GitLab

R=pfeldman,apavlov

Review URL: https://codereview.chromium.org/214443005

git-svn-id: svn://svn.chromium.org/blink/trunk@170317 bbb929c8-8fbe-4397-9dbb-9b2b20218538

R=pfeldman@chromium.org, pfeldman

Review URL: https://codereview.chromium.org/216193003

git-svn-id: svn://svn.chromium.org/blink/trunk@170316 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The use of filecmp:
* speeds up r-b-t by 23% (if no diffs): (885 ms vs. 1.150 sec)
* uses only 1 process (if no diffs): 30 => 1
* if diffs, same speed
(can't measure difference, because filecmp is basically instantaneous on my machine)

Also some cleanup:
* remove dead code
* simplify temp file handling (caused build bot failures previously)
* don't catch exceptions, so easier to debug errors in CG
(because it's just one process now)

Currently r-b-t launches a separate diff(1) process for each file it compares.
Python has a built-in diff facility, which is extremely fast and avoids other processes.
This should make a big difference on Windows, meaning faster uploads for
Windows developers and less load on build bots.

Also a lot of the code is no longer necessary, after Terry's simplifications.
We can eliminate all the redundant temp file handling, and just need a simple
temp dir class.

R=haraken
BUG=341748

Review URL: https://codereview.chromium.org/216753002

git-svn-id: svn://svn.chromium.org/blink/trunk@170315 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This CL enables Media Query evaluation inside the preload scanner, without joining the main thread.
That is to enable resource fetching based on the media attribute, without hampering future resource fetching improvements.

Major changes:
* Added a MediaValues class that "freezes" the various media values needed for MQE when it is initialized.
* Initialize MediaValues and provide them to the PreloadScanner during its initialization.
* Run MQE inside the PreloadScanner, according to provided MediaValues.
* Removed mediaAttribute passing to the HTMLResourcePreloader, that runs on the main thread.

BUG=

Review URL: https://codereview.chromium.org/201813002

git-svn-id: svn://svn.chromium.org/blink/trunk@170314 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch adds a new flag to begin experimentation with the Subresource
Integrity specification[1], and begins adding the minimal amount of
functionality so we can test that the flag is working. That turns out to
be the 'integrity' attribute on various HTML elements that we'll use
later in order to do the actual verification.

Intent to Implement at [2].

[1]: http://w3c.github.io/webappsec/specs/subresourceintegrity/
[2]: https://groups.google.com/a/chromium.org/d/msg/blink-dev/hTDUpMk_TV8/t_rjlkKfgGgJ

BUG=355467

Review URL: https://codereview.chromium.org/208423011

git-svn-id: svn://svn.chromium.org/blink/trunk@170313 bbb929c8-8fbe-4397-9dbb-9b2b20218538

In r170168 I made pushing the parent scopes lazy for Elements, but I
didn't make it lazy for ShadowRoot. This meant that you could trigger a
style recalc on a single element in a ShadowRoot and then we'd push the
scopes out of order, where we'd push all the ShadowRoots on the way down
and then push all the elements at the end.

I wasn't able to show any rendering errors, but by inspection it's wrong
and it appears it was making us go through the
ScopedStyleTree::resolveStyleCache code path more often than needed.

Review URL: https://codereview.chromium.org/216283003

git-svn-id: svn://svn.chromium.org/blink/trunk@170312 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This CL fixes a problem where Resources can be taken out of the cache but not
deleted (if their handle count is not 0 for example).

BUG=290193

Review URL: https://codereview.chromium.org/179943002

git-svn-id: svn://svn.chromium.org/blink/trunk@170311 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This will be tested by a Chromium browsertest. Service Worker plumbing that would make this possible to test in Blink isn't fitted yet.

BUG=356021

Review URL: https://codereview.chromium.org/214413004

git-svn-id: svn://svn.chromium.org/blink/trunk@170310 bbb929c8-8fbe-4397-9dbb-9b2b20218538

DevTools: Make SourcesView shortcut work when it is shown in drawer and move "save" shortcut into it.

R=pfeldman@chromium.org, apavlov, lushnikov, pfeldman
BUG=322632, 349428

Review URL: https://codereview.chromium.org/212623013

git-svn-id: svn://svn.chromium.org/blink/trunk@170309 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TimelineFlameChart: fix for exception in highlightTimeRange which happens when entryIndex is incorrect.

It happens when newly recorded timeline has fewer elements than the old timeline.
So we need to reset FlameChart when we start the new recording session.

BUG=
R=pfeldman@chromium.org

Review URL: https://codereview.chromium.org/216713002

git-svn-id: svn://svn.chromium.org/blink/trunk@170308 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=352557
NOTRY=true

Review URL: https://codereview.chromium.org/216693002

git-svn-id: svn://svn.chromium.org/blink/trunk@170307 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Continuing the work to make StyleAdjuster cheaper we can skip all the work
related to overflow adjustment in the common case when overflow is visible.

Review URL: https://codereview.chromium.org/214413011

git-svn-id: svn://svn.chromium.org/blink/trunk@170306 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Implements transform-style as defined in CSS Transforms[1]. Behavior is
identical to -webkit-transform-style.

Note, this property is guarded by the CSSTransformsUnprefixed runtime feature
(enabled only for testing).

[1] http://www.w3.org/TR/css3-transforms/

BUG=154772

Review URL: https://codereview.chromium.org/211123002

git-svn-id: svn://svn.chromium.org/blink/trunk@170305 bbb929c8-8fbe-4397-9dbb-9b2b20218538

These functions are always called with string literals, so having the
functions accept Strings means all callers have to perform implicit
conversions.
By modifying the signatures to use 'const char*' for these cases, these
conversion can be avoided (in the callers) and only be performed in the
functions themselves.
This reduces code size by ~70kB (Linux, x86-64, gcc-4.8.2).

Review URL: https://codereview.chromium.org/216633002

git-svn-id: svn://svn.chromium.org/blink/trunk@170304 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The removal has caused more trouble than expected on Android, because
usage is higher there and the UseCounter data from Windows was used in
the "Intent to Remove" thread.

Restore the API behind a flag so that it can more easily be turned on
and off. It's off by default, but likely it will have to be enabled on
at least the upcoming M35 branch.

BUG=346236

Review URL: https://codereview.chromium.org/213193010

git-svn-id: svn://svn.chromium.org/blink/trunk@170303 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Logic to split keyframe animations into multiple players was required before
Web Animations supported per-keyframe easings.

CSS Animations were updated to use per-keyframe easings in r169119[1], but this
logic remained.

[1] https://src.chromium.org/viewvc/blink?view=rev&revision=169119

BUG=346590

Review URL: https://codereview.chromium.org/214603002

git-svn-id: svn://svn.chromium.org/blink/trunk@170302 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Inside method updateImplicitGridColumn() of test
fast/css-grid-layout/implicit-position-dynamic-change.html we define a
variable gridItem that is not used. This patch removes it.

Review URL: https://codereview.chromium.org/213423009

git-svn-id: svn://svn.chromium.org/blink/trunk@170301 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Enable V8 Promises.
This CL sets --harmony-promises flag.
When V8 Promises are shipped, I will remove it.

BUG=352597, 352552

Review URL: https://codereview.chromium.org/209853010

git-svn-id: svn://svn.chromium.org/blink/trunk@170300 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch removes the code branch handling "numCharacters == 1"
case leaving only the 'memcpy' call inside the 'WTF::copyChars'
function's body.

The reason for it is that the branch made the code more complex
giving the minor performance benefit: there was no benefit if the
'source' arg was static, otherwise the difference was just two
extra asm calls ('testq' and 'je') which should not give impact
to the real use cases performance (tested with GCC 4.6.3).

Review URL: https://codereview.chromium.org/213073002

git-svn-id: svn://svn.chromium.org/blink/trunk@170299 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This CL changes OwnPtr<OwnPtr<InterporableValue>[]> in InterpolableValueList to Vector<OwnPtr<InterporableValue>>. As far as I see the code, I don't think this change will affect performance.

BUG=340522

Review URL: https://codereview.chromium.org/212213007

git-svn-id: svn://svn.chromium.org/blink/trunk@170298 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Oilpan: Replace all remaining adoptRef()s in the Event hierarchy with adoptRefWillBeRefCountedGarbageCollected()

This CL just fixes left-over from r170164.

BUG=340522
NOTRY=true

Review URL: https://codereview.chromium.org/216113003

git-svn-id: svn://svn.chromium.org/blink/trunk@170297 bbb929c8-8fbe-4397-9dbb-9b2b20218538

togglePlayState() handles the MediaController case correctly, which the
updated code paths did not.

Introduce togglePlayStateWillPlay() to replace canPlay() and paused() in
certain contexts. The readyState < HAVE_METADATA in canPlay() condition
is not replicated because the hasSource() checks in paintMediaPlayButton
and paintMediaOverlayPlayButton disables the buttons so that they can't
be pressed at all. Some corner case seems likely to exist, but was not
found in ad-hoc testing.

Drop the ASSERT(controls()) in tooglePlayState() since it is now used in
contexts not directly tied to MediaControls. The ASSERT doesn't fail
yet, but the same ASSERT in togglePlayStateWillPlay() would, since
MediaControlPlayButtonElement::updateDisplayType() is called when the
controls are created for text track rendering but are not shown.

These changes made it possible to reduce the size of
MediaControllerInterface a little bit more.

BUG=341813
TEST=none, no observable changes intended

Review URL: https://codereview.chromium.org/214793005

git-svn-id: svn://svn.chromium.org/blink/trunk@170296 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Adds a runtime flag (enabled only for testing) and defines property names
for each property defined in CSS Transforms[1]. Future patches will make
these properties functional.

[1] http://www.w3.org/TR/css3-transforms/

BUG=154772

Review URL: https://codereview.chromium.org/211033002

git-svn-id: svn://svn.chromium.org/blink/trunk@170295 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=356616

Review URL: https://codereview.chromium.org/214453003

git-svn-id: svn://svn.chromium.org/blink/trunk@170294 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=
R=pfeldman@chromium.org

Review URL: https://codereview.chromium.org/216613002

git-svn-id: svn://svn.chromium.org/blink/trunk@170293 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=351956

Review URL: https://codereview.chromium.org/214733002

git-svn-id: svn://svn.chromium.org/blink/trunk@170292 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Align with Firefox and return false in the case where the
Range method leads to an exception. Previously, we returned
true.

The behavior of these selection APIs are not in any specs
and they are only implemented by Blink and Firefox.

R=haraken@chromium.org, nbarth@chromium.org
BUG=

Review URL: https://codereview.chromium.org/212873002

git-svn-id: svn://svn.chromium.org/blink/trunk@170291 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=pfeldman

Review URL: https://codereview.chromium.org/214793004

git-svn-id: svn://svn.chromium.org/blink/trunk@170290 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Revert of Remove applyAuthorStyles flag from ShadowRoot. (https://codereview.chromium.org/215313003/)

Reason for revert:
w3c/web-platform-tests/shadow-dom/elements-and-dom-objects/shadowroot-object/shadowroot-attributes/test-004.html started to fail.

http://test-results.appspot.com/dashboards/flakiness_dashboard.html#group=%40ToT%20Blink&tests=w3c/web-platform-tests/shadow-dom/elements-and-dom-objects/shadowroot-object/shadowroot-attributes/test-004.html


Original issue's description:
> Remove applyAuthorStyles flag from ShadowRoot.
> 
> This re-applies https://src.chromium.org/viewvc/blink?revision=165673&view=revision.
> 
> BUG=357151
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=170271

TBR=esprehn@chromium.org,tasak@google.com,hayato@chromium.org,dglazkov@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=357151

Review URL: https://codereview.chromium.org/216543002

git-svn-id: svn://svn.chromium.org/blink/trunk@170289 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Perf tests in PerformanceTests/Canvas usually run 6 times; 1 for warm-up, 5 for data.
Currently, upload-canvas-2d-to-texture.html takes ~72 sec on Intel Ivy Bridge. If we have
Android perf bot, this perf test takes huge time.

upload-canvas-2d-to-texture.html calls glTexImage2d() 1000 times per one run. Ivy Bridge
without Accelerated 2D Canvas can call glTexImage2d() ~85 times/sec, so one run takes
12 sec, and 6 runs take 72 sec.

This CL limits 1 sec per one run for all canvas perf tests. To apply common logic to 6 tests,
this CL introduces canvas_runner.js, which is a wrapper of PerformanceTests/resources/runner.js.
canvas_runner.js makes sure that one run takes less than 1 sec. The algorithm is 
min(1 sec, time to call api 1000 times).

This CL effects test results slightly, because now upload-canvas-2d-to-texture.html takes
6 sec, instead of 72 sec.

BUG=242244
NOTRY=true

Review URL: https://codereview.chromium.org/200253002

git-svn-id: svn://svn.chromium.org/blink/trunk@170288 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This consolidates the test cases post-rewrite.

Specifically:
* Remove old Perl TestInterface and TestObject (-2 files)
* Move special operations into TestInterface{,2,3} (-2 files)
(2 TestSpecialOperations*.idl left b/c 5 variants total, if squeezed so overlap)
* Remove TestEventTarget, as covered by TestInterfaceEventTarget (-1 file)
* Rename s/(.*)Python.idl/\1.idl

TBR=haraken
BUG=239771
NOTRY=true

Review URL: https://codereview.chromium.org/213543004

git-svn-id: svn://svn.chromium.org/blink/trunk@170287 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Both transition and will-change should produce string values for unknown
properties, but were producing identifiers. transition requires follow on
work as it only currently considers valid property names.

NOTRY=true

Review URL: https://codereview.chromium.org/216003002

git-svn-id: svn://svn.chromium.org/blink/trunk@170286 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=
R=pfeldman@chromium.org

Review URL: https://codereview.chromium.org/214783002

git-svn-id: svn://svn.chromium.org/blink/trunk@170285 bbb929c8-8fbe-4397-9dbb-9b2b20218538

These tests come from WebKit's LayoutTests/crypto, with only slight modifications to work in Blink.

BUG=245025

Review URL: https://codereview.chromium.org/215763002

git-svn-id: svn://svn.chromium.org/blink/trunk@170284 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=335712

Review URL: https://codereview.chromium.org/214523004

git-svn-id: svn://svn.chromium.org/blink/trunk@170283 bbb929c8-8fbe-4397-9dbb-9b2b20218538

+minor refactoring.

BUG=354131

Review URL: https://codereview.chromium.org/212933011

git-svn-id: svn://svn.chromium.org/blink/trunk@170282 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://src.chromium.org/viewvc/blink?view=revision&revision=170250

TBR=vollick@chromium.org

Review URL: https://codereview.chromium.org/216483002

git-svn-id: svn://svn.chromium.org/blink/trunk@170281 bbb929c8-8fbe-4397-9dbb-9b2b20218538

We have ASSERT_DISABLED macro, and it is typically used as

#if !ASSERT_DISABLED
...
#endif

Double negative is not good for readability. We can write it with ASSERT_ENABLED
as following:

#if ASSERT_ENABLED
...
#endif

TEST=none; no behavior changes
BUG=

Review URL: https://codereview.chromium.org/216313002

git-svn-id: svn://svn.chromium.org/blink/trunk@170280 bbb929c8-8fbe-4397-9dbb-9b2b20218538

All forwarded methods, expect didReceiveResponse, contains m_client check.

Added check to didReceiveResponse for consistency.

BUG=356654

Review URL: https://codereview.chromium.org/213383008

git-svn-id: svn://svn.chromium.org/blink/trunk@170279 bbb929c8-8fbe-4397-9dbb-9b2b20218538

<select>, or <input> tags via Autofill.

This more closely matches the spec behavior as "input" events should be fired
whenever the text is changed.  From the spec[1]:

"""
When the user agent is to change an input element's value on behalf of the user
(e.g. as part of a form prefilling feature), the user agent must queue a task to
first update the value accordingly, then fire a simple event that bubbles named
input at the input element, then fire a simple event that bubbles named change
at the input element.
"""

BUG=353691
R=tkent@chromium.org

[1] http://www.whatwg.org/specs/web-apps/current-work/multipage/common-input-element-attributes.html#event-input-input

Review URL: https://codereview.chromium.org/215903003

git-svn-id: svn://svn.chromium.org/blink/trunk@170278 bbb929c8-8fbe-4397-9dbb-9b2b20218538