blob: f3f21b2c3e956f1490e40d113eb8a4f7b75a5d86 [file] [log] [blame]
2010-05-24 Ojan Vafai <ojan@chromium.org>
Reviewed by Eric Seidel.
add tests to ensure that --git-commit ranges are exclusive of the start of the range
https://bugs.webkit.org/show_bug.cgi?id=39612
* Scripts/webkitpy/common/checkout/scm_unittest.py:
2010-05-24 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
Add a temporary script for testing the html5 parser until it can run more layout tests
https://bugs.webkit.org/show_bug.cgi?id=39611
* Scripts/test-html5-parser: Added.
2010-05-24 Eric Seidel <eric@webkit.org>
Reviewed by Chris Jerdonek.
webkit-patch needs --verbose flag to enable DEBUG logging
https://bugs.webkit.org/show_bug.cgi?id=39208
I also added some code to print out how long commands take to run.
* Scripts/webkit-patch:
- Add hackish -v/--verbose parsing (similar to check-webkit-style)
* Scripts/webkitpy/common/system/executive.py:
- Log how long commands take to run.
* Scripts/webkitpy/tool/main.py:
- Add -v/--verbose option to global options.
2010-05-23 Eric Seidel <eric@webkit.org>
Reviewed by Daniel Bates.
Split PatchReader out into its own file
https://bugs.webkit.org/show_bug.cgi?id=39576
This is in preparation for making check-webkit-style
support being passed paths to patch files on the command line.
* Scripts/check-webkit-style:
* Scripts/webkitpy/style/checker.py:
* Scripts/webkitpy/style/checker_unittest.py:
* Scripts/webkitpy/style/patchreader.py: Added.
* Scripts/webkitpy/style/patchreader_unittest.py: Added.
* Scripts/webkitpy/style_references.py:
2010-05-23 Adam Barth <abarth@webkit.org>
Reviewed by Daniel Bates.
webkit-patch should let you add a comment when uploading a patch
https://bugs.webkit.org/show_bug.cgi?id=39552
As requested by Dan "the man" Bates.
* Scripts/webkitpy/tool/steps/options.py:
* Scripts/webkitpy/tool/steps/postdiff.py:
2010-05-23 Adam Barth <abarth@webkit.org>
Reviewed by Eric Seidel.
webkit-patch should assign newly created bugs to their creator
https://bugs.webkit.org/show_bug.cgi?id=39548
As requested on webkit-dev.
* Scripts/webkitpy/common/net/bugzilla.py:
2010-05-23 Adam Barth <abarth@webkit.org>
Reviewed by Eric Seidel.
Hide prepare and post commands for webkit-patch
https://bugs.webkit.org/show_bug.cgi?id=39539
It turns out these commands aren't very popular and they confuse new
users. They'll still be there for advanced users, however.
* Scripts/webkitpy/tool/commands/upload.py:
2010-05-23 Jesus Sanchez-Palencia <jesus@webkit.org>
Reviewed by Laszlo Gombos.
[Qt] QtTestBrowser has two graphicsview options that aren't enabled correctly
https://bugs.webkit.org/show_bug.cgi?id=39491
Making toggleResizesToContents and toggleTiledBackingStore checkable when
QtTestBrowser is started on graphics view mode.
* QtTestBrowser/main.cpp:
(LauncherWindow::createChrome):
2010-05-23 Jesus Sanchez-Palencia <jesus@webkit.org>
Reviewed by Laszlo Gombos.
[Qt] QtTestBrowser is still called QtLauncher in the code
https://bugs.webkit.org/show_bug.cgi?id=39488
Finish the name change of QtLauncher to QtTestBrowser.
* QtTestBrowser/main.cpp:
(LauncherApplication::LauncherApplication):
(LauncherApplication::handleUserOptions):
* QtTestBrowser/mainwindow.cpp:
(MainWindow::MainWindow):
* QtTestBrowser/useragentlist.txt:
2010-05-23 Marcus Bulach <bulach@chromium.org>
Reviewed by Kent Tamura.
[chromium] Adds Geolocation support to DumpRenderTree.
https://bugs.webkit.org/show_bug.cgi?id=39440
Existing LayoutTests/fast/dom/Geolocation/* should pass.
* DumpRenderTree/chromium/LayoutTestController.cpp:
(LayoutTestController::LayoutTestController):
(LayoutTestController::setGeolocationPermission):
(LayoutTestController::setMockGeolocationPosition):
(LayoutTestController::setMockGeolocationError):
* DumpRenderTree/chromium/LayoutTestController.h:
* DumpRenderTree/chromium/TestShell.cpp:
(TestShell::TestShell):
* DumpRenderTree/chromium/WebViewHost.cpp:
(WebViewHost::geolocationService):
* DumpRenderTree/chromium/WebViewHost.h:
2010-05-22 Adam Barth <abarth@webkit.org>
Unreviewed.
Re-order Yong's email addresses because his gmail account is the one he
uses for bugs.webkit.org.
* Scripts/webkitpy/common/config/committers.py:
2010-05-22 Daniel Bates <dbates@rim.com>
Reviewed by Chris Jerdonek.
Add infrastructure to parse SVN property changes
https://bugs.webkit.org/show_bug.cgi?id=38885
Adds function VCSUtils::parseSvnDiffFooter to parse an SVN footer
that consists of one or more properties.
Note, the first line of an SVN footer begins with "Property changes on".
* Scripts/VCSUtils.pm:
- Added function parseSvnDiffFooter. Will use this function
towards resolving Bug #39409 <https://bugs.webkit.org/show_bug.cgi?id=39409>.
- Removed FIXME comment above function parseSvnProperty, since
it is being used by parseSvnDiffFooter.
* Scripts/webkitperl/VCSUtils_unittest/parseSvnDiffFooter.pl: Added.
- Added unit tests.
2010-05-22 Eric Seidel <eric@webkit.org>
Unreviewed, fixing test results only.
Disable compositing tests on the commit-queue as a workaround for bug 38912
https://bugs.webkit.org/show_bug.cgi?id=39067
* Scripts/webkitpy/tool/steps/steps_unittest.py:
- Update test results after my previous change.
2010-05-22 Eric Seidel <eric@webkit.org>
Unreviewed.
Disable compositing tests on the commit-queue as a workaround for bug 38912
https://bugs.webkit.org/show_bug.cgi?id=39067
* Scripts/webkitpy/tool/steps/runtests.py:
- Disable all of compositing, not just compositing/iframes
2010-05-22 Kent Tamura <tkent@chromium.org>
Reviewed by Dimitri Glazkov.
[DRT/Chromium] Link resources and load Ahem font for Windows
https://bugs.webkit.org/show_bug.cgi?id=39473
* DumpRenderTree/chromium/DumpRenderTree.cpp:
(main): Call platformInit().
* DumpRenderTree/chromium/TestShell.h:
Declare platformInit(). It is not related to TestShell class, but the
implementation of paltformInit() is placed at TestShell*.{cpp,mm}.
* DumpRenderTree/chromium/TestShellGtk.cpp:
(platformInit):
* DumpRenderTree/chromium/TestShellMac.mm:
(platformInit):
* DumpRenderTree/chromium/TestShellWin.cpp:
(platformInit):
- Make stdout/stderr binary mode
- Load Ahem font
2010-05-21 Eric Seidel <eric@webkit.org>
Unreviewed, EWS build fix only.
QueueStatusServer returns 500 error when EWS bots post empty queues
https://bugs.webkit.org/show_bug.cgi?id=39523
Mac python seems to have some built-in timezone support
however other python installs don't. So we need to ignore
timezones in our parsing.
Date parsing is tested by existing unit tests.
* QueueStatusServer/handlers/updateworkitems.py:
- Fix typo causing exception on server.
* Scripts/webkitpy/common/net/bugzilla.py:
- Fix exception due to python's lack of timezone support.
2010-05-21 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
QueueStatusServer returns 500 error when EWS bots post empty queues
https://bugs.webkit.org/show_bug.cgi?id=39523
updateworkitems handler was raising an exception because
int() couldn't convert "" to a number.
I attempted to unit test this but we don't yet have a system by
which to load unit tests for appengine classes which depend on
google.appengine libraries which are not in the python default install.
We'll need to write a wrapper script to load those into the python path
and then run the unit test files.
* QueueStatusServer/handlers/statusbubble.py:
- Hide cr-win-ews since we're not currently running this bot.
* QueueStatusServer/handlers/updateworkitems.py:
- Fix the parsing logic to be able to understand "".
* Scripts/webkitpy/common/net/statusserver.py:
- Only log the work items posted to the server to the debug log channel.
2010-05-21 Eric Seidel <eric@webkit.org>
Unreviewed, fixing the commit-queue to run again.
Make the EWSes report queue position in white bubbles
https://bugs.webkit.org/show_bug.cgi?id=39519
* Scripts/webkitpy/common/net/statusserver.py:
- Fix exception in _post_work_items_to_server when passed
integers. Unfortunately we have no good way to mock
the Browser object yet, and after several attempts I was
not able to create a good one, so no tests. :(
2010-05-14 Ojan Vafai <ojan@chromium.org>
Reviewed by Eric Seidel.
webkit-patch land --squash commits too much if branch is not up to date
https://bugs.webkit.org/show_bug.cgi?id=38852
* Scripts/webkitpy/common/checkout/scm.py:
* Scripts/webkitpy/common/checkout/scm_unittest.py:
2010-05-21 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
Make the EWSes report queue position in white bubbles
https://bugs.webkit.org/show_bug.cgi?id=39519
This also fixes sorting of commit-queue patches
to be in order of patch attachment.
https://bugs.webkit.org/show_bug.cgi?id=33395
This makes the various Queues post what patches they are about to process
so that we can display a list of patches on status server pages, as well
as report queue position in status bubbles.
This is the first step towards creating a control-channel for the queues.
Next step will be to have them read back the patches in order from the server
and finally we will add the ability for the server to control that order.
* Scripts/webkitpy/common/net/bugzilla.py:
- Teach bugzilla how to parse attach_date for attachments.
* Scripts/webkitpy/common/net/bugzilla_unittest.py:
- Test that we're parsing dates correctly.
This may have timezone issues for non-PST contributers, unsure.
* Scripts/webkitpy/common/net/statusserver.py:
- Post work items to the status server for display.
* Scripts/webkitpy/tool/bot/patchcollection.py:
- Call StatusServer.update_work_items
* Scripts/webkitpy/tool/commands/earlywarningsystem_unittest.py:
- Update unit test results now that we're posting work item list.
* Scripts/webkitpy/tool/commands/queues.py:
- Call StatusSever.update_work_items
- Sort patches so that the server's list understands
that the commit-queue gives priority to rollout patches.
- I also fixed patch sorting per bug 33395 while I was here.
* Scripts/webkitpy/tool/commands/queues_unittest.py:
- Update results after update_work_items changes.
- Test attachment sorting.
* Scripts/webkitpy/tool/mocktool.py:
- Add mock for update_work_items
2010-05-21 Robin Cao <robin.cao@torchmobile.com.cn>
Reviewed by Adam Roben.
fast/dom/HTMLObjectElement/children-changed.html times out on Windows run-webkit-tests
https://bugs.webkit.org/show_bug.cgi?id=31315
* DumpRenderTree/win/FrameLoadDelegate.cpp:
(FrameLoadDelegate::didFailProvisionalLoadWithError): Need to invoke locationChangeDone here as mac port does.
2010-05-20 Mark Rowe <mrowe@apple.com>
Rubber-stamped by Dan Bernstein.
<rdar://problem/7848154> Remove the dependency on Foundation's private __COCOA_FORMAL_PROTOCOLS_2__ define.
* DumpRenderTree/mac/LayoutTestControllerMac.mm:
2010-05-20 Tony Gentilcore <tonyg@chromium.org>
Reviewed by Daniel Bates.
Look in /proc/registry64 for the Platform SDK on 64-bit Windows.
https://bugs.webkit.org/show_bug.cgi?id=39296
The build-webkit script failed for me on Vista 64. A web search turned
up this blog post with a patch that worked for me:
http://www.nicholaswilson.me.uk/2010/04/hacking-webkit-fail/
* Scripts/webkitdirs.pm:
2010-05-20 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
ThreadedMessageQueue should use with_statement for exception safety
https://bugs.webkit.org/show_bug.cgi?id=39233
* Scripts/webkitpy/common/thread/threadedmessagequeue.py:
2010-05-20 Diego Gonzalez <diegohcg@webkit.org>
Reviewed by Kenneth Rohde Christiansen.
[Qt] QtTestBrowser does not support websites which requires HTTP Authentication via dialogs
https://bugs.webkit.org/show_bug.cgi?id=38456
* QtTestBrowser/webpage.cpp:
(WebPage::WebPage):
(WebPage::authenticationRequired):
* QtTestBrowser/webpage.h:
2010-05-20 Martin Robinson <mrobinson@igalia.com>
Reviewed by Eric Seidel.
editingBehavior settings needs to be set back to a reasonable default between tests
https://bugs.webkit.org/show_bug.cgi?id=39433
For now, hard code the default setting during reset, so that the serialized
version of the setting stays in sync with expectations.
* DumpRenderTree/mac/DumpRenderTree.mm:
(resetDefaultsToConsistentValues): Reset editing behavior to the appropriate platform default.
* DumpRenderTree/win/DumpRenderTree.cpp:
(resetDefaultsToConsistentValues): Ditto.
2010-05-20 Brent Fulgham <bfulgham@webkit.org>
Build fix. No review.
The WebKitAPITest targets do not use the "_debug" suffix needed
by the WinCairo port. Added Debug_Cairo target to correct this.
* WebKitAPITest/WebKitAPITest.vcproj:
2010-05-20 Martin Robinson <mrobinson@igalia.com>
Reviewed by Ojan Vafai.
editing/selection/extend-selection-after-double-click.html crashes on the Leopard Intel release bot
https://bugs.webkit.org/show_bug.cgi?id=39431
* DumpRenderTree/mac/LayoutTestControllerMac.mm:
(LayoutTestController::setEditingBehavior):
Prevent a double-free by not having this variable be in the auto-release pool.
2010-05-20 Martin Robinson <mrobinson@webkit.org>
Reviewed by Ojan Vafai.
Expose the editing behavior setting in DRT to test all editing code paths
https://bugs.webkit.org/show_bug.cgi?id=38603
* DumpRenderTree/LayoutTestController.cpp:
(setEditingBehaviorCallback): Added.
(LayoutTestController::staticFunctions): Expose the setEditingBehaviorCallback function.
* DumpRenderTree/LayoutTestController.h:
* DumpRenderTree/chromium/LayoutTestController.cpp: Add callback method for setting editing behavior.
* DumpRenderTree/chromium/LayoutTestController.h: Declaration for this method.
* DumpRenderTree/gtk/LayoutTestControllerGtk.cpp:
(LayoutTestController::setEditingBehavior): Implementation of editing behavior control.
* DumpRenderTree/mac/LayoutTestControllerMac.mm:
(LayoutTestController::setEditingBehavior): Ditto
* DumpRenderTree/qt/LayoutTestControllerQt.cpp:
(LayoutTestController::setEditingBehavior): Added stub implementation of editing behavior control.
* DumpRenderTree/qt/LayoutTestControllerQt.h:
(LayoutTestController::setEditingBehavior): Add slot for controlling editor behavior.
* DumpRenderTree/win/LayoutTestControllerWin.cpp:
(LayoutTestController::setEditingBehavior): Implementation of editing behavior control.
* DumpRenderTree/wx/LayoutTestControllerWx.cpp:
(LayoutTestController::setEditingBehavior): Added stub implementation of editing behavior control.
2010-05-20 Kent Tamura <tkent@chromium.org>
Reviewed by Eric Seidel.
[DRT/Chromium] Increase the time out value
https://bugs.webkit.org/show_bug.cgi?id=39203
Change the time out value of Chromium DRT to 30 seconds, which is
the same as other ports.
If a DRT process exits before new-run-webkit-tests detects time
out, new-run-webkit-tests assumes the DRT process crashed.
* DumpRenderTree/chromium/TestShell.cpp:
(TestShell::layoutTestTimeout):
Change the time out value from 10 seconds to 30 seconds.
2010-05-20 Chris Evans <cevans@google.com>
Unreviewed.
Marking myself as a committer.
* Scripts/webkitpy/common/config/committers.py: Add cevans@google.com.
2010-05-20 Simon Hausmann <simon.hausmann@nokia.com>
Reviewed by Tor Arne Vestbo.
[Qt] Weekly binary builds on Mac OS X don't work when launched in the Finder
https://bugs.webkit.org/show_bug.cgi?id=37273
* QtTestBrowser/QtTestBrowser.pro: Build QtLauncher as bundle in package builds
2010-05-20 Fumitoshi Ukai <ukai@chromium.org>
Unreviewed fix for websocket test failures.
* Scripts/new-run-webkit-websocketserver:
options is named parameter for factory.get().
2010-05-20 Fumitoshi Ukai <ukai@chromium.org>
Reviewed by Shinichiro Hamaji.
Chromium: Add --chromium option to new-run-webkit-websocketserver
https://bugs.webkit.org/show_bug.cgi?id=37664
Missed to pass options to factory.get() in r59595
* Scripts/new-run-webkit-websocketserver:
Pass options to factory.get().
2010-05-19 Eric Seidel <eric@webkit.org>
Reviewed by Adam Roben.
WinEWS should build Debug instead of Release
https://bugs.webkit.org/show_bug.cgi?id=39242
This is a workaround for
https://bugs.webkit.org/show_bug.cgi?id=39197
Adam Roben and Brian Weinstein believe this may
also make building faster since Debug builds
take less time to link.
* Scripts/webkitpy/tool/commands/earlywarningsystem.py:
2010-05-19 Dirk Pranke <dpranke@chromium.org>
Unreviewed, build fix.
* DumpRenderTree/chromium/NotificationPresenter.cpp:
(NotificationPresenter::show):
2010-05-19 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
user.py throws exception when readline module is missing
https://bugs.webkit.org/show_bug.cgi?id=39239
* Scripts/webkitpy/common/system/user.py:
- The error handling path requires the "sys" module,
so added an import sys at the top of the file.
2010-05-18 Kent Tamura <tkent@chromium.org>
Reviewed by Dimitri Glazkov.
[DRT/Chromium] Fix a repaint issue and textarea tests
https://bugs.webkit.org/show_bug.cgi?id=39054
* DumpRenderTree/chromium/WebViewHost.cpp:
(WebViewHost::canvas): Remove m_paintRect initialization in canvas().
This line updated m_paintRect unexpectedly during paintRect().
We don't need to initialize m_paintRect because show() does it.
2010-05-18 Tony Chang <tony@chromium.org>
Reviewed by Kent Tamura.
[chromium] new-run-webkit-tests --use-drt should run on Linux
https://bugs.webkit.org/show_bug.cgi?id=37845
* Scripts/webkitpy/layout_tests/port/chromium.py:
* Scripts/webkitpy/layout_tests/port/chromium_linux.py:
* Scripts/webkitpy/layout_tests/port/chromium_mac.py:
* Scripts/webkitpy/layout_tests/port/chromium_unittest.py:
* Scripts/webkitpy/layout_tests/port/chromium_win.py:
2010-05-18 Kent Tamura <tkent@chromium.org>
Reviewed by Dimitri Glazkov.
[DRT/Chromium] Fix some initialization/reset issues
https://bugs.webkit.org/show_bug.cgi?id=39281
* DumpRenderTree/chromium/LayoutTestController.cpp:
(LayoutTestController::reset): Reset m_userStyleSheetLocation.
* DumpRenderTree/chromium/LayoutTestController.h:
Remove unused variable, m_workQueueFrozen.
(LayoutTestController::WorkQueue::WorkQueue): Initialize m_frozen.
* DumpRenderTree/chromium/TestShell.cpp:
(TestShell::resetTestController): Reset WebSettings too.
2010-05-18 Dirk Pranke <dpranke@chromium.org>
Reviewed by Ojan Vafai.
new-run-webkit-tests: implement '--reset-results' flag to complement
the '--new-baseline' flag. '--new-baseline' will always write the
results into the platform directory; '--reset-results' will update the
existing baseline wherever it happens to be. Both sets of behavior
are useful in different circumstances.
https://bugs.webkit.org/show_bug.cgi?id=38879
* Scripts/webkitpy/layout_tests/data/image/canvas-bg.html: Added.
* Scripts/webkitpy/layout_tests/data/image/canvas-zoom-expected.checksum: Added.
* Scripts/webkitpy/layout_tests/data/image/canvas-zoom-expected.png: Added.
* Scripts/webkitpy/layout_tests/data/image/canvas-zoom-expected.txt: Added.
* Scripts/webkitpy/layout_tests/data/image/canvas-zoom.html: Added.
* Scripts/webkitpy/layout_tests/data/misc/crash-expected.txt: Added.
* Scripts/webkitpy/layout_tests/data/misc/crash.html: Added.
* Scripts/webkitpy/layout_tests/data/misc/missing-expectation.html: Added.
* Scripts/webkitpy/layout_tests/data/misc/passing-expected.txt: Added.
* Scripts/webkitpy/layout_tests/data/misc/passing.html: Added.
* Scripts/webkitpy/layout_tests/data/platform/test/image/canvas-bg-expected.checksum: Added.
* Scripts/webkitpy/layout_tests/data/platform/test/image/canvas-bg-expected.png: Added.
* Scripts/webkitpy/layout_tests/data/platform/test/image/canvas-bg-expected.txt: Added.
* Scripts/webkitpy/layout_tests/data/platform/test/test_expectations.txt: Added.
* Scripts/webkitpy/layout_tests/data/text/article-element-expected.txt: Added.
* Scripts/webkitpy/layout_tests/data/text/article-element.html: Added.
* Scripts/webkitpy/layout_tests/layout_package/test_expectations_unittest.py:
* Scripts/webkitpy/layout_tests/port/test.py:
* Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
* Scripts/webkitpy/layout_tests/layout_package/dump_render_tree_thread.py:
* Scripts/webkitpy/layout_tests/layout_package/printing_unittest.py:
* Scripts/webkitpy/layout_tests/run_webkit_tests.py:
* Scripts/webkitpy/layout_tests/test_types/image_diff.py:
* Scripts/webkitpy/layout_tests/test_types/test_type_base.py:
* Scripts/webkitpy/layout_tests/test_types/text_diff.py:
* Scripts/webkitpy/layout_tests/port/test.py:
* Scripts/webkitpy/layout_tests/port/dryrun.py:
2010-05-18 Eric Seidel <eric@webkit.org>
Reviewed by Adam Roben.
Add an --html5-parser option to DumpRenderTree to allow testing the new HTML5Lexer
https://bugs.webkit.org/show_bug.cgi?id=39311
This flag allows us to run the new HTML5Lexer code.
Right now all documents parse as empty documents, but
now that we're able to run the code we can fix that.
Once we're able to lex a few basic documents I'll add
an --html5-parser flag to run-webkit-tests so that we test
running all of the layout tests with the HTML5 parser.
* DumpRenderTree/mac/DumpRenderTree.mm:
(resetDefaultsToConsistentValues):
(initializeGlobalsFromCommandLineOptions):
2010-05-08 Robert Hogan <robert@roberthogan.net>
Reviewed by Simon Hausmann.
[Qt] Fix http/tests/xmlhttprequest/cross-origin-no-authorization.html
and http/tests/xmlhttprequest/cross-origin-authorization.html
QHttpNetworkRequest adds Authorization and Cookie headers to XHRs
without knowing if this is valid behaviour or not. In order to allow
Qt to decide whether Cookie/Authorization headers should be added
to an XHR QtWebKit needs to use an attribute added to QNetworkRequest.
These new attributes are: QNetworkRequest::CookieLoadControlAttribute,
QNetworkRequest::CookieSaveControlAttribute,and
QNetworkRequest::AuthenticationReuseControlAttribute.
In order to properly support the tests, Qt's DRT needs to use one
NetworkAccessManager for all pages. This allows it to use cached
credentials where appropriate.
The tests now pass when run individually but there seems to be a problem with
leaking the results of requests across tests when run with the others in
http/tests. This will be addressed in a separate patch.
https://bugs.webkit.org/show_bug.cgi?id=32967
* DumpRenderTree/qt/DumpRenderTreeQt.cpp:
(WebCore::WebPage::WebPage):
(WebCore::DumpRenderTree::DumpRenderTree):
* DumpRenderTree/qt/DumpRenderTreeQt.h:
2010-05-18 Fumitoshi Ukai <ukai@chromium.org>
Reviewed by Eric Seidel.
run_webkit_tests_unittest fails on SnowLeopard
https://bugs.webkit.org/show_bug.cgi?id=39279
* Scripts/webkitpy/layout_tests/port/base.py:
Return copy of os.environ.
* Scripts/webkitpy/layout_tests/port/chromium_win_unittest.py:
Check os.environ was not modified.
2010-05-18 Fumitoshi Ukai <ukai@chromium.org>
Reviewed by Eric Seidel.
Chromium: new-run-webkit-httpd fails to setup_mount
https://bugs.webkit.org/show_bug.cgi?id=39257
* Scripts/webkitpy/common/system/executive.py:
Assert type of args in run_command.
* Scripts/webkitpy/common/system/executive_unittest.py:
Add test_run_command_args_type
* Scripts/webkitpy/layout_tests/port/chromium_win.py:
Executive.run_command takes array for command line.
* Scripts/webkitpy/layout_tests/port/chromium_win_unittest.py:
Test if setup_environ_for_server() run setup_mount.bat.
2010-05-17 Kent Tamura <tkent@chromium.org>
Reviewed by David Levin.
Chromium Windows build system does not rebuild correctly when
enabling/disabling a feature
https://bugs.webkit.org/show_bug.cgi?id=38926
Add a workaround of this issue.
* Scripts/update-webkit:
Chromium-Windows only: If WebKit/chromium/features.gyp has been
updated, remove WebKit/chromium/Debug and WebKit/chromium/Release.
2010-05-17 Sheriff Bot <webkit.review.bot@gmail.com>
Unreviewed, rolling out r59631.
http://trac.webkit.org/changeset/59631
https://bugs.webkit.org/show_bug.cgi?id=39255
chromium canaries can no longer run webkit_tests, suspect this
change. (Requested by atwilson on #webkit).
* Scripts/webkitpy/layout_tests/layout_package/dump_render_tree_thread.py:
* Scripts/webkitpy/layout_tests/run_webkit_tests.py:
* Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
2010-05-17 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
Attempt to make new-run-webkit-tests --help more sane
https://bugs.webkit.org/show_bug.cgi?id=37836
* Scripts/webkitpy/layout_tests/layout_package/dump_render_tree_thread.py:
- Add a FIXME about options.singly and options.batch_size being different.
* Scripts/webkitpy/layout_tests/run_webkit_tests.py:
- Add support for hidden options.
- Add option groupings to attempt to simplify --help.
- Fix a bunch of option helps to start with a capitalized verb.
- Hide a bunch of options which make no sense to users.
- Sort options in --help.
* Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
- Add tests for option sorting.
2010-05-17 Robert Hogan <robert@webkit.org>
Reviewed by Kenneth Rohde Christiansen.
[Qt] Disable Icon Database by default in Qt DRT
Unskip:
http/tests/misc/favicon-loads-with-images-disabled.html
http/tests/security/XFrameOptions/x-frame-options-deny-meta-tag-in-body.html
http/tests/security/XFrameOptions/x-frame-options-deny-meta-tag-parent-same-origin-deny.html
http/tests/security/XFrameOptions/x-frame-options-deny-meta-tag.html
https://bugs.webkit.org/show_bug.cgi?id=37382
Add support for layoutTestController.setIconDatabaseEnabled and layoutTestController.disableImageLoading().
The XFrameOptions tests were failing because of an extra resource load callback for favicon.ico requests.
These extra callbacks are removed by supporting both of the above layoutTestContoller commands.
* DumpRenderTree/qt/DumpRenderTreeQt.cpp:
(WebCore::WebPage::resetSettings):
(WebCore::DumpRenderTree::DumpRenderTree):
* DumpRenderTree/qt/DumpRenderTreeQt.h:
(WebCore::DumpRenderTree::drtStoragePath):
* DumpRenderTree/qt/LayoutTestControllerQt.cpp:
(LayoutTestController::reset):
(LayoutTestController::disableImageLoading):
(LayoutTestController::setIconDatabaseEnabled):
* DumpRenderTree/qt/LayoutTestControllerQt.h:
2010-05-17 Fumitoshi Ukai <ukai@chromium.org>
Reviewed by Eric Seidel.
Chromium: Add --chromium option to new-run-webkit-websocketserver
https://bugs.webkit.org/show_bug.cgi?id=37664
os.environ setup and setup_mount for cygwin are moved in ChromiumWinPort.setup_environ_for_server.
* Scripts/new-run-webkit-httpd:
Remove passing register_cygwin.
* Scripts/new-run-webkit-websocketserver:
Add --chromium flag.
Remove passing register_cygwin.
Create port object using options.
* Scripts/webkitpy/layout_tests/port/base.py:
Add setup_environ_for_server().
* Scripts/webkitpy/layout_tests/port/chromium_win.py:
Ditto.
* Scripts/webkitpy/layout_tests/port/factory_unittest.py: Added.
* Scripts/webkitpy/layout_tests/port/http_server.py:
Remove register_cygwin_parameter.
Call setup_environ_for_server().
* Scripts/webkitpy/layout_tests/port/websocket_server.py:
Ditto.
2010-05-16 Daniel Bates <dbates@rim.com>
Reviewed by Chris Jerdonek.
https://bugs.webkit.org/show_bug.cgi?id=39184
Adds function VCSUtils::parseSvnProperty to parse an SVN property with
either a single-line or multi-line value change.
* Scripts/VCSUtils.pm:
- Added function parseSvnProperty. We will use this function
towards resolving Bug #38885 <https://bugs.webkit.org/show_bug.cgi?id=38885>.
- Removed FIXME comment above function parseSvnPropertyValue, since
it is being used by parseSvnProperty.
- Modified function parseSvnPropertyValue to break out of "while (<$fileHandle>)"
loop when it encounters the start of the next property so that it can be
processed by its caller, parseSvnPropertyValue. We reference this bullet below
by (*).
* Scripts/webkitperl/VCSUtils_unittest/parseSvnProperty.pl: Added.
- Added unit tests.
* Scripts/webkitperl/VCSUtils_unittest/parseSvnPropertyValue.pl:
- Changed the name of the unit test "simple multi-line '-' change" to
"single-line '-' change followed by empty line" since the former was an
incorrect description of this test.
- Added unit test "single-line '-' change followed by the next property", and
"multi-line '-' change followed by the next property" to test (*) above.
2010-05-16 Tony Chang <tony@chromium.org>
Not reviewed, fixing layout test.
Don't output Inspect Element since this is not enabled on the bots,
but most developers probably have it installed.
* DumpRenderTree/mac/EventSendingController.mm:
(-[EventSendingController contextClick:]):
2010-05-12 Tony Chang <tony@chromium.org>
Reviewed by Darin Adler.
Spellcheck disabling does not disable context menu
https://bugs.webkit.org/show_bug.cgi?id=25639
* DumpRenderTree/mac/EventSendingController.mm:
(+[EventSendingController isSelectorExcludedFromWebScript:]):
(+[EventSendingController webScriptNameForSelector:]):
(-[EventSendingController contextClick:]): add a bool parameter that
when true, dumps the context menu items to stdout.
2010-05-16 Robert Hogan <robert@webkit.org>
Reviewed by Kenneth Rohde Christiansen.
[Qt] Unskip fast/loader/main-document-url-for-non-http-loads.html
Update Qt DRT to use frame loader, editor client and notification presenter
functions in DumpRenderTreeSupportQt.
https://bugs.webkit.org/show_bug.cgi?id=38867
* DumpRenderTree/qt/DumpRenderTreeQt.cpp:
(WebCore::DumpRenderTree::open):
(WebCore::DumpRenderTree::dump):
* DumpRenderTree/qt/LayoutTestControllerQt.cpp:
(LayoutTestController::LayoutTestController):
(LayoutTestController::reset):
(LayoutTestController::dumpEditingCallbacks):
(LayoutTestController::dumpFrameLoadCallbacks):
(LayoutTestController::dumpResourceLoadCallbacks):
(LayoutTestController::setWillSendRequestReturnsNullOnRedirect):
(LayoutTestController::setWillSendRequestReturnsNull):
(LayoutTestController::setWillSendRequestClearHeader):
2010-05-16 Eric Seidel <eric@webkit.org>
Unreviewed, rolling out r59571.
http://trac.webkit.org/changeset/59571
https://bugs.webkit.org/show_bug.cgi?id=39054
Broke Cr Win, but we didn't notice immediately due to
https://bugs.webkit.org/show_bug.cgi?id=38926. It's possible
that this didn't actually break Cr Win, but rather that bug
38926 necessitates a clean compile after this and sucessive
checkins only produced a partial recompile and thus failed to
build.
* DumpRenderTree/chromium/WebViewHost.cpp:
(WebViewHost::canvas):
2010-05-16 Robert Hogan <robert@webkit.org>
Reviewed by Kenneth Rohde Christiansen.
[Qt]Unskip security/set-form-autocomplete-attribute.html
Add support for layoutTestController.elementDoesAutoCompleteForElementWithId().
https://bugs.webkit.org/show_bug.cgi?id=38859
* DumpRenderTree/qt/LayoutTestControllerQt.cpp:
(LayoutTestController::elementDoesAutoCompleteForElementWithId):
* DumpRenderTree/qt/LayoutTestControllerQt.h:
2010-05-16 Kent Tamura <tkent@chromium.org>
Reviewed by Dimitri Glazkov.
[DRT/Chromium] Fix repaint, WebGL, textarea tests
https://bugs.webkit.org/show_bug.cgi?id=39054
* DumpRenderTree/chromium/WebViewHost.cpp:
(WebViewHost::canvas): Remove m_paintRect initialization in canvas().
This line updated m_paintRect unexpectedly during paintRect().
We don't need to initialize m_paintRect because show() does it.
2010-05-16 Sergio Villar Senin <svillar@igalia.com>
Reviewed by Gustavo Noronha Silva.
[GTK] Improve reporting of frame loader callbacks in DRT
https://bugs.webkit.org/show_bug.cgi?id=36454
* DumpRenderTree/gtk/DumpRenderTree.cpp:
(webViewLoadStatusNotified):
(createWebView): added connection to notify::load-status and
signal callback
2010-05-15 Daniel Bates <dbates@rim.com>
Reviewed by Chris Jerdonek.
https://bugs.webkit.org/show_bug.cgi?id=39170
Add function parseSvnPropertyValue to parse single-line and multi-line
property values of an SVN property change.
* Scripts/VCSUtils.pm:
Added function parseSvnPropertyValue. We will use this as part of
Bug #38885 <https://bugs.webkit.org/show_bug.cgi?id=38885>.
* Scripts/webkitperl/VCSUtils_unittest/parseSvnPropertyValue.pl: Added.
2010-05-15 Jochen Eisinger <jochen@chromium.org>
Reviewed by Dmitry Titov.
Add allowDatabase method to TestWebWorker.
https://bugs.webkit.org/show_bug.cgi?id=38742
* DumpRenderTree/chromium/TestWebWorker.h:
(TestWebWorker::allowDatabase):
2010-05-15 Simon Hausmann <simon.hausmann@nokia.com>
Rubber-stamped by Antti Koivisto.
[Qt] Rename QtLauncher to QtTestBrowser
https://bugs.webkit.org/show_bug.cgi?id=37665
Forgot to remove the original directory after the rename.
* QtLauncher: Removed.
2010-05-15 Sheriff Bot <webkit.review.bot@gmail.com>
Unreviewed, rolling out r59544.
http://trac.webkit.org/changeset/59544
https://bugs.webkit.org/show_bug.cgi?id=39165
Cased LayoutTest to start crashing (Requested by abarth on
#webkit).
* Scripts/old-run-webkit-tests:
2010-05-15 Mario Sanchez Prada <msanchez@igalia.com>
Reviewed by Xan Lopez.
Ensure DRT loads GAIL (Gtk+ module), for a11y tests
https://bugs.webkit.org/show_bug.cgi?id=38648
Add the GTK_MODULES envvar (set to "gail") to the clean
environment when running DRT for the Gtk+ port
* Scripts/old-run-webkit-tests:
2010-05-15 Eric Seidel <eric@webkit.org>
Reviewed by Maciej Stachowiak.
Add script to check for minimum python version and install if missing on Tiger
https://bugs.webkit.org/show_bug.cgi?id=38886
Per Maciej's request on webkit-dev:
https://lists.webkit.org/pipermail/webkit-dev/2010-May/012785.html
provide a script which can automatically install Python on Tiger where
the system provided version is too old to be of use.
Note this uses the official Mac Python installer from python.org.
This installs a copy of Python in /Library/Frameworks/Python.framework.
It also makes symlinks from /usr/local/bin to the Python.framework/bin.
I have tested this script on Leopard and it worked fine. I have not
tested it on Tiger as I do not have access to a Tiger machine. In
either case this should provide a great starting point for someone
wishing to upgrade their copy of Python on Tiger.
Future patches can make our scripts depend on a success return from this
script and then they can either skip tests/sections for which python is
insufficient, or they can fail themselves.
* Scripts/ensure-valid-python: Added.
2010-05-15 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
Display queue position inside EWS bubbles
https://bugs.webkit.org/show_bug.cgi?id=38979
This ended up being a lot of clean-up to our status server code.
Added a new WorkItems model, a form with which to fill it,
and taught the Attachment class how to calculate the current queue
postion for an attachment using the data in WorkItems.
I also finally made statusbubble.* and dashboard.* not use copy-paste code.
The Attachment class has this summary() method which is very
controller/view-like and does not belong in a model class.
This patch got rid of all direct uses of summary().
* QueueStatusServer/handlers/dashboard.py:
- Build row objects to hand off to the view instead of handing off a
summary object and expecting the view to process it directly.
* QueueStatusServer/handlers/statusbubble.py:
- Build bubble object and hand them off to the view.
* QueueStatusServer/handlers/updatestatus.py:
- Code cleanup. Just move some code into _queue_status_from_request
to make the main put() handler easier to read.
* QueueStatusServer/handlers/updateworkitems.py: Added.
- Controller to handle storing WorkItems model objects.
* QueueStatusServer/main.py:
- Add route for /update-work-items
* QueueStatusServer/model/attachment.py:
- Add new methods to replace direct summary() access.
- Teach attachment how to calculate queue positions from WorkItems data.
* QueueStatusServer/model/queues.py:
- Move name_with_underscores here for easier re-use.
* QueueStatusServer/model/workitems.py: Added.
- New model for storing what items are currently queue for any bot.
* QueueStatusServer/templates/dashboard.html:
- Kill the copy/paste code!
* QueueStatusServer/templates/statusbubble.html:
- Kill the copy/paste code!
* QueueStatusServer/templates/updateworkitems.html: Added.
- Simple form for updating a queue's current work items.
2010-05-15 Joanmarie Diggs <joanmarie.diggs@gmail.com>
Reviewed by Xan Lopez.
https://bugs.webkit.org/show_bug.cgi?id=30500
[Gtk] Find a way for WebKit to "announce" itself so that ATs can readily distinguish it from true Gtk/Gail
The "announcement" is now made in the form of an object attribute
associated with the AtkObject.
* DumpRenderTree/gtk/AccessibilityUIElementGtk.cpp:
(AccessibilityUIElement::allAttributes):
(attributeSetToString):
2010-05-15 Mario Sanchez Prada <msanchez@igalia.com>
Reviewed by Xan Lopez.
Accessibility: Implement isSelected in DRT for GTK
https://bugs.webkit.org/show_bug.cgi?id=31018
Implement AccessibilityUIElement::isSelected() for Gtk
* DumpRenderTree/gtk/AccessibilityUIElementGtk.cpp:
(AccessibilityUIElement::isSelected):
2010-05-14 Adam Barth <abarth@webkit.org>
Reviewed by Eric Seidel.
scm.py should use self.run instead of run_command
https://bugs.webkit.org/show_bug.cgi?id=38957
We've wanted to do this for a while, but it's a prerequiste for running
SVN from the cwd instead of the checkout_root.
* Scripts/webkitpy/common/checkout/scm.py:
2010-05-14 Anton Muhin <antonm@chromium.org>
Reviewed by Darin Fisher.
[Chromium] Consider implementing addOriginAccessWhitelistEntry method
https://bugs.webkit.org/show_bug.cgi?id=37578
Remove deprecated methods.
* DumpRenderTree/chromium/LayoutTestController.cpp
* DumpRenderTree/chromium/LayoutTestController.h
2010-05-14 Kenneth Russell <kbr@google.com>
Reviewed by Darin Adler.
Rename WebGLArray types to TypedArray types
https://bugs.webkit.org/show_bug.cgi?id=39091
Extended functionality of do-webcore-rename script and used it to
rename the WebGLArray types to the TypedArray naming convention.
The only source files which were touched by hand, and which are
being manually reviewed, are:
WebCore/page/DOMWindow.idl
WebCore/bindings/generic/RuntimeEnabledFeatures.h (script's changes undone)
WebKit/WebCore/bindings/js/JSDOMWindowCustom.cpp
WebKit/WebCore/bindings/v8/custom/V8DOMWindowCustom.cpp
These only needed to be touched to update the aliases between the
WebGLArray and TypedArray names introduced in bug 39036. (It was
not feasible to have do-webcore-rename handle this as it would
introduce circular renamings.) These aliases will be removed in
roughly a month once existing WebGL content has been updated.
No new tests; covered under existing WebGL tests. Updated
constructed-objects-prototypes and prototype-inheritance-2 tests.
Ran all layout tests in Safari and all WebGL tests in Chromium.
* Scripts/do-webcore-rename:
Handle the case where some renames are substrings of others.
Support renaming files containing custom JS bindings. If
isDOMTypeRename is non-zero, expand the regexp which rewrites
the file's contents in order to support custom JS bindings.
2010-05-14 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
webkit-patch rollout throws exception if bug is already open
https://bugs.webkit.org/show_bug.cgi?id=38803
This was caused by someone incorrectly wrapping the code. :p
I'm going to have to start demanding unit tests for wrapping changes...
I also fixed the code to be able to reopen bugs which were never confirmed.
Before it regressed, the code would just log in that case. Now it actually
will reopen the bug, but there is a FIXME about how the logic is a bit backwards.
* Scripts/webkitpy/common/net/bugzilla.py:
* Scripts/webkitpy/common/net/bugzilla_unittest.py:
- Yay testing!
2010-05-14 Simon Hausmann <simon.hausmann@nokia.com>
Rubber-stamped by Antti Koivisto.
[Qt] Rename QtLauncher to QtTestBrowser
* QtTestBrowser: Copied from WebKitTools/QtLauncher.
* QtTestBrowser/QtLauncher.pro: Removed.
* QtTestBrowser/QtLauncher.qrc: Removed.
* QtTestBrowser/QtTestBrowser.pro: Copied from WebKitTools/QtLauncher/QtLauncher.pro.
* QtTestBrowser/QtTestBrowser.qrc: Copied from WebKitTools/QtLauncher/QtLauncher.qrc.
* Scripts/run-launcher:
* Scripts/webkitdirs.pm:
2010-05-14 Adam Barth <abarth@webkit.org>
Reviewed by Eric Seidel.
Stop CCing webkit-bot-watchers
https://bugs.webkit.org/show_bug.cgi?id=39020
webkit-bot-watchers is somewhat of a failed experiment. No one
subscribed to the list (not even me). Removing it from the code
because wms says it bounces email sometimes.
* Scripts/webkitpy/tool/commands/queues.py:
* Scripts/webkitpy/tool/commands/sheriffbot_unittest.py:
2010-05-14 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
REGRESSION: webkit-patch commit-message throws exception
https://bugs.webkit.org/show_bug.cgi?id=38997
* Scripts/webkitpy/tool/commands/upload.py:
- Fix to respect and pass the --squash and --git-commit options.
* Scripts/webkitpy/tool/commands/upload_unittest.py:
- Fix the unit test to use the central MockCheckout instead of
its own custom Mock which didn't require enough parameters.
2010-05-13 Diego Gonzalez <diegohcg@webkit.org>
Reviewed by Kenneth Rohde Christiansen.
[Qt] Improve QtLauncher user agent dialog resize
https://bugs.webkit.org/show_bug.cgi?id=39062
* QtLauncher/main.cpp:
(LauncherWindow::showUserAgentDialog):
2010-05-13 Eric Seidel <eric@webkit.org>
Reviewed by Ojan Vafai.
Disable compositing/iframes tests on the commit-queue as a workaround for bug 38912
https://bugs.webkit.org/show_bug.cgi?id=39067
Fix yet another typo in my original hack.
I also added another unit test for this fix.
* Scripts/webkitpy/common/config/ports.py:
* Scripts/webkitpy/common/config/ports_unittest.py:
2010-05-13 Eric Seidel <eric@webkit.org>
Reviewed by Ojan Vafai.
Disable compositing/iframes tests on the commit-queue as a workaround for bug 38912
https://bugs.webkit.org/show_bug.cgi?id=39067
My previous (unreviewed) hack didn't actually work due to checking "mac" instead of "Mac".
This change fixes my hack, and unit tests it.
* Scripts/webkitpy/tool/mocktool.py:
- Make it possible to make run_command log too.
* Scripts/webkitpy/tool/steps/runtests.py:
- Fix my hack to actually work.
* Scripts/webkitpy/tool/steps/steps_unittest.py:
- Test my hack (and basic RunTests behavior as well).
2010-05-13 Antonio Gomes <tonikitoo@webkit.org>, Yi Shen <yi.4.shen@nokia.com>
Reviewed by Kenneth Christiansen.
[Qt] Add LayoutTestController interface: computedStyleIncludingVisitedInfo
https://bugs.webkit.org/show_bug.cgi?id=37759
WebKitTools:
* DumpRenderTree/qt/LayoutTestControllerQt.cpp:
(LayoutTestController::computedStyleIncludingVisitedInfo):
* DumpRenderTree/qt/LayoutTestControllerQt.h:
2010-05-10 Adam Roben <aroben@apple.com>
Convert status bar text to UTF-8 before logging it on Windows
We were previously logging the text using printf("%S", bstr). This
converts the UTF-16 BSTR to a multibyte string using wctomb, which
uses the codepage for the current locale to perform the conversion.
The conversion was failing, causing printf to bail and truncate the
string. By converting to UTF-8 manually before logging, we avoid this
issue (and also end up with UTF-8 output, which is what the expected
results contain). We may have to do this in other places in DRT,
eventually.
Fixes <http://webkit.org/b/38849> REGRESSION (r59016):
plugins/set-status.html fails on Windows
Reviewed by Alexey Proskuryakov.
* DumpRenderTree/win/DumpRenderTree.cpp:
(toUTF8): Moved this here from FrameLoadDelegate. Renamed from
BSTRtoString.
* DumpRenderTree/win/DumpRenderTreeWin.h: Added declaration of toUTF8.
* DumpRenderTree/win/FrameLoadDelegate.cpp: Removed BSTRtoString.
(descriptionSuitableForTestResult): Updated for rename.
* DumpRenderTree/win/UIDelegate.cpp:
(UIDelegate::setStatusText): Convert the status bar text to UTF-8
before logging it so that Windows won't try (and fail) to convert it
to the current locale's codepage.
2010-05-13 Eric Seidel <eric@webkit.org>
Unreviewed hack, attempting to get the commit-queue running again.
Disable compositing/iframes tests on the commit-queue as a workaround for bug 38912
https://bugs.webkit.org/show_bug.cgi?id=39067
I had this hack locally on the commit-queue, but it's fragile
and broke this morning. It will work much better if committed
to the repository. In either case it's temporary while we
work up a real fix for bug 38912.
* Scripts/webkitpy/common/config/ports.py:
- Expose an is_leopard() method.
* Scripts/webkitpy/tool/steps/runtests.py:
- Don't run compositing/iframes tests on the commit-queue under leopard.
2010-05-13 Fumitoshi Ukai <ukai@chromium.org>
Unreviewed fix. Revert previous change.
Even if with-statments are changed in websocket_server.py, it also claims syntax error for with statement in http_server.py.
Until python 2.5 is installed on tiger bot, skips websocket/tests on tiger.
* Scripts/webkitpy/layout_tests/port/websocket_server.py: use with statement
2010-05-13 Fumitoshi Ukai <ukai@chromium.org>
Unreviewed fix.
On tiger bot, it claims syntax error for with statement.
* Scripts/webkitpy/layout_tests/port/websocket_server.py: rewrite with statement with try-finally.
2010-05-13 Fumitoshi Ukai <ukai@chromium.org>
Reviewed by David Levin.
WebSocket: pywebsocket 0.5
https://bugs.webkit.org/show_bug.cgi?id=38034
Remove pywebsocket from webkitpy/thirdparty.
Make pywebsocket autoinstalled.
* Scripts/new-run-webkit-websocketserver:
Add --output-dir option.
* Scripts/old-run-webkit-tests:
Use new-run-webkit-websocketserver, rather than directly run pywebsocket's standalone.py
* Scripts/run-webkit-websocketserver:
Ditto.
* Scripts/webkitpy/layout_tests/port/websocket_server.py:
Use autoinstalled pywebsocket.
* Scripts/webkitpy/thirdparty/__init__.py:
Autoinstall pywebsocket
* Scripts/webkitpy/thirdparty/pywebsocket: Removed.
2010-05-12 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Daniel Bates.
Fixed a bug in svn-apply whereby the reviewer would not get set if
the portion of the patch for the ChangeLog contains "NOBODY (**PS!)"
in the leading junk.
https://bugs.webkit.org/show_bug.cgi?id=38998
* Scripts/VCSUtils.pm:
- Added the $changeLogTimeZone variable from svn-apply.
- Added setChangeLogDateAndReviewer() from svn-apply.
- Added a localTimeInProjectTimeZone() subroutine.
- In setChangeLogDateAndReviewer():
- Added an $epochTime parameter to make the subroutine more testable.
- Made the "NOBODY (**PS!)" regular expression more specific so that
it will not apply to text in the leading junk.
- Updated to call localTimeInProjectTimeZone().
* Scripts/svn-apply:
- Removed the $changeLogTimeZone file variable.
- Added an $epochTime file variable to represent the current time.
- Removed the setChangeLogDateAndReviewer() subroutine.
- Updated patch() to use the new setChangeLogDateAndReviewer() syntax.
* Scripts/webkitperl/VCSUtils_unittest/setChangeLogDateAndReviewer.pl: Added.
- Added unit tests.
2010-05-12 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
test-webkitpy fails on windows due to lack of readline module
https://bugs.webkit.org/show_bug.cgi?id=38884
Win32 Python does not have a readline module, so we should
not exit(1) if the import fails.
Also the failure message is mac-specific and doesn't need to be.
Only print the mac-specific install instructions on mac.
* Scripts/webkitpy/common/system/user.py:
2010-05-12 Kent Tamura <tkent@chromium.org>
Reviewed by Nate Chapin.
[DRT/Chromium] Add a missing Sans-serif font setting
https://bugs.webkit.org/show_bug.cgi?id=38981
* DumpRenderTree/chromium/TestShell.cpp:
(TestShell::resetWebSettings):
Set "Helvetica" for Sans-serif. It is the same as the default setting of test_shell.
2010-05-10 Ojan Vafai <ojan@chromium.org>
Reviewed by Adam Barth.
webkit-patch upload --fancy-review now uses the bugzilla bug ID as the rietveld ID
https://bugs.webkit.org/show_bug.cgi?id=38866
wkrietveld.appspot.com has already been updated to allow you to
pass --issue for the issue creation as well as subsequent uploads.
Also, remove the extra code for trying to read out the issue ID
from the changelog description since we just use the bugzilla ID now.
* Scripts/webkitpy/common/config/__init__.py:
* Scripts/webkitpy/common/net/rietveld.py:
* Scripts/webkitpy/tool/steps/postcodereview.py:
* Scripts/webkitpy/tool/steps/postdiff.py:
2010-05-12 Ojan Vafai <ojan@chromium.org>
No review needed.
Marking myself as a reviewer.
* Scripts/webkitpy/common/config/committers.py:
2010-05-12 Dirk Pranke <dpranke@chromium.org>
Reviewed by Eric Seidel.
new-run-webkit-tests: add a --print default option so that you can
easily get the default output plus something (e.g., you can say
'--print default,config' instead of '--print misc,one-line-progress,
one-line-summary,unexpected,unexpected-results,updates,config'.
Also, add more unit tests for --verbose, --print everything, etc.
https://bugs.webkit.org/show_bug.cgi?id=38877
* Scripts/webkitpy/layout_tests/layout_package/printing.py:
* Scripts/webkitpy/layout_tests/layout_package/printing_unittest.py:
2010-05-12 Csaba Osztrogonác <ossy@webkit.org>
Reviewed by Adam Barth.
Qt build failures cause SheriffBot false positives
https://bugs.webkit.org/show_bug.cgi?id=38969
Add Qt bots back to the core builders, because bug fixed by r59261.
* Scripts/webkitpy/common/net/buildbot.py:
* Scripts/webkitpy/common/net/buildbot_unittest.py:
2010-05-12 Csaba Osztrogonác <ossy@webkit.org>
Reviewed by Kenneth Rohde Christiansen.
[Qt] Slave losts cause build break on bots
https://bugs.webkit.org/show_bug.cgi?id=38980
* Scripts/build-webkit: Remove 0 byte sized files from productDir before build.
2010-05-12 Csaba Osztrogonác <ossy@webkit.org>
Unreviewed trivial fix after r59254.
* Scripts/old-run-webkit-tests:
2010-05-12 Csaba Osztrogonác <ossy@webkit.org>
Reviewed by Eric Seidel.
run-webkit-tests --exit-after-N-failures should not count new tests as failures
https://bugs.webkit.org/show_bug.cgi?id=31829
* Scripts/old-run-webkit-tests:
2010-05-12 James Robinson <jamesr@chromium.org>
Reviewed by Simon Fraser.
Disable smooth scrolling on OS X when running tests
https://bugs.webkit.org/show_bug.cgi?id=38964
Some tests (like fast/repaint/fixed-move-after-keyboard-scroll.html)
depend on smooth scrolling behavior. Since this is off by default
in Leopard and on by default in Snow Leopard, DRT should turn it
off always to ensure a consistent test environment.
* DumpRenderTree/mac/DumpRenderTree.mm:
(resetDefaultsToConsistentValues):
2010-05-10 Rodrigo Belem <rodrigo.belem@openbossa.org>
Reviewed by Kenneth Christiansen , Simon Hausmann and Gustavo Noronha.
[Qt, Gtk] Allows build-webkit script to receive an install prefix as parameter
https://bugs.webkit.org/show_bug.cgi?id=26224
Added more parameters to build-webkit script, the --prefix for gkt
and --install-libs, --install-headers for qt. Now it is possible
to change the install prefix for gtk and install path for qt.
* Scripts/build-webkit:
* Scripts/webkitdirs.pm:
2010-05-12 Philippe Normand <pnormand@igalia.com>
Unreviewed, added my IRC nickname.
* Scripts/webkitpy/common/config/committers.py:
2010-05-11 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Eric Seidel.
Fixed two FIXME's in svn-apply: eliminated the unnecessary %copiedFiles
variable and changed gitKnowsOfFile() to use exitStatus().
https://bugs.webkit.org/show_bug.cgi?id=38862
* Scripts/svn-apply:
2010-05-11 Eric Seidel <eric@webkit.org>
Unreviewed, just fixing python typo.
EWS bots should poll more often than every 5 minutes
https://bugs.webkit.org/show_bug.cgi?id=38968
Typo in _now(), add a unittest to prove I fixed it.
* Scripts/webkitpy/tool/bot/queueengine.py:
* Scripts/webkitpy/tool/bot/queueengine_unittest.py:
2010-05-11 Adam Barth <abarth@webkit.org>
Reviewed by Eric Seidel.
Qt build failures cause SheriffBot false positives
https://bugs.webkit.org/show_bug.cgi?id=38969
The Qt buildbot randomly fails to compile occasionally because its
network connection causes SVn to leave zero-byte files around. These
compile failures confuse SheriffBot into thinking someone's patch
caused a build break.
In this patch, I've temporarily removed Qt from the list of core
builders. Ossy is working on a script to clean up the zero byte files.
Once that goes in, we can add Qt back to the core builders.
* Scripts/webkitpy/common/net/buildbot.py:
* Scripts/webkitpy/common/net/buildbot_unittest.py:
2010-05-11 Adam Barth <abarth@webkit.org>
Reviewed by Eric Seidel.
sheriffbot is spammy.
https://bugs.webkit.org/show_bug.cgi?id=38936
Reduce sheriffbot spam by not warning about new blameworthy revisions
that can be explained by previously blamed revisions. This might cause
us to not warn about some real failures, but we're getting too much
spam from slow builders that have large blamelists (and we've already
poked the responsible folks using data from a fast builder).
* Scripts/webkitpy/tool/bot/sheriff_unittest.py:
- Removed unneeded import
* Scripts/webkitpy/tool/commands/sheriffbot.py:
- The logic change
* Scripts/webkitpy/tool/commands/sheriffbot_unittest.py:
- Tests of the change
* Scripts/webkitpy/tool/mocktool.py:
- Merged two declarations of MockBuilder
2010-05-11 Tony Chang <tony@chromium.org>
Reviewed by David Levin.
[chromium] detect num processors to pass to make -j on Linux
https://bugs.webkit.org/show_bug.cgi?id=38833
* Scripts/webkitdirs.pm:
2010-05-11 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
EWS bots should poll more often than every 5 minutes
https://bugs.webkit.org/show_bug.cgi?id=38968
We'll make them poll every 2 minutes to start with.
I'm going to re-write how polling works soon, so this is
a stop-gap to try and make the bots more responsive.
If Bill notices any additional load on bugzilla we'll drop
the polling frequency back to 5 minutes. He's historically said
that the EWS bots appear to be a drop in the bucket and thus
should be able to poll much more frequently w/o causing trouble.
* Scripts/webkitpy/tool/bot/queueengine.py:
* Scripts/webkitpy/tool/bot/queueengine_unittest.py:
- Add a test for sleep_message
2010-05-11 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
commit-queue should commit rollouts without running tests
https://bugs.webkit.org/show_bug.cgi?id=38940
Most of this change is improving our test coverage for the commit-queue.
The only functional change is removing the --test flag when the commit-queue
is running in rollouts mode.
I added test coverage for status updates, and updated the commit-queue status
messages to distinguish rollout vs. normal landing mode in its empty queue
and land patch messages.
* Scripts/webkitpy/tool/commands/earlywarningsystem_unittest.py:
- Got rid of a bunch of copy/paste code using _default_expected_stderr
* Scripts/webkitpy/tool/commands/queues.py:
- Moved rollout patch filtering out of _validate_patches_in_commit_queue
so that we only have to check if the builders are green in one place.
- Make the "empty queue" message note which queue it is referring to.
- Don't pass --text to land-attachment when in rollout mode.
* Scripts/webkitpy/tool/commands/queues_unittest.py:
- Update results now that we're testing status updates.
- Test _can_build_and_test since I made a typo in that call while
writing this change and the unit tests failed to catch it!
* Scripts/webkitpy/tool/mocktool.py:
- Log status updates to make them testable in our unit tests.
2010-05-11 Kent Tamura <tkent@chromium.org>
Reviewed by Dimitri Glazkov.
[DRT/Chromium] Apply recent changes of test_shell to DRT
https://bugs.webkit.org/show_bug.cgi?id=38895
Port the changes to test_shell during (r40492, r46810] of Chromium.
Highlights:
- Introduce NotificationPresenter
- Fix parameter mismatch of WebViewClient::startDragging()
This change fixes dozens of unexpected behaviors.
* DumpRenderTree/DumpRenderTree.gypi:
Add NotificationPresenter.{cpp,h}
* DumpRenderTree/chromium/EventSender.cpp:
(applyKeyModifier):
(EventSender::EventSender):
(EventSender::reset):
(EventSender::webview):
(EventSender::doDragDrop):
(EventSender::keyDown):
(EventSender::addTouchPoint):
(EventSender::clearTouchPoints):
(EventSender::releaseTouchPoint):
(EventSender::setTouchModifier):
(EventSender::updateTouchPoint):
(EventSender::cancelTouchPoint):
(EventSender::sendCurrentTouchEvent):
(EventSender::touchEnd):
(EventSender::touchMove):
(EventSender::touchStart):
(EventSender::touchCancel):
* DumpRenderTree/chromium/EventSender.h:
* DumpRenderTree/chromium/LayoutTestController.cpp:
(LayoutTestController::LayoutTestController):
(LayoutTestController::reset):
(LayoutTestController::showWebInspector):
(LayoutTestController::setAuthorAndUserStylesEnabled):
(LayoutTestController::setScrollbarPolicy):
(LayoutTestController::setWillSendRequestClearHeader):
(LayoutTestController::callShouldCloseOnWebView):
(LayoutTestController::grantDesktopNotificationPermission):
(LayoutTestController::removeOriginAccessWhitelistEntry):
(LayoutTestController::addUserScript):
(LayoutTestController::addUserStyleSheet):
* DumpRenderTree/chromium/LayoutTestController.h:
* DumpRenderTree/chromium/NotificationPresenter.cpp: Added.
* DumpRenderTree/chromium/NotificationPresenter.h: Added.
* DumpRenderTree/chromium/TestShell.cpp:
(TestShell::TestShell):
(TestShell::runFileTest):
(TestShell::resetTestController):
* DumpRenderTree/chromium/TestShell.h:
(TestShell::eventSender):
(TestShell::notificationPresenter):
(TestShell::showDevTools):
* DumpRenderTree/chromium/WebViewHost.cpp:
(WebViewHost::createView):
(WebViewHost::createPopupMenu):
(WebViewHost::startDragging):
(WebViewHost::notificationPresenter):
(WebViewHost::createApplicationCacheHost):
(WebViewHost::willSendRequest):
(WebViewHost::updateAddressBar):
(WebViewHost::updateURL):
* DumpRenderTree/chromium/WebViewHost.h:
(WebViewHost::addClearHeader):
(WebViewHost::clearHeaders):
2010-05-10 Dirk Pranke <dpranke@chromium.org>
Reviewed by Eric Seidel.
new-run-webkit-tests: fix a path-handling bug that was breaking the
dryrun ports on windows, and add a comment about why we don't run
the chromium dryrun tests by default on every port.
https://bugs.webkit.org/show_bug.cgi?id=38796
* Scripts/webkitpy/layout_tests/port/dryrun.py:
* Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
2010-05-11 Mark Rowe <mrowe@apple.com>
Reviewed by Darin Adler.
<http://webkit.org/b/38941> build-webkit shouldn't always override ENABLE_FOO Xcode configuration settings
* Scripts/build-webkit: Don't pass the ENABLE setting to xcodebuild if the value matches the default.
This will lead to xcodebuild picking up the settings from FeatureDefines.xcconfig, and will aid in
revealing problems that are due to inconsistent values for settings across projects.
2010-05-11 Mark Rowe <mrowe@apple.com>
Build fix.
* DumpRenderTree/mac/LayoutTestControllerMac.mm:
2010-05-10 Ojan Vafai <ojan@chromium.org>
Reviewed by David Levin.
CheckStyle was eating script errors when there were local-commits and working copy changes
https://bugs.webkit.org/show_bug.cgi?id=38880
* Scripts/webkitpy/tool/steps/checkstyle.py:
* Scripts/webkitpy/tool/steps/checkstyle_unittest.py: Added.
2010-05-11 Jian Li <jianli@chromium.org>
Reviewed by Dmitry Titov.
Expose FileReader interface.
https://bugs.webkit.org/show_bug.cgi?id=38609
* Scripts/build-webkit: turn on building FileReader for Apple's WebKit.
2010-05-10 Tony Chang <tony@chromium.org>
Reviewed by Kent Tamura.
[chromium] Build DRT when running build-webkit --chromium
https://bugs.webkit.org/show_bug.cgi?id=38730
* DumpRenderTree/DumpRenderTree.gyp/DumpRenderTree.gyp: Removed.
* Scripts/build-dumprendertree: This does nothing now.
2010-05-10 Jon Honeycutt <jhoneycutt@apple.com>
REGRESSION(r59100): Added test is broken on many platforms.
https://bugs.webkit.org/show_bug.cgi?id=38881
Reviewed by Eric Seidel.
* DumpRenderTree/win/TestNetscapePlugin/main.cpp:
(NPP_Destroy):
Use pluginLog, rather than printf, to match other platforms.
2010-05-10 Dirk Pranke <dpranke@chromium.org>
Unreviewed, build fix.
Fix a build failure caused by assuming the default platform in
a unit test for new-run-webkit-tests instead of specifying
--platform test.
* Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
2010-05-10 Dirk Pranke <dpranke@chromium.org>
Unreviewed, build-fix
new-run-webkit-tests: fix test failure caused by me failing to update
the expected output. Also, run '--platform test' instead of
'platform dryrun' since the former is guaranteed to work and the
latter isn't.
* Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
2010-05-10 Dirk Pranke <dpranke@chromium.org>
Reviewed by David Levin.
new-run-webkit-tests: looks like the unicode conversion broke
--print-last-failures and --retest-last-failures. Fixing.
* Scripts/webkitpy/layout_tests/run_webkit_tests.py:
* Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
2010-05-10 Dirk Pranke <dpranke@chromium.org>
Reviewed by David Levin.
Re-attempt to fix 38616 - newline handling in new-run-webkit-tests.
I didn't handle some cases correctly before and the solution was
confusing. The new patch assumes all calls to the printing module
don't have newlines, and will append newlines where necessary, just
like logging does.
https://bugs.webkit.org/show_bug.cgi?id=38790
* Scripts/webkitpy/layout_tests/layout_package/metered_stream.py:
* Scripts/webkitpy/layout_tests/layout_package/metered_stream_unittest.py:
* Scripts/webkitpy/layout_tests/layout_package/printing.py:
* Scripts/webkitpy/layout_tests/layout_package/printing_unittest.py:
* Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
2010-05-10 Kent Tamura <tkent@chromium.org>
Reviewed by Adam Barth.
Make tkent a reviewer
https://bugs.webkit.org/show_bug.cgi?id=38875
* Scripts/webkitpy/common/config/committers.py:
2010-05-07 Jon Honeycutt <jhoneycutt@apple.com>
Crash closing window containing Flash plug-in
https://bugs.webkit.org/show_bug.cgi?id=38797
<rdar://problem/7935266>
Reviewed by Eric Seidel.
* DumpRenderTree/TestNetscapePlugIn.subproj/PluginObject.cpp:
(pluginAllocate):
Initialize new member to false.
* DumpRenderTree/TestNetscapePlugIn.subproj/PluginObject.h:
Added a new member.
* DumpRenderTree/win/TestNetscapePlugin/main.cpp:
(NPP_New):
Check whether the element has the "testGetURLOnDestroy" attribute, and
record that.
(NPP_Destroy):
If "testGetURLOnDestroy" is set, perform a load while destroying the
plug-in.
2010-05-10 Eric Seidel <eric@webkit.org>
Reviewed by David Levin.
Executive.kill_* do not work with windows python
https://bugs.webkit.org/show_bug.cgi?id=38872
http://trac.webkit.org/changeset/57444 is where the original
breakage occurred.
http://trac.webkit.org/changeset/58314 is where the regression
started affecting chromium.
I have since learned that sys.platform has no "windows" value.
"win32" is always the value, under 32 or 64 bit windows
The tests for this code are not run anywhere because
test-webkitpy does not yet work on "win32". Mostly due to
depending on unixisms like "cat" and "yes".
* Scripts/webkitpy/common/system/executive.py:
* Scripts/webkitpy/common/system/executive_unittest.py:
2010-05-10 Dirk Pranke <dpranke@chromium.org>
Reviewed by Eric Seidel.
Fix a bug in rebaseline-chromium-webkit-tests where we would crash
instead of logging an error and exiting if it couldn't find either a
debug or a release build of the image diff binary.
https://bugs.webkit.org/show_bug.cgi?id=38692
* Scripts/webkitpy/layout_tests/rebaseline_chromium_webkit_tests.py:
* Scripts/webkitpy/layout_tests/rebaseline_chromium_webkit_tests_unittest.py: Added.
2010-05-10 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
new-run-webkit-tests fails when run under sys.platform == "windows" due to undefined signal.SIGKILL
https://bugs.webkit.org/show_bug.cgi?id=38861
* Scripts/webkitpy/common/system/executive.py:
* Scripts/webkitpy/common/system/executive_unittest.py:
2010-05-10 Jer Noble <jer.noble@apple.com>
Unreviewed.
Adding myself as a committer.
* Scripts/webkitpy/common/config/committers.py:
2010-05-10 Hans Wennborg <hans@chromium.org>
Reviewed by Jeremy Orlow.
[Chromium] Add quota parameter to WebViewClient::createSessionStorageNamespace()
https://bugs.webkit.org/show_bug.cgi?id=38750
Put a per-origin quota on session storage since it is using memory in
the browser process, and should not be allowed to grow arbitrarily
large. See also http://trac.webkit.org/changeset/58828.
* DumpRenderTree/chromium/WebViewHost.cpp:
(WebViewHost::createSessionStorageNamespace):
* DumpRenderTree/chromium/WebViewHost.h:
2010-05-10 Csaba Osztrogonác <ossy@webkit.org>
Rubber-stamped by Simon Hausmann.
[Qt] Roll-out r59020 and r59021, because the Qt part of these changes
haven't been landed in Qt trunk yet. Should be rolled-in again after the merge.
https://bugs.webkit.org/show_bug.cgi?id=32967
* DumpRenderTree/qt/DumpRenderTreeQt.cpp:
(WebCore::WebPage::WebPage):
(WebCore::DumpRenderTree::DumpRenderTree):
* DumpRenderTree/qt/DumpRenderTreeQt.h:
2010-05-09 Daniel Bates <dbates@rim.com>
Reviewed by Chris Jerdonek.
https://bugs.webkit.org/show_bug.cgi?id=38812
Make the regular expressions for parsing the start of an SVN
and Git header global variables since they are used throughout
VCSUtils.pm.
* Scripts/VCSUtils.pm:
2010-05-09 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Daniel Bates.
Finished moving the header-parsing logic from svn-apply and -unapply
to VCSUtils.pm's parsing methods.
https://bugs.webkit.org/show_bug.cgi?id=38802
* Scripts/VCSUtils.pm:
- Added to parseGitDiffHeader() the ability to parse and store
whether a file is new or deleted.
- Also reordered in parseGitDiffHeader() some of the else statements
to a more readable ordering.
- Added to parseSvnDiffHeader() the ability to parse and store
whether a file is new.
* Scripts/svn-apply:
- Changed handleGitBinaryChange() to use the new "isNew" and "isDeletion"
diffHash key-values.
- Changed patch() to use the new "isNew" diffHash key-value.
* Scripts/svn-unapply:
- Changed patch() to use the new "isNew" and "isDeletion" diffHash key-values.
* Scripts/webkitperl/VCSUtils_unittest/parseDiff.pl:
- Added unit tests for new and deleted files.
* Scripts/webkitperl/VCSUtils_unittest/parseGitDiffHeader.pl:
- Updated the unit tests as necessary.
- Added a unit test for a deleted file.
* Scripts/webkitperl/VCSUtils_unittest/parseSvnDiffHeader.pl:
- Updated the unit tests as necessary.
2010-05-08 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Daniel Bates.
Added to svn-apply support for git renames and copies with similarity
index less than 100%.
https://bugs.webkit.org/show_bug.cgi?id=32834
* Scripts/VCSUtils.pm:
- Added to parseGitDiffHeader() support for renames and similarity
index less than 100%.
- Added to parseDiff() support for processing renames, renames with
changes, and copies with changes.
- Added to parsePatch() the ability to process multiple return
values from parseDiff().
* Scripts/svn-apply:
- Added to patch() the ability to process diff hashes with the
isDeletion key-value set.
* Scripts/webkitperl/VCSUtils_unittest/parseDiff.pl:
- Updated the unit tests as necessary.
- Added unit tests for rename with similarity index 100%,
rename with similarity index < 100%, and rename with a change
to the executable bit.
* Scripts/webkitperl/VCSUtils_unittest/parseGitDiffHeader.pl:
- Added unit tests for rename with similarity index 100%,
rename with similarity index < 100%, and rename with a change
to the executable bit.
* Scripts/webkitperl/VCSUtils_unittest/parsePatch.pl:
- Updated the unit tests as necessary.
* Scripts/webkitperl/VCSUtils_unittest/parseSvnDiffHeader.pl:
- Simplified the carriage-return unit test to more narrowly test
only carriage returns.
2010-05-08 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Daniel Bates.
This revision suppresses the misleading "error: pathspec..." messages
when using svn-apply to add a binary file from a Git diff.
https://bugs.webkit.org/show_bug.cgi?id=38047
When adding a new binary file from a Git diff, svn-apply prints
a misleading error of the form -- "error: pathspec '<filename>' did
not match any file(s) known to git. Did you forget to 'git add'?"
This patch suppresses these messages since they are normal.
* Scripts/VCSUtils.pm:
Added the callSilently() subroutine from runPatchCommand.pl, which
executes a Perl function while suppressing STDERR.
* Scripts/svn-apply:
Refactored the Git portion of scmKnowsOfFile() into a
gitKnowsOfFile(), and called this new subroutine using callSilently().
* Scripts/webkitperl/VCSUtils_unittest/runPatchCommand.pl:
Removed callSilently() since it was moved to VCSUtils.pm in this patch.
2010-05-08 Victor Wang <victorw@chromium.org>
Reviewed by Adam Barth.
[chromium]: Upload test results json files to app engine server
Add an option to run_webkit_tests.py to upload generated
JSON files to app engine server. These JSON files will be used
by chromium layout test falkiness dashboard.
https://bugs.webkit.org/show_bug.cgi?id=36063
* Scripts/webkitpy/layout_tests/layout_package/test_results_uploader.py: Added.
* Scripts/webkitpy/layout_tests/run_webkit_tests.py:
2010-05-08 Laszlo Gombos <laszlo.1.gombos@nokia.com>
Reviewed by Kenneth Rohde Christiansen.
[Qt] Build the ImageDiff tool for all platforms including Windows and Symbian
https://bugs.webkit.org/show_bug.cgi?id=38706
Use qmath.h instead of math.h to make it portable.
* DumpRenderTree/qt/ImageDiff.cpp:
(main):
2010-05-08 Robert Hogan <robert@roberthogan.net>
Reviewed by Simon Hausmann.
[Qt] Fix http/tests/xmlhttprequest/cross-origin-no-authorization.html
and http/tests/xmlhttprequest/cross-origin-authorization.html
QHttpNetworkRequest adds Authorization and Cookie headers to XHRs
without knowing if this is valid behaviour or not. In order to allow
Qt to decide whether Cookie/Authorization headers should be added
to an XHR QtWebKit needs to use an attribute added to QNetworkRequest.
These new attributes are: QNetworkRequest::CookieLoadControlAttribute,
QNetworkRequest::CookieSaveControlAttribute,and
QNetworkRequest::AuthenticationReuseControlAttribute.
In order to properly support the tests, Qt's DRT needs to use one
NetworkAccessManager for all pages. This allows it to use cached
credentials where appropriate.
The tests now pass when run individually but there seems to be a problem with
leaking the results of requests across tests when run with the others in
http/tests. This will be addressed in a separate patch.
https://bugs.webkit.org/show_bug.cgi?id=32967
* DumpRenderTree/qt/DumpRenderTreeQt.cpp:
(WebCore::WebPage::WebPage):
(WebCore::DumpRenderTree::DumpRenderTree):
* DumpRenderTree/qt/DumpRenderTreeQt.h:
2010-05-04 Robert Hogan <robert@webkit.org>
Reviewed by Eric Seidel.
[Qt] unskip http/tests/plugins/npapi-response-headers.html
Turns out this failed because run-webkit-tests was eating the first occurrence
of 'Content-Type: text/plain' in the test output as a header. Strange but true.
So do as Chromium does and preface all text dumps with the
'Content-Type: text/plain' header.
https://bugs.webkit.org/show_bug.cgi?id=38541
* DumpRenderTree/qt/DumpRenderTreeQt.cpp:
(WebCore::DumpRenderTree::dump):
2010-05-08 Eric Seidel <eric@webkit.org>
Reviewed by Daniel Bates.
test-webkitpy fails under cygwin
https://bugs.webkit.org/show_bug.cgi?id=38718
* Scripts/webkitpy/common/system/executive.py:
- Add _KILL_PROCESS_KILLED_PROCESS_EXIT_CODE and
_KILL_ALL_KILLED_PROCESS_EXIT_CODE to store the expected
exit codes of processes killed by kill_process and kill_all.
These two constants are only used by the unit tests but are
stored in executive.py so they can be right next to the platform ifs.
- Remove unnecessary str() conversion, run_command does that for us.
- Make os.kill retry on cygwin on EAGAIN. It's unclear why CYGWIN
throws EAGAIN, but it only does so sometimes. 3 may not be enough
retries, but we'll try it to start with.
- Add _windows_image_name to automatically convert "yes" to "yes.exe"
for use with taskkill.exe /im. Various callers to kill_all could
be updated to remove the .exe, but that can be done in another patch.
- Use taskkill.exe for killall on cygwin.
* Scripts/webkitpy/common/system/executive_unittest.py:
- Use the new *_KILLED_PROCESS_EXIT_CODE constants which are correctly
set to 0 on windows/cygwin systems where taskkill.exe is used.
- Test _windows_image_name
* Scripts/webkitpy/layout_tests/port/chromium_win.py:
- Add FIXME about including mac-snowleopard in baseline_search_path.
* Scripts/webkitpy/layout_tests/port/webkit.py:
- Make default_configuration actually read from the Configuration file.
* Scripts/webkitpy/layout_tests/port/webkit_unittest.py: Added.
- Test default_configuration
* Scripts/webkitpy/layout_tests/port/win.py:
- Need a basic baseline_search_path if --platform dryrun is to work.
2010-05-08 Eric Seidel <eric@webkit.org>
Unreviewed, just reverting commit.
REGRESSION(59000): r59000 contained all sorts of changes it should not have, needs revert.
https://bugs.webkit.org/show_bug.cgi?id=38798
* Scripts/webkitpy/common/system/executive.py:
* Scripts/webkitpy/common/system/executive_unittest.py:
* Scripts/webkitpy/layout_tests/port/chromium_linux.py:
* Scripts/webkitpy/layout_tests/port/chromium_mac.py:
* Scripts/webkitpy/layout_tests/port/chromium_win.py:
* Scripts/webkitpy/layout_tests/port/mac.py:
* Scripts/webkitpy/layout_tests/port/webkit.py:
* Scripts/webkitpy/layout_tests/port/webkit_unittest.py: Removed.
* Scripts/webkitpy/layout_tests/port/win.py:
* Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
* Scripts/webkitpy/tool/commands/download.py:
* Scripts/webkitpy/tool/commands/download_unittest.py:
2010-05-06 Eric Seidel <eric@webkit.org>
Reviewed by Daniel Bates.
test-webkitpy fails under cygwin
https://bugs.webkit.org/show_bug.cgi?id=38718
* Scripts/webkitpy/common/system/executive.py:
- Add _KILL_PROCESS_KILLED_PROCESS_EXIT_CODE and
_KILL_ALL_KILLED_PROCESS_EXIT_CODE to store the expected
exit codes of processes killed by kill_process and kill_all.
These two constants are only used by the unit tests but are
stored in executive.py so they can be right next to the platform ifs.
- Remove unnecessary str() conversion, run_command does that for us.
- Make os.kill retry on cygwin on EAGAIN. It's unclear why CYGWIN
throws EAGAIN, but it only does so sometimes. 3 may not be enough
retries, but we'll try it to start with.
- Add _windows_image_name to automatically convert "yes" to "yes.exe"
for use with taskkill.exe /im. Various callers to kill_all could
be updated to remove the .exe, but that can be done in another patch.
- Use taskkill.exe for killall on cygwin.
* Scripts/webkitpy/common/system/executive_unittest.py:
- Use the new *_KILLED_PROCESS_EXIT_CODE constants which are correctly
set to 0 on windows/cygwin systems where taskkill.exe is used.
- Test _windows_image_name
* Scripts/webkitpy/layout_tests/port/chromium_win.py:
- Add FIXME about including mac-snowleopard in baseline_search_path.
* Scripts/webkitpy/layout_tests/port/webkit.py:
- Make default_configuration actually read from the Configuration file.
* Scripts/webkitpy/layout_tests/port/webkit_unittest.py: Added.
- Test default_configuration
* Scripts/webkitpy/layout_tests/port/win.py:
- Need a basic baseline_search_path if --platform dryrun is to work.
2010-05-07 Eric Seidel <eric@webkit.org>
Reviewed by Nikolas Zimmermann.
DryrunTest fails on every platform other than mac
https://bugs.webkit.org/show_bug.cgi?id=38796
The test uses the port detection logic to find a suitable
port to use results from. However that detection logic assumes
chromium on linux, which requires a chromium checkout which the
bots don't have. The test is broken and we'll need to fix it.
For now I'm just going to disable the test on all platforms besides mac.
* Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
2010-05-07 Eric Seidel <eric@webkit.org>
Reviewed by Daniel Bates.
Clean up baseline_search_path to use map to reduce copy/paste code
https://bugs.webkit.org/show_bug.cgi?id=38792
Reading which portnames a port falls back to is easier if
we convert port names to paths with map instead of using copy/paste code.
* Scripts/webkitpy/layout_tests/port/chromium_linux.py:
* Scripts/webkitpy/layout_tests/port/chromium_mac.py:
* Scripts/webkitpy/layout_tests/port/chromium_win.py:
* Scripts/webkitpy/layout_tests/port/mac.py:
2010-05-07 Eric Seidel <eric@webkit.org>
Reviewed by Daniel Bates.
rollout commands fail when commit is missing bug number
https://bugs.webkit.org/show_bug.cgi?id=38791
* Scripts/webkitpy/tool/commands/download.py:
- _commit_info failed to actually return the CommitInfo in the no-bug-id case.
* Scripts/webkitpy/tool/commands/download_unittest.py:
- Test that the fix worked.
2010-05-07 Darin Fisher <darin@chromium.org>
Fix build bustage: toElement<T> should be to<T>.
* DumpRenderTree/chromium/LayoutTestController.cpp:
(LayoutTestController::elementDoesAutoCompleteForElementWithId):
2010-05-07 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Daniel Bates.
Refactored VCSUtils.pm's parse-related methods to leave inapplicable
hash values unset instead of setting them to "undef".
https://bugs.webkit.org/show_bug.cgi?id=38724
Preferring "not set" over "undef" keeps the unit tests smaller and
easier to maintain. Otherwise, we would have to update every unit
test case each time we add support for a new key-value pair --
instead of just the relevant ones.
* Scripts/VCSUtils.pm:
- In parseGitDiffHeader(), adjusted the handling of these key-values:
executableBitDelta and isBinary.
- In parseSvnDiffHeader(), adjusted the handling of these key-values:
copiedFromPath, isBinary, and sourceRevision.
- In parseDiffHeader(), adjusted the handling of these key-values:
isGit and isSvn.
- In parseDiff(), adjusted the handling of these key-values:
isBinary, isGit, isSvn, and sourceRevision.
* Scripts/webkitperl/VCSUtils_unittest/parseDiff.pl:
- Updated the unit tests as necessary.
* Scripts/webkitperl/VCSUtils_unittest/parseDiffHeader.pl:
- Updated the unit tests as necessary.
* Scripts/webkitperl/VCSUtils_unittest/parseGitDiffHeader.pl:
- Updated the unit tests as necessary.
* Scripts/webkitperl/VCSUtils_unittest/parseSvnDiffHeader.pl:
- Updated the unit tests as necessary.
2010-05-06 Mark Rowe <mrowe@apple.com>
Rubber-stamped by Dan Bernstein.
Exclude leaks due to <rdar://problem/7815391> from the output.
* Scripts/old-run-webkit-tests:
2010-05-06 Sheriff Bot <webkit.review.bot@gmail.com>
Unreviewed, rolling out r58933.
http://trac.webkit.org/changeset/58933
https://bugs.webkit.org/show_bug.cgi?id=38717
"Broke all websocket tests on Tiger" (Requested by eseidel on
#webkit).
* Scripts/new-run-webkit-websocketserver:
* Scripts/old-run-webkit-tests:
* Scripts/run-webkit-websocketserver:
* Scripts/webkitpy/layout_tests/port/websocket_server.py:
* Scripts/webkitpy/thirdparty/__init__.py:
* Scripts/webkitpy/thirdparty/pywebsocket/COPYING: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/MANIFEST.in: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/README: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/README.webkit: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/example/echo_client.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/example/echo_wsh.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/example/handler_map.txt: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/mod_pywebsocket/__init__.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/mod_pywebsocket/dispatch.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/mod_pywebsocket/handshake.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/mod_pywebsocket/headerparserhandler.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/mod_pywebsocket/memorizingfile.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/mod_pywebsocket/msgutil.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/mod_pywebsocket/standalone.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/mod_pywebsocket/util.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/setup.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/config.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/mock.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/run_all.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/test_dispatch.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/test_handshake.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/test_memorizingfile.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/test_mock.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/test_msgutil.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/test_util.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/testdata/README: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/testdata/handlers/blank_wsh.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/testdata/handlers/origin_check_wsh.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/testdata/handlers/sub/exception_in_transfer_wsh.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/testdata/handlers/sub/no_wsh_at_the_end.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/testdata/handlers/sub/non_callable_wsh.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/testdata/handlers/sub/plain_wsh.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/testdata/handlers/sub/wrong_handshake_sig_wsh.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/testdata/handlers/sub/wrong_transfer_sig_wsh.py: Added.
* Scripts/webkitpy/thirdparty/pywebsocket/test/testdata/hello.pl: Added.
2010-05-06 Fumitoshi Ukai <ukai@chromium.org>
Reviewed by David Levin.
WebSocket: pywebsocket 0.5
https://bugs.webkit.org/show_bug.cgi?id=38034
Remove pywebsocket from webkitpy/thirdparty.
Make pywebsocket autoinstalled.
* Scripts/new-run-webkit-websocketserver:
Add --output-dir option.
* Scripts/old-run-webkit-tests:
Use new-run-webkit-websocketserver, rather than directly run pywebsocket's standalone.py
* Scripts/run-webkit-websocketserver:
Ditto.
* Scripts/webkitpy/layout_tests/port/websocket_server.py:
Use autoinstalled pywebsocket.
* Scripts/webkitpy/thirdparty/__init__.py:
Autoinstall pywebsocket
* Scripts/webkitpy/thirdparty/pywebsocket: Removed.
2010-05-06 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Daniel Bates.
svn-apply now understands the Git diff "copy from" syntax when the
similarity index is 100%.
https://bugs.webkit.org/show_bug.cgi?id=38628
* Scripts/VCSUtils.pm:
- Adjusted parseGitDiffHeader() to parse the "copy from" and
"similarity index" lines.
* Scripts/svn-unapply:
- Adjusted the patch() subroutine so that copies are recognized
as file additions.
* Scripts/webkitperl/VCSUtils_unittest/parseGitDiffHeader.pl:
- Added unit tests for the cases of a copy with similarity index
100% and less than 100%.
2010-05-06 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Daniel Bates.
Removed the need for svn-apply and -unapply to re-parse whether
a diff is binary or not.
https://bugs.webkit.org/show_bug.cgi?id=38320
* Scripts/VCSUtils.pm:
- Adjusted parseGitDiffHeader() to set the isBinary key.
- Adjusted parseSvnDiffHeader() to set the isBinary key.
- Adjusted parseDiffHeader() to set the isBinary key.
- Changed the scmFormat key set by parseDiffHeader() to
isGit and isSvn keys.
- Adjusted parseDiff() to set the isBinary, isGit, and isSvn keys.
* Scripts/svn-apply:
- Updated the patch() method to use the isBinary, isGit, and
isSvn keys.
* Scripts/svn-unapply:
- Updated the patch() method to use the isBinary and isSvn keys.
* Scripts/webkitperl/VCSUtils_unittest/parseDiff.pl:
- Updated the unit tests as necessary.
- Added a test case to test that the isBinary key is getting set properly.
* Scripts/webkitperl/VCSUtils_unittest/parseDiffHeader.pl:
- Updated the unit tests as necessary.
* Scripts/webkitperl/VCSUtils_unittest/parseGitDiffHeader.pl:
- Updated the unit tests as necessary.
* Scripts/webkitperl/VCSUtils_unittest/parseSvnDiffHeader.pl:
- Updated the unit tests as necessary.
2010-05-06 Kent Tamura <tkent@chromium.org>
Reviewed by Dimitri Glazkov.
[DRT/Chromium] Add support for resources on Mac
https://bugs.webkit.org/show_bug.cgi?id=38637
Repack webkit_chromium_resources.pak, webkit_strings_en-US.pak,
and webkit_resources.pak, and put them as Mac bundle resource.
The 'actions' section is almost same as a part of test_shell.gypi.
* DumpRenderTree/DumpRenderTree.gyp/DumpRenderTree.gyp:
2010-05-06 Csaba Osztrogonác <ossy@webkit.org>
[Qt] Unreviewed buildfix after r58917.
* DumpRenderTree/qt/LayoutTestControllerQt.h: Missing function declaration added.
2010-05-06 Anders Carlsson <andersca@apple.com>
Reviewed by Darin Adler and Dan Bernstein..
REGRESSION (r51617): when plugins are disabled, plugins show up as garbage characters
https://bugs.webkit.org/show_bug.cgi?id=38698
<rdar://problem/7942075>
Add a 'setPluginsEnabled' layoutTestController function for disabling plug-ins. This is only implemented on Mac currently
because the bug that needs this functionality is mac specific.
* DumpRenderTree/LayoutTestController.cpp:
(setPluginsEnabledCallback):
(LayoutTestController::staticFunctions):
* DumpRenderTree/LayoutTestController.h:
* DumpRenderTree/gtk/LayoutTestControllerGtk.cpp:
(LayoutTestController::setPluginsEnabled):
* DumpRenderTree/mac/DumpRenderTree.mm:
(resetDefaultsToConsistentValues):
* DumpRenderTree/mac/LayoutTestControllerMac.mm:
(LayoutTestController::setPluginsEnabled):
* DumpRenderTree/qt/LayoutTestControllerQt.cpp:
(LayoutTestController::setPluginsEnabled):
* DumpRenderTree/win/LayoutTestControllerWin.cpp:
(LayoutTestController::setPluginsEnabled):
* DumpRenderTree/wx/LayoutTestControllerWx.cpp:
(LayoutTestController::setPluginsEnabled):
2010-05-06 Jochen Eisinger <jochen@chromium.org>
Reviewed by Dimitri Glazkov.
Make ImageDiff depend on WebKit. When compiled from within Chromium, WTF is not a standalone dynamic library but depends on WebKit.
https://bugs.webkit.org/show_bug.cgi?id=38632
* DumpRenderTree/DumpRenderTree.gyp/DumpRenderTree.gyp:
2010-05-06 Simon Hausmann <simon.hausmann@nokia.com>
Reviewed by Kenneth Rohde Christiansen.
[Qt] Replace public inspector url with private property for QtLauncher
https://bugs.webkit.org/show_bug.cgi?id=35340
Replace the public API with a private dynamic property until this feature
is ready.
* QtLauncher/main.cpp:
(LauncherWindow::init):
2010-05-05 Joseph Pecoraro <joepeck@webkit.org>
Reviewed by Pavel Feldman.
Web Inspector: build-webkit --inspector-frontend Should Exclude *.re2js
https://bugs.webkit.org/show_bug.cgi?id=38449
* Scripts/webkitdirs.pm:
2010-05-05 Charles Wei <charles.wei@torchmobile.com.cn>
Reviewed by George Staikos
This patch adds WCSS -wap-input-format and -wap-input-required support to WebKit
Make the test cases in fast/wcss optionional only when WCSS is enabled.
https://bugs.webkit.org/show_bug.cgi?id=37848
* Scripts/old-run-webkit-tests:
* Scripts/webkitperl/features.pm:
2010-05-05 Kent Tamura <tkent@chromium.org>
Reviewed by Eric Seidel.
[DRT/Chromium] Remove InitWebCoreSystemInterface() call
https://bugs.webkit.org/show_bug.cgi?id=38624
Chromium r45167 <http://src.chromium.org/viewvc/chrome?view=rev&revision=45167>
added InitWebCoreSystemInterface() to webkit/support/platform_support_mac.mm.
So we don't need to call it from DumpRenderTree.cpp anymore.
* DumpRenderTree/DumpRenderTree.gyp/DumpRenderTree.gyp:
* DumpRenderTree/chromium/DumpRenderTree.cpp:
(main): Remove InitWebCoreSystemInterface().
2010-05-05 Dirk Pranke <dpranke@chromium.org>
Reviewed by Eric Seidel.
new-run-webkit-tests: clean up newline handling in printing
The new printing module seems to handle newlines somewhat
inconsistently, especially in --verbose mode. This change cleans up
the code to make things more consistent and adds a bunch of unit tests.
https://bugs.webkit.org/show_bug.cgi?id=38616
* Scripts/webkitpy/common/array_stream.py: Added.
* Scripts/webkitpy/common/array_stream_unittest.py: Added.
* Scripts/webkitpy/layout_tests/layout_package/metered_stream.py:
* Scripts/webkitpy/layout_tests/layout_package/metered_stream_unittest.py: Added.
* Scripts/webkitpy/layout_tests/layout_package/printing.py:
* Scripts/webkitpy/layout_tests/layout_package/printing_unittest.py:
* Scripts/webkitpy/layout_tests/run_webkit_tests.py:
* Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
2010-05-05 James Robinson <jamesr@chromium.org>
Reviewed by Adam Barth.
Ban the single letter 'l' as an identifier name
http://trac.webkit.org/changeset/58844
Add a lint rule to ban the single letter 'l' as an identifier name
since it is very easy to confuse with the numeral '1', especially
in code like WebCore/css/CSSHelper.cpp.
See http://trac.webkit.org/changeset/58844 as an example of a bug
caused by confusing short variable names.
* Scripts/webkitpy/style/checkers/cpp.py:
* Scripts/webkitpy/style/checkers/cpp_unittest.py:
2010-05-04 Eric Seidel <eric@webkit.org>
Reviewed by Shinichiro Hamaji.
PrettyPatch.pretty_diff("") should not hang
https://bugs.webkit.org/show_bug.cgi?id=38552
Also found a bug in PrettyPatch.pretty_diff where it would
hang when passed "" as input.
I suspect there may be bugs in prettify.rb (or our use there-of)
where it can hang, which would then cause the testing thread to hang.
* Scripts/webkitpy/common/prettypatch.py:
- Don't hang when passed ""
* Scripts/webkitpy/common/prettypatch_unittest.py:
- Test that we don't hang when passed ""
* Scripts/webkitpy/layout_tests/port/base.py:
- Add a FIXME that we should share code with prettypatch.rb
2010-05-04 Dirk Pranke <dpranke@chromium.org>
Reviewed by Eric Seidel.
new-run-webkit-tests: turn off threading on the Chromium Mac port until
we can stabilize the port more and figure out why it is hanging so
frequently.
https://bugs.webkit.org/show_bug.cgi?id=38553
* Scripts/webkitpy/layout_tests/port/chromium_mac.py:
- override default_child_processes() and log a warning
* Scripts/webkitpy/layout_tests/run_webkit_tests.py:
- fix a typo that caused us to print a method object instead of the
value the method object returns in the case where there is only
one child process.
* Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
- Add unit tests for the output of run_webkit_tests - in this case,
the handling of --child-processes and --print config
2010-05-04 Timothy Hatcher <timothy@apple.com>
Fix the find command in extract-localizable-strings so skip
directories are correctly skipped and header files are searched.
https://bugs.webkit.org/show_bug.cgi?id=38545
rdar://problem/7941295
Reviewed by Darin Adler.
* Scripts/extract-localizable-strings: Append -o after each -prune
so -and isn't implied. Surround all the -name arguments so they get
an implied -print action. Removed check for "icu". Skip any header that
ends in LocalizableStrings.h, so SafariLocalizableStrings.h is skipped.
* Scripts/update-webkit-localizable-strings: Add the two icu directories
to the skip list.
2010-05-04 Jesus Sanchez-Palencia <jesus@webkit.org>
Reviewed by Eric Seidel.
Wrong documentation on 'webkit-patch help land'.
https://bugs.webkit.org/show_bug.cgi?id=37871
Small fix on the help documentation for webkit-patch
land.
* Scripts/webkitpy/tool/commands/download.py:
2010-05-04 Laszlo Gombos <laszlo.1.gombos@nokia.com>
Reviewed by Kenneth Rohde Christiansen.
[Qt] QWebPage viewMode property
https://bugs.webkit.org/show_bug.cgi?id=38119
Rename the property from wrt_viewMode to _q_viewMode.
* DumpRenderTree/qt/LayoutTestControllerQt.cpp:
(LayoutTestController::setViewModeMediaFeature):
2010-05-04 Jochen Eisinger <jochen@chromium.org>
Reviewed by Jeremy Orlow.
Fix typo in run_webkit_tests.py: s/_print\./_printer./
https://bugs.webkit.org/show_bug.cgi?id=38515
* Scripts/webkitpy/layout_tests/run_webkit_tests.py:
2010-05-04 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Eric Seidel.
This revision completes the terminology change from "processor" to
"checker" for the CarriageReturnProcessor, CppProcessor,
PythonProcessor, and TextProcessor classes, etc.
https://bugs.webkit.org/show_bug.cgi?id=38262
The word "processor" currently has two uses in our code. This
revision renames the lower-level use to "checker" and preserves the
word "processor" for higher-level, more general uses. This
revision also makes whatever other name changes that logically
followed from this change.
* Scripts/check-webkit-style:
- Updated references to PatchChecker.
* Scripts/webkitpy/style/checker.py:
- Renamed the StyleCheckerConfiguration class to
StyleProcessorConfiguration.
- Renamed the ProcessorDispatcher class to CheckerDispatcher, and
made similar changes for the class's method names.
- Renamed the PatchChecker class to PatchReader.
- Updated the file as necessary to accommodate the other class
renames in this patch.
* Scripts/webkitpy/style/checker_unittest.py:
- Updated the unit test code as necessary.
* Scripts/webkitpy/style/checkers/common.py:
- Renamed the CarriageReturnProcessor class to CarriageReturnChecker,
and changed its process() method to check().
* Scripts/webkitpy/style/checkers/common_unittest.py:
- Updated the unit test code as necessary.
* Scripts/webkitpy/style/checkers/cpp.py:
- Renamed the CppProcessor class to CppChecker, and renamed its
process() method to check().
* Scripts/webkitpy/style/checkers/cpp_unittest.py:
- Updated the unit test code as necessary.
* Scripts/webkitpy/style/checkers/python.py:
- Renamed the PythonProcessor class to PythonChecker, and renamed
its process() method to check().
* Scripts/webkitpy/style/checkers/python_unittest.py:
- Updated the unit test code as necessary.
* Scripts/webkitpy/style/checkers/text.py:
- Renamed the TextProcessor class to TextChecker, and renamed
its process() method to check().
* Scripts/webkitpy/style/checkers/text_unittest.py:
- Updated the unit test code as necessary.
* Scripts/webkitpy/style/error_handlers.py:
- Updated the code as necessary.
* Scripts/webkitpy/style/error_handlers_unittest.py:
- Updated the unit test code as necessary.
2010-05-04 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Eric Seidel.
Adjusted svn-apply and -unapply to accept git diffs generated
using the --no-prefix flag.
https://bugs.webkit.org/show_bug.cgi?id=32438
* Scripts/VCSUtils.pm:
- Loosened the regular expression for the "diff --git" line to
match when the --no-prefix flag is used with "git diff".
- Also refactored the code parsing the first line so that the
script exits with an error message if the first line cannot
be parsed.
* Scripts/webkitperl/VCSUtils_unittest/parseGitDiffHeader.pl:
- Added a unit test case for the --no-prefix case.
2010-05-04 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Daniel Bates.
Changed VCSUtils.pm's parseDiffHeader() to call the new
parseGitDiffHeader() method.
https://bugs.webkit.org/show_bug.cgi?id=38454
This revision makes more progress towards adding executable-bit
support to svn-apply and svn-unapply. It also makes more progress
towards refactoring the code into a more maintainable form.
* Scripts/VCSUtils.pm:
- Removed gitdiff2svndiff().
- Removed the Git-specific logic from parseDiffHeader() and
renamed it parseSvnDiffHeader().
- Added a new parseDiffHeader() subroutine which calls
parseSvnDiffHeader() or parseGitDiffHeader() depending on
the format of the first header line.
* Scripts/webkitperl/VCSUtils_unittest/gitdiff2svndiff.pl: Removed.
* Scripts/webkitperl/VCSUtils_unittest/parseDiffHeader.pl:
- Removed most of the test cases since these cases are now
covered by the unit tests for parseSvnDiffHeader() and
parseGitDiffHeader().
* Scripts/webkitperl/VCSUtils_unittest/parseSvnDiffHeader.pl: Added.
- Copied the SVN unit tests from parseDiffHeader.pl and updated
them as necessary.
2010-05-03 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
Builders should run the perl and python unit tests on every commit
https://bugs.webkit.org/show_bug.cgi?id=37976
The unit tests take a few seconds to run, so they should not
have any noticable effect on builder speed.
We're running the tests everywhere but Tiger as Tiger's
Python version is ancient.
I would have rather have detected the python version of the
slave but I couldn't find any API to do that, and I didn't want
to hack version detection into test-webkitpy (at least not yet).
* BuildSlaveSupport/build.webkit.org-config/master.cfg:
2010-05-03 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Adam Barth.
Fixed a recent REGRESSION that caused svn-apply and -unapply to
skip over changes to the first file in a diff if leading junk was
present (like in an e-mail diff) and if the --force option was used.
https://bugs.webkit.org/show_bug.cgi?id=38507
* Scripts/svn-apply:
- Removed the bit of code at the beginning of the patch()
subroutine that checks for the "Index:" line at the beginning
of a file diff (since the parsePatch() subroutine already
checks this).
* Scripts/svn-unapply:
- Removed the bit of code at the beginning of the patch()
subroutine that checks for the "Index:" line at the beginning
of a file diff (since the parsePatch() subroutine already
checks this).
2010-05-03 Chris Jerdonek <cjerdonek@webkit.org>
Unreviewed.
Adjusted the ChangeLog entry below for r58732 (bug 35804) to reflect
the fact that the change will not become active until the patch
for bug 38454 lands.
2010-05-03 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Eric Seidel.
Added code for svn-apply and -unapply to display an instructive error
message if the --binary flag is left off the "git diff" command
for diffs containing binary file differences.
https://bugs.webkit.org/show_bug.cgi?id=35804
This change will become active when parseDiffHeader() is modified
to call parseGitDiffHeader (see bug 38454).
* Scripts/VCSUtils.pm:
- Adjusted parseDiffHeader() to exit with an appropriate error message
if it encounters a line of the form "Binary files <path1> and
<path2> differ".
2010-05-03 Dirk Pranke <dpranke@chromium.org>
Unreviewed, build fix.
new-run-webkit-tests: r58728 broke the buildbot logic for parsing the
output of the log; specifying --verbose should basically be equivalent
to --print everything, but instead it was equivalent to not specifying
--print and getting the default set. Now, --verbose acts as if
--print everything was implicitly specified as the default (you can
still override it if you specify both; this is a somewhat debatable
call).
https://bugs.webkit.org/show_bug.cgi?id=38504
* Scripts/webkitpy/layout_tests/layout_package/printing.py:
2010-05-03 Dirk Pranke <dpranke@chromium.org>
Unreviewed, build fix.
new-run-webkit-tests: Fix minor precedence bug introduced in r58728 where we printed
"-\n" 78 times instead of "-" 78 times followed by a single "\n".
* Scripts/webkitpy/layout_tests/layout_package/printing.py:
2010-05-03 Dirk Pranke <dpranke@chromium.org>
Reviewed by Eric Seidel.
new-run-webkit-tests: refactor a large chunk of the printing/logging
code out of run-webkit-tests py (almost 300 lines out of 1900).
This change also renames --log to --print (to be slightly more
descriptive). I've also added better help messages for printing.
The new code has unit tests!
There is still more code to be moved, but this patch is big enough as
it is. Namely, still to move are the printing of the actual results
and the timing statistics, which should move another 300-400 lines
out of the file.
Notable changes to run_webkit_tests.py beyond code simply moving:
* MeteredStream is now hidden under the new printing.Printer class.
All the references to self._meter now point to self._printer.
* All logging configuration is done in printing.configure_logging()
* Instead of using write() lambdas to control what is and isn't
printed, we use separate methods on the printer object. This will
make it easier to grep which print statements are printed
under protection of each flag.
* The print_results flag I added a few revs back to suppress printing
in the unit tests has been replaced with --print nothing.
* The ResultSummary class now stores the entire TestResult for each
test, not just the actual result type.
* summarize_unexpected_results() got moved from a method on TestRunner
to a standalone function. This should move into a separate file
along with the ResultSummary class and the TestResult class
* The --trace option added recently has been replaced by
'--print trace-everything' and '--print trace-unexpected'
https://bugs.webkit.org/show_bug.cgi?id=38018
* Scripts/new-run-webkit-tests:
- update to new entry points in run_webkit_tests.py
* Scripts/webkitpy/layout_tests/layout_package/metered_stream.py:
- fix a minor nit where we were printing an empty string where
we didn't need to
* Scripts/webkitpy/layout_tests/layout_package/printing.py: Added.
* Scripts/webkitpy/layout_tests/layout_package/printing_unittest.py: Added
* Scripts/webkitpy/layout_tests/port/test.py:
- implement relative_test_filename() and expected_filename() so
we can test printing unexpected results in a platform-neutral
way
* Scripts/webkitpy/run_webkit_test.py:
- move a lot of the printing code into printing.py
- change the signatures of the exported entry points for easier
unit testing
* Scripts/webkitpy/run_webkit_tests_unittest.py:
- update w/ changes to run_webkit_tests entry points.
2010-05-03 Kevin Watters <kevinwatters@gmail.com>
Reviewed by Kevin Ollivier.
[wx] Build and use Mac's ComplexTextController to support complex text in wx.
https://bugs.webkit.org/show_bug.cgi?id=38482
* wx/build/settings.py:
2010-05-03 Abhishek Arya <inferno@chromium.org>
Reviewed by Adam Barth.
Add support for controlling clipboard access from javascript.
Clipboard access from javascript is enabled in test framework.
https://bugs.webkit.org/show_bug.cgi?id=27751
* DumpRenderTree/LayoutTestController.cpp:
(setJavaScriptCanAccessClipboardCallback):
(LayoutTestController::staticFunctions):
* DumpRenderTree/LayoutTestController.h:
* DumpRenderTree/chromium/LayoutTestController.cpp:
(LayoutTestController::LayoutTestController):
(LayoutTestController::setJavaScriptCanAccessClipboard):
(LayoutTestController::overridePreference):
* DumpRenderTree/chromium/LayoutTestController.h:
* DumpRenderTree/chromium/TestShell.cpp:
(TestShell::resetWebSettings):
* DumpRenderTree/gtk/DumpRenderTree.cpp:
(resetDefaultsToConsistentValues):
* DumpRenderTree/gtk/LayoutTestControllerGtk.cpp:
(LayoutTestController::setJavaScriptCanAccessClipboard):
* DumpRenderTree/mac/DumpRenderTree.mm:
(resetDefaultsToConsistentValues):
* DumpRenderTree/mac/LayoutTestControllerMac.mm:
(LayoutTestController::setJavaScriptCanAccessClipboard):
* DumpRenderTree/qt/DumpRenderTreeQt.cpp:
(WebCore::WebPage::WebPage):
(WebCore::WebPage::resetSettings):
* DumpRenderTree/qt/LayoutTestControllerQt.cpp:
(LayoutTestController::setJavaScriptCanAccessClipboard):
* DumpRenderTree/qt/LayoutTestControllerQt.h:
* DumpRenderTree/win/DumpRenderTree.cpp:
(resetDefaultsToConsistentValues):
* DumpRenderTree/win/LayoutTestControllerWin.cpp:
(LayoutTestController::setJavaScriptCanAccessClipboard):
* DumpRenderTree/wx/LayoutTestControllerWx.cpp:
(LayoutTestController::setJavaScriptCanAccessClipboard):
2010-05-03 Anders Carlsson <andersca@apple.com>
Reviewed by Adam Roben.
Maui MiniBrowser: Add an option to show/hide the web view
https://bugs.webkit.org/show_bug.cgi?id=38486
* MiniBrowser/win/BrowserWindow.cpp:
(BrowserWindow::createWindow):
Set the background brush to something other than null.
(BrowserWindow::onCommand):
Show and hide the web view accordingly.
* MiniBrowser/win/MiniBrowser.rc:
* MiniBrowser/win/resource.h:
Add new menu item.
2010-05-03 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Daniel Bates.
Added a parseGitDiffHeader() subroutine to VCSUtils.pm that
parses any changes to the executable bit in a Git diff.
https://bugs.webkit.org/show_bug.cgi?id=38425
This revision is more preparation towards adding "executable bit"
support to svn-apply and svn-unapply. No code is going "live" in
this change except for the new unit tests in test-webkitperl.
* Scripts/VCSUtils.pm:
- Added isExecutable() to determine whether a file mode has the
executable bit set or not.
- Added parseGitDiffHeader() to parse the header of a Git diff.
* Scripts/webkitperl/VCSUtils_unittest/parseGitDiffHeader.pl: Added.
- Added unit tests for parseGitDiffHeader().
2010-05-03 Daniel Bates <dbates@rim.com>
Reviewed by Chris Jerdonek.
https://bugs.webkit.org/show_bug.cgi?id=38447
Refactor the unit tests in VCSUtils_unittest/parseDiff.pl to use
Test::More::is_deeply like we do in VCSUtils_unittest/parseDiffHeader.pl.
* Scripts/webkitperl/VCSUtils_unittest/parseDiff.pl:
2010-05-02 Chris Jerdonek <cjerdonek@webkit.org>
Unreviewed.
Corrected a file path in the ChangeLog entry for r58663 (bug 38319) below.
2010-05-02 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Eric Seidel.
https://bugs.webkit.org/show_bug.cgi?id=38319
* Scripts/VCSUtils.pm:
- In parseDiffHeader()--
- Added an "scmFormat" hash key to the return value to represent
whether the diff is Git or SVN formatted.
- Adjusted the code so the value of "copiedFromPath" will
be undef rather than "does not exist" if the file was not
copied.
* Scripts/webkitperl/VCSUtils_unittest/parseDiff.pl:
- Added a FIXME to refactor these unit tests to use is_deeply().
* Scripts/webkitperl/VCSUtils_unittest/parseDiffHeader.pl:
- Updated the unit tests to test the "scmFormat" value.
- Simplified the unit tests by refactoring them to use is_deeply().
2010-05-01 Daniel Bates <dbates@rim.com>
Reviewed by Chris Jerdonek.
https://bugs.webkit.org/show_bug.cgi?id=38423
Adds infrastructure to change the file mode of a file using
the appropriate SCM-specific routines.
No functionality was changed, so no new tests.
* Scripts/VCSUtils.pm: Added subroutines scmToggleExecutableBit,
scmAddExecutableBit, and scmRemoveExecutableBit.
* Scripts/svn-apply: Check for the hash key executableBitDelta
and toggle the executable bit.
* Scripts/svn-unapply: Ditto.
2010-04-30 Chris Marrin <cmarrin@apple.com>
Reviewed by Simon Fraser.
Enabled accelerated compositing in DRT for Windows
https://bugs.webkit.org/show_bug.cgi?id=38404
* DumpRenderTree/win/DumpRenderTree.cpp:
(main):
2010-04-30 Anders Carlsson <andersca@apple.com>
Try to fix GTK+ build.
* DumpRenderTree/unix/TestNetscapePlugin/TestNetscapePlugin.cpp:
(webkit_test_plugin_new_stream):
2010-04-30 Anders Carlsson <andersca@apple.com>
Fix build.
* DumpRenderTree/unix/TestNetscapePlugin/TestNetscapePlugin.cpp:
2010-04-30 Anders Carlsson <andersca@apple.com>
Reviewed by Timothy Hatcher.
Next step towards fixing
https://bugs.webkit.org/show_bug.cgi?id=20784
move npapi.h to C99 integer types
Use the C99 types everywhere. The "old" types are still around but will be removed
in a subsequent commit.
* DumpRenderTree/TestNetscapePlugIn.subproj/PluginObject.cpp:
(int32VariantToIdentifier):
(doubleVariantToIdentifier):
(testIdentifierToInt):
(testGetIntIdentifier):
* DumpRenderTree/TestNetscapePlugIn.subproj/main.cpp:
(NPP_New):
(NPP_NewStream):
(NPP_WriteReady):
(NPP_Write):
(NPP_HandleEvent):
* DumpRenderTree/win/TestNetscapePlugin/main.cpp:
(NPP_New):
(NPP_NewStream):
(NPP_WriteReady):
(NPP_Write):
(NPP_HandleEvent):
2010-04-30 Steve Block <steveblock@google.com>
Reviewed by Jeremy Orlow.
Changed Steve Block from committer to reviewer.
* Scripts/webkitpy/common/config/committers.py:
2010-04-28 Ojan Vafai <ojan@chromium.org>
Reviewed by Eric Seidel.
webkit-patch doesn't work if a git repo is tracking multiple svn repos
https://bugs.webkit.org/show_bug.cgi?id=38290
Getting the tests to pass required getting our SVN repo to more closely
match the real svn.webkit.org repo by having a trunk directory.
That involved adding an extra commit at the beginning and thus changing
all the commit numbers in the tests.
* Scripts/webkitpy/common/checkout/scm.py:
2010-04-29 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
new-run-webkit-tests can deadlock with Chromium's TestShell
https://bugs.webkit.org/show_bug.cgi?id=38298
Fix _write_command_and_read_line to never send unicode() to
test_shell, instead to always encode as utf-8. This was causing
random hangs because if test_shell ever encounters a \0 in the
stream it can deadlock with NRWT.
There is still a deadlock bug to fix in NRWT/test_shell design, however
this fix should make the deadlock occur less often.
* Scripts/webkitpy/layout_tests/port/chromium.py:
* Scripts/webkitpy/layout_tests/port/chromium_unittest.py:
2010-04-29 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Eric Seidel.
Refactored svn-apply and svn-unapply to use the new
parsePatch() subroutine.
https://bugs.webkit.org/show_bug.cgi?id=34033
* Scripts/VCSUtils.pm:
- Consolidated %diffHash documentation.
- Added prepareParsedPatch().
* Scripts/svn-apply:
- Replaced main while loop with calls to parsePatch() and
prepareParsedPatch().
* Scripts/svn-unapply:
- Replaced main while loop with calls to parsePatch() and
prepareParsedPatch().
* Scripts/test-webkitperl:
- Changed to render relative test paths rather than absolute
test paths.
* Scripts/webkitperl/VCSUtils_unittest/prepareParsedPatch.pl: Added.
- Added unit tests for prepareParsedPatch().
2010-04-28 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Eric Seidel.
Added to VCSUtils's parseDiffHeader() support for binary patches.
https://bugs.webkit.org/show_bug.cgi?id=38094
The parseDiffHeader() function is part of new patch-parsing code
for svn-apply and svn-unapply that will go live in a subsequent
revision.
* Scripts/VCSUtils.pm:
- Added logic to parseDiffHeader() to recognize the ending of
the header portion of a binary diff.
* Scripts/webkitperl/VCSUtils_unittest/parseDiffHeader.pl:
- Added unit test cases for SVN and Git binary diffs.
2010-04-28 Eric Seidel <eric@webkit.org>
Reviewed by David Levin.
Document that subprocess.poll/wait are not threadsafe
https://bugs.webkit.org/show_bug.cgi?id=38289
* Scripts/webkitpy/common/system/executive.py:
* Scripts/webkitpy/layout_tests/layout_package/dump_render_tree_thread.py:
* Scripts/webkitpy/layout_tests/port/chromium.py:
* Scripts/webkitpy/layout_tests/port/http_server.py:
* Scripts/webkitpy/layout_tests/port/server_process.py:
* Scripts/webkitpy/layout_tests/port/websocket_server.py:
2010-04-28 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Daniel Bates.
Removed the dividing line (i.e. "====...") logic from the code
called by svn-apply and svn-unapply.
https://bugs.webkit.org/show_bug.cgi?id=38093
The dividing line logic is unnecessary. Removing it simplifies the
code and makes some subsequent changes easier.
* Scripts/VCSUtils.pm:
- Removed the logic in gitdiff2svndiff() to convert the git
"index" line to an SVN dividing line.
- Adjusted the logic similarly in parseDiffHeader().
* Scripts/webkitperl/VCSUtils_unittest/gitdiff2svndiff.pl:
- Updated the unit tests as necessary.
* Scripts/webkitperl/VCSUtils_unittest/parseDiff.pl:
- Updated the unit tests as necessary.
- Corrected an error in the unit tests whereby all elements
of an array were referencing the same element rather than
distinct elements -- causing unit test failures to be masked.
* Scripts/webkitperl/VCSUtils_unittest/parseDiffHeader.pl:
- Updated the unit tests as necessary.
- Made the same unit test correction as above for parseDiff.pl.
2010-04-28 Evan Stade <estade@chromium.org>
Unreviewed.
* Scripts/webkitpy/common/config/committers.py: adding myself as a committer
2010-04-28 Sam Weinig <sam@webkit.org>
Reviewed by Mark Rowe.
Add MiniBrowser to the lists of modules to build and only build it on
SnowLeopard and later.
* Makefile:
* MiniBrowser/Makefile:
2010-04-28 Sam Weinig <sam@webkit.org>
Reviewed by Mark Rowe.
Make running MiniBrowser.app without explicitly setting DYLD_FRAMEWORK_PATH.
* MiniBrowser/MiniBrowser.xcodeproj/project.pbxproj:
* MiniBrowser/mac/make-launchable.sh: Added.
2010-04-28 Eric Seidel <eric@webkit.org>
Reviewed by David Levin.
Audit all uses of subprocess in webkitpy
https://bugs.webkit.org/show_bug.cgi?id=38284
After further discussions with Jeffrey Yasskin
about http://bugs.python.org/issue2320
and related issues of using subprocess from
multiple threads, I have learned that subprocess
is known to be non-threadsafe through recent
Python 2.7 builds.
I'm attempting to lessen our exposure to these
subprocess bugs by auditing each use of subprocess
in webkitpy. I did not find any unsafe calls
in my audit, but I did remove numerous unneeded
import subprocess lines.
* Scripts/webkitpy/common/checkout/api.py:
* Scripts/webkitpy/common/net/bugzilla.py:
* Scripts/webkitpy/common/system/deprecated_logging_unittest.py:
* Scripts/webkitpy/common/system/user.py:
* Scripts/webkitpy/layout_tests/port/base.py:
* Scripts/webkitpy/layout_tests/port/chromium_linux.py:
* Scripts/webkitpy/layout_tests/port/chromium_mac.py:
* Scripts/webkitpy/layout_tests/port/chromium_mac_unittest.py: Added.
* Scripts/webkitpy/layout_tests/port/chromium_win.py:
* Scripts/webkitpy/layout_tests/port/gtk.py:
* Scripts/webkitpy/layout_tests/port/mac.py:
* Scripts/webkitpy/layout_tests/port/qt.py:
* Scripts/webkitpy/layout_tests/port/webkit.py:
* Scripts/webkitpy/layout_tests/port/win.py:
2010-04-28 Darin Adler <darin@apple.com>
Ignore a directory the Python tools creates.
* Scripts/webkitpy/style: Modified property svn:ignore.
2010-04-28 Darin Adler <darin@apple.com>
* Scripts/extract-localizable-strings: Fix minor mistake in
argument checking.
2010-04-28 Luiz Agostini <luiz.agostini@openbossa.org>
Reviewed by Kenneth Rohde Christiansen.
[Qt] QWebPage viewMode property
https://bugs.webkit.org/show_bug.cgi?id=38119
Replacing method qt_wrt_setViewMode by wrt_viewMode property.
* DumpRenderTree/qt/LayoutTestControllerQt.cpp:
(LayoutTestController::setViewModeMediaFeature):
2010-04-28 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by Shinichiro Hamaji.
Adjusted check-webkit-style so that files with file type NONE
are automatically skipped without warning.
https://bugs.webkit.org/show_bug.cgi?id=38197
This change simplifies configuring which files to skip. It also
addresses an issue whereby check-webkit-style was unintentionally
checking .vcproj files for carriage returns.
* Scripts/webkitpy/style/checker.py:
- Moved the C++, Python, and text file extensions to new
module-level configuration variables.
- Removed .pyc from the _SKIPPED_FILES_WITHOUT_WARNING configuration
variable.
- Changed the numeric values of the FileType enum so that
FileType.NONE evaluates to False.
- For ProcessorDispatcher.should_skip_without_warning():
- Changed the method to return True for FileType.NONE files.
- Made ChangeLog files an exception to getting skipped.
- Changed the StyleProcessor.process() method to raise an
exception if given a file path that should not be processed.
* Scripts/webkitpy/style/checker_unittest.py:
- Updated the unit tests and added more test cases as necessary.
2010-04-28 Eric Seidel <eric@webkit.org>
Reviewed by Jeremy Orlow.
webkitpy: ScriptError('Failed to run "[u\'taskkill.exe\', u\'/f\', u\'/im\', u\'httpd.exe\']" exit_code: 128',)
https://bugs.webkit.org/show_bug.cgi?id=38248
The previous code did not check the return code of taskkill.
When I moved that callsite from using subprocess.call to
Executive.run_command having a non-zero return code became an error.
In this change I've centralized our killall handling in executive,
and added tests for it to make sure it works.
Currently kill_process and kill_all swallow exceptions in the cases
where the process(es) to be killed do(es) not exist.
* Scripts/webkitpy/common/system/executive.py:
* Scripts/webkitpy/common/system/executive_unittest.py:
* Scripts/webkitpy/layout_tests/port/chromium_linux.py:
* Scripts/webkitpy/layout_tests/port/chromium_mac.py:
* Scripts/webkitpy/layout_tests/port/chromium_win.py:
* Scripts/webkitpy/layout_tests/port/gtk.py:
* Scripts/webkitpy/layout_tests/port/mac.py:
* Scripts/webkitpy/layout_tests/port/qt.py:
* Scripts/webkitpy/layout_tests/port/win.py:
2010-04-28 Eric Seidel <eric@webkit.org>
Reviewed by Shinichiro Hamaji.
wdiff_text throws ScriptError because wdiff returns non-zero when files differ
https://bugs.webkit.org/show_bug.cgi?id=38246
wdiff returns 0 when files are the same, 1 when they differ.
run_command by default raises ScriptError if the return code is non-zero.
Fixed this by adding a custom error handler which only raises if the
return code is not 1.
I broke up the huge wdiff_text() method into little pieces
for easier unit testing. There is only one functional change here
and that is the addition of the custom error handler.
* Scripts/webkitpy/layout_tests/port/base.py:
* Scripts/webkitpy/layout_tests/port/base_unittest.py:
2010-04-28 Fumitoshi Ukai <ukai@chromium.org>
Unreviewed build fix.
* DumpRenderTree/chromium/TestShell.cpp:
(TestShell::dumpImage): format '%u' expects type 'unsigned int', but argument 2 has type 'size_t'.
* DumpRenderTree/chromium/WebViewHost.cpp:
(WebViewHost::willSendRequest): too few arguments for format.
2010-04-27 Shinichiro Hamaji <hamaji@chromium.org>
Reviewed by Darin Adler and Eric Seidel.
Add layoutTestController.setPrinting()
https://bugs.webkit.org/show_bug.cgi?id=37203
* DumpRenderTree/LayoutTestController.cpp:
(LayoutTestController::LayoutTestController):
(setPrintingCallback):
(LayoutTestController::staticFunctions):
* DumpRenderTree/LayoutTestController.h:
(LayoutTestController::isPrinting):
(LayoutTestController::setIsPrinting):
* DumpRenderTree/mac/DumpRenderTree.mm:
(dump):
2010-04-27 Michael Nordman <michaeln@google.com>
Reviewed by Dmitry Titov.
[Chromium] Add two things to the webkit API to support appcaches in workers.
1) WebURLRequest TargetTypes for worker and shared worker main resources.
2) Factory method on class WebCommonWorkerClient to
createApplicationCacheHost() for the associated worker.
https://bugs.webkit.org/show_bug.cgi?id=38147
* DumpRenderTree/chromium/TestWebWorker.h add a stub impl of the factory method
(TestWebWorker::createApplicationCacheHost):
2010-04-27 Adam Barth <abarth@webkit.org>
Reviewed by Eric Seidel.
run-bindings-tests doesn't work in an SVN checkout
https://bugs.webkit.org/show_bug.cgi?id=38225
Previously detect_scm_system needed an absolute path for SVN. Now we
accept a relative path.
* Scripts/webkitpy/common/checkout/scm.py:
* Scripts/webkitpy/common/checkout/scm_unittest.py:
2010-04-27 Sam Weinig <sam@webkit.org>
Reviewed by Stephanie Lewis.
Always build WebKit2 when building on SnowLeopard and later.
* Scripts/build-webkit:
2010-04-27 Sam Weinig <sam@webkit.org>
Reviewed by Maciej Stachowiak.
Fix for https://bugs.webkit.org/show_bug.cgi?id=38238
Allow both WebKit and WebKit2 to link to the same WebCore.framework
* Scripts/build-webkit: Remove the UMBRELLA_LDFLAGS override when building
WebCore for WebKit2, it is no longer necessary.
2010-04-27 James Robinson <jamesr@chromium.org>
Reviewed by David Levin.
Fix a typo in chromium.py that causes NRWT to fail in --verbose
mode in a Chromium checkout.
https://bugs.webkit.org/show_bug.cgi?id=38234
* Scripts/webkitpy/layout_tests/port/chromium.py:
2010-04-27 Darin Fisher <darin@chromium.org>
Reviewed by Dimitri Glazkov.
[Chromium] Remove deprecated form of didChangeLocationWithinPage
https://bugs.webkit.org/show_bug.cgi?id=38178
Switch over to implementing didNavigateWithinPage.
* DumpRenderTree/chromium/WebViewHost.cpp:
(TestWebViewDelegate::didNavigateWithinPage):
(WebViewHost::didChangeLocationWithinPage):
* DumpRenderTree/chromium/WebViewHost.h:
2010-04-27 Evan Martin <evan@chromium.org>
Unreviewed.
Adding myself to commmitters list.
* Scripts/webkitpy/common/config/committers.py:
2010-04-27 Adam Barth <abarth@webkit.org>
Reviewed by Darin Adler.
run-bindings-tests should use --reset-results instead of --overwrite
https://bugs.webkit.org/show_bug.cgi?id=38200
As requested by Ojan.
* Scripts/run-bindings-tests:
2010-04-27 Adam Barth <abarth@webkit.org>
Reviewed by Eric Seidel.
REGRESSION(r58261): webkit-patch edit-changelogs is broken
https://bugs.webkit.org/show_bug.cgi?id=38204
In 58261, we added code to abstract step that interrogates the squash
and git_commit options, but it doesn't declare that it uses those
options. That means any command that doesn't happen to declare those
options might be broken if it uses the cached_lookup mechanism.
* Scripts/webkitpy/tool/steps/abstractstep.py:
* Scripts/webkitpy/tool/steps/applypatch.py:
* Scripts/webkitpy/tool/steps/applypatchwithlocalcommit.py:
* Scripts/webkitpy/tool/steps/build.py:
* Scripts/webkitpy/tool/steps/checkstyle.py:
* Scripts/webkitpy/tool/steps/cleanworkingdirectory.py:
* Scripts/webkitpy/tool/steps/closebug.py:
* Scripts/webkitpy/tool/steps/closebugforlanddiff.py:
* Scripts/webkitpy/tool/steps/commit.py:
* Scripts/webkitpy/tool/steps/confirmdiff.py:
* Scripts/webkitpy/tool/steps/createbug.py:
* Scripts/webkitpy/tool/steps/ensurebuildersaregreen.py:
* Scripts/webkitpy/tool/steps/ensurelocalcommitifneeded.py:
* Scripts/webkitpy/tool/steps/obsoletepatches.py:
* Scripts/webkitpy/tool/steps/postcodereview.py:
* Scripts/webkitpy/tool/steps/postdiff.py:
* Scripts/webkitpy/tool/steps/preparechangelog.py:
* Scripts/webkitpy/tool/steps/runtests.py:
* Scripts/webkitpy/tool/steps/update.py:
* Scripts/webkitpy/tool/steps/updatechangelogswithreviewer.py:
* Scripts/webkitpy/tool/steps/validatereviewer.py:
2010-04-27 Eric Seidel <eric@webkit.org>
Reviewed by David Levin.
prepare-ChangeLog --bug fails on some CYGWIN installs due to missing certs
https://bugs.webkit.org/show_bug.cgi?id=38212
* Scripts/prepare-ChangeLog:
- Pass --insecure to curl to work around CYGWIN missing certs.
2010-04-27 Sam Weinig <sam@webkit.org>
Rubber-stamped by Beth Dakin.
It is no longer necessary to set ENABLE_EXPERIMENTAL_SINGLE_VIEW_MODE
or WTF_USE_WEB_THREAD. Remove them.
* Scripts/build-webkit:
2010-04-27 Chris Jerdonek <cjerdonek@webkit.org>
Reviewed by David Levin.
Changed Chris Jerdonek from committer to reviewer.
* Scripts/webkitpy/common/config/committers.py:
2010-04-27 Benjamin Poulain <benjamin.poulain@nokia.com>
Reviewed by Kenneth Rohde Christiansen.
[Qt] QGraphicsWebView: Arrow keys scroll the graphics-view instead of the web-page
https://bugs.webkit.org/show_bug.cgi?id=35834
The scene should always have the size of the web view otherwhise it is
possible to scroll the graphics view.
* QtLauncher/webview.cpp:
(WebViewGraphicsBased::resizeEvent):
2010-04-27 Diego Gonzalez <diegohcg@webkit.org>
Unreviewed.
Adding myself to committers.py
* Scripts/webkitpy/common/config/committers.py:
2010-04-27 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
[chromium] new-run-webkit-tests hangs on Chromium Bots (OS X and Linux)
https://bugs.webkit.org/show_bug.cgi?id=37987
After further research, I believe the hang is caused by:
http://bugs.python.org/issue2320
Basically Popen() is not reentrant.
The workaround is to pass close_fds=True to Popen() on Mac/Linux.
I fixed our main Popen wrapper "Executive.run_command" to use close_fds=True
when appropriate.
I audited all places we call Popen() and either moved them to run_command
or left a FIXME that they are not thread safe. A few places I added the
close_fds workaround there and left an explanitory note.
* Scripts/webkitpy/common/checkout/scm_unittest.py:
- Added note that this Popen use is not threadsafe.
* Scripts/webkitpy/common/system/executive.py:
- Fixed our Executive.run_* to workaround python bug 2320.
* Scripts/webkitpy/common/system/user.py:
_ Added note that this Popen use is not threadsafe.
* Scripts/webkitpy/layout_tests/layout_package/json_results_generator.py: ditto.
* Scripts/webkitpy/layout_tests/port/apache_http_server.py: ditto.
* Scripts/webkitpy/layout_tests/port/base.py:
- Change wdiff back to using run_command now that we believe it
to be threadsafe.
* Scripts/webkitpy/layout_tests/port/chromium.py:
- Fix to use Executive in places.
- Pass self._executive down to the Driver for easier unit testing.
* Scripts/webkitpy/layout_tests/port/chromium_win.py:
- Re-factor to use a _kill_all method.
- Made the _kill_all method use run_command to be threadsafe.
* Scripts/webkitpy/layout_tests/port/http_server.py:
- Add FIXME about using Executive.
* Scripts/webkitpy/layout_tests/port/server_process.py:
- Use Executive to be threadsafe.
* Scripts/webkitpy/layout_tests/port/webkit.py:
- Pass self._executive down to the Driver.
* Scripts/webkitpy/layout_tests/port/websocket_server.py:
- Add note about Popen not being threadsafe.
* Scripts/webkitpy/layout_tests/rebaseline_chromium_webkit_tests.py:
- Move one caller to run_command add notes about moving others.
2010-04-27 Adam Barth <abarth@webkit.org>
Reviewed by Maciej Stachowiak.
REGRESSION(r58261): webkit-patch upload does not work in an SVN checkout.
https://bugs.webkit.org/show_bug.cgi?id=38186
Unfortunately, we don't have a good way of testing this change because
our test coverage of the scm.py API is poor...
* Scripts/webkitpy/common/checkout/scm.py:
* Scripts/webkitpy/tool/steps/preparechangelog.py:
2010-04-26 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
Remove unused code from text_diff.py
https://bugs.webkit.org/show_bug.cgi?id=38170
* Scripts/webkitpy/layout_tests/test_types/text_diff.py:
- Remove is_render_tree_dump which appears unused.
2010-04-26 Adam Barth <abarth@webkit.org>
Reviewed by Eric Seidel.
webkit-patch pretty-diff is broken
https://bugs.webkit.org/show_bug.cgi?id=38172
We need to register for these options because they're used when we look
up the diff.
* Scripts/webkitpy/tool/steps/confirmdiff.py:
2010-04-26 Adam Barth <abarth@webkit.org>
Reviewed by Eric Seidel.
Add ObjC and GObject to run-bindings-test
https://bugs.webkit.org/show_bug.cgi?id=38168
* Scripts/run-bindings-tests:
2010-04-26 Adam Barth <abarth@webkit.org>
Reviewed by Eric Seidel.
Add testing infrastructure for JSC bindings generator
https://bugs.webkit.org/show_bug.cgi?id=38167
Add support for testing more than one bindings. Also, converted the
script to PEP8 style.
* Scripts/run-bindings-tests:
2010-04-26 Eric Seidel <eric@webkit.org>
Reviewed by Adam Barth.
Refactor results.html generation out into a new method and test it
https://bugs.webkit.org/show_bug.cgi?id=38164
Hopefully this results in no change in functionality.
* Scripts/webkitpy/layout_tests/run_webkit_tests.py:
* Scripts/webkitpy/layout_tests/run_webkit_tests_unittest.py:
2010-04-26 Adam Barth <abarth@webkit.org>
Unreviewed, rolling out r58265.
http://trac.webkit.org/changeset/58265
https://bugs.webkit.org/show_bug.cgi?id=38021
This change prevents me from uploading patches. It also breaks sheriff-bot.
* Script