GitLab

This is a speculative revert to address reported crashes in the wild.

> [HTML Import] Let script block by pending resources created by lifecycle callback
> 
> This change tells the HTML parser know that lifecycle callbacks,
> that is triggered through Microtask checkpoint, can create resources that
> blocks scripts.
> 
> It also streamlines the unblock notification from Document to the parser
> into didLoadAllScriptBlockingResources(). Now both didRemoveAllPendingStylesheets()
> and didLoadAllImports() checks resource readiness and invokes the unified notification.
> 
> TEST=custom-element-style-block.html
> BUG=346408
> R=dglazkov@chromium.org, dominicc@chromium.org
> 
> Review URL: https://codereview.chromium.org/202813002

TBR=morrita@chromium.org

Review URL: https://codereview.chromium.org/201673005

git-svn-id: svn://svn.chromium.org/blink/trunk@169769 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Verify that a very large shape-margin value combined with a very small
shape-outside image defines a shape that matches the shape-outside
element's margin-box.

This is a merge of http://trac.webkit.org/changeset/166019 by
Hans Muller <hmuller@adobe.com>.

BUG=354917

Review URL: https://codereview.chromium.org/208423003

git-svn-id: svn://svn.chromium.org/blink/trunk@169768 bbb929c8-8fbe-4397-9dbb-9b2b20218538

There's no need to constantly recompute this data. It's
already determined by the style information, so we might as
well just compute it once during style recalc.

Note: We're still computing all the direct compositing
reasons during style recalc (via updateLayerCompositingState in RenderLayer::styleChanged), but now we skip recomputing
the style determined reasons again when updating layout
and compositing. ASSERTs verify that we get the same answer
as before.

This CL reduces the amount of time spent in the compositing
update for dragging the drawer in the Polymer calculator by
3%.

Review URL: https://codereview.chromium.org/207513002

git-svn-id: svn://svn.chromium.org/blink/trunk@169766 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The feature has been at status=experimental for over a month, and all known bugs have been fixed.

BUG=335247

Review URL: https://codereview.chromium.org/208153007

git-svn-id: svn://svn.chromium.org/blink/trunk@169765 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Currently, it only guarantees no two siblings have the same unique name. This
patch tries much harder to guarantee that no two frames in the same page have
the same unique name.

BUG=352651
TEST=http/tests/history/frameset-repeated-name.html

Review URL: https://codereview.chromium.org/207143002

git-svn-id: svn://svn.chromium.org/blink/trunk@169764 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This CL teaches CompositedLayerMapping::updateCompositedBounds about the
m_needToUpdateGeometry dirty bit. Now we only update the composited bounds if
they could have possibly changed.

This CL saves 14% of the compositing update in Polymer's calculator sample app.

R=ojan@chromium.org

Review URL: https://codereview.chromium.org/208643002

git-svn-id: svn://svn.chromium.org/blink/trunk@169762 bbb929c8-8fbe-4397-9dbb-9b2b20218538

It's trivial to replace the local AnimationClock in SMILTimeContainer
with the one from Document after making the observation that the only
current manipulations (updateTime/unfreeze) are already performed by
PageAnimator (DocumentAnimations::updateAnimationTimingForAnimationFrame)
and RenderLayerCompositor (DocumentAnimations::startPendingAnimations).

With this change it become possible to push the DiscardScope into
updateAnimationsAndScheduleFrameIfNeeded - so do that to simplify the
code (by reducing the number of places that instantiate a DiscardScope.)

BUG=231576

Review URL: https://codereview.chromium.org/198723005

git-svn-id: svn://svn.chromium.org/blink/trunk@169761 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Revert r169675 (commit aca3f9e60c129c49e4e48cc9097f80d7f697c417) as it
might have caused the Mac ASAN bot to fail.

R=vandebo@chromium.org
BUG=

Review URL: https://codereview.chromium.org/208073005

git-svn-id: svn://svn.chromium.org/blink/trunk@169760 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Step 1 of refactoring LayoutState to be a more general
quick-lookup while walking the tree. No behavior change,
just better encapsulation.

BUG=343896

Review URL: https://codereview.chromium.org/207223003

git-svn-id: svn://svn.chromium.org/blink/trunk@169759 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The existing compositing/video-frame-size-change.html layout test provides the coverage for the compositing path.

BUG=354168
R=enne@chromium.org

Review URL: https://codereview.chromium.org/208703002

git-svn-id: svn://svn.chromium.org/blink/trunk@169758 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Unreviewed gardening.

BUG=344497,352628
TBR=jcgregorio@chromium.org

Review URL: https://codereview.chromium.org/208373002

git-svn-id: svn://svn.chromium.org/blink/trunk@169756 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The null-attribute and 'indefinite' cases were not cached, which made
this method do some unnecessary work for each animation update.

BUG=354520

Review URL: https://codereview.chromium.org/206013003

git-svn-id: svn://svn.chromium.org/blink/trunk@169755 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=

Review URL: https://codereview.chromium.org/203443004

git-svn-id: svn://svn.chromium.org/blink/trunk@169754 bbb929c8-8fbe-4397-9dbb-9b2b20218538

NavigationAction class before this patch had a RefPtr<Event> as its member. The event retained the previous document longer than necessary, resulting in a soft leak. This patch removes the member from the NavigationAction class and replaces it with DOMTimeStamp of the event which is the only information may be retrieved afterward.

BUG=None

Review URL: https://codereview.chromium.org/206453010

git-svn-id: svn://svn.chromium.org/blink/trunk@169753 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Contrary to the assertions on the bug, the current code is correct. As
the user changes the selection, there will naturally be a visual
affordance to indicate the selection changes. By the time we scroll,
RL::hasBlockSelectionGapsBounds will be current and correct.

That said, the hope is that in the future paint will have no side
effects at all, nor do we want layout tests to require repaints during
their execution. So even if this is technically correct it's undesirable
on a number of levels.

This cl replaces the use of hasBlockSelectionGapsBounds with non-paint
related code (in fact, that function could be removed entirely).

R=abarth@chromium.org
BUG=353827

Review URL: https://codereview.chromium.org/208203002

git-svn-id: svn://svn.chromium.org/blink/trunk@169752 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Add a helper function 'svgRootElement' and use that where appropriate.

Review URL: https://codereview.chromium.org/208503002

git-svn-id: svn://svn.chromium.org/blink/trunk@169751 bbb929c8-8fbe-4397-9dbb-9b2b20218538

SSIA. Considering the memory sensitiveness of strings in Blink
we should make sure that no change can accidentally increase
their size.

Review URL: https://codereview.chromium.org/205583005

git-svn-id: svn://svn.chromium.org/blink/trunk@169750 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=353997

Review URL: https://codereview.chromium.org/203193015

git-svn-id: svn://svn.chromium.org/blink/trunk@169749 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Changes to the LRU lists are hidden behind a call to MemoryCache::update(), which adjusts sizes, access count, and LRU list position as appropriate. All of these are to some degree interrelated, since we keep multiple LRU lists and select one to use based access count and resource size.

Also, remove the Resource::inCache() bit in favor of a
MemoryCache::contains()

BUG=none

Review URL: https://codereview.chromium.org/164333008

git-svn-id: svn://svn.chromium.org/blink/trunk@169748 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Simplify VectorMover<true, T>::swap() - use std::swap_ranges instead
of creating the loop ourselves, plus the patch removes a "FIXME"
comment - compiler already produces the most efficient code for POD
types swap, so there is no need to introduce compex algorithms for
that (checked with GCC 4.7.2).

Review URL: https://codereview.chromium.org/208253003

git-svn-id: svn://svn.chromium.org/blink/trunk@169747 bbb929c8-8fbe-4397-9dbb-9b2b20218538

{,un}setPendingActivity relies on RefCounted, which is going to be deprecated in oilpan. This CL removes usage of these methods in SpeechRecognition.cpp, and provide hasPendingActivity() implementation as replacement

BUG=354845
TEST=fast/speech

Review URL: https://codereview.chromium.org/208273002

git-svn-id: svn://svn.chromium.org/blink/trunk@169746 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=353090

Review URL: https://codereview.chromium.org/203683005

git-svn-id: svn://svn.chromium.org/blink/trunk@169745 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Make this EventTarget-derived object a garbage collected object.

R=haraken@chromium.org
BUG=340522

Review URL: https://codereview.chromium.org/208083003

git-svn-id: svn://svn.chromium.org/blink/trunk@169744 bbb929c8-8fbe-4397-9dbb-9b2b20218538

As a drive-by, shadow DOM editing tests have been extracted into a new HTML

R=aandrey, pfeldman, vsevik
BUG=353910
NOTRY=true

Review URL: https://codereview.chromium.org/206503002

git-svn-id: svn://svn.chromium.org/blink/trunk@169743 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This CL uses some memoization to avoid unnecessary tree walking when
computing scroll parent. Why memoization and not just passing down the
values? Because sad. It turns out that we recur on the stacking tree (we
have to - squashing demands processing stuff in paint order), but we
walk up the render tree asking questions. Different tree topologies; we
could very easily visit a RenderLayer before its great great
grandparent.

No functionality change -- covered by existing tests.
BUG=None

Review URL: https://codereview.chromium.org/206283003

git-svn-id: svn://svn.chromium.org/blink/trunk@169741 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=354791
TBR=vsevik

Review URL: https://codereview.chromium.org/208053005

git-svn-id: svn://svn.chromium.org/blink/trunk@169740 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=354849
TBR=alancutter

Review URL: https://codereview.chromium.org/207613006

git-svn-id: svn://svn.chromium.org/blink/trunk@169739 bbb929c8-8fbe-4397-9dbb-9b2b20218538

HeapSnapshotView should be unsubscribed from events it has been subscribed to, otherwise it will leak when corresponding profile is removed.

BUG=None

Review URL: https://codereview.chromium.org/208053003

git-svn-id: svn://svn.chromium.org/blink/trunk@169738 bbb929c8-8fbe-4397-9dbb-9b2b20218538

All old SVGProperties implementations are now all removed. The "New" prefix for distinguishing two implementations are no longer needed.

Changes automatically generated from: sed -i 's/NewSVG/SVG/g' `find . -name \*.cpp -or -name \*.h`

BUG=349370

Review URL: https://codereview.chromium.org/208133002

git-svn-id: svn://svn.chromium.org/blink/trunk@169737 bbb929c8-8fbe-4397-9dbb-9b2b20218538

When URL of an iframe changes from "unset" to "set", we don't want to
create new history entry. This worked for URLs without fragment
identifier but failed with it present.

This change takes the idea from the old code, before it was rewritten
in https://codereview.chromium.org/126453005

BUG=353096

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=169665

Review URL: https://codereview.chromium.org/201773002

git-svn-id: svn://svn.chromium.org/blink/trunk@169736 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Review URL: https://codereview.chromium.org/166273018

Review URL: https://codereview.chromium.org/166273018

git-svn-id: svn://svn.chromium.org/blink/trunk@169735 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=354405
R=ager@chromium.org

Review URL: https://codereview.chromium.org/206333003

git-svn-id: svn://svn.chromium.org/blink/trunk@169734 bbb929c8-8fbe-4397-9dbb-9b2b20218538

With heap supplement{able}s support now in place, follow up and turn
the supplements of these two into garbage allocated objects also.

R=haraken@chromium.org
BUG=340522

Review URL: https://codereview.chromium.org/200373003

git-svn-id: svn://svn.chromium.org/blink/trunk@169733 bbb929c8-8fbe-4397-9dbb-9b2b20218538

- HeapSnapshotNode, HeapSnapshotEdge and HeapSnapshotRetainerEdge now store only reference to HeapSnapshot and global index of corresponding item.
- all index boundaries were moved out of these classes into corresponding iterators

BUG=None

Review URL: https://codereview.chromium.org/207523004

git-svn-id: svn://svn.chromium.org/blink/trunk@169732 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=haraken@chromium.org, oilpan-reviews@chromium.org, yhirano@chromium.org
NOTRY=true

Review URL: https://codereview.chromium.org/205693004

git-svn-id: svn://svn.chromium.org/blink/trunk@169731 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=rwlbuis
BUG=354791

Review URL: https://codereview.chromium.org/198183004

git-svn-id: svn://svn.chromium.org/blink/trunk@169730 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=354787
TBR=tdresser@chromium.org

Review URL: https://codereview.chromium.org/207613003

git-svn-id: svn://svn.chromium.org/blink/trunk@169729 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=fs@opera.com
BUG=354785
NOTRY=true

Review URL: https://codereview.chromium.org/207773002

git-svn-id: svn://svn.chromium.org/blink/trunk@169728 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Use Blink Promise instead of MIDIAccessPromise.
BUG=320984
R=toyoshim@chromium.org

Review URL: https://codereview.chromium.org/77773003

git-svn-id: svn://svn.chromium.org/blink/trunk@169727 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Reason for revert:
This have broken browser_tests according to my bisect:
http://build.chromium.org/p/chromium.webkit/builders/Linux%20Tests%20%28dbg%29/builds/2034/steps/browser_tests/logs/TabDestroyed

Original issue's description:
> id of iframe incorrectly sets window name
>
> As per the specification, the iframe's name should be an empty string if unset:
> http://www.whatwg.org/specs/web-apps/current-work/multipage/the-iframe-element.html#attr-iframe-name
>
> Instead, Chromium was using the iframe's id as window name if the name was
> unset. This change makes us behave like Firefox 27 and IE 11, verified using:
> http://jsfiddle.net/xf5H7/9/
>
> This CL sets the name attribute iframe in a lot of layout tests so that
> testRunner.dumpChildFramesAsText() keeps printing the same result.
>
> R=arv@chromium.org, tkent@chromium.org
> BUG=347169
> TEST=fast/frames/iframe-no-name.html
>
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=168553
>
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=168801
>
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=169304
>
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=169683

TBR=abarth@chromium.org,arv@chromium.org,tkent@chromium.org,ch.dumez@samsung.com
NOTREECHECKS=true
NOTRY=true
BUG=347169

Review URL: https://codereview.chromium.org/198183003

git-svn-id: svn://svn.chromium.org/blink/trunk@169726 bbb929c8-8fbe-4397-9dbb-9b2b20218538