GitLab

WebServiceWorkerProviderClient::setCurrentServiceWorker passes the
ownership of a WebServiceWorker to the container. If the execution
context is gone, delete the WebServiceWorker instead of dropping it on
the floor.

BUG=363967

Review URL: https://codereview.chromium.org/304023002

git-svn-id: svn://svn.chromium.org/blink/trunk@175023 bbb929c8-8fbe-4397-9dbb-9b2b20218538

r175007 broke oilpan builds.

TBR=tkent

Review URL: https://codereview.chromium.org/301073002

git-svn-id: svn://svn.chromium.org/blink/trunk@175022 bbb929c8-8fbe-4397-9dbb-9b2b20218538

scroll-top.html is intended to test performance of DOM attributes that return small integers.
Due to a recent change to Element.scrollTop, Element.scrollTop no longer returns small integers
(it returns a double value). So this CL updates the performance test so that it uses Node.nodeType.

BUG=371429

Review URL: https://codereview.chromium.org/306783003

git-svn-id: svn://svn.chromium.org/blink/trunk@175021 bbb929c8-8fbe-4397-9dbb-9b2b20218538

In theory, a simultaneous caller to Partitions::getBufferPartition()
could see the flag as being true, and return the uninitialized buffer
partition, while another thread is initializing it.

BUG=367672

Review URL: https://codereview.chromium.org/268673012

git-svn-id: svn://svn.chromium.org/blink/trunk@175020 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Check nullable value before using it in ReplaceSelectionCommand::removeRedundantStylesAndKeepStyleSpanInline

This patch changes removeRedundantStylesAndKeepStyleSpanInline() to check
nullable value |element->parentNode()| before using it. L529, if-statement
just after this change, checks |element->parentNode()| too.

BUG=369422
TEST=LayoutTests/editing/execCommand/insert-html-to-document-element-crash.html

Review URL: https://codereview.chromium.org/300143012

git-svn-id: svn://svn.chromium.org/blink/trunk@175019 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Currently we force page scale factor to 1.0 in fullscreen mode (whether it is
a video or not). The assumption was broken when switching orientation. Adding
extra check in screen resize code to make sure the constraint is enforced.

R=aelias,qinmin
BUG=376408

Review URL: https://codereview.chromium.org/308513002

git-svn-id: svn://svn.chromium.org/blink/trunk@175018 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Update Element::updateFocusAppearance() so that calling focus() on an iframe
no longer triggers a 'focusout' event. The issue was that Element:focus() would
call focusController().setFocusedElement() to set focus to the given element
and then call updateFocusAppearance() which would potentially steal the focus
and give it to another Element. This is because updateFocusAppearance() calls
FrameSelection::setSelection() which set focus unless explicitely told
otherwise using the DoNotSetFocus option.

This CL updates updateFocusAppearance() to pass DoNotSetFocus option when
calling FrameSelection::setSelection() to maintain the focused Element
previously set by Element::focus().

The new behavior is in line with Firefox 29 and IE11.

R=eseidel@chromium.org, ojan@chromium.org
TEST=fast/frames/frame-focus-no-focusout-event.html
BUG=372777

Review URL: https://codereview.chromium.org/284823005

git-svn-id: svn://svn.chromium.org/blink/trunk@175017 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This CL adds the following metrics to UMAs:

- Duration of time consumed in Heap::collectGarbage().
- Duration of time consumed in ThreadState::performPendingSweep().
- The total size of object space in all threads when Heap::collectGarbage() is triggered.

BUG=340522

Review URL: https://codereview.chromium.org/304673003

git-svn-id: svn://svn.chromium.org/blink/trunk@175016 bbb929c8-8fbe-4397-9dbb-9b2b20218538

In basically all cases it was calling containerForRepaint(),
so no need to pass that extra argument.

BUG=370664

Review URL: https://codereview.chromium.org/307713002

git-svn-id: svn://svn.chromium.org/blink/trunk@175015 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Hixie asked in bug 378163:
"Should we drop focusin/focusout?"
"Is there a compat need to support the focusin/focusout events?"

To answer that question, add a UseCounter for focusin/focusout.
Instead of counting all listeners, count cases where the event is
actually fired with a listener attached, which is a better estimate of
the compatibility risk. Make the same change to the DOMFocusInOut
counter to make the two comparable.

BUG=378163

Review URL: https://codereview.chromium.org/307613002

git-svn-id: svn://svn.chromium.org/blink/trunk@175014 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=esprehn@chromium.org
BUG=164167

Review URL: https://codereview.chromium.org/303993002

git-svn-id: svn://svn.chromium.org/blink/trunk@175013 bbb929c8-8fbe-4397-9dbb-9b2b20218538

We currently correctly send a GET in the POST->replaceState->reload case, but
DocumentLoader::m_request still shows a method of POST, and as a result we
incorrectly report it as a POST to the browser process. This contributes to
unnecessarily showing a form resubmission warning in that case.

BUG=138324

Review URL: https://codereview.chromium.org/302543008

git-svn-id: svn://svn.chromium.org/blink/trunk@175011 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This CL adds to RenderMenuList::setTextFromOption(int) so that when the
associated HTML <select> element has the multiple attribute, the <select>
element's options are iterated over and a count of the number of selected
options is determined. When the number of selected options is 0 or 2 or
more, then the RenderMenuList's text is set to "SELECTED_COUNT selected"
(e.g. "2 selected"). When exactly one option is selected, then the text
is set to the selected option's text.

This is similar to iOS. When 0 or 2 or more options are selected, then
the text of the menulist button is set to "SELECTED_COUNT Items". When
exactly one option is selected, then the text is set to the selected
option's text.

Depends on:
- https://codereview.chromium.org/280483002/
- https://codereview.chromium.org/295933003/

BUG=314704

Review URL: https://codereview.chromium.org/302473003

git-svn-id: svn://svn.chromium.org/blink/trunk@175010 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Review URL: https://codereview.chromium.org/304693002

git-svn-id: svn://svn.chromium.org/blink/trunk@175009 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://test-results.appspot.com/dashboards/flakiness_dashboard.html#group=%40ToT%20Blink&tests=compositing%2Frtl%2F&showExpectations=true

BUG=338794,192172
TBR=ojan@chromium.org,trchen@chromium.org

Review URL: https://codereview.chromium.org/300373002

git-svn-id: svn://svn.chromium.org/blink/trunk@175008 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This is a revision of r173794, which introduced extra complexity
to CustomElementMicrotaskStep classes.

This change moves the responsibility that introduces the complexity
to HTMLImportChild. This localizes the tricky part in the single class
and helps overall code sanity.

TEST=no
BUG=371654
R=dominicc@chromium.org, dglazkov@chromium.org

Review URL: https://codereview.chromium.org/288323004

git-svn-id: svn://svn.chromium.org/blink/trunk@175007 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Revert of Set proper referrer for css resource fetching (https://codereview.chromium.org/239353007/)

Reason for revert:
Caused regression

Original issue's description:
> Set proper referrer for css resource fetching
> 
> Fetching css resources are initiated as follows:
>   - Create FetchRequest
>   - Call ResourceFetcher::fetchXXX()
> 
> git-grep shows me that CSSImageValue, CSSImageSetValue and
> CSSFontFaceSrcValue are all of the resource fetch initiators
> which live in css/. I think that setting referrer in these
> classes should cover all css resource fetching.
> 
> Since resource fetching is delayed until the resource is actually
> needed, we need to hold the referrer somewhere.
> This CL modifies the css parser to set referrer to corresponding
> CSSValue (CSSImageValue and CSSFontFaceSrcValue) when it sees url(...).
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=172433

Review URL: https://codereview.chromium.org/304603006

git-svn-id: svn://svn.chromium.org/blink/trunk@175004 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=374398

Review URL: https://codereview.chromium.org/302873002

git-svn-id: svn://svn.chromium.org/blink/trunk@175003 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This is just moving code around in preparation for
making HTMLImportController not being a root HTMLImport.

R=morrita
BUG=

Review URL: https://codereview.chromium.org/303643005

git-svn-id: svn://svn.chromium.org/blink/trunk@175002 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=esprehn@chromium.org
BUG=164167

Review URL: https://codereview.chromium.org/307723002

git-svn-id: svn://svn.chromium.org/blink/trunk@174999 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Just giving up on this test, it tried to work around windows oddity
but we clearly still don't understand what windows is doing
so just skipping the test.

Review URL: https://codereview.chromium.org/305613003

git-svn-id: svn://svn.chromium.org/blink/trunk@174998 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Per spec, after an error the |result| should always be |undefined|. Chromium was yielding |null| or an obsolete value instead.

BUG=378411

Review URL: https://codereview.chromium.org/307653004

git-svn-id: svn://svn.chromium.org/blink/trunk@174997 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=

Review URL: https://codereview.chromium.org/303653008

git-svn-id: svn://svn.chromium.org/blink/trunk@174996 bbb929c8-8fbe-4397-9dbb-9b2b20218538

There used to be a similar cache before r168864. The old cache
system did not survive the refactoring, which caused a measurable
performance degradation on several mobile platforms.  This change
reinstates it in a refactored form.

BUG=354565
TEST=jsgamebench telemetry on android

Review URL: https://codereview.chromium.org/297293004

git-svn-id: svn://svn.chromium.org/blink/trunk@174995 bbb929c8-8fbe-4397-9dbb-9b2b20218538

When porting the original srcset parser from WebKit, the code was split from
HTMLParserIdioms into a separate file - HTMLSrcsetParser.
The Apple copyright notice of that code was mistakenly not copied to the new file.
This CL corrects this.

BUG=
TBR=pdr

Review URL: https://codereview.chromium.org/305653007

git-svn-id: svn://svn.chromium.org/blink/trunk@174994 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=361045

Review URL: https://codereview.chromium.org/305623007

git-svn-id: svn://svn.chromium.org/blink/trunk@174993 bbb929c8-8fbe-4397-9dbb-9b2b20218538

When subpixel font scaling is enabled text runs are positioned at
subpixel boundaries on the x-axis and thus there is no reason to snap
the x value.
We still round the y-axis to ensure consistent line heights.

BUG=164167
R=leviw@chromium.org, mstensho@opera.com

Review URL: https://codereview.chromium.org/300623007

git-svn-id: svn://svn.chromium.org/blink/trunk@174992 bbb929c8-8fbe-4397-9dbb-9b2b20218538

I'd like to drive this number down to zero. Having a list of issues to
drive down is the starting place.

Mostly this just requires exposing a few more bits from builders and
loader. Otherwise, it's just going through the motions of loading
each result for testtype/group combination and seeing which ones
fail.

NOTRY=true

Review URL: https://codereview.chromium.org/293393006

git-svn-id: svn://svn.chromium.org/blink/trunk@174991 bbb929c8-8fbe-4397-9dbb-9b2b20218538

In theory, this case should never happen, but clearly there's
a bug somewhere. I don't have time to dig into the bug right now,
so just bandaid over it by returning the empty string. This will
mean the old data is lost, since we basically treat this as an
empty file, but at least we start collecting the new data.

We'll still log the error though, so we can grep the log to find
these cases in the future and fix them.

BUG=377594
NOTRY=true

Review URL: https://codereview.chromium.org/298163003

git-svn-id: svn://svn.chromium.org/blink/trunk@174990 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The under-repaint occurs in non-RAL mode when an object needs both
repaint and non-full layout. https://codereview.chromium.org/265703012/
caused this because it skips the direct repaint() call in the above
case.

BUG=376306
TEST=fast/repaint/change-text-content-and-background-color.html

Review URL: https://codereview.chromium.org/298973011

git-svn-id: svn://svn.chromium.org/blink/trunk@174989 bbb929c8-8fbe-4397-9dbb-9b2b20218538

When click() method is called, previously we change the active status
of the element to true and then back to false. This doesn't change
anything for a button (or other elements that status doesn't change
after click()) but causes a repaint.

Don't change active status when click() method is called. (Elements
that status changes after click() call (e.g. checkboxes) will still
repaint.)

This doesn't change how we handle real mouse down/up/click events.

BUG=369358
TEST=fast/repaint/button-checkbox-click-method-repaint.html
TEST=fast/events/button-mouse-active.html # ensures :active is correct on real mouse down events.

Review URL: https://codereview.chromium.org/291563009

git-svn-id: svn://svn.chromium.org/blink/trunk@174988 bbb929c8-8fbe-4397-9dbb-9b2b20218538

V8 now implements Isolate::RunMicrotasks() in C++, which means its callers
no longer need to enter V8 Context::Scope before calling it. This simplifies
WebCore::Microtask::performCheckpoint, and gets us one step closer to ditching
V8PerIsolateData::ensureDomInJSContext().

This patch also gets rid of most of the microtask-related logic in
ScriptPromiseResolverWithContext, allowing WebCore::Microtask::performCheckpoint
to be the only code in Blink that calls Isolate::RunMicrotasks().
Instead of triggering microtask delivery directly, or via the RunMicrotasksTask,
ScriptPromiseResolverWithContext now utilizes V8RecursionScope to "simulate"
a JavaScript stack (from Blink's point of view). In workers, this will
result in a RunMicrotasks() call when the V8RecursionScope is destructed.

There are still two FIXMEs:
  - The V8RecursionScope discussed above would be unnecessary if there were
    a WebThread::TaskObserver on worker threads to handle microtask-running
    at the end of each task.
  - ensureDomInJSContext() is still needed in order to instantiate the v8::Functions
    that are passed into Isolate::EnqueueMicrotask(). Fixing this will require
    more work on the V8 side to allow passing in a C++ callback directly.

BUG=374772

Review URL: https://codereview.chromium.org/303693005

git-svn-id: svn://svn.chromium.org/blink/trunk@174987 bbb929c8-8fbe-4397-9dbb-9b2b20218538

When creating a new GraphicsContextState the constructor initializes the state
with various default values. In GraphicsContext::realizePaintState we immediately
overwrite these values after creation. This patch adds a copy constructor to
GraphicsContextState to avoid that.

Even though this seems like a clear win, the benefits are not visible when
profiling. Looks like the overhead of the redundant initialization is
not very large. Still seems worth fixing though.

BUG=377687

Review URL: https://codereview.chromium.org/303613002

git-svn-id: svn://svn.chromium.org/blink/trunk@174985 bbb929c8-8fbe-4397-9dbb-9b2b20218538

In the destruction of HTMLCanvasElement, we call
HTMLCanvasElement::existingDrawingContext(), whose behavior was changed
incorrectly in r162330 and was triggering the allocation of an ImageBuffer
during tear-down.  This is not only wasteful, but it can trigger a
garbage collection because the creation of an ImageBuffer will increase
the amount of externallay allocated memory that is declared to V8.
Triggering a GC during the finalization of a garbage collected object
is bad.

BUG=378289

Review URL: https://codereview.chromium.org/303803002

git-svn-id: svn://svn.chromium.org/blink/trunk@174984 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Change the way clip rects are determined for multicol by using enclosing
IntRect instead of pixelSnappedIntRect to ensure that the colum contents
fully fits within the clip rect.

R=leviw@chromium.org, mstensho@opera.com

Review URL: https://codereview.chromium.org/300843010

git-svn-id: svn://svn.chromium.org/blink/trunk@174983 bbb929c8-8fbe-4397-9dbb-9b2b20218538

We don't really need these classes anymore, we can just use methods on
ContainerNode. This gets us one step closer to completely eliminating
ContainerNodeAlgorithms which now only contains one shared method related
to SVGElementInstance and cleanup code that Oilpan will eliminate.

Review URL: https://codereview.chromium.org/305603003

git-svn-id: svn://svn.chromium.org/blink/trunk@174982 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Debug assertions in the GraphicsContext destructor were
dependent on object destruction order, which made them
incompatible with oilpan. The state validation logic
was refactored to make the state validation checks more
useful.

BUG=370793

Review URL: https://codereview.chromium.org/293963009

git-svn-id: svn://svn.chromium.org/blink/trunk@174981 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This reduces methods exposed on the throttle class and makes
throttling details hidden to XMLHttpRequest class.

- Use dispatchProgressEvent also for loadstart event
- Remove "throttled" from method names
- Move if-clause to switch how to dispatch a ProgressEvent based on its
  name to XMLHttpRequestProgressEventThrottle::dispatchProgressEvent()
- Remove dispatchThrottledProgressEvent and just call
  m_progressEventThrottle.dispatchProgressEvent for each event
- Make XMLHttpRequestProgressEventThrottle::dispatchEvent() private

BUG=none

Review URL: https://codereview.chromium.org/301643002

git-svn-id: svn://svn.chromium.org/blink/trunk@174979 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Forgot to do this during my last gardening stint :(

BUG=369600
TBR=ojan

Review URL: https://codereview.chromium.org/305623006

git-svn-id: svn://svn.chromium.org/blink/trunk@174978 bbb929c8-8fbe-4397-9dbb-9b2b20218538

From private conversation with annevk - make the properties of the
TextDecoder object inspectable.

BUG=375847

Review URL: https://codereview.chromium.org/292243003

git-svn-id: svn://svn.chromium.org/blink/trunk@174977 bbb929c8-8fbe-4397-9dbb-9b2b20218538