GitLab

We have RefPtr for preventing XHR from being destroyed in
internalAbort(). But we're touching XHR's member variables in the caller
of internalAbort(). We must have the RefPtr for protection there.

BUG=421504

Review URL: https://codereview.chromium.org/663693003

git-svn-id: svn://svn.chromium.org/blink/trunk@183859 bbb929c8-8fbe-4397-9dbb-9b2b20218538

A false bug report led me to the realization that srcset's unit testing didn't include URLs with encoded spaces in them.
Added such a test here.

BUG=422914

Review URL: https://codereview.chromium.org/643703005

git-svn-id: svn://svn.chromium.org/blink/trunk@183858 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The current parser allows an arbitrary number of %s (e.g. left: 10%%%),
but we shouldn't depend on it in our unit tests.

Review URL: https://codereview.chromium.org/641343004

git-svn-id: svn://svn.chromium.org/blink/trunk@183857 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Strings are not valid keyframes/animation-name identifiers. This change
tracks the occurrence of these strings in the wild to provide data
on the impact dropping support for this would have.

BUG=423424

Review URL: https://codereview.chromium.org/651123003

git-svn-id: svn://svn.chromium.org/blink/trunk@183856 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Before this CL, we allocate memory for IDL dictionaries each time we use
them. For performance sensitive methods, this may not be acceptable.
Avoid allocating them in the oilpan heap. Instead, put them on stack.
Dictionaries can be a member of collections, so we use
ALLOW_ONLY_INLINE_ALLOCATION().

Side effect of this change:
If an impl class implements an IDL method which returns a dictionary,
the impl method needs to take an argument for return value.

This CL also add blank lines in generated dictionary impl classes for
readability.

BUG=321462

Review URL: https://codereview.chromium.org/656073002

git-svn-id: svn://svn.chromium.org/blink/trunk@183855 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The AffineTransformation for animateMotion is pretty rare
and just makes SVGGraphicsElement take up more memory, so
move it to SVGElementRareData.

Review URL: https://codereview.chromium.org/660653004

git-svn-id: svn://svn.chromium.org/blink/trunk@183854 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The code to cache preferredMainAxisContentExtentForChild when possible
and re-use to avoid forcing layout was broken in the case that a child's
flex basis changed to cause it to be dependent on layout, triggering
an assert in debug. Just pushing it down the normal path fixes the
issue.

BUG=366185

Review URL: https://codereview.chromium.org/655053002

git-svn-id: svn://svn.chromium.org/blink/trunk@183853 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=sof@opera.com

Review URL: https://codereview.chromium.org/656403002

git-svn-id: svn://svn.chromium.org/blink/trunk@183852 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The scrollCount field is unused.

The unacceleratedScrollingDeltaX/Y field always equal deltaX/Y (by
inspection, then by adding an assert and testing), so use those
directly.

BUG=133097

Review URL: https://codereview.chromium.org/658953003

git-svn-id: svn://svn.chromium.org/blink/trunk@183851 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Currently, chromium hung bots and blink hung bots are shown in their
corresponding views. With this change, all hung bots will show up in the Trooper
view. This includes the chromium and blink ones, but also others other trees
(e.g. non-closers).

BUG=412189

Review URL: https://codereview.chromium.org/661463004

git-svn-id: svn://svn.chromium.org/blink/trunk@183850 bbb929c8-8fbe-4397-9dbb-9b2b20218538

We are about to support Promise<T> syntax in the Web IDL parser. As it is
specified in the Web IDL spec, I would like to implement it in
tools/idl_parser. This CL is a preparation for the change in chromium side,
in order not to break the existing behavior when the change lands.

BUG=421539

Review URL: https://codereview.chromium.org/654193002

git-svn-id: svn://svn.chromium.org/blink/trunk@183849 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=423071

Review URL: https://codereview.chromium.org/653303003

git-svn-id: svn://svn.chromium.org/blink/trunk@183848 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Add a new hidden devtools experiment for animation inspection within the
inspector. A new inspector domain is created for Animations.

CSS Animations, CSS Transitions and Web Animations can be viewed in an
Animations pane when an animation is in effect and targeting the element
being viewed in the Elements pane. Each animation can be paused, played
and scrubbed based on time position. Animation properties are shown for
all animations.

BUG=419269

Review URL: https://codereview.chromium.org/620783002

git-svn-id: svn://svn.chromium.org/blink/trunk@183847 bbb929c8-8fbe-4397-9dbb-9b2b20218538

(only used in unit tests)

R=schenney@chromium.org,senorblanco@chromium.org

Review URL: https://codereview.chromium.org/657293004

git-svn-id: svn://svn.chromium.org/blink/trunk@183846 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Alerts history will return a list of IDs, which can then be used to get actual alerts from history. This allows to avoid exceeding memory and response size limits when processing the query.

BUG=416182
R=ojan@chromium.org

Review URL: https://codereview.chromium.org/664563002

git-svn-id: svn://svn.chromium.org/blink/trunk@183845 bbb929c8-8fbe-4397-9dbb-9b2b20218538

When crossing an insertion point boundary from style recalc, we mark the
distributed nodes for subtree style recalc. The reason is that the
distributed nodes will have their style recalculated in the context of
where they are distributed from. In the case below, changing class=a on
content will affect the style of class=b. The invalidation starts on
class=a and will not reach class=b in the invalidation traversal.

The bug was that InsertionPoint::willRecalcStyle had an early return
if the parent style change was at least an Inherit, but that is not the
case if the class change does not affect anything at, or above, the
insertion point, like in the case below.

<div>
  <:shadow-root>
    <style>.a::content .b { ... }</style>
    <content class="a"></content>
  </:shadow-root>
  <div class="b"></div>
</div>

R=esprehn@chromium.org,chrishtr@chromium.org
BUG=423293

Review URL: https://codereview.chromium.org/637273003

git-svn-id: svn://svn.chromium.org/blink/trunk@183844 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Content editable when mixed in direction rtl, it was not moving caret because editing boundary was not calculated correctly. In RTL scenario, nodes need to be calculated from after or before for left and right respectively. Based on the direction of the block calculates appropriate editing boundary.

R=yosin
BUG=409517
TEST=editing/selection/skip-non-editable-rtl.html

Review URL: https://codereview.chromium.org/602423002

git-svn-id: svn://svn.chromium.org/blink/trunk@183843 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Update the test to not call generateRequest() for either stream
until both "onencrypted" events are received. This is to avoid
the occasional race condition where the first "message" event
arrives before the second "onencrypted" event, and thus messes
up the expected-output log comparison.

BUG=407993
TEST=modified test still pass

Review URL: https://codereview.chromium.org/658633002

git-svn-id: svn://svn.chromium.org/blink/trunk@183842 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Prevents annoying accidental seeks to the beginning of content for
streaming content; non-streaming content always exposes (0, duration)
as the seekable range so it's unaffected by this change.

BUG=412562
TEST=new layout test.

Review URL: https://codereview.chromium.org/614263002

git-svn-id: svn://svn.chromium.org/blink/trunk@183841 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This doesn't really invoke all the machinery involved in swapping
frames (a real frame swap would likely have a provisional local
frame before it committed into the frame tree, etc.) but it's
still helpful for catching problematic code.

BUG=346764

Review URL: https://codereview.chromium.org/409823002

git-svn-id: svn://svn.chromium.org/blink/trunk@183840 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This will eventually go away, but it doesn't belong on ChromeClient
in the meantime.

BUG=339659
TEST=No behavior change.

Review URL: https://codereview.chromium.org/642823006

git-svn-id: svn://svn.chromium.org/blink/trunk@183839 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Rename it to generateReferrer() to better match its return type.

Also, make setHTTPReferrer() remove the header if the value parameter is empty. There shouldn't ever be a reason to send an empty referrer header (rather than not sending a referrer header at all), so this enables us to enforce this property in one place, rather than several.

BUG=

Review URL: https://codereview.chromium.org/650023003

git-svn-id: svn://svn.chromium.org/blink/trunk@183838 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=oilpan-reviews,haraken
BUG=423536
NOTRY=true

Review URL: https://codereview.chromium.org/640813003

git-svn-id: svn://svn.chromium.org/blink/trunk@183837 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch modifies GraphicsContext::drawPicture to make it
avoid the use of image filters for presenting canvas
contents, and just use a plain save layer when drawing
contents at 1:1 pixel scale.

BUG=386601

Review URL: https://codereview.chromium.org/662673002

git-svn-id: svn://svn.chromium.org/blink/trunk@183836 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Introduced accidentally by copy-paste.

Review URL: https://codereview.chromium.org/658073003

git-svn-id: svn://svn.chromium.org/blink/trunk@183835 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=423536

Review URL: https://codereview.chromium.org/655063003

git-svn-id: svn://svn.chromium.org/blink/trunk@183834 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The logs for these EME tests expect all the "onencrypted" events to
occur before the "message" event is received. Modify the tests to
create the session on the second event rather than the first, to avoid
the "message" event occasionally happening before the second
"onencrypted" event".

BUG=407993
TEST=modified tests still pass

Review URL: https://codereview.chromium.org/634243005

git-svn-id: svn://svn.chromium.org/blink/trunk@183833 bbb929c8-8fbe-4397-9dbb-9b2b20218538

There are 101 different classes in the RenderObject class tree and
they share 150-250 virtual functions each. That scales to 60-200 KB
of vtbls (depending on architecture), which cost space in the
binary and time during startup since they need to be copied 
to RAM and adjusted. In fact, some RenderObject vtbls are the
largest vtbls in Chrome.

There doesn't have to be anything wrong with any of this, but the
use of 60+ virtual isXXX functions which were the same for 100 out of
101 classes could easily be replaced by a single virtual function, saving 20-50 KB of the space mentioned above (again depending on architecture).

When the class is known, the isOfType functions will constant fold, just
like the isFoo methods.

clang x64 space info:
Total change: -47075 bytes
==========================
  78 added, totalling +4731 bytes across 70 sources
  142 removed, totalling -4008 bytes across 70 sources
  397 grown, for a net change of +5064 bytes (296097 bytes before, 301161 bytes after) across 128 sources
  115 shrunk, for a net change of -52862 bytes (209130 bytes before, 156268 bytes after) across 5 sources

gcc x64 space into:
Total change: -49078 bytes
==========================
  141 added, totalling +19000 bytes across 90 sources
  200 removed, totalling -21292 bytes across 90 sources
  371 grown, for a net change of +17491 bytes (387125 bytes before, 404616 bytes after) across 164 sources
  261 shrunk, for a net change of -64277 bytes (387358 bytes before, 323081 bytes after) across 98 sources

R=esprehn@chromium.org

Review URL: https://codereview.chromium.org/613783002

git-svn-id: svn://svn.chromium.org/blink/trunk@183832 bbb929c8-8fbe-4397-9dbb-9b2b20218538

r183289 caused a new overwriting alerts bug because it didn't replicate
the way builder_alerts handled null failure reasons. Using step::reason
won't actually be unique if the reason is null since builder_alerts
doesn't merge failures with null reasons.

R=leviw@chromium.org,jyasskin@chromium.org
NOTRY=true

Review URL: https://codereview.chromium.org/638333002

git-svn-id: svn://svn.chromium.org/blink/trunk@183831 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Here is the OWP tracking bug: https://code.google.com/p/chromium/issues/detail?id=422984
Here is the link to the entry on the feature dashboard: https://www.chromestatus.com/features/4871530282483712

Currently canvas clips are not antialiased which differs what what other
browsers are doing and what the users expect. Skia supports antialiased clipping
but Skia's gpu backend (which is used for canvas) can be significantly slower
for concave clips. 

This CL starts surfacing the desired antialiased clipping behavior behind a flag
so developers can start assessing the capability. There is a lot more work that
needs to occur in Skia before we have a consistently good story for gpu-backed
antialiased clipping so we will not be ready to switch behaviors for a while.

BUG=crbug.com/7508
BUG=crbug.com/422984

junov: core & platform
dglazkov: Source/web & public/web

Review URL: https://codereview.chromium.org/652483003

git-svn-id: svn://svn.chromium.org/blink/trunk@183830 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This fixes the FrameViewer json serialization of annotatedInvalidationRects.
This CL fixes the format from (x, y, maxX, maxY) to (x, y, width, height).

BUG=402033

Review URL: https://codereview.chromium.org/659003002

git-svn-id: svn://svn.chromium.org/blink/trunk@183829 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Rather than NULL-checking it directly, I've instead plumbed through
ActiveDOMCallback::canInvokeCallback(), which handles all the logic
MutationObserver was doing previously and also handles the NULL check.

The test case is crazy-looking because it was generated by ClusterFuzz.
I've simplified it as much as possible.

BUG=410755

Review URL: https://codereview.chromium.org/657983002

git-svn-id: svn://svn.chromium.org/blink/trunk@183827 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=373552
R=eroman
Tests= crypto/ LayoutTests

Review URL: https://codereview.chromium.org/595873003

git-svn-id: svn://svn.chromium.org/blink/trunk@183826 bbb929c8-8fbe-4397-9dbb-9b2b20218538

R=haraken
BUG=
NOTRY=true

Review URL: https://codereview.chromium.org/661793002

git-svn-id: svn://svn.chromium.org/blink/trunk@183825 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=392606
TEST=RecordingImageBufferSurfaceTest

Review URL: https://codereview.chromium.org/659873002

git-svn-id: svn://svn.chromium.org/blink/trunk@183824 bbb929c8-8fbe-4397-9dbb-9b2b20218538

When we decide to promote all children of an anonymous block and then
nuke it, ignore any continuation set on the new parent of the children,
to make sure that the children are in fact inserted as children of said
new parent, and not somewhere else in the tree.

This is a hack that's needed as long as we keep the old multicol
implementation. The root cause is probably buggy block continuation
handling, but fixing that would be risky.

BUG=421720

Review URL: https://codereview.chromium.org/656143003

git-svn-id: svn://svn.chromium.org/blink/trunk@183823 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This reverts commit 9c188b20fbb4d4c2006ab1a611ed8a9b971e7cee.

Broke several browser_tests which depended on the previous behavior:
BrowserEncodingTest.TestOverrideEncoding
EncodingAliases/BrowserEncodingTest.TestEncodingAliasMapping/16
BrowserEncodingTest.TestEncodingAutoDetect
EncodingAliases/BrowserEncodingTest.TestEncodingAliasMapping/3

E.g.
http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=BrowserEncodingTest.TestOverrideEncoding&testType=browser_tests

TBR=jshin@chromium.org
BUG=412053
NOTRY=true

Review URL: https://codereview.chromium.org/658193002

git-svn-id: svn://svn.chromium.org/blink/trunk@183822 bbb929c8-8fbe-4397-9dbb-9b2b20218538

There are some legal tricks used to encode JSON. But JSON.parse rejects
to parse it, whereas XHR response is correctly transformed to JSON.

BUG=406900

Review URL: https://codereview.chromium.org/654083006

git-svn-id: svn://svn.chromium.org/blink/trunk@183821 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Currently, SASSSourceMapping polls CSS for 5 seconds after the SASS
change, and this turns out to be insufficient in case of complex
css-processing build steps.

Instead of adding one more setting "css polling duration", this patch
tries to solve most of the problems by increasing polling duration
to 30 seconds.

BUG=421925
R=vsevik, apavlov

Review URL: https://codereview.chromium.org/658123002

git-svn-id: svn://svn.chromium.org/blink/trunk@183819 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This is a bug that I introduced in r164832.
This caused SMIL animation targeting <marker>'s refY to animate refX attribute instead.

TEST=svg/markers/animate-refY.svg
BUG=None

Review URL: https://codereview.chromium.org/661733002

git-svn-id: svn://svn.chromium.org/blink/trunk@183818 bbb929c8-8fbe-4397-9dbb-9b2b20218538