and not try to kill us, then wait
297         for the web process to finish crashing so a crash log will have time to be saved.
298         (WTR::toWK): Simple hepler function.
299         (WTR::TestController::platformInitializeContext): Pass along the name of the event the web
300         process should use to tell us it is crashing in the context's initialization user data.
301
302 2011-03-01  Dimitri Glazkov  <dglazkov@chromium.org>
303
304         Reviewed by Tony Gentilcore.
305
306         Move the checks to exclude Python/Perl tests for Chromium Win build to the right place.
307         https://bugs.webkit.org/show_bug.cgi?id=55476
308
309         Chromium Win uses BuildAndTestFactory, not TestFactory.
310
311         * BuildSlaveSupport/build.webkit.org-config/master.cfg: Moved to the right place.
312
313 2011-03-01  Adam Roben  <aroben@apple.com>
314
315         Test that the WebKit2 responsiveness timer doesn't fire too early
316
317         Test for <http://webkit.org/b/55417> <rdar://problem/9065287> RunLoop::Timer fires 1000x too
318         early on Windows
319
320         Reviewed by Anders Carlsson.
321
322         * TestWebKitAPI/PlatformUtilities.h: Added sleep().
323
324         * TestWebKitAPI/Tests/WebKit2/ResponsivenessTimerDoesntFireEarly.cpp: Added.
325         (TestWebKitAPI::didReceiveMessageFromInjectedBundle):
326         (TestWebKitAPI::didFinishLoadForFrame):
327         (TestWebKitAPI::processDidBecomeUnresponsive):
328         (TestWebKitAPI::setInjectedBundleClient):
329         (TestWebKitAPI::setPageLoaderClient):
330         Simple helper functions.
331
332         (TestWebKitAPI::TEST): Load an HTML file to make sure the web process is initialized. Then
333         tell the web process to pause and press the spacebar key. The spacebar keypress should cause
334         the responsiveness timer to start, but the web process should unpause before it has a chance
335         to fire. Run until the web process has unpaused, and assert that the timer didn't fire.
336
337         * TestWebKitAPI/Tests/WebKit2/ResponsivenessTimerDoesntFireEarly_Bundle.cpp: Added.
338         (TestWebKitAPI::ResponsivenessTimerDoesntFireEarlyTest::ResponsivenessTimerDoesntFireEarlyTest):
339         Call up to the base class.
340         (TestWebKitAPI::ResponsivenessTimerDoesntFireEarlyTest::didReceiveMessage): When asked to
341         pause, sleep for 0.5 seconds, then send back a message saying we paused.
342
343         * TestWebKitAPI/mac/PlatformUtilitiesMac.mm:
344         (TestWebKitAPI::Util::sleep): Added. Calls through to usleep.
345         * TestWebKitAPI/win/PlatformUtilitiesWin.cpp:
346         (TestWebKitAPI::Util::sleep): Added. Calls through to ::Sleep.
347
348         * TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj:
349         * TestWebKitAPI/win/TestWebKitAPI.vcproj:
350         * TestWebKitAPI/win/TestWebKitAPIInjectedBundle.vcproj:
351         Added new files.
352
353 2011-03-01  Andras Becsi  <abecsi@webkit.org>
354
355         Reviewed by Csaba Osztrogonác.
356
357         [Qt] Clean up the project files and move common options to WebKit.pri.
358
359         * WebKitTestRunner/qt/WebKitTestRunner.pro: Deduplicate.
360
361
362 2011-03-01  Adam Roben  <aroben@apple.com>
363
364         Skip another multiprocessing test on Windows
365
366         It is sometimes failing (and may be leaving child Python processes around).
367         <http://webkit.org/b/55087> tracks the failure.
368
369         * Scripts/webkitpy/layout_tests/layout_package/manager_worker_broker_unittest.py:
370         (FuncitonTests.test_get__processes): Skip this test on Windows.
371
372 2011-03-01  Ilya Tikhonovsky  <loislo@chromium.org>
373
374         Reviewed by Pavel Feldman.
375
376         Web Inspector: reduce inspector tests flakiness.
377         https://bugs.webkit.org/show_bug.cgi?id=55401
378
379         * DumpRenderTree/chromium/TestShell.cpp:
380         (TestShell::runFileTest):
381         * DumpRenderTree/chromium/WebPreferences.cpp:
382         (WebPreferences::reset):
383
384 2011-02-28  Mihai Parparita  <mihaip@chromium.org>
385
386         Reviewed by James Robinson.
387
388         NRWT should still upload results when it exits early due to too many failures
389         https://bugs.webkit.org/show_bug.cgi?id=55432
390         
391         Only keyboard interrupts should prevent results from being uploaded.
392
393         * Scripts/webkitpy/layout_tests/layout_package/test_runner.py:
394         * Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
395
396 2011-02-28  Alice Liu  <alice.liu@apple.com>
397
398         Reviewed by Adam Roben and Gavin Barraclough.
399
400         Test for https://bugs.webkit.org/show_bug.cgi?id=54898
401         InjectedBundleNodeHandle dies too early in WKBundleHitTestResultGetNodeHandle.
402
403         * TestWebKitAPI/InjectedBundleController.h:
404         (TestWebKitAPI::InjectedBundleController::bundle):
405         * TestWebKitAPI/PlatformWebView.h:
406         * TestWebKitAPI/Tests/WebKit2/HitTestResultNodeHandle.cpp: Added.
407         (TestWebKitAPI::didFinishLoadForFrame):
408         (TestWebKitAPI::didReceiveMessageFromInjectedBundle):
409         (TestWebKitAPI::setPageLoaderClient):
410         (TestWebKitAPI::TEST):
411         * TestWebKitAPI/Tests/WebKit2/HitTestResultNodeHandle_Bundle.cpp: Added.
412         (TestWebKitAPI::HitTestResultNodeHandleTest::HitTestResultNodeHandleTest):
413         (TestWebKitAPI::HitTestResultNodeHandleTest::getContextMenuFromDefaultMenu):
414         (TestWebKitAPI::HitTestResultNodeHandleTest::didCreatePage):
415         * TestWebKitAPI/mac/PlatformWebViewMac.mm:
416         (TestWebKitAPI::PlatformWebView::simulateRightClick):
417         * TestWebKitAPI/win/PlatformWebViewWin.cpp:
418         (TestWebKitAPI::PlatformWebView::simulateRightClick):
419         
420         Added files to:
421         * TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj:
422         * TestWebKitAPI/win/TestWebKitAPI.vcproj:
423         * TestWebKitAPI/win/TestWebKitAPIInjectedBundle.vcproj:
424
425 2011-02-28  Adam Roben  <aroben@apple.com>
426
427         Start running InjectedBundle API tests on Windows
428
429         A new project, TestWebKitAPIInjectedBundle, builds the injected bundle DLL.
430
431         Fixes <http://webkit.org/b/55420> run-api-tests doesn't run injected bundle tests on Windows
432
433         Reviewed by Darin Adler.
434
435         * TestWebKitAPI/Configurations/TestWebKitAPIDebug.vsprops: Added.
436         * TestWebKitAPI/Configurations/TestWebKitAPIDebugAll.vsprops: Added.
437         * TestWebKitAPI/Configurations/TestWebKitAPIDebugCairoCFLite.vsprops: Added.
438         * TestWebKitAPI/Configurations/TestWebKitAPIRelease.vsprops: Added.
439         * TestWebKitAPI/Configurations/TestWebKitAPIReleaseCairoCFLite.vsprops: Added.
440         These are just extracted from TestWebKitAPI.vcproj.
441
442         * TestWebKitAPI/Configurations/TestWebKitAPIInjectedBundleCommon.vsprops: Added. Just sets
443         up the DLL's filename.
444
445         * TestWebKitAPI/win/PlatformUtilitiesWin.cpp: Fixed DLL name.
446         (TestWebKitAPI::Util::createInjectedBundlePath): Changed to use kCFAllocatorNull instead of
447         0 (== kCFAllocatorDefault) so that CF won't try to deallocate our C string.
448
449         * TestWebKitAPI/win/TestWebKitAPI.sln: Added TestWebKitAPIInjectedBundle. Removed the bogus
450         "all" configuration.
451
452         * TestWebKitAPI/win/TestWebKitAPI.vcproj: Changed to use the new .vsprops files. Added
453         injected bundle tests. Let VS resort things.
454
455         * TestWebKitAPI/win/TestWebKitAPIInjectedBundle.vcproj: Copied from Tools/TestWebKitAPI/win/TestWebKitAPI.vcproj.
456
457 2011-02-28  Adam Roben  <aroben@apple.com>
458
459         Add a test for the paint rect passed via WM_PAINT to windowless plugins
460
461         Test for <http://webkit.org/b/55365> <rdar://problem/9031089> REGRESSION (r79040): WebKit2:
462         Large portions of pages with plugins paint black
463
464         Reviewed by Anders Carlsson.
465
466         * DumpRenderTree/TestNetscapePlugIn/PluginTest.cpp:
467         (PluginTest::NPP_HandleEvent):
468         * DumpRenderTree/TestNetscapePlugIn/PluginTest.h:
469         Added NPP_HandleEvent. Just returns 0 at this level.
470
471         * DumpRenderTree/TestNetscapePlugIn/Tests/win/WindowlessPaintRectCoordinates.cpp: Added.
472         (WindowlessPaintRectCoordinates::WindowlessPaintRectCoordinates): Call up to the base class.
473         (WindowlessPaintRectCoordinates::NPP_New): Mark ourselves as windowless.
474         (WindowlessPaintRectCoordinates::NPP_HandleEvent): Check that the paint rect passed via
475         WM_PAINT has the expected coordinates.
476
477         * DumpRenderTree/TestNetscapePlugIn/main.cpp:
478         (NPP_HandleEvent): Give the PluginTest a chance to handle the event before doing anything
479         else.
480
481         * DumpRenderTree/TestNetscapePlugIn/win/TestNetscapePlugin.vcproj: Added new file, let VS
482         resort things.
483
484         * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp:
485         (WTR::InjectedBundlePage::didFinishLoadForFrame): Added a call to WKBundlePageForceRepaint.
486         This is roughly equivalent to the call to -[WebView displayIfNeeded] in -[FrameLoadDelegate
487         webView:didFinishLoadForFrame:] in DumpRenderTree.
488
489         * WebKitTestRunner/win/PlatformWebViewWin.cpp:
490         (WTR::PlatformWebView::PlatformWebView): Tell the WKView it's in a window so that plugins
491         can start running.
492
493 2011-02-28  Adam Roben  <aroben@apple.com>
494
495         Tell MiniBrowser's WKViews that they're in a window
496
497         Fixes <http://webkit.org/b/55364> Plugins don't work in MiniBrowser on Windows
498
499         Reviewed by Anders Carlsson.
500
501         * MiniBrowser/win/BrowserView.cpp:
502         (BrowserView::create): Call WKViewSetIsInWindow after creating the view.
503
504 2011-02-28  Alexis Menard  <alexis.menard@openbossa.org>
505
506         Reviewed by Csaba Osztrogonác.
507
508         The new QML tests require QML_IMPORT_PATH variable to be set in order to
509         load the proper plugin.
510
511         * Scripts/run-qtwebkit-tests:
512
513 2011-02-28  Joone Hur  <joone.hur@collabora.co.uk>
514
515         Unreviewed, fix spelling mistake.
516
517         hardware-accelated => hardware-accelerated.
518
519         * Scripts/webkitpy/layout_tests/run_webkit_tests.py:
520
521 2011-02-28  Kristian Amlie  <kristian.amlie@nokia.com>
522
523         Reviewed by Andreas Kling.
524
525         Made sure that the build-webkit --qmake argument is always respected.
526
527         Previously, it would be respected during the build, but not during
528         the early feature detection. The build would also fail if qmake was
529         not in your path, even if you specified --qmake with a valid qmake
530         binary.
531
532         [Qt] WebKit patches required to work with a modularized version of Qt
533         https://bugs.webkit.org/show_bug.cgi?id=53916
534
535         * Scripts/build-webkit:
536         * Scripts/webkitdirs.pm:
537
538 2011-02-27  Ojan Vafai  <ojan@chromium.org>
539
540         Fix failing tests from http://trac.webkit.org/changeset/79837.
541
542         * Scripts/webkitpy/layout_tests/layout_package/printing.py:
543         This was just a mistake that would break if someone tried printing with this configuration.
544         * Scripts/webkitpy/layout_tests/layout_package/printing_unittest.py:
545         Method signature changed. Didn't notice that this method was being calling in a unittest.
546
547 2011-02-24  Hayato Ito  <hayato@chromium.org>
548
549         Reviewed by Eric Seidel.
550
551         [NRWT] Remove test_types/* classes.
552
553         This is the last patch in the series of efforts which tried to remove test_types classes.
554         A single_test_runner module will take over responsibilities of test_type classes.
555
556         From this patch, we won't display statistics per test_type even if a '--verbose' option is given.
557
558         https://bugs.webkit.org/show_bug.cgi?id=55123
559
560         * Scripts/webkitpy/layout_tests/layout_package/single_test_runner.py:
561         * Scripts/webkitpy/layout_tests/layout_package/test_results.py:
562         * Scripts/webkitpy/layout_tests/layout_package/test_results_unittest.py:
563         * Scripts/webkitpy/layout_tests/layout_package/test_runner.py:
564         * Scripts/webkitpy/layout_tests/layout_package/worker_mixin.py:
565         * Scripts/webkitpy/layout_tests/test_types/__init__.py: Removed.
566         * Scripts/webkitpy/layout_tests/test_types/image_diff.py: Removed.
567         * Scripts/webkitpy/layout_tests/test_types/test_type_base.py: Removed.
568         * Scripts/webkitpy/layout_tests/test_types/test_type_base_unittest.py: Removed.
569         * Scripts/webkitpy/layout_tests/test_types/text_diff.py: Removed.
570
571 2011-02-25  Ojan Vafai  <ojan@chromium.org>
572
573         Reviewed by Tony Chang.
574
575         Change results.json format to the one used by unexpected_results.json
576         https://bugs.webkit.org/show_bug.cgi?id=52267
577
578         Also add runtimes in milliseconds to the JSON and make the output format more compact.
579         Named the file full_results.json to avoid conflicting with the results.json
580         file the test-results server currently serves up.
581
582         * Scripts/webkitpy/layout_tests/layout_package/json_layout_results_generator.py:
583         * Scripts/webkitpy/layout_tests/layout_package/json_results_generator.py:
584         * Scripts/webkitpy/layout_tests/layout_package/json_results_generator_unittest.py:
585         * Scripts/webkitpy/layout_tests/layout_package/printing.py:
586         * Scripts/webkitpy/layout_tests/layout_package/result_summary.py:
587         * Scripts/webkitpy/layout_tests/layout_package/test_runner.py:
588
589 2011-02-27  Adam Roben  <aroben@apple.com>
590
591         Followup fix to r79827
592
593         * DumpRenderTree/win/UIDelegate.cpp:
594         (UIDelegate::webViewAddMessageToConsole): Replaced call to urlSuitableForTestResult with
595         lastPathComponent. This matches what we did here before r79827, and matches Mac.
596
597 2011-02-27  Dan Bernstein  <mitz@apple.com>
598
599         Reviewed by Adam Roben.
600
601         Make bisect-builds print a trac link for the regression range
602         https://bugs.webkit.org/show_bug.cgi?id=55331
603
604         * Scripts/bisect-builds:
605
606 2011-02-27  Adam Roben  <aroben@apple.com>
607
608         Use iswalpha instead of isalpha when dealing with wchar_ts in EditingDelegate
609
610         Fixes <http://webkit.org/b/55062> <rdar://problem/9059907> Crash beneath
611         EditingDelegate::checkSpellingOfString when running fast/forms/input-text-maxlength.html or
612         fast/forms/input-text-paste-maxlength.html on Windows with full page heap enabled
613
614         Reviewed by Anders Carlsson.
615
616         * DumpRenderTree/win/EditingDelegate.cpp:
617         (indexOfFirstWordCharacter):
618         (wordLength):
619         Use iswalpha instead of isalpha, since TCHAR is really just wchar_t.
620
621 2011-02-27  Adam Roben  <aroben@apple.com>
622
623         Make ResourceLoadDelegate print URLs relative to the main resource on Windows
624
625         This matches Mac. (In fact, the code was ported from Mac!)
626
627         Fixes <http://webkit.org/b/55328> URLs printed by ResourceLoadDelegate on Windows don't
628         match Mac
629
630         Reviewed by Anders Carlsson.
631
632         * DumpRenderTree/win/DumpRenderTree.cpp:
633         (substringFromIndex): Added. Emulates -[NSString substringFromIndex:].
634         (urlSuitableForTestResult): Ported code from -[NSURL(DRTExtras)
635         _drt_descriptionSuitableForTestResult].
636         (cfStringRefToWString): Moved here from LayoutTestControllerWin.cpp.
637
638         * DumpRenderTree/win/DumpRenderTreeWin.h: Added declaration of cfStringRefToWString.
639
640         * DumpRenderTree/win/LayoutTestControllerWin.cpp: Moved cfStringRefToWString to
641         DumpRenderTree.cpp.
642
643 2011-02-27  Adam Roben  <aroben@apple.com>
644
645         Change the order of Chang Shu's email addresses so the first one is their Bugzilla username
646
647         This makes Bugzilla autocompletion work correctly.
648
649         * Scripts/webkitpy/common/config/committers.py:
650
651 2011-02-27  Patrick Gansterer  <paroga@webkit.org>
652
653         Reviewed by Andreas Kling.
654
655         [Qt] fast/loader/user-stylesheet-fast-path.html fails
656         https://bugs.webkit.org/show_bug.cgi?id=50911
657
658         * DumpRenderTree/qt/LayoutTestControllerQt.cpp:
659         (LayoutTestController::setUserStyleSheetLocation):
660         Use QUrl::fromEncoded since the passed string is already encoded.
661
662 2011-02-27  Adam Roben  <aroben@apple.com>
663
664         Ensure $testResultsDirectory is an absolute path before setting up Windows crash log saving
665
666         Windows 7 (and maybe Vista?) launches the post-mortem debugger with a working directory that
667         is different from old-run-webkit-test's. Because of this, the path to the crash log file
668         (which is relative to $testResultsDirectory) must be an absolute path. See the bug for more
669         details.
670
671         Fixes <http://webkit.org/b/55318> Crash logs aren't saved on Windows 7 when a relative
672         --results-directory path is used (like on the test slaves)
673
674         Reviewed by Dan Bernstein.
675
676         * Scripts/old-run-webkit-tests: Moved the call to setUpWindowsCrashLogSaving after we've
677         made $testResultsDirectory an absolute path.
678
679 2011-02-26  Dominic Mazzoni  <dmazzoni@google.com>
680
681         Reviewed by Adam Barth.
682
683         m_dumpAccessibilityNotifications should be initialized.
684         https://bugs.webkit.org/show_bug.cgi?id=54874
685
686         * DumpRenderTree/chromium/AccessibilityController.cpp:
687         (AccessibilityController::AccessibilityController):
688
689 2011-02-26  Vsevolod Vlasov  <vsevik@chromium.org>
690
691         Reviewed by Pavel Feldman.
692
693         DumpRenderTree should reset frame opener between tests.
694         https://bugs.webkit.org/show_bug.cgi?id=54874
695
696         Added clearOpener() calls between tests.
697         Some tests are using window.open() to change current window location.
698         This changes window.opener property whose nullity is checked before
699         showing xml viewer.
700
701         * DumpRenderTree/chromium/TestShell.cpp:
702         (TestShell::resetTestController):
703         * DumpRenderTree/gtk/DumpRenderTree.cpp:
704         (resetDefaultsToConsistentValues):
705         * DumpRenderTree/mac/DumpRenderTree.mm:
706         (resetWebViewToConsistentStateBeforeTesting):
707         * DumpRenderTree/qt/DumpRenderTreeQt.cpp:
708         (WebCore::DumpRenderTree::resetToConsistentStateBeforeTesting):
709         * DumpRenderTree/win/DumpRenderTree.cpp:
710         (resetWebViewToConsistentStateBeforeTesting):
711         * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp:
712         (WTR::InjectedBundlePage::reset):
713
714 2011-02-26  Yael Aharon  <yael.aharon@nokia.com>
715
716         Reviewed by Andreas Kling.
717
718         [Qt] Fix the focus in MiniBrowser.
719         https://bugs.webkit.org/show_bug.cgi?id=55288
720
721         Give focus to the web view when a load starts.
722
723         * MiniBrowser/qt/BrowserWindow.cpp:
724         (BrowserWindow::urlChanged):
725
726 2011-02-26  Gyuyoung Kim  <gyuyoung.kim@samsung.com>
727
728         Reviewed by Adam Barth.
729
730         [EFL] Add coding style exceptions for EFL port
731         https://bugs.webkit.org/show_bug.cgi?id=54733
732
733         The EFL APIs use EFL naming style, which includes both lower-cased and camel-cased,
734         underscore-sparated values. But, style checking rule have found parameter_name errors
735         in efl directories so far.
736
737         * Scripts/webkitpy/style/checker.py: Add prameter_name rule's exception and efl directories.
738
739 2011-02-26  David Levin  <levin@chromium.org>
740
741         Reviewed by Darin Adler.
742
743         check-webkit-style should check for spaces after periods in comments
744         https://bugs.webkit.org/show_bug.cgi?id=55253
745
746         * Scripts/webkitpy/style/checkers/cpp.py: Added the check.
747         * Scripts/webkitpy/style/checkers/cpp_unittest.py: Added tests
748           and fixed the name of one related test.
749
750 2011-02-26  Kenneth Rohde Christiansen  <kenneth@webkit.org>
751
752         Reviewed by Andreas Kling.
753
754         Make it possible to test the targetdensity-dpi support
755         https://bugs.webkit.org/show_bug.cgi?id=55142
756
757         Test the viewport meta tag feature targetdensity-dpi by
758         adding extra arguments to dumpConfigurationForViewport
759
760         Add the new arguments to the shared header and all
761         implementation.
762
763         * DumpRenderTree/LayoutTestController.cpp:
764         (dumpConfigurationForViewportCallback):
765         * DumpRenderTree/LayoutTestController.h:
766         * DumpRenderTree/gtk/LayoutTestControllerGtk.cpp:
767         (LayoutTestController::dumpConfigurationForViewport):
768         * DumpRenderTree/mac/LayoutTestControllerMac.mm:
769         (LayoutTestController::dumpConfigurationForViewport):
770         * DumpRenderTree/win/LayoutTestControllerWin.cpp:
771         (LayoutTestController::dumpConfigurationForViewport):
772         * DumpRenderTree/wx/LayoutTestControllerWx.cpp:
773         (LayoutTestController::dumpConfigurationForViewport):
774         * DumpRenderTree/qt/LayoutTestControllerQt.cpp:
775         (LayoutTestController::dumpConfigurationForViewport):
776         * DumpRenderTree/qt/LayoutTestControllerQt.h:
777
778 2011-02-26  Sheriff Bot  <webkit.review.bot@gmail.com>
779
780         Unreviewed, rolling out r79764.
781         http://trac.webkit.org/changeset/79764
782         https://bugs.webkit.org/show_bug.cgi?id=55295
783
784         "broke Chromium builds" (Requested by rniwa on #webkit).
785
786         * DumpRenderTree/chromium/TestShell.cpp:
787         (TestShell::resetTestController):
788         * DumpRenderTree/gtk/DumpRenderTree.cpp:
789         (resetDefaultsToConsistentValues):
790         * DumpRenderTree/mac/DumpRenderTree.mm:
791         (resetWebViewToConsistentStateBeforeTesting):
792         * DumpRenderTree/qt/DumpRenderTreeQt.cpp:
793         (WebCore::DumpRenderTree::resetToConsistentStateBeforeTesting):
794         * DumpRenderTree/win/DumpRenderTree.cpp:
795         (resetWebViewToConsistentStateBeforeTesting):
796         * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp:
797         (WTR::InjectedBundlePage::reset):
798
799 2011-02-26  Yongjun Zhang  <yongjun_zhang@apple.com>
800
801         Reviewed by David Kilzer.
802
803         https://bugs.webkit.org/show_bug.cgi?id=48781
804
805         Add a resource load delegate method to query if WebCore should paint the default broken image for failed images.
806
807         Add a new resource load client method (shouldPaintBrokenImage).  WebKit client can decide if WebCore
808         should paint the default broken image when an image fails to load or decode.
809
810         * DumpRenderTree/LayoutTestController.cpp:
811         (LayoutTestController::LayoutTestController):
812         (setShouldPaintBrokenImageCallback):
813         (LayoutTestController::staticFunctions):
814         (LayoutTestController::setShouldPaintBrokenImage):
815         * DumpRenderTree/LayoutTestController.h:
816         (LayoutTestController::shouldPaintBrokenImage):
817         * DumpRenderTree/mac/ResourceLoadDelegate.mm:
818         (-[ResourceLoadDelegate webView:shouldPaintBrokenImageForURL:]):
819
820 2011-02-26  Vsevolod Vlasov  <vsevik@chromium.org>
821
822         Reviewed by Pavel Feldman.
823
824         DumpRenderTree should reset frame opener between tests.
825         https://bugs.webkit.org/show_bug.cgi?id=54874
826
827         Added clearOpener() calls between tests.
828         Some tests are using window.open() to change current window location.
829         This changes window.opener property whose nullity is checked before
830         showing xml viewer.
831
832         * DumpRenderTree/chromium/TestShell.cpp:
833         (TestShell::resetTestController):
834         * DumpRenderTree/gtk/DumpRenderTree.cpp:
835         (resetDefaultsToConsistentValues):
836         * DumpRenderTree/mac/DumpRenderTree.mm:
837         (resetWebViewToConsistentStateBeforeTesting):
838         * DumpRenderTree/qt/DumpRenderTreeQt.cpp:
839         (WebCore::DumpRenderTree::resetToConsistentStateBeforeTesting):
840         * DumpRenderTree/win/DumpRenderTree.cpp:
841         (resetWebViewToConsistentStateBeforeTesting):
842         * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp:
843         (WTR::InjectedBundlePage::reset):
844
845 2011-02-25  Lukasz Slachciak  <l.slachciak@samsung.com>
846
847         Reviewed by Kenneth Rohde Christiansen.
848
849         [EFL] Added EWebLauncher command line option to turn on/off frame flattening
850         https://bugs.webkit.org/show_bug.cgi?id=54809
851
852         * EWebLauncher/main.c: Handling -f option added.
853         (on_key_down):
854         (browserCreate):
855         (main):
856
857 2011-02-25  Sheriff Bot  <webkit.review.bot@gmail.com>
858
859         Unreviewed, rolling out r79324.
860         http://trac.webkit.org/changeset/79324
861         https://bugs.webkit.org/show_bug.cgi?id=55260
862
863         Leopard Debug is once again running out of address space while
864         linking (Requested by aroben on #webkit).
865
866         * Scripts/webkitpy/common/net/buildbot/buildbot.py:
867         * Scripts/webkitpy/common/net/buildbot/buildbot_unittest.py:
868
869 2011-02-25  Tony Chang  <tony@chromium.org>
870
871         Reviewed by James Robinson.
872
873         Clean up temp files left by crashing DRTs on chromium-win
874         https://bugs.webkit.org/show_bug.cgi?id=55182
875
876         * BuildSlaveSupport/build.webkit.org-config/master.cfg:
877         * BuildSlaveSupport/chromium/remove-crash-logs: Added.
878
879 2011-02-25  Adam Roben  <aroben@apple.com>
880
881         Double-check that modifying the registry worked
882
883         On Windows Vista/7 with UAC enabled, regtool will fail to modify the registry, but it will
884         still return a successful exit code. So we double-check here that the value we tried to
885         write to the registry was really written.
886
887         Fixes <http://webkit.org/b/55227> old-run-webkit-tests falsely claims to be saving crash
888         logs on Windows Vista/7
889
890         Reviewed by Darin Adler.
891
892         * Scripts/old-run-webkit-tests:
893         (writeRegistryString): After trying to write the string value, check that it was really
894         written.
895
896 2011-02-25  Adam Roben  <aroben@apple.com>
897
898         Set the 32-bit post-mortem debugger on 64-bit OSes
899
900         The post-mortem debugger is controlled by some registry values in
901         HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\AeDebug. On 64-bit OSes, you can either
902         set the values in that key, which will be used for 64-bit apps, or in the equivalent key
903         under Wow6432Node, which will be used for 32-bit apps. Since DumpRenderTree and
904         WebKitTestRunner are 32-bit, we want to use the latter location.
905
906         Fixes <http://webkit.org/b/55225> old-run-webkit-tests sets the post-mortem debugger for
907         64-bit apps, but should set it for 32-bit apps instead
908
909         Reviewed by Darin Adler.
910
911         * Scripts/old-run-webkit-tests:
912         (readRegistryString):
913         (writeRegistryString):
914         Pass --wow32 to regtool so it will get/set values beneath Wow6432Node, which is where
915         Windows looks for the post-mortem debugger values when a 32-bit app crashes on a 64-bit OS.
916         Passing --wow32 on a 32-bit OS will not have any effect.
917
918 2011-02-25  Adam Roben  <aroben@apple.com>
919
920         Clean up code to set up crash log saving on Windows
921
922         Fixes <http://webkit.org/b/55224> old-run-webkit-test's Windows registry code is ugly
923
924         Reviewed by David Kilzer.
925
926         * Scripts/old-run-webkit-tests:
927         (readRegistryString):
928         (writeRegistryString):
929         Moved code to read strings from and write strings to the registry here...
930
931         (setUpWindowsCrashLogSaving): ...from here.
932
933 2011-02-25  Adam Roben  <aroben@apple.com>
934
935         Work around Cygwin's crash-suppression behavior
936
937         Cygwin calls ::SetErrorMode(SEM_FAILCRITICALERRORS), which any processes it launches will
938         inherit. This is bad for testing/debugging, as it causes the post-mortem debugger not to be
939         invoked. (Cygwin does this because it makes crashes more UNIX-y.) We reset the error mode
940         when our test apps launch to work around Cygwin's behavior.
941
942         Fixes <http://webkit.org/b/55222> Test apps crash silently (without invoking post-mortem
943         debugger) when launched from Cygwin 1.7
944
945         Reviewed by Darin Adler.
946
947         * DumpRenderTree/win/DumpRenderTree.cpp:
948         (main):
949         * TestWebKitAPI/win/main.cpp:
950         (main):
951         * WebKitAPITest/main.cpp:
952         (main):
953         * WebKitTestRunner/win/TestControllerWin.cpp:
954         (WTR::TestController::platformInitialize):
955         Call ::SetErrorMode(0) to undo Cygwin's folly.
956
957 2011-02-24  Jocelyn Turcotte  <jocelyn.turcotte@nokia.com>
958
959         Reviewed by Andreas Kling.
960
961         [Qt] Revert the support for QNAM affined to a different thread.
962         https://bugs.webkit.org/show_bug.cgi?id=55149
963
964         Qt 4.8 will have QNAM use its own thread internally by default,
965         no need to keep this complexity in WebKit.
966
967         This mainly reverts:
968         http://trac.webkit.org/changeset/73710
969         http://trac.webkit.org/changeset/73712
970
971         * QtTestBrowser/launcherwindow.cpp:
972         (LauncherWindow::~LauncherWindow):
973         (LauncherWindow::initializeView):
974         (LauncherWindow::createChrome):
975         * QtTestBrowser/launcherwindow.h:
976         (WindowOptions::WindowOptions):
977         * QtTestBrowser/webpage.cpp:
978         (WebPage::WebPage):
979         * QtTestBrowser/webpage.h:
980
981 2011-02-24  Adam Barth  <abarth@webkit.org>
982
983         Reviewed by Eric Seidel.
984
985         Add a script for creating gypi files
986         https://bugs.webkit.org/show_bug.cgi?id=55113
987
988         This is a trivial script to create gypi files from the source code in
989         the current directory.
990
991         * Scripts/make-gypi: Added.
992
993 2011-02-24  Sheriff Bot  <webkit.review.bot@gmail.com>
994
995         Unreviewed, rolling out r79628.
996         http://trac.webkit.org/changeset/79628
997         https://bugs.webkit.org/show_bug.cgi?id=55195
998
999         "broke chromium win tests" (Requested by dpranke on #webkit).
1000
1001         * Scripts/webkitpy/layout_tests/port/chromium_win.py:
1002         * Scripts/webkitpy/layout_tests/port/chromium_win_unittest.py:
1003
1004 2011-02-24  Dirk Pranke  <dpranke@chromium.org>
1005
1006         Reviewed by Mihai Parparita.
1007
1008         rebaseline-chromium-webkit-tests can't handle GPU variants
1009         https://bugs.webkit.org/show_bug.cgi?id=55002
1010
1011         * Scripts/webkitpy/layout_tests/port/chromium_gpu.py:
1012         * Scripts/webkitpy/layout_tests/rebaseline_chromium_webkit_tests.py:
1013
1014 2011-02-24  Dirk Pranke  <dpranke@chromium.org>
1015
1016         Reviewed by Mihai Parparita.
1017
1018         nrwt: log command used to start web server to --verbose
1019
1020         https://bugs.webkit.org/show_bug.cgi?id=55188
1021
1022         * Scripts/webkitpy/layout_tests/port/apache_http_server.py:
1023         * Scripts/webkitpy/layout_tests/port/http_server.py:
1024
1025 2011-02-24  Victoria Kirst  <vrk@google.com>
1026
1027         Reviewed by Mihai Parparita.
1028
1029         [chromium] Move media-file.js and video-test.js back to media/ for UILayoutTests
1030         https://bugs.webkit.org/show_bug.cgi?id=55089
1031
1032         This adds an alias into the media/ directory so that http/tests/media
1033         tests can access the media resources when running in an httpd process.
1034
1035         * Scripts/webkitperl/httpd.pm:
1036         * Scripts/webkitpy/layout_tests/port/apache_http_server.py:
1037         * Scripts/webkitpy/layout_tests/port/http_server.py:
1038
1039 2011-02-24  Dirk Pranke  <dpranke@chromium.org>
1040
1041         Reviewed by Mihai Parparita.
1042
1043         nrwt: enable multiple processes by default on Chromium Win.
1044
1045         https://bugs.webkit.org/show_bug.cgi?id=55163
1046
1047         * Scripts/webkitpy/layout_tests/port/chromium_win.py:
1048         * Scripts/webkitpy/layout_tests/port/chromium_win_unittest.py:
1049
1050 2011-02-24  Sam Weinig  <sam@webkit.org>
1051
1052         Attempt to fix Qt.
1053
1054         * WebKitTestRunner/TestInvocation.cpp:
1055
1056 2011-02-24  James Robinson  <jamesr@chromium.org>
1057
1058         Unreviewed, rolling out r79598.
1059         http://trac.webkit.org/changeset/79598
1060         https://bugs.webkit.org/show_bug.cgi?id=55089
1061
1062         Appears to be causing trouble with httpd on windows
1063
1064         * Scripts/webkitperl/httpd.pm:
1065         * Scripts/webkitpy/layout_tests/port/apache_http_server.py:
1066         * Scripts/webkitpy/layout_tests/port/http_server.py:
1067
1068 2011-02-24  Sam Weinig  <sam@webkit.org>
1069
1070         Reviewed by Darin Adler.
1071
1072         Add basic pixel testing support to WebKitTestRunner
1073         https://bugs.webkit.org/show_bug.cgi?id=55161
1074
1075         * Scripts/old-run-webkit-tests:
1076         Don't print color space information when using WebKitTestRunner since
1077         we don't need to change the screen.
1078
1079         * WebKitTestRunner/InjectedBundle/InjectedBundle.cpp:
1080         (WTR::InjectedBundle::initialize):
1081         Add call to initializePlatformDefaults.
1082         
1083         (WTR::InjectedBundle::done):
1084         Return a dictionary on completion with both the text output and pixel dump override.
1085
1086         * WebKitTestRunner/InjectedBundle/InjectedBundle.h:
1087         Add declaration for initializePlatformDefaults.
1088
1089         * WebKitTestRunner/InjectedBundle/LayoutTestController.cpp:
1090         (WTR::LayoutTestController::LayoutTestController):
1091         (WTR::LayoutTestController::dumpAsText):
1092         * WebKitTestRunner/InjectedBundle/LayoutTestController.h:
1093         (WTR::LayoutTestController::shouldDumpPixels):
1094         Make setting dumpAsText override the dump pixels preference.
1095
1096         * WebKitTestRunner/InjectedBundle/mac/InjectedBundleMac.mm: Added.
1097         (WTR::InjectedBundle::initializePlatformDefaults):
1098         Set platform defaults matching DumpRenderTree.
1099
1100         * WebKitTestRunner/InjectedBundle/qt/InjectedBundleQt.cpp: Added.
1101         (WTR::InjectedBundle::initializePlatformDefaults):
1102         * WebKitTestRunner/InjectedBundle/win/InjectedBundleWin.cpp: Added.
1103         (WTR::InjectedBundle::initializePlatformDefaults):
1104         Add stubs for initializePlatformDefaults.
1105
1106         * WebKitTestRunner/TestController.cpp:
1107         (WTR::TestController::runTest):
1108         (WTR::TestController::runTestingServerLoop):
1109         (WTR::TestController::run):
1110         Parse expected pixel results out of stdin and setup the invocation with it.
1111
1112         * WebKitTestRunner/TestInvocation.cpp:
1113         (WTR::TestInvocation::TestInvocation):
1114         Initialize new members.
1115
1116         (WTR::TestInvocation::~TestInvocation):
1117         (WTR::TestInvocation::setIsPixelTest):
1118         (WTR::sizeWebViewForCurrentTest):
1119         (WTR::TestInvocation::invoke):
1120         (WTR::TestInvocation::dump):
1121         (WTR::TestInvocation::didReceiveMessageFromInjectedBundle):
1122         * WebKitTestRunner/TestInvocation.h:
1123         Dump pixels in addition to text by calling dumpPixelsAndCompareWithExpected.
1124
1125         * WebKitTestRunner/mac/PlatformWebViewMac.mm:
1126         (WTR::PlatformWebView::PlatformWebView):
1127         Use -[NSColorSpace genericRGBColorSpace] instead of the main screen color space.
1128
1129         * WebKitTestRunner/mac/TestInvocationMac.mm: Added.
1130         (WTR::createCGContextFromPlatformView):
1131         (WTR::computeMD5HashStringForContext):
1132         (WTR::dumpBitmap):
1133         (WTR::forceRepaintFunction):
1134         (WTR::TestInvocation::dumpPixelsAndCompareWithExpected):
1135         Add mac specific pixel dumping support. Right now, we always pull pixels
1136         from the window server.
1137
1138         * WebKitTestRunner/WebKitTestRunner.xcodeproj/project.pbxproj:
1139         * WebKitTestRunner/InjectedBundle/qt/InjectedBundle.pro:
1140         * WebKitTestRunner/qt/WebKitTestRunner.pro:
1141         * WebKitTestRunner/win/InjectedBundle.vcproj:
1142         * WebKitTestRunner/win/WebKitTestRunner.vcproj:
1143         Add new files.
1144
1145 2011-02-24  Chang Shu  <cshu@webkit.org>
1146
1147         Reviewed by Kenneth Rohde Christiansen.
1148
1149         [Qt] Launch MiniBrowser if the test run is WebKit2. 
1150         https://bugs.webkit.org/show_bug.cgi?id=55145
1151
1152         * Scripts/old-run-webkit-tests:
1153         * Scripts/run-launcher:
1154         * Scripts/webkitdirs.pm:
1155
1156 2011-02-24  Victoria Kirst  <vrk@google.com>
1157
1158         Reviewed by Mihai Parparita.
1159
1160         [chromium] Move media-file.js and video-test.js back to media/ for UILayoutTests
1161         https://bugs.webkit.org/show_bug.cgi?id=55089
1162
1163         This adds an alias into the media/ directory so that http/tests/media
1164         tests can access the media resources when running in an httpd process.
1165
1166         * Scripts/webkitperl/httpd.pm:
1167         * Scripts/webkitpy/layout_tests/port/apache_http_server.py:
1168         * Scripts/webkitpy/layout_tests/port/http_server.py:
1169
1170 2011-02-24  Andrew Wilson  <atwilson@chromium.org>
1171
1172         Unreviewed, rolling out r79570.
1173         http://trac.webkit.org/changeset/79570
1174         https://bugs.webkit.org/show_bug.cgi?id=54874
1175
1176         Breaks chromium build because glue/mocks/mock_web_frame.h/cc
1177         was not updated
1178
1179         * DumpRenderTree/chromium/TestShell.cpp:
1180         (TestShell::resetTestController):
1181         * DumpRenderTree/gtk/DumpRenderTree.cpp:
1182         (resetDefaultsToConsistentValues):
1183         * DumpRenderTree/mac/DumpRenderTree.mm:
1184         (resetWebViewToConsistentStateBeforeTesting):
1185         * DumpRenderTree/qt/DumpRenderTreeQt.cpp:
1186         (WebCore::DumpRenderTree::resetToConsistentStateBeforeTesting):
1187         * DumpRenderTree/win/DumpRenderTree.cpp:
1188         (resetWebViewToConsistentStateBeforeTesting):
1189         * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp:
1190         (WTR::InjectedBundlePage::reset):
1191
1192 2011-02-24  Mahesh Kulkarni  <mahesh.kulkarni@nokia.com>
1193
1194         Reviewed by Laszlo Gombos.
1195
1196         [Qt] add android 2.2 user agent to QtTestBrowser
1197         https://bugs.webkit.org/show_bug.cgi?id=55085
1198
1199         Credit to Forrest Hodgkins to find right user-agent which works for youtube.
1200         * QtTestBrowser/useragentlist.txt:
1201
1202 2011-02-24  Adam Roben  <aroben@apple.com>
1203
1204         Change FrameLoadDelegate to support any number of delegates with delayed work to process
1205
1206         This makes our behavior match Mac more closely, and allows us to remove an incorrect
1207         assertion that was firing during some tests. (The assertion was claiming that there was
1208         never more than one delegate with delayed work to process, but that was not the case.)
1209
1210         Fixes <http://webkit.org/b/55146> Assertion failure in FrameLoadDelegate::locationChangeDone
1211         when running http/tests/navigation/back-twice-without-commit.html
1212
1213         Reviewed by Eric Carlson.
1214
1215         * DumpRenderTree/win/FrameLoadDelegate.cpp:
1216         (delegatesWithDelayedWork): Added. Returns all FrameLoadDelegates that have delayed work to
1217         process. A single delegate may appear in this Vector more than once (just as, on Mac, a
1218         single delegate may have multiple performSelector requests).
1219         (processWorkTimer): Pass the HWND to ::KillTimer, for pedantic brownie points. Added an
1220         assertion that the timer firing is the shared process work timer. Instead of using the
1221         single, global "delegate waiting for timer" delegate, give all delegates that have delayed
1222         work to process a chance to process their work.
1223         (FrameLoadDelegate::locationChangeDone): If we don't already have an active timer for
1224         processing delayed work, create one. Then add ourselves to the delegatesWithDelayedWork
1225         Vector so our processWork function will be called when the timer fires.
1226
1227 2011-02-24  Vsevolod Vlasov  <vsevik@chromium.org>
1228
1229         Reviewed by Alexey Proskuryakov.
1230
1231         DumpRenderTree should reset frame opener between tests.
1232         https://bugs.webkit.org/show_bug.cgi?id=54874
1233
1234         Added clearOpener() calls between tests.
1235         Some tests are using window.open() to change current window location.
1236         This changes window.opener property whose nullity is checked before
1237         showing xml viewer.
1238
1239         * DumpRenderTree/chromium/TestShell.cpp:
1240         (TestShell::resetTestController):
1241         * DumpRenderTree/gtk/DumpRenderTree.cpp:
1242         (resetDefaultsToConsistentValues):
1243         * DumpRenderTree/mac/DumpRenderTree.mm:
1244         (resetWebViewToConsistentStateBeforeTesting):
1245         * DumpRenderTree/qt/DumpRenderTreeQt.cpp:
1246         (WebCore::DumpRenderTree::resetToConsistentStateBeforeTesting):
1247         * DumpRenderTree/win/DumpRenderTree.cpp:
1248         (resetWebViewToConsistentStateBeforeTesting):
1249         * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp:
1250         (WTR::InjectedBundlePage::reset):
1251
1252 2011-02-24  Adam Roben  <aroben@apple.com>
1253
1254         Don't trigger Windows builds for changes to WebCore.exp.in or the top-level GNUmakefile.am
1255
1256         Fixes <http://webkit.org/b/55144> Windows builders should not have built r79343 or r79440,
1257         but did
1258
1259         Reviewed by Anders Carlsson.
1260
1261         * Scripts/webkitpy/common/config/build.py:
1262         (_should_file_trigger_build): Loosened the pattern for GNUmakefile.am to include the
1263         top-level GNUmakefile.am, and loosened the pattern for .exp files to include WebCore.exp.in.
1264
1265         * Scripts/webkitpy/common/config/build_unittest.py:
1266         (ShouldBuildTest): Added some more tests.
1267
1268 2011-02-24  Andras Becsi  <abecsi@webkit.org>
1269
1270         Reviewed by Laszlo Gombos.
1271
1272         [Qt] MinGW build fails to link
1273         https://bugs.webkit.org/show_bug.cgi?id=55050
1274
1275         Prepend the libraries of subcomponents instead of appending them
1276         to fix the library order according to the dependency of the libraries
1277
1278         * WebKitTestRunner/InjectedBundle/qt/InjectedBundle.pro:
1279         * WebKitTestRunner/qt/WebKitTestRunner.pro:
1280
1281 2011-02-24  Gabor Loki  <loki@webkit.org>
1282
1283         Reviewed by Csaba Osztrogonác.
1284
1285         [Qt] Remove ARMv5 Release buildbot.
1286
1287         Remove ARMv5 Release buildbot, because there is very little difference
1288         between ARMv5 and ARMv7 Release bots, and there is also very little
1289         interest in ARMv5 build slave (compared to ARMv7).
1290
1291         * BuildSlaveSupport/build.webkit.org-config/config.json:
1292         * Scripts/webkitpy/common/net/buildbot/buildbot_unittest.py:
1293
1294 2011-02-22  Ojan Vafai  <ojan@chromium.org>
1295
1296         Reviewed by Tony Chang.
1297
1298         fix revision number finding when creating git patches
1299         https://bugs.webkit.org/show_bug.cgi?id=55024
1300
1301         HEAD~n does not walk merge commits correctly. git log however does.
1302
1303         * Scripts/webkitpy/common/checkout/scm.py:
1304         * Scripts/webkitpy/common/checkout/scm_unittest.py:
1305
1306 2011-02-23  Kenneth Russell  <kbr@google.com>
1307
1308         Unreviewed, rolling out r79387.
1309         https://bugs.webkit.org/show_bug.cgi?id=54885
1310
1311         Reapply the code changes from r79038 since they weren't the root
1312         cause of the test timeouts.
1313
1314         * DumpRenderTree/chromium/WebViewHost.cpp:
1315         (WebViewHost::reset):
1316
1317 2011-02-23  Siddharth Mathur  <siddharth.mathur@nokia.com>
1318
1319         Reviewed by Laszlo Gombos.
1320
1321         [Qt] Make sure Symbian binary UIDs are unique
1322
1323         * MiniBrowser/qt/MiniBrowser.pro: Resolve the UID collision with
1324         QtTestBrowser.pro by changing the UID.
1325
1326 2011-02-23  Patrick Gansterer  <paroga@webkit.org>
1327
1328         Reviewed by Darin Adler.
1329
1330         Rename PLATFORM(CF) to USE(CF)
1331         https://bugs.webkit.org/show_bug.cgi?id=53540
1332
1333         * DumpRenderTree/config.h:
1334
1335 2011-02-23  Dirk Pranke  <dpranke@chromium.org>
1336
1337         Reviewed by Ojan Vafai.
1338
1339         nrwt: turn on multiprocessing on chromium_linux, clean up unit tests for worker model defaults
1340         https://bugs.webkit.org/show_bug.cgi?id=55016
1341
1342         * Scripts/webkitpy/layout_tests/port/base.py:
1343         * Scripts/webkitpy/layout_tests/port/chromium_linux.py:
1344         * Scripts/webkitpy/layout_tests/port/chromium_linux_unittest.py: Added.
1345         * Scripts/webkitpy/layout_tests/port/chromium_mac.py:
1346         * Scripts/webkitpy/layout_tests/port/chromium_mac_unittest.py:
1347         * Scripts/webkitpy/layout_tests/port/chromium_win_unittest.py:
1348         * Scripts/webkitpy/layout_tests/port/mac.py:
1349         * Scripts/webkitpy/layout_tests/port/mac_unittest.py:
1350         * Scripts/webkitpy/layout_tests/port/mock_drt_unittest.py:
1351         * Scripts/webkitpy/layout_tests/port/port_testcase.py:
1352
1353 2011-02-23  Tony Chang  <tony@chromium.org>
1354
1355         Reviewed by Kent Tamura.
1356
1357         [chromium] DumpRenderTree --check-layout-test-sys-deps leaks 4 temp dirs
1358         https://bugs.webkit.org/show_bug.cgi?id=55004
1359
1360         * DumpRenderTree/chromium/DumpRenderTree.cpp:
1361         (WebKitSupportTestEnvironment::WebKitSupportTestEnvironment):
1362         (WebKitSupportTestEnvironment::~WebKitSupportTestEnvironment):
1363         (main): Use a C++ object to ensure that TearDownTestEnvironment
1364             gets called when --check-layout-test-sys-deps is called
1365
1366 2011-02-23  Gopal Raghavan  <gopal.1.raghavan@nokia.com>
1367
1368         Reviewed by Anders Carlsson.
1369
1370         [Qt] [Qt] check webkit style fails in qt/tests
1371         https://bugs.webkit.org/show_bug.cgi?id=54998
1372
1373         check-webkit-style was complaining about missing config.h in WebKit/qt/test/qdeclarativewebview. This patch was originally part of 50222. Reviewer suggested to split and create a separate patch for checker.
1374
1375         * Scripts/webkitpy/style/checker.py:
1376
1377 2011-02-22  Vsevolod Vlasov  <vsevik@chromium.org>
1378
1379         Reviewed by Pavel Feldman.
1380
1381         DumpRenderTree should dump as text if test path contains "dumpAsText/"
1382         https://bugs.webkit.org/show_bug.cgi?id=54873
1383
1384         * DumpRenderTree/chromium/LayoutTestController.h:
1385         (LayoutTestController::setShouldDumpAsText):
1386         (LayoutTestController::setShouldGeneratePixelResults):
1387         * DumpRenderTree/chromium/TestShell.cpp:
1388         (TestShell::runFileTest):
1389         * DumpRenderTree/gtk/DumpRenderTree.cpp:
1390         (shouldDumpAsText):
1391         (dump):
1392         (runTest):
1393         * DumpRenderTree/mac/DumpRenderTree.mm:
1394         (shouldDumpAsText):
1395         (runTest):
1396         * DumpRenderTree/qt/DumpRenderTreeQt.cpp:
1397         (WebCore::isDumpAsTextTest):
1398         (WebCore::DumpRenderTree::open):
1399         * DumpRenderTree/win/DumpRenderTree.cpp:
1400         (shouldDumpAsText):
1401         (runTest):
1402         * WebKitTestRunner/InjectedBundle/InjectedBundlePage.cpp:
1403         (WTR::InjectedBundlePage::dump):
1404
1405 2011-02-22  Kenneth Russell  <kbr@google.com>
1406
1407         Reviewed by James Robinson.
1408
1409         Many tests started timing out on the chrome mac canaries around r79035
1410         https://bugs.webkit.org/show_bug.cgi?id=54885
1411
1412         Revert the code changes, but not the Chromium DEPS roll, from
1413         r79038 to see whether this clears up the timing-out tests. Was not
1414         able to reproduce the timeouts locally with a Release mode build
1415         of DRT (on Snow Leopard).
1416
1417         * DumpRenderTree/chromium/WebViewHost.cpp:
1418         (WebViewHost::reset):
1419
1420 2011-02-22  Dirk Pranke  <dpranke@chromium.org>
1421
1422         Reviewed by James Robinson.
1423
1424         nrwt: --platform chromium-gpu doesn't work with --worker-model=processes
1425         https://bugs.webkit.org/show_bug.cgi?id=55009
1426
1427         * Scripts/webkitpy/layout_tests/port/chromium_gpu.py:
1428
1429 2011-02-18  Mihai Parparita  <mihaip@chromium.org>
1430
1431         Reviewed by Ojan Vafai.
1432
1433         Default to 500 for --exit-after-n-failures
1434         https://bugs.webkit.org/show_bug.cgi?id=54773
1435
1436         Make 500 the default value for --exit-after-n-failures (ORWT bots
1437         already default to this as of r75726). Similarly, change the default for
1438         --exit-after-n-crashes-or-timeouts to 20.
1439
1440         * Scripts/webkitpy/layout_tests/run_webkit_tests.py:
1441
1442 2011-02-02  Martin Robinson  <mrobinson@igalia.com>
1443
1444         Reviewed by Xan Lopez.
1445
1446         [GTK] DRT needs an implementation of LayoutTestController.setSelectTrailingWhitespace
1447         https://bugs.webkit.org/show_bug.cgi?id=53603
1448
1449         Add support for LayoutTestController.setSelectTrailingWhitespace by calling through
1450         to DumpRenderTreeSupportGtk for this functionality.
1451
1452         * DumpRenderTree/gtk/DumpRenderTree.cpp:
1453         (resetDefaultsToConsistentValues): Always call LayoutTestController.setSelectTrailingWhitespace to
1454         false when reseting values between tests.
1455         * DumpRenderTree/gtk/LayoutTestControllerGtk.cpp:
1456         (LayoutTestController::setSelectTrailingWhitespaceEnabled): Call LayoutTestController.setSelectTrailingWhitespace
1457         with the appropriate value.
1458
1459 2011-02-22  Adam Roben  <aroben@apple.com>
1460
1461         Don't trigger a Windows build when the top-level ChangeLog changes
1462
1463         r79320 is an example of a change that shouldn't have built on Windows, but did, because of
1464         this bug.
1465
1466         Fixes <http://webkit.org/b/54957> Windows bots build when top-level ChangeLog changes, but
1467         shouldn't
1468
1469         Reviewed by Anders Carlsson.
1470
1471         * Scripts/webkitpy/common/config/build.py:
1472         (_should_file_trigger_build): Modified the ChangeLog pattern to match the top-level
1473         ChangeLog, and re-sorted it in the list.
1474
1475         * Scripts/webkitpy/common/config/build_unittest.py:
1476         (ShouldBuildTest): Added a test that ChangeLogs don't trigger builds.
1477
1478 2011-02-22  Sheriff Bot  <webkit.review.bot@gmail.com>
1479
1480         Unreviewed, rolling out r78691.
1481         http://trac.webkit.org/changeset/78691
1482         https://bugs.webkit.org/show_bug.cgi?id=54956
1483
1484         Leopard Debug is successfully building again (Requested by
1485         aroben on #webkit).
1486
1487         * Scripts/webkitpy/common/net/buildbot/buildbot.py:
1488         * Scripts/webkitpy/common/net/buildbot/buildbot_unittest.py:
1489
1490 2011-02-22  Ilya Tikhonovsky  <loislo@chromium.org>
1491
1492         Reviewed by Pavel Feldman.
1493
1494         Web Inspector: flakyness of inspector tests.
1495         https://bugs.webkit.org/show_bug.cgi?id=54729
1496
1497         As far as we have the protocol with sequence numbers we can simplify test harness support
1498         and drop out chomium specific methods.
1499
1500         * DumpRenderTree/DumpRenderTree.gypi:
1501         * DumpRenderTree/chromium/DRTDevToolsAgent.cpp:
1502         (DRTDevToolsAgent::sendMessageToInspectorFrontend):
1503         (DRTDevToolsAgent::runtimePropertyChanged):
1504         (DRTDevToolsAgent::asyncCall):
1505         (DRTDevToolsAgent::call):
1506         * DumpRenderTree/chromium/DRTDevToolsAgent.h:
1507         * DumpRenderTree/chromium/DRTDevToolsCallArgs.cpp: Removed.
1508         * DumpRenderTree/chromium/DRTDevToolsCallArgs.h: Removed.
1509         * DumpRenderTree/chromium/DRTDevToolsClient.cpp:
1510         (DRTDevToolsClient::sendMessageToBackend):
1511         (DRTDevToolsClient::asyncCall):
1512         (DRTDevToolsClient::call):
1513         * DumpRenderTree/chromium/DRTDevToolsClient.h:
1514
1515 2011-02-22  Andras Becsi  <abecsi@webkit.org>
1516
1517         Reviewed by Laszlo Gombos.
1518
1519         [Qt] Redesign the build system
1520         https://bugs.webkit.org/show_bug.cgi?id=51339
1521
1522         Part 2.
1523
1524         Build WebCore as a static library, compile the WebKit API and WebKit2 API
1525         in a final step and link to WebKit2, WebCore and JSC libraries to fix
1526         linking issues resulting from stripped away symbols.
1527
1528         * WebKitTestRunner/InjectedBundle/qt/InjectedBundle.pro: Remove superfluous includepaths.
1529
1530 2011-02-21  Lukasz Slachciak  <l.slachciak@samsung.com>
1531
1532         Reviewed by Antonio Gomes.
1533
1534         [EFL] Restored spatial navigation option in EWebLauncher since #18662 bug is closed
1535         https://bugs.webkit.org/show_bug.cgi?id=54806
1536
1537         * EWebLauncher/main.c:
1538         (on_key_down): calling ewk_view_setting_spatial_navigation_set when "F12" pressed added.
1539
1540 2011-02-21  Dirk Pranke  <dpranke@chromium.org>
1541
1542         Reviewed by Kent Tamura.
1543
1544         nrwt: revert behavior on 10.5 to "old-threads". It's possible
1545         that r79062 has introduced a bunch of flakiness, although I
1546         can't explain why right now. This change should restore the
1547         old behavior, and then we can look at the build logs to see
1548         if the tests stop timing out or being flaky.
1549
1550         https://bugs.webkit.org/show_bug.cgi?id=54925
1551
1552         * Scripts/webkitpy/layout_tests/port/chromium_mac.py:
1553         * Scripts/webkitpy/layout_tests/port/chromium_mac_unittest.py:
1554         * Scripts/webkitpy/layout_tests/port/mac.py:
1555         * Scripts/webkitpy/layout_tests/port/mac_unittest.py:
1556
1557 2011-02-21  Dirk Pranke  <dpranke@chromium.org>
1558
1559         Reviewed by Dimitri Glazkov.
1560
1561         nrwt: fix typo causing us to try and repeatedly shut down the servers
1562         https://bugs.webkit.org/show_bug.cgi?id=54904
1563
1564         * Scripts/webkitpy/layout_tests/layout_package/worker_mixin.py:
1565
1566 2011-02-21  Alejandro G. Castro  <alex@igalia.com>
1567
1568         Reviewed by Csaba Osztrogonác.
1569
1570         REGRESSION(r79157):
1571         plugins/get-url-with-javascript-destroying-plugin.html make the
1572         next test crash
1573         https://bugs.webkit.org/show_bug.cgi?id=54863
1574
1575         * GNUmakefile.am:
1576
1577 2011-02-20  Dirk Pranke  <dpranke@chromium.org>
1578
1579         Reviewed by Ojan Vafai.
1580
1581         test-webkitpy: stop skipping a bunch of tests on win32
1582         https://bugs.webkit.org/show_bug.cgi?id=54788
1583
1584         * Scripts/webkitpy/common/net/testoutputset_unittest.py:
1585         * Scripts/webkitpy/test/main.py:
1586
1587 2011-02-20  Dirk Pranke  <dpranke@chromium.org>
1588
1589         Reviewed by Ojan Vafai.
1590
1591         Fix various bugs related to handling of dir separators and real
1592         vs. fake filesystems that were causing test-webkitpy to fail
1593         tests on win32. These bugs have a variety of causes but are
1594         lumped together because they should be easy to review.
1595
1596         https://bugs.webkit.org/show_bug.cgi?id=54700
1597
1598         * Scripts/webkitpy/common/net/testoutput.py:
1599           Here we were using os.path.sep but not escaping it prior to
1600           using in a regexp, which  caused bad things on win32.
1601         * Scripts/webkitpy/common/net/testoutput_unittest.py:
1602           Same here.
1603         * Scripts/webkitpy/layout_tests/port/chromium_unittest.py:
1604           Here we were using the native filesystem instead of the mock
1605           filesystem, making some tests unnecessarily platform dependent.
1606           Also we weren't setting the default configuration for the
1607           TestChromiumWinPort.
1608         * Scripts/webkitpy/layout_tests/update_webgl_conformance_tests.py:
1609           Here we *shouldn't* use os.path.join(), just so the code becomes
1610           less platform dependent.
1611         * Scripts/webkitpy/style/checker.py:
1612           Here we should be using os.path.sep instead of "/"; the tests
1613           require the native filesystem.
1614         * Scripts/webkitpy/style/checkers/cpp_unittest.py:
1615           Here the underlying code is using os.path.sep, so we probably
1616           should as well.
1617         * Scripts/webkitpy/style/test_expectations_unittest.py:
1618           Here we should be using the filesystem hanging of the port
1619           object (which is the mocked filesystem) rather than the native
1620           one.
1621
1622 2011-02-20  Anders Carlsson  <andersca@apple.com>
1623
1624         Reviewed by Maciej Stachowiak.
1625
1626         Crash when a plug-in requests a javascript: url that destroys the plug-in
1627         https://bugs.webkit.org/show_bug.cgi?id=54837
1628         <rdar://problem/9005475>
1629
1630         Add new plug-in test.
1631
1632         * DumpRenderTree/DumpRenderTree.gypi:
1633         * DumpRenderTree/DumpRenderTree.xcodeproj/project.pbxproj:
1634         * DumpRenderTree/TestNetscapePlugIn/PluginTest.cpp:
1635         (PluginTest::NPN_GetURL):
1636         * DumpRenderTree/TestNetscapePlugIn/PluginTest.h:
1637         * DumpRenderTree/TestNetscapePlugIn/Tests/GetURLWithJavaScriptURLDestroyingPlugin.cpp: Added.
1638         (GetURLWithJavaScriptURLDestroyingPlugin::GetURLWithJavaScriptURLDestroyingPlugin):
1639         (GetURLWithJavaScriptURLDestroyingPlugin::NPP_New):
1640         * DumpRenderTree/TestNetscapePlugIn/win/TestNetscapePlugin.vcproj:
1641         * DumpRenderTree/qt/TestNetscapePlugin/TestNetscapePlugin.pro:
1642
1643 2011-02-19  Patrick Gansterer  <paroga@webkit.org>
1644
1645         Reviewed by Andreas Kling.
1646
1647         Make WinCE a core builder
1648         https://bugs.webkit.org/show_bug.cgi?id=54801
1649
1650         * Scripts/webkitpy/common/net/buildbot/buildbot.py:
1651         * Scripts/webkitpy/common/net/buildbot/buildbot_unittest.py:
1652
1653 2011-02-19  Lukasz Slachciak  <l.slachciak@samsung.com>
1654
1655         Reviewed by Andreas Kling.
1656
1657         Fixed problem with launching EWebLauncher with "run-launcher --efl". Incorrect path referring to non-existing .libs directory ws set.
1658         https://webkit.org/b/54778
1659
1660         * Scripts/webkitdirs.pm:
1661
1662 2011-02-18  David Levin  <levin@chromium.org>
1663
1664         Reviewed by Eric Seidel.
1665
1666         check-webkit-style: Misses brace style error when the line with the { has a } in it.
1667         https://bugs.webkit.org/show_bug.cgi?id=54769
1668
1669         * Scripts/webkitpy/style/checkers/cpp.py: Fixed the check for the close brace
1670           to only look after the last open brace, so that the open brace in this line
1671           "} else {" will still be able to trigger the error.
1672         * Scripts/webkitpy/style/checkers/cpp_unittest.py: Added related unit tests.
1673
1674 2011-02-18  Zan Dobersek  <zandobersek@gmail.com>
1675
1676         Reviewed by Martin Robinson.
1677
1678         [Gtk] Re-enable meter tag support
1679         https://bugs.webkit.org/show_bug.cgi?id=54762
1680
1681         Meter tag is supported on the Gtk port, enable it by default.
1682
1683         * Scripts/build-webkit:
1684
1685 2011-02-18  David Levin  <levin@chromium.org>
1686
1687         Reviewed by Eric Seidel.
1688
1689         check-webkit-style falsely complains about WebKitGTK+ public headers
1690         https://bugs.webkit.org/show_bug.cgi?id=54650
1691
1692         * Scripts/webkitpy/style/checker.py: Add some exceptions for the gtk files.
1693         * Scripts/webkitpy/style/checkers/cpp.py: Restrict the checks for WEBKIT_API
1694           to the chromium directory (and improved the checks slightly).
1695         * Scripts/webkitpy/style/checkers/cpp_unittest.py: Added corresponding tests.
1696
1697 2011-02-18  Dirk Pranke  <dpranke@chromium.org>
1698
1699         Reviewed by James Robinson.
1700
1701         new-run-webkit-tests: enable multiple processes by default on
1702         mac. This change removes the artificial restrictions we placed
1703         on the # of child processes we used with the old threading
1704         model, and switches to the new message-based model and multiple
1705         processes, where available. If multiple processes are not
1706         available (Leopard / Python 2.5), then we use the 'inline' model
1707         instead of the 'old-threads' model on the 'mac' port or one
1708         process on the 'chromium-mac' port. We need additional testing
1709         to see if the new 'threads' model works reliably and is worth
1710         supporting, or if we should just live with things being slightly
1711         slow.
1712
1713         https://bugs.webkit.org/show_bug.cgi?id=54596
1714
1715         * Scripts/webkitpy/layout_tests/port/chromium_mac.py:
1716         * Scripts/webkitpy/layout_tests/port/chromium_mac_unittest.py:
1717         * Scripts/webkitpy/layout_tests/port/mac.py:
1718         * Scripts/webkitpy/layout_tests/port/mac_unittest.py:
1719         * Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
1720
1721 2011-02-18  Andrew Wilson  <atwilson@chromium.org>
1722
1723         Unreviewed, rolling out r79047.
1724         http://trac.webkit.org/changeset/79047
1725         https://bugs.webkit.org/show_bug.cgi?id=54596
1726
1727         Broke canary bots - please watch the canaries the next time
1728         you land this
1729
1730         * Scripts/webkitpy/layout_tests/port/chromium_mac.py:
1731         * Scripts/webkitpy/layout_tests/port/mac.py:
1732
1733 2011-02-18  Emil A Eklund  <eae@chromium.org>
1734
1735         Reviewed by Darin Adler.
1736
1737         Crash in EventHandler::sendContextMenuEventForKey
1738         https://bugs.webkit.org/show_bug.cgi?id=54495
1739
1740         Add support for the context menu key (VK_APPS) to EventSender for
1741         platforms that support the key.
1742
1743         * DumpRenderTree/chromium/EventSender.cpp:
1744         (EventSender::keyDown):
1745         * DumpRenderTree/gtk/EventSender.cpp:
1746         (keyDownCallback):
1747         * DumpRenderTree/qt/EventSenderQt.cpp:
1748         (EventSender::keyDown):
1749         * DumpRenderTree/win/EventSender.cpp:
1750         (keyDownCallback):
1751
1752 2011-02-18  Dirk Pranke  <dpranke@chromium.org>
1753
1754         Reviewed by Ojan Vafai.
1755
1756         new-run-webkit-tests: enable multiple processes by default on mac
1757         https://bugs.webkit.org/show_bug.cgi?id=54596
1758
1759         * Scripts/webkitpy/layout_tests/port/chromium_mac.py:
1760         * Scripts/webkitpy/layout_tests/port/mac.py:
1761
1762 2011-02-18  Dirk Pranke  <dpranke@chromium.org>
1763
1764         Reviewed by Adam Roben.
1765
1766         Skip pretty patch unit tests if ruby isn't installed.
1767
1768         https://bugs.webkit.org/show_bug.cgi?id=54699
1769
1770         * Scripts/webkitpy/common/prettypatch_unittest.py:
1771
1772 2011-02-18  Dirk Pranke  <dpranke@chromium.org>
1773
1774         Reviewed by Adam Roben.
1775
1776         Skip mac.test_skipped_file_paths() when running on win32; the
1777         test has hardcoded '/' directory separators and since it's a
1778         port-specific test and we have coverage on other ports it's not
1779         worth it to make the test more generic.
1780
1781         https://bugs.webkit.org/show_bug.cgi?id=54698
1782
1783         * Scripts/webkitpy/layout_tests/port/mac_unittest.py:
1784
1785 2011-02-18  Dirk Pranke  <dpranke@chromium.org>
1786
1787         Reviewed by Adam Roben.
1788
1789         Fix webkitpy.layout_tests.port.config_unittest to work on Win32.
1790
1791         https://bugs.webkit.org/show_bug.cgi?id=54703
1792
1793         * Scripts/webkitpy/layout_tests/port/config_unittest.py:
1794
1795 2011-02-18  Dirk Pranke  <dpranke@chromium.org>
1796
1797         Reviewed by Adam Roben.
1798
1799         Change the default port we're mocking out on win32 when we
1800         run mock_drt_unittest. Normally we'd default to the
1801         port-specific default, but that would be "win", which doesn't
1802         currently work.
1803
1804         https://bugs.webkit.org/show_bug.cgi?id=54708
1805
1806         * Scripts/webkitpy/layout_tests/port/mock_drt_unittest.py:
1807
1808 2011-02-18  Kenneth Russell  <kbr@google.com>
1809
1810         Reviewed by James Robinson.
1811
1812         [chromium] Clean up initialization of minimum timer interval
1813         https://bugs.webkit.org/show_bug.cgi?id=54772
1814
1815         Use new GetForegroundTabTimerInterval function in webkit_support
1816         to reset the minimum timer interval before each test.
1817
1818         * DumpRenderTree/chromium/WebViewHost.cpp:
1819         (WebViewHost::reset):
1820
1821 2011-02-18  Dirk Pranke  <dpranke@chromium.org>
1822
1823         Reviewed by Adam Roben.
1824
1825         r78494 fixed the chromium_win unittest failing under win32,
1826         but not cygwin. This fixes cygwin as well.
1827
1828         https://bugs.webkit.org/show_bug.cgi?id=54652
1829
1830         * Scripts/webkitpy/layout_tests/port/chromium_win.py:
1831
1832 2011-02-18  Dirk Pranke  <dpranke@chromium.org>
1833
1834         Reviewed by Ojan Vafai, Mihai Parparita.
1835
1836         new-run-webkit-tests: r78522 made it impossible to create a
1837         'chromium-win' or 'chromium-mac' port; they would always get the
1838         version tacked on to the end. It turns out that we actually need
1839         to be able to create these ports in order for
1840         rebaseline-chromium-webkit-tests to work correctly given the
1841         broken coverage of our layout bots. As soon as we get SL and
1842         Win 7 bots running reliably on the canaries, we can get rid of
1843         this change. Note that we only need to change the chromium_mac
1844         and win ports because linux doesn't yet use versions and that
1845         script doesn't work with non-chromium ports.
1846
1847         Also, this change fixes a bug where we were using
1848         "platform/chromium-win-win7" and
1849         "platform/chromium-mac-snowleopard" as the baseline dirs for
1850         Win 7 and SL; we should've been using the generic directories
1851         instead since they are the newest versions of the os's.
1852
1853         https://bugs.webkit.org/show_bug.cgi?id=54691
1854
1855         * Scripts/webkitpy/layout_tests/port/chromium_mac.py:
1856         * Scripts/webkitpy/layout_tests/port/chromium_mac_unittest.py:
1857         * Scripts/webkitpy/layout_tests/port/chromium_win.py:
1858         * Scripts/webkitpy/layout_tests/port/chromium_win_unittest.py:
1859         * Scripts/webkitpy/layout_tests/rebaseline_chromium_webkit_tests.py:
1860
1861 2011-02-16  MORITA Hajime  <morrita@google.com>
1862
1863         Reviewed by Kent Tamura.
1864
1865         [Chromium][DRT] some PostDelayedTask() accesses possibly dangling this pointer.
1866         http://webkit.org/b/53899
1867
1868         Replaced static invocation wrappers with MethodTask objects,
1869         which are removed on WebViewHost destructor.
1870         
1871         * DumpRenderTree/chromium/WebViewHost.cpp:
1872         (WebViewHost::requestCheckingOfText):
1873         (WebViewHost::scheduleAnimation):
1874         (WebViewHost::closeWidgetSoon):
1875         * DumpRenderTree/chromium/WebViewHost.h:
1876         (WebViewHost::HostMethodTask::HostMethodTask):
1877         (WebViewHost::HostMethodTask::runIfValid):
1878         (WebViewHost::taskList):
1879
1880 2011-02-07  Hayato Ito  <hayato@chromium.org>
1881
1882         Reviewed by Ojan Vafai.
1883
1884         Get rid of code which writes test results from test_type's
1885         compare_output() method.
1886         Writing test results now happens in single_test_runner, using
1887         newly introduced test_result_writer module.
1888
1889         Subsequent patches will eliminate test_type/* classes.
1890
1891         https://bugs.webkit.org/show_bug.cgi?id=53908
1892
1893         * Scripts/webkitpy/layout_tests/layout_package/single_test_runner.py:
1894         * Scripts/webkitpy/layout_tests/layout_package/test_result_writer.py: Added.
1895         * Scripts/webkitpy/layout_tests/test_types/image_diff.py:
1896         * Scripts/webkitpy/layout_tests/test_types/test_type_base.py:
1897         * Scripts/webkitpy/layout_tests/test_types/text_diff.py:
1898
1899 2011-02-17  Dirk Pranke  <dpranke@chromium.org>
1900
1901         Reviewed by Adam Roben.
1902
1903         Add blacklist of remaining webkitpy/* modules that fail on
1904         Win32, so that test-webkitpy will at least run cleanly.
1905
1906         https://bugs.webkit.org/show_bug.cgi?id=54709
1907
1908         * Scripts/webkitpy/layout_tests/test/main.py:
1909
1910 2011-02-17  Simon Fraser  <simon.fraser@apple.com>
1911
1912         Reviewed by Sam Weinig.
1913
1914         Set a preference to ensure that scrollbars are in a predictable
1915         state for DRT.
1916
1917         * DumpRenderTree/mac/DumpRenderTree.mm:
1918         (resetDefaultsToConsistentValues):
1919
1920 2011-02-17  Dirk Pranke  <dpranke@chromium.org>
1921
1922         Reviewed by Adam Roben.
1923
1924         test-webkitpy doesn't work with the multiprocessing module on win32
1925         or cygwin, so we skip it for now.
1926
1927         https://bugs.webkit.org/show_bug.cgi?id=54520
1928
1929         * Scripts/webkitpy/layout_tests/layout_package/manager_worker_broker_unittest.py:
1930         * Scripts/webkitpy/layout_tests/layout_package/run_webkit_tests_unittest.py:
1931
1932 2011-02-17  Dirk Pranke  <dpranke@chromium.org>
1933
1934         Reviewed by Adam Roben.
1935
1936         Fix version, baseline_search_path for chromium-win-win7. Also
1937         fix chromium_win_unittests, and fix port_testcase.make_port()
1938         to correctly use the options passed into it and only set
1939         results_directory() if necessary.
1940
1941         https://bugs.webkit.org/show_bug.cgi?id=54652
1942
1943         * Scripts/webkitpy/layout_tests/port/chromium_win.py:
1944         * Scripts/webkitpy/layout_tests/port/chromium_win_unittest.py:
1945         * Scripts/webkitpy/layout_tests/port/port_testcase.py:
1946
1947 2011-02-17  Dirk Pranke  <dpranke@chromium.org>
1948
1949         Reviewed by Ojan Vafai.
1950
1951         Fix a hang in new-run-webkit-tests when run with
1952         --dry-run --worker-model=processes
1953
1954         https://bugs.webkit.org/show_bug.cgi?id=54595
1955
1956         * Scripts/webkitpy/layout_tests/layout_package/test_runner2.py:
1957         * Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
1958
1959 2011-02-17  Kevin Ollivier  <kevino@theolliviers.com>
1960
1961         [wx] Build fixes after recent changes.
1962
1963         * DumpRenderTree/wx/LayoutTestControllerWx.cpp:
1964         (LayoutTestController::setMinimumTimerInterval):
1965         * wx/build/settings.py:
1966
1967 2011-02-17  Csaba Osztrogonác  <ossy@webkit.org>
1968
1969         Reviewed by Andreas Kling.
1970
1971         [Qt] ImageDiff sometimes hangs
1972         https://bugs.webkit.org/show_bug.cgi?id=54641
1973
1974         * DumpRenderTree/qt/ImageDiff.cpp: Clear actualImage and baselineImage after all comparison.
1975         (main):
1976
1977 2011-02-17  Peter Varga  <pvarga@webkit.org>
1978
1979         Reviewed by Csaba Osztrogonác.
1980
1981         The run-sunspider --v8-suite doesn't work
1982         https://bugs.webkit.org/show_bug.cgi?id=54660
1983
1984         * Scripts/run-sunspider:
1985
1986 2011-02-17  Gabor Rapcsanyi  <rgabor@webkit.org>
1987
1988         Reviewed by Csaba Osztrogonác.
1989
1990         [NRWT] Fix environment setup on Qt port.
1991         https://bugs.webkit.org/show_bug.cgi?id=54549
1992
1993         * Scripts/webkitpy/layout_tests/port/qt.py:
1994
1995 2011-02-16  Yael Aharon  <yael.aharon@nokia.com>
1996
1997         Reviewed by Andreas Kling.
1998
1999         [Qt] Enable tiled backing store by default in MiniBrowser.
2000         https://bugs.webkit.org/show_bug.cgi?id=54581
2001
2002         Instead of a flag to turn it on, a new flag is defined to turn
2003         tiled backing store off.
2004
2005         * MiniBrowser/qt/MiniBrowserApplication.cpp:
2006         (MiniBrowserApplication::handleUserOptions):
2007         * MiniBrowser/qt/MiniBrowserApplication.h:
2008         (WindowOptions::WindowOptions):
2009
2010 2011-02-16  Eric Seidel  <eric@webkit.org>
2011
2012         Reviewed by Mihai Parparita.
2013
2014         flaky test archive uploads are always 22 bytes long
2015         https://bugs.webkit.org/show_bug.cgi?id=54593
2016
2017         Speculative fix for the 22-bytes bug.
2018
2019         * Scripts/webkitpy/tool/bot/flakytestreporter.py:
2020
2021 2011-02-16  Anna Cavender  <annacc@chromium.org>
2022
2023         Reviewed by Mihai Parparita.
2024
2025         http/tests/media should be served over HTTP (not from local file)
2026         https://bugs.webkit.org/show_bug.cgi?id=54028
2027
2028         Removed special file:// treatment of http/tests/media
2029
2030         * Scripts/old-run-webkit-tests:
2031         * Scripts/webkitpy/layout_tests/port/base.py:
2032
2033 2011-02-16  Mihai Parparita  <mihaip@chromium.org>
2034
2035         Reviewed by Ojan Vafai.
2036
2037         ChangeLog.parse_latest_entry_from_file does not handle rolled over ChangeLogs
2038         https://bugs.webkit.org/show_bug.cgi?id=54609
2039         
2040         r78737 rolled over ChangeLogs, and webkit-patch complains when landing
2041         patches such as this one that are the first to add an entry to the
2042         ChangeLog (since they don't see a second date line to know where the
2043         entry ends).
2044
2045         * Scripts/webkitpy/common/checkout/changelog.py:
2046         * Scripts/webkitpy/common/checkout/changelog_unittest.py:
2047
2048 == Rolled over to ChangeLog-2011-02-16 ==