GitLab

TBR=dcheng
NOTRY=true

Review URL: https://codereview.chromium.org/702003003

git-svn-id: svn://svn.chromium.org/blink/trunk@184866 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=none

Review URL: https://codereview.chromium.org/673163004

git-svn-id: svn://svn.chromium.org/blink/trunk@184865 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The revalidation headers ("if-modified-since", "if-none-match") should not be exposed to the ServiceWorker.

BUG=429570
TEST=http/tests/serviceworker/resources/fetch-request-no-freshness-headers-iframe.html

Review URL: https://codereview.chromium.org/703813002

git-svn-id: svn://svn.chromium.org/blink/trunk@184864 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Move the v8::Isolate* parameter to the first parameter of various binding methods in third_party/WebKit/Source/bindings

This step is to move the v8::Isolate* parameter to the first parameter of various binding methods in third_party/WebKit/Source/bindings
(Ref: binding-methods changed:
static v8::Handle<v8::Object> associateObjectWithWrapperNonTemplate(ScriptWrappable*, const WrapperTypeInfo*, v8::Handle<v8::Object>, v8::Isolate*)
static v8::Handle<v8::Object> associateObjectWithWrapperNonTemplate(Node*, const WrapperTypeInfo*, v8::Handle<v8::Object>, v8::Isolate*)
static v8::Handle<v8::Object> associateObjectWithWrapper(PassRefPtr<T>, const WrapperTypeInfo*, v8::Handle<v8::Object>, v8:Isolate*)
static v8::Handle<v8::Object> associateObjectWithWrapper(T*, const WrapperTypeInfo*, v8::Handle<v8::Object>, v8::Isolate*))

BUG=424446

Review URL: https://codereview.chromium.org/695773002

git-svn-id: svn://svn.chromium.org/blink/trunk@184863 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=429176
R=apavlov

Review URL: https://codereview.chromium.org/706523003

git-svn-id: svn://svn.chromium.org/blink/trunk@184862 bbb929c8-8fbe-4397-9dbb-9b2b20218538

There is a bug due to which source ranges in inherited styles sections
and pseudo elements style sections are not updated on the front-end
as the editing happens. This patch fixes it.

BUG=430129
R=vsevik

Review URL: https://codereview.chromium.org/698113006

git-svn-id: svn://svn.chromium.org/blink/trunk@184861 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Window has custom toV8 function, and Window inherits from EventTarget.
So, EventTarget also has custom toV8 because an EventTarget can be a
Window.

crrev.com/646523004 made Node to call EventTarget-version of
v8SetReturnValueForMainWorld because Node inherits from EventTarget,
and it calls toV8 function every time. It caused a perf regression.

This CL adds a new version of v8SetReturnValueForMainWorld for Node
and its subclasses so that they can use the fastest path.

class hierarchy:
    ScriptWrappable <-- EventTarget <--+-- Node <-- ...
                                       +-- Window
overloads:
    v8SetReturnValueForMainWorld(ScriptWrappable*)
        Optimized and very fast.
    v8SetReturnValueForMainWorld(EventTarget*)
        Uses custom toV8 function and slow.
    v8SetReturnValueForMainWorld(Node*)  <----------- NEW!!
        Optimized and very fast.
    v8SetReturnValueForMainWorld(Window*)
        Uses custom toV8 function and slow.

Before this CL:
Element matches EventTarget version of
v8SetReturnValueForMainWorld, and it's very slow.

After this CL:
Element matches Node version of
v8SetReturnValueForMainWorld, and it's very fast.

BUG=428707,235436

Review URL: https://codereview.chromium.org/684763004

git-svn-id: svn://svn.chromium.org/blink/trunk@184860 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Pesky overrun issues.

BUG=430351

Review URL: https://codereview.chromium.org/705663003

git-svn-id: svn://svn.chromium.org/blink/trunk@184859 bbb929c8-8fbe-4397-9dbb-9b2b20218538

According to the grid and flexbox specs:
"the ::first-line and ::first-letter pseudo-elements do not apply to
grid/flex containers".
http://dev.w3.org/csswg/css-grid/#grid-containers
http://dev.w3.org/csswg/css-flexbox/#flex-containers

This was almost working right, except in the case that an ancestor was
setting the ::first-letter pseudo-element.

Modified RenderBlock::updateFirstLetter() in order to stop looking for
the first text child when you reach a grid or flexbox.
Added a few more cases to the current tests in order to check this
behavior.

TEST=css3/flexbox/flexbox-ignore-container-firstLetter.html
TEST=fast/css-grid-layout/grid-container-ignore-first-letter.html

BUG=430099

Review URL: https://codereview.chromium.org/686173006

git-svn-id: svn://svn.chromium.org/blink/trunk@184858 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The patch fixes CSS.getMatchedStylesForNode protocol method so that
it starts to report matched selectors for pseudo style matches.

R=vsevik, apavlov

Review URL: https://codereview.chromium.org/697773004

git-svn-id: svn://svn.chromium.org/blink/trunk@184857 bbb929c8-8fbe-4397-9dbb-9b2b20218538

DevTools shows "there are unsaved changes" dialog if there's at least one
dirty uiSourceCode. However, saving a network-originated javascript file will
not prevent you from loosing changes as the devtools will close.

This patch changes the behavior so that the alerting message pops up only
if there are file system uiSourceCodes with unsaved changes.

R=vsevik
BUG=426783

Review URL: https://codereview.chromium.org/698053003

git-svn-id: svn://svn.chromium.org/blink/trunk@184856 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Layout test not needed.

Review URL: https://codereview.chromium.org/691493005

git-svn-id: svn://svn.chromium.org/blink/trunk@184855 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Before this series of patches, registration succeeds even if the worker script
has a syntax error because the installing sequence does not check the result of
the script evaluation.

To enable the installing sequence to wait for the script evaluation, this adds
WorkerReportingProxy::didEvaluteWorkerScript() to be called when the
evaluation is completed and to propagate the result up to the Chromium-side.

Patch dependency:
[1] Blink: THIS PATCH
[2] Chromium: https://codereview.chromium.org/697593002/
[3] Blink: https://codereview.chromium.org/697833004/

BUG=426344
TEST=compile (test will be added by [3])

Review URL: https://codereview.chromium.org/692003002

git-svn-id: svn://svn.chromium.org/blink/trunk@184854 bbb929c8-8fbe-4397-9dbb-9b2b20218538

A previous CL allowed word spacing to be added around inline-blocks.
However the line width calculation wasn't consistent with children placement,
and contents will overflow on the right side of the line if there was
inline-block induced word spacing.

This CL makes RenderBlockFlow::computeInlineDirectionPositionsForSegment to be
consistent with InlineFlowBox::placeBoxRangeInInlineDirection.

BUG=429655

Review URL: https://codereview.chromium.org/704693003

git-svn-id: svn://svn.chromium.org/blink/trunk@184853 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This is based on the current heuristic in
plugins::PluginPlaceholder::HidePlugin
(components/plugins/renderer/plugin_placeholder.cc).

Tests are added to ensure that the placeholder closing code works as
expected, and to verify that the newly-added WebPluginPlaceholder method
is respected.

A crude UI is currently used; showing the X graphics currently used is blocked
on deciding how to load image resources in this environment.

BUG=364716

Review URL: https://codereview.chromium.org/698533003

git-svn-id: svn://svn.chromium.org/blink/trunk@184852 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Run-time flag enableImageColorProfiles defaults to off. Add a window
InternalSettings method to enable the flag during layout tests.

BUG=369787

Review URL: https://codereview.chromium.org/643713006

git-svn-id: svn://svn.chromium.org/blink/trunk@184851 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Hide isSpecialElement since it is only used in
htmlediting.cpp. And since the two callers pass
a non-null node, turn the null check into an ASSERT.

Review URL: https://codereview.chromium.org/694893003

git-svn-id: svn://svn.chromium.org/blink/trunk@184850 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=425651
R=eroman@chromium.org

Review URL: https://codereview.chromium.org/692763002

git-svn-id: svn://svn.chromium.org/blink/trunk@184849 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Revert of Widen GPU trigger condition (patchset #3 id:40001 of https://codereview.chromium.org/660133005/)

Reason for revert:
There's content that depends on the existing trigger in order to
opt-out of GPU rasterization. Since we're close to the branch point,
let's revert this and then try to expand again in M41.

Original issue's description:
> Widen GPU trigger condition
> 
> GPU rasterization will be triggered when viewports meet either of the 
> following conditions:
> 1) width=device-width, minimum-scale=X, where X >= 1.0
> 2) width=device-width, user-scalable=no
> 
> BUG=424469
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=184566

BUG=424469

Review URL: https://codereview.chromium.org/685003008

git-svn-id: svn://svn.chromium.org/blink/trunk@184848 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Some specs use union type attributes. For example, TrackEvent[1]
has 'track' attribute, of which type is
(VideoTrack or AudioTrack or TextTrack). Let's support union type
attributes so that we can reduce custom bindings. Follow-up CLs
will replace custom bindings if possible.

BUG=240176

Review URL: https://codereview.chromium.org/704503002

git-svn-id: svn://svn.chromium.org/blink/trunk@184847 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://src.chromium.org/viewvc/blink?view=revision&revision=184840

BUG=428189
TBR=bokan@chromium.org

Review URL: https://codereview.chromium.org/701783006

git-svn-id: svn://svn.chromium.org/blink/trunk@184846 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch devirtualizes animatedLocalTransform and renames it
calculateAnimatedLocalTransform to emphasize that it is not cheap.

SVGGraphicsElement::animatedLocalTransform was virtual to support
SVGTextElement's special transform code. The text-specific
transform code is only in an infrequently-taken codepath
(style && style->hasTransform()) so we can actually improve performance
by devirtalizing and combining the two transform codepaths. The real
benefit is code simplification though.

Secondly, I've renamed this function to make it clear that it is not
a simple lookup.

BUG=429551

Review URL: https://codereview.chromium.org/693423002

git-svn-id: svn://svn.chromium.org/blink/trunk@184845 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Review URL: https://codereview.chromium.org/695563008

git-svn-id: svn://svn.chromium.org/blink/trunk@184844 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This code just is the hasOneTextChild logic, so
to avoid confusion use that instead.

Review URL: https://codereview.chromium.org/692153002

git-svn-id: svn://svn.chromium.org/blink/trunk@184843 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://src.chromium.org/viewvc/blink?view=revision&revision=184838

BUG=416535
TBR=wangxianzhu@chromium.org

Review URL: https://codereview.chromium.org/703473005

git-svn-id: svn://svn.chromium.org/blink/trunk@184842 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://src.chromium.org/viewvc/blink?view=revision&revision=184836

BUG=425872
TBR=chrishtr@chromium.org

Review URL: https://codereview.chromium.org/685243003

git-svn-id: svn://svn.chromium.org/blink/trunk@184841 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The svg test turned out to simply need a rebaseline as the scrollbars should
no longer show up in layout tests.

The autoresize test was off by 1 pixel. This was happening because in
virtual viewport mode we size the main FrameView's width to the content
size and the height by the aspect ratio. This meant that the height would
have some tiny error in some aspect ratios which would get blown up to a
full pixel when we returned an expandedIntSize in
PageScaleConstraintsSet::mainFrameSize. This is fixed by explicitly setting
the FrameView size equal to the viewport size if the content width matches
the viewport width.

BUG=428189

Review URL: https://codereview.chromium.org/699253002

git-svn-id: svn://svn.chromium.org/blink/trunk@184840 bbb929c8-8fbe-4397-9dbb-9b2b20218538

In three patches the initial mock implementation of bluetooth.requestDevice
is implemented across blink & content. This allows layout tests to specify the
data responses the mock implemented in content should return via
testRunner.SetBluetoothMockDataSet and call bluetooth.requestDevice with the
expected results.

crrev.com/650613005 blink::WebBluetooth & WebBluetoothError interfaces.
crrev.com/702593002 content::WebBluetoothImpl & testRunner.SetBluetoothMockDataSet.
crrev.com/686813003 blink::BluetoothDiscovery::requestDevice implemented.

BUG=420284

Review URL: https://codereview.chromium.org/650613005

git-svn-id: svn://svn.chromium.org/blink/trunk@184839 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This reverts commit 20feb3edc199015a96b39c081d50a083cf39ac87.

We still can't clearly distinguish invalidations for container layer
and scrolling contents layer. For example, crbug.com/429845, the shadow
div of an editor is scrollable, but the editor invalidates itself to
update selection, caret and content; With the change, the invalidation
will happen on the container only.

Revert first and try to find a good solution.

Perhaps we can just wait for slimming paint.

Still keep the layout tests added in the original change but will be
rebaselined.

TBR=vollick@chromium.org,chrishtr@chromium.org,dsinclair@chromium.org
BUG=416535,429845

Review URL: https://codereview.chromium.org/701873002

git-svn-id: svn://svn.chromium.org/blink/trunk@184838 bbb929c8-8fbe-4397-9dbb-9b2b20218538

These tests appear to be running normally now on the Mac Oilpan clients.

BUG=370586
TEST=flakiness dashboard all green for these tests

Review URL: https://codereview.chromium.org/685983006

git-svn-id: svn://svn.chromium.org/blink/trunk@184837 bbb929c8-8fbe-4397-9dbb-9b2b20218538

To do this, add a method to request full paint invalidation on FrameView. This also requires moving the full invalidation code into invalidateTreeIfNeeded() instead of layout(), since otherwise the invalidation requested will not always happen if layout does not occur.

In any case, invalidation should be in invalidateTreeIfNeeded(), not layout().

BUG=425872

Review URL: https://codereview.chromium.org/683373002

git-svn-id: svn://svn.chromium.org/blink/trunk@184836 bbb929c8-8fbe-4397-9dbb-9b2b20218538

"Threadable" in class names came from core/loader (e.g. ThreadableLoader),
but it doesn't have any meaning in modules/websockets. This CL drops
that adjective from class names.

R=tyoshino,mkwst
BUG=NONE

Review URL: https://codereview.chromium.org/701753002

git-svn-id: svn://svn.chromium.org/blink/trunk@184835 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Using C++11 range-based loop.
Using ">>" instead of "> >" in templates.
Using nullptr instead of 0.


BUG=none
TEST=no layout test failures

Review URL: https://codereview.chromium.org/656723005

git-svn-id: svn://svn.chromium.org/blink/trunk@184834 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This was broken in crrev.com/653303003.

There is no test for the latter, but that code is not live yet. Punting to 
mstensho@opera.com.

BUG=425030

Review URL: https://codereview.chromium.org/694753004

git-svn-id: svn://svn.chromium.org/blink/trunk@184831 bbb929c8-8fbe-4397-9dbb-9b2b20218538

For example, vendor, renderer, driver version, etc.This
way a developer can gather data on what platforms the app
fails to run. This needs changes on both blink and chromium.

Below steps are followed to implement this

1. Add basic headers in blink.
2. Add the necessary apis in chromium.
3. Start using this in WebglRenderingContxt.cpp.

This patch represents step-3 in this process.

BUG=412440

Review URL: https://codereview.chromium.org/695563003

git-svn-id: svn://svn.chromium.org/blink/trunk@184830 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=386601

Review URL: https://codereview.chromium.org/689153004

git-svn-id: svn://svn.chromium.org/blink/trunk@184829 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Window.idl declares that Window is ImplementedAs=LocalDOMWindow.
It also declares self/window/frames/opener/parent/top attributes
return Window, so LocalDOMWindow::self/window/frames/opener/parent/
top member functions must return LocalDOMWindow, not DOMWindow.

dcheng@, you can change Window.idl to declare ImplementedAs=DOMWindow
instead of LocalDOMWindow, if necessary.  Otherwise, please keep the
consistency here.

BUG=425623

Review URL: https://codereview.chromium.org/700803002

git-svn-id: svn://svn.chromium.org/blink/trunk@184827 bbb929c8-8fbe-4397-9dbb-9b2b20218538

It's possible that Resource::encoding() changes once we receive data.

This requires moving the creation of v8::ScriptCompiler::StreamedSource and
SourceStream forward in time (we need to know the encoding to create them).

Additional fix: m_cachedDataType was not initialized properly.

BUG=428137

Review URL: https://codereview.chromium.org/700543002

git-svn-id: svn://svn.chromium.org/blink/trunk@184826 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The only notable change is setting column-gap explicitly.

R=mstensho@opera.com
BUG=386567

Review URL: https://codereview.chromium.org/696273002

git-svn-id: svn://svn.chromium.org/blink/trunk@184825 bbb929c8-8fbe-4397-9dbb-9b2b20218538

(ArrayBuffer or ArrayBufferView) is a common union type usage. For example,
CSS font loading[1] and fetch API[2] use it. Implement and use them for
FontFace constructor.

Change for FontFace is covered by fast/css/fontface-arraybuffer.html

This CL also fixed two minor bugs:
- Collect union types from constructors (we didn't before)
- We need to use 'implemented_as' for forward class declarations, instead of
  'name'.

[1] http://dev.w3.org/csswg/css-font-loading/#fontface-interface
[2] https://fetch.spec.whatwg.org/#body-mixin

BUG=240176

Review URL: https://codereview.chromium.org/699713003

git-svn-id: svn://svn.chromium.org/blink/trunk@184824 bbb929c8-8fbe-4397-9dbb-9b2b20218538