GitLab

http://src.chromium.org/viewvc/blink?view=revision&revision=170239

BUG=345027
TBR=enne@chromium.org

Review URL: https://codereview.chromium.org/215913003

git-svn-id: svn://svn.chromium.org/blink/trunk@170273 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Remove unused selection related methods from WebView that got moved to WebFrame in r169968 now that Chrome uses the new ones.

BUG=304341

Review URL: https://codereview.chromium.org/215703002

git-svn-id: svn://svn.chromium.org/blink/trunk@170272 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This re-applies https://src.chromium.org/viewvc/blink?revision=165673&view=revision.

BUG=357151

Review URL: https://codereview.chromium.org/215313003

git-svn-id: svn://svn.chromium.org/blink/trunk@170271 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Since the unprefixed API isn't event implemented yet removal is nowhere
in sight, but it's still interesting to track these features over time.

BUG=none

Review URL: https://codereview.chromium.org/214873002

git-svn-id: svn://svn.chromium.org/blink/trunk@170270 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Review URL: https://codereview.chromium.org/214523005

git-svn-id: svn://svn.chromium.org/blink/trunk@170269 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This was done in r165624 but was wrong, you cannot ever synchronously
dispatch the load or error event from inside didFinish since it means
you run can end up running script inside ::insertedInto.

It's not clear from the bug what the issue was, but the timing in the
test that I updated was also wrong since it had <link> elements firing
their load events synchronously while parsing.

There might be a real timing bug in imports, but firing the event
synchronously inside ::didFinish() isn't how to fix it.

BUG=336113, 357435, 357445

Review URL: https://codereview.chromium.org/214513011

git-svn-id: svn://svn.chromium.org/blink/trunk@170268 bbb929c8-8fbe-4397-9dbb-9b2b20218538

To eventually remove the ChildNodesLazySnapshot we need to stop allowing
script to run inside ::insertedInto. It's also important since even with
the lazy snapshot we still keep getting security bugs down there.

This starts the process of removal by switching HTMLMetaElement to use
a post insertion callback.

R=ojan@chromium.org,abarth@chromium.org
BUG=357435

Review URL: https://codereview.chromium.org/214003003

git-svn-id: svn://svn.chromium.org/blink/trunk@170267 bbb929c8-8fbe-4397-9dbb-9b2b20218538

I forgot to replace a couple of adoptRef()s in r170164.
SimulatedMouseEvent's adoptRef() was another left-over.

BUG=340522
TBR=ager

Review URL: https://codereview.chromium.org/216283002

git-svn-id: svn://svn.chromium.org/blink/trunk@170266 bbb929c8-8fbe-4397-9dbb-9b2b20218538

- w3c/web-platform-tests/html-templates/parsing-html-templates/creating-an-element-for-the-token/template-owner-document.html [ Pass Timeout ]
- virtual/implsidepainting/inspector/timeline/timeline-decode-resize.html [ Crash Pass Slow ]
- inspector-protocol/page/enable-disable.html [ Failure Pass ]
- virtual/threaded/animations/3d/matrix-transform-type-animation.html [ Pass Timeout ]
- fast/events/touch/gesture/touch-gesture-scroll-div-removed.html [ Crash Pass ]
- fast/mediastream/RTCPeerConnection-AddRemoveStream.html [ Crash Pass ]
- http/tests/security/contentSecurityPolicy/1.1/frame-ancestors/frame-ancestors-nested-cross-in-cross-none-block.html [ Failure Pass ]
- virtual/softwarecompositing/visibility/visibility-composited-incremental-update.html [ Pass Timeout ]
- virtual/threaded/animations/keyframe-timing-functions-multiple-animations.html [ Pass Timeout ]
- css3/flexbox/repaint.html [ Pass Failure ]

BUG=330086,318490,316349,357459,357460,357464,357464,357465,357465,355463
TBR=enne@chromium.org

Review URL: https://codereview.chromium.org/215983003

git-svn-id: svn://svn.chromium.org/blink/trunk@170265 bbb929c8-8fbe-4397-9dbb-9b2b20218538

window.atob() returns wrong value when given a string container only white
spaces. The reason was that base64DecodeInternal() was calling Vector::grow()
before the loop but failed to call Vector::shrink() in case of early return.

R=
BUG=357332

Review URL: https://codereview.chromium.org/215833002

git-svn-id: svn://svn.chromium.org/blink/trunk@170264 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Now that all the tests have been moved off using shouldRejectPromiseWithNull() these temporary helpers can be removed.

BUG=245025

Review URL: https://codereview.chromium.org/212173007

git-svn-id: svn://svn.chromium.org/blink/trunk@170263 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://src.chromium.org/viewvc/blink?view=revision&revision=170232

TBR=senorblanco@chromium.org

Review URL: https://codereview.chromium.org/216253002

git-svn-id: svn://svn.chromium.org/blink/trunk@170262 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Now that it's guaranteed that DOMWrapperWorld is not 0 at any given time, we can make it a reference instead of a pointer.

BUG=341032

Review URL: https://codereview.chromium.org/209713003

git-svn-id: svn://svn.chromium.org/blink/trunk@170261 bbb929c8-8fbe-4397-9dbb-9b2b20218538

the implementation of shape-outside, it's worth to remove shape-inside code at this point for now.

A list of reasons for the removal:

- Shape-inside is only part of Shapes Level 2, which needs to be improved on some topics.
- Shape-inside is lack of new shapes support (e.g. inset).
- Deprecated shapes are removed from the code (e.g. rectangle), which affects shape-inside.
- The current shape-inside code spreads across the layout code.
- The current shape-inside implementation is experimental in some areas, and the partially implemented code can have security implications.
- Removal of shape-inside opens possibilities for code complexity and performance optimizations for shape-outside. (e.g. simpler geometry code) 

BUG=356778

Review URL: https://codereview.chromium.org/209443007

git-svn-id: svn://svn.chromium.org/blink/trunk@170260 bbb929c8-8fbe-4397-9dbb-9b2b20218538

According to the spec (http://www.whatwg.org/specs/web-apps/current-work/multipage/the-button-element.html#the-select-element) the idl for select.add is
void add((HTMLOptionElement or HTMLOptGroupElement) element, optional (HTMLElement or long)? before = null);

select.add should accept an optgroup as the first argument.
select.add should support an index as the second argument.

BUG=345345

Review URL: https://codereview.chromium.org/213773003

git-svn-id: svn://svn.chromium.org/blink/trunk@170259 bbb929c8-8fbe-4397-9dbb-9b2b20218538

interleaved-fragments.html, invalid-continuation.html, zero-length-text.html,
connect-src-websocket-allowed.html and handshake-error.html now pass.

All other WebSocket tests which were set to "Skip" have been changed to
"Failure", except for multiple-connections.html, which times out.

Tests of functionality which is not implemented in the new implementation
remain skipped. These are
FIXME-handshake-fail-by-null-subprotocol-selection-whitelist.html,
handshake-fail-by-no-cr.html and deflate-frame-*

BUG=339373
TEST=layout tests

Review URL: https://codereview.chromium.org/214473002

git-svn-id: svn://svn.chromium.org/blink/trunk@170258 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch sorts animations in the stack so that they will be applied in the
correct order. Note that due to the way compositor integration currently is
implemented, we don't know the exact time when new animations will be started,
but expect it will be very close to the current timeline time.

http://dev.w3.org/fxtf/web-animations/#the-animation-stack

BUG=

Review URL: https://codereview.chromium.org/210703002

git-svn-id: svn://svn.chromium.org/blink/trunk@170257 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Currently derefBase() does not do the decrement from 1 to 0 because an object that has 0 refcount is going to die immediately (there is no need to do the decrement).

However, in the oilpan world, it's possible that an object that has 0 refcount is resurrected, because the object can still be retained through oilpan handles. Thus we need to manage the refcount correctly even when the refcount is going to be 0.

I'm making this change separately from oilpan changes, since this change might regress performance of something. derefBase() is performance-sensitive.

BUG=340522

Review URL: https://codereview.chromium.org/214453005

git-svn-id: svn://svn.chromium.org/blink/trunk@170256 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Now we're implementing the DOM-object leak detector. When the number of
DOM objects is increased after a test, we can assume that the test
causes a leak. This feature is available at content_shell with the flag
--enable-leak-detection.

This patch enables this leak detection feature at run_webkit_test.py
with the flag --enable-leak-detection.

content_shell is assumed to report a detected leak with the following
format:

  #LEAK - renderer pid PID (DETAIL_IN_JSON_FORMAT)

and run_webkit_test.py will treat this as a kind of error by this patch.

See also:
https://docs.google.com/a/chromium.org/document/d/1sFAsZxeISKnbGdXoLZlB2tDZ8pvO102ePFQx6TX4X14/edit

BUG=332630
TEST=test-webkitpy

Review URL: https://codereview.chromium.org/148153009

git-svn-id: svn://svn.chromium.org/blink/trunk@170255 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The callback interface code was the first code and template we wrote.
Since then we've improved style and template usage; this simplifies
the code and brings it up-to-date.

TBR=haraken
BUG=345503

Review URL: https://codereview.chromium.org/215993002

git-svn-id: svn://svn.chromium.org/blink/trunk@170254 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=333040

Review URL: https://codereview.chromium.org/214413010

git-svn-id: svn://svn.chromium.org/blink/trunk@170253 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=245025

Review URL: https://codereview.chromium.org/215153005

git-svn-id: svn://svn.chromium.org/blink/trunk@170252 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://src.chromium.org/viewvc/blink?view=revision&revision=170214

BUG=345027
TBR=enne@chromium.org

Review URL: https://codereview.chromium.org/216153002

git-svn-id: svn://svn.chromium.org/blink/trunk@170251 bbb929c8-8fbe-4397-9dbb-9b2b20218538

We used to squash RenderLayers even if they reported that
subtreeIsInvisible. This caused the creation of needlessly enormous
layers that contained nothing. This CL excludes those layers from
the list of squashed layers.

R=abarth@chromium.org
BUG=None

Review URL: https://codereview.chromium.org/212923005

git-svn-id: svn://svn.chromium.org/blink/trunk@170250 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This was attempted long ago but then reverted because it broke some tests
related to text transforms and disabled states [1]. Those tests don't seem
to fail anymore though and I can't think of any reason why sharing style
should change the rendering of RenderMenuList.

Enabling this makes http://jsfiddle.net/p4saV/1/ 56% faster where you have
5000 <option>'s and all of bootstrap's CSS.

[1] https://bugs.webkit.org/show_bug.cgi?id=88405

Review URL: https://codereview.chromium.org/215723002

git-svn-id: svn://svn.chromium.org/blink/trunk@170249 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Revert of Revert of Make start/stop loading notifications per-frame (https://codereview.chromium.org/215533004/)

Reason for revert:
r170192 was not a culprit.


Original issue's description:
> Revert of Make start/stop loading notifications per-frame (https://codereview.chromium.org/183793002/)
> 
> Reason for revert:
> Possibly causing content_browsertests EndToEnd to fail.
> 
> See: http://build.chromium.org/p/chromium.webkit/builders/Linux%20Tests%20%28dbg%29/builds/2095/steps/content_browsertests/logs/stdio
> 
> Output:
> [20917:20917:0327/122353:3268961829:INFO:CONSOLE(0)] "Uncaught TypeError: Cannot convert undefined or null to object", source: http://127.0.0.1:34379/files/web_ui_mojo.html (0)
> [20917:20917:0327/122353:3269022437:INFO:CONSOLE(28)] "Uncaught TypeError: undefined is not a function", source: http://127.0.0.1:34379/files/web_ui_mojo.js (28)
> BrowserTestBase signal handler received SIGTERM. Backtrace:
> #0 0x000000a593be base::debug::StackTrace::StackTrace()
> #1 0x000002bfc17b content::(anonymous namespace)::DumpStackTraceSignalHandler()
> #2 0x7f9c3e4ab4a0 <unknown>
> #3 0x7f9c3e55da43 __poll
> #4 0x7f9c42aabff6 <unknown>
> #5 0x7f9c42aac124 g_main_context_iteration
> #6 0x000000aaaf5f base::MessagePumpGlib::RunWithDispatcher()
> #7 0x000000a8e802 base::RunLoop::Run()
> #8 0x00000058256e content::(anonymous namespace)::WebUIMojoTest_EndToEnd_Test::RunTestOnMainThread()
> 
> Original issue's description:
> > Make start/stop loading notifications per-frame
> > 
> > Also, make ProgressTracker a frame-level concept rather than a page-level concept and merge it with FrameLoader's FrameProgressTracker helper. Send per-frame progress change notifications accordingly.
> > 
> > This depends on https://codereview.chromium.org/180113003/ landing in chromium.
> > 
> > BUG=347643
> > 
> > Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=170192
> 
> TBR=abarth@chromium.org,japhet@chromium.org
> NOTREECHECKS=true
> NOTRY=true
> BUG=347643
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=170218

TBR=abarth@chromium.org,japhet@chromium.org,enne@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=347643

Review URL: https://codereview.chromium.org/216083002

git-svn-id: svn://svn.chromium.org/blink/trunk@170248 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=

Review URL: https://codereview.chromium.org/212983012

git-svn-id: svn://svn.chromium.org/blink/trunk@170247 bbb929c8-8fbe-4397-9dbb-9b2b20218538

We need to change
  HashMap<ExecutionContext*, RefPtr<DatabaseContext>>
to
  HashMap<ExecutionContext*, OwnPtr<Persistent<DatabaseContext>>>
Because we can't use PersistentHeapHashMap due to mutiple threads, and can't put
Persistent<> to HashMap directly.

BUG=347902

Review URL: https://codereview.chromium.org/215063005

git-svn-id: svn://svn.chromium.org/blink/trunk@170246 bbb929c8-8fbe-4397-9dbb-9b2b20218538

In C++, it's not legal to call member functions on objects that are null
pointers.

R=esprehn@chromium.org

Review URL: https://codereview.chromium.org/215773002

git-svn-id: svn://svn.chromium.org/blink/trunk@170245 bbb929c8-8fbe-4397-9dbb-9b2b20218538

R=adamk@chromium.org
BUG=355658,178319,161068

Review URL: https://codereview.chromium.org/209363002

git-svn-id: svn://svn.chromium.org/blink/trunk@170243 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Controlled by explicit API now, per
https://codereview.chromium.org/210243004 and must land after https://codereview.chromium.org/212953007/ .

R=eae@chromium.org, jamesr@chromium.org
BUG=333029

Review URL: https://codereview.chromium.org/213173003

git-svn-id: svn://svn.chromium.org/blink/trunk@170242 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This lets us re-land http://crrev.com/208393008 without breaking
AwSettingsTest.testLayoutAlgorithmWithTwoViews.

Without http://crrev.com/208393008, style recalc would always reset the
autosizing multiplier to 1, so the bug in recalculateMultipliers was not
noticed.

BUG=353309,356960

Review URL: https://codereview.chromium.org/215063004

git-svn-id: svn://svn.chromium.org/blink/trunk@170241 bbb929c8-8fbe-4397-9dbb-9b2b20218538

- web-animations-api/change-in-animation-frame.html [ Failure Pass ]
- compositing/plugins/invalidate_rect.html [ Failure Pass ]
- virtual/softwarecompositing/plugins/invalidate_rect.html [ Failure Pass ]
- svg/custom/js-late-pattern-and-object-creation.svg [ ImageOnlyFailure Pass ]

BUG=357437,357438,357439
TBR=enne@chromium.org

Review URL: https://codereview.chromium.org/216023002

git-svn-id: svn://svn.chromium.org/blink/trunk@170240 bbb929c8-8fbe-4397-9dbb-9b2b20218538

These are the remaining tests that call display() directly and also do
not need actually need a paint phase to test what they intend to.

BUG=345027

Review URL: https://codereview.chromium.org/196573042

git-svn-id: svn://svn.chromium.org/blink/trunk@170239 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Chromium now sends ctrl+wheel events to webkit first before using them for zooming. Rather than prevent scrolling for these events in EventHandler::handleWheel, push that check a couple levels deeper into ScrollableArea::handleWheelEvent.  In addition to EventHandler, this code is also called by WebPluginScrollbarImpl and so will fix some additional cases.

This is mostly already tested by the fast/event/wheelevent-ctrl.html test I added in http://crrev.com/183403002.  I've extended that test to cover document and div scrolling cases and ensured it fails reliably if either the ScrollableArea or EventHandler::defaultWheelEventHandler check is removed. 

BUG=352474, 111059

Review URL: https://codereview.chromium.org/212913002

git-svn-id: svn://svn.chromium.org/blink/trunk@170238 bbb929c8-8fbe-4397-9dbb-9b2b20218538

All of these tests don't seem to depend on minimizing repaints and are
more trying to make sure that something has repainted.  Therefore,
change them all to just use rAF rather than repaint.js and make them
pixel tests where applicable.

BUG=345027

Review URL: https://codereview.chromium.org/207383002

git-svn-id: svn://svn.chromium.org/blink/trunk@170237 bbb929c8-8fbe-4397-9dbb-9b2b20218538

- inspector/sources/debugger/script-formatter-breakpoints-2.html [ Pass Timeout ]

- http/tests/workers/terminate-during-sync-operation.html [ Crash Pass Slow Timeout ]

BUG=357004,357427
TBR=enne@chromium.org

Review URL: https://codereview.chromium.org/215153008

git-svn-id: svn://svn.chromium.org/blink/trunk@170236 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Per request of haraken:
https://codereview.chromium.org/212593008/#msg4

TBR=haraken

Review URL: https://codereview.chromium.org/214283002

git-svn-id: svn://svn.chromium.org/blink/trunk@170235 bbb929c8-8fbe-4397-9dbb-9b2b20218538

When calling RenderLayerClipper::backgroundClipRect on the
RenderLayerClipper object corresponding to the RenderLayer
which is equal to clipRectsContext.rootLayer, it first calls
parentClipRects on the clipper of that RenderLayer's parent.
In turn, this starts a recursive walk up the RenderLayer
tree in updateClipRects. This is incorrect, since a
RenderLayerClipper object is unable to clip above rootLayer
in the render tree. In particular, the recursion condition
in updateClipRects
depends on the recursion starting at a layer which is below
or equal to rootLayer.

BUG=336563

Review URL: https://codereview.chromium.org/211683003

git-svn-id: svn://svn.chromium.org/blink/trunk@170234 bbb929c8-8fbe-4397-9dbb-9b2b20218538

A bunch of code was needlessly null-checking enclosingLayer. That function can
only return zero for RenderObjects that are outside of the render tree.

R=esprehn@chromium.org

Review URL: https://codereview.chromium.org/215843002

git-svn-id: svn://svn.chromium.org/blink/trunk@170233 bbb929c8-8fbe-4397-9dbb-9b2b20218538