GitLab

The scrollbar margin on the overlay scrollbar theme is used to distance
the thumb from the edge of the screen on all sides, not just the
widthwise side.  This was already computed properly in the theme's
TrackRect, but it wasn't communicated to the WebScrollbarLayer.

Also, shrink the margin to 3 DIPs on Android, which on closer inspection
better matches the system theme.

Chromium-side change at https://codereview.chromium.org/234603002/
NOTRY=true
BUG=361923

Review URL: https://codereview.chromium.org/234613002

git-svn-id: svn://svn.chromium.org/blink/trunk@171404 bbb929c8-8fbe-4397-9dbb-9b2b20218538

There is no big reason that NavigatorContentUtils should use RefCountedSupplement. This CL marks the NavigatorContentUtils to use Supplement instead of RefCountedSupplement. Additionally this cl adds a tracing method for supplement.

BUG=none

Review URL: https://codereview.chromium.org/235023003

git-svn-id: svn://svn.chromium.org/blink/trunk@171403 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Certain type of resources must have their clients notified
asynchronously.

BUG=352043

Review URL: https://codereview.chromium.org/231873006

git-svn-id: svn://svn.chromium.org/blink/trunk@171402 bbb929c8-8fbe-4397-9dbb-9b2b20218538

It seems not to needed. 
The variable "paths" is not used.

Review URL: https://codereview.chromium.org/235933005

git-svn-id: svn://svn.chromium.org/blink/trunk@171401 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Chrome side is: https://codereview.chromium.org/231763003/

This isn't perfect (using the family name directly rather than
doing a name/style/etc. fallback based lookup), but it's simple
and works sufficiently for layout tests and local developer use
of custom fonts. If we want to in the future, the same interface
could be used in conjunction with some skia changes to support
custom font collections, which would get the full style matching
support.

TBR=dglazkov@chromium.org
R=eae@chromium.org
BUG=333029

Review URL: https://codereview.chromium.org/235553003

git-svn-id: svn://svn.chromium.org/blink/trunk@171400 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Instead of using SkTypeface directly which goes through the
default Skia FontMgr, pass ours through so that when enabling
DirectWrite we don't use the GDI one.

R=eae@chromium.org
BUG=362694

Review URL: https://codereview.chromium.org/235753003

git-svn-id: svn://svn.chromium.org/blink/trunk@171399 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This CL removes some compositing chicken/egg work by deferring
updateCompositingRequirementsState until the compositing update. This CL is the
first step towards properly integrating this work with the rest of the
compositing update.

Review URL: https://codereview.chromium.org/235933002

git-svn-id: svn://svn.chromium.org/blink/trunk@171398 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This also moves a few static methods that don't make sense on WebFrame
to WebLocalFrame.

BUG=346764
R=eseidel@chromium.org

Review URL: https://codereview.chromium.org/232133004

git-svn-id: svn://svn.chromium.org/blink/trunk@171396 bbb929c8-8fbe-4397-9dbb-9b2b20218538

run-webkit-tests normally is robust against trying to run invalid or
missing tests, but it turns out that when run with the --run-chunk
argument there was a code path that tried to divide by zero because
we didn't err out early enough.

R=thestig@chromium.org
BUG=362749

Review URL: https://codereview.chromium.org/235863005

git-svn-id: svn://svn.chromium.org/blink/trunk@171395 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This SVG1.1[1] property has been removed from SVG2[2] and can be removed.

[1] http://www.w3.org/TR/SVG11/single-page.html#text-KerningProperty
[2] https://svgwg.org/svg2-draft/single-page.html

Review URL: https://codereview.chromium.org/234453002

git-svn-id: svn://svn.chromium.org/blink/trunk@171393 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Add a performance test that emulates regions but does not actually use them.

This is a merge of http://trac.webkit.org/changeset/167022 by Zoltan Horvath <zoltan@webkit.org>.

NOTRY=true
BUG=349093

Review URL: https://codereview.chromium.org/231943002

git-svn-id: svn://svn.chromium.org/blink/trunk@171392 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Previously it was running a number of concurrent operations, however none of those operations were using the same key.

Now each of the encrypt/decrypt/exportKey, sign/verify/exportKey sequences for a particular key are also run in parallel.

BUG=245025
R=abarth@chromium.org

Review URL: https://codereview.chromium.org/224823003

git-svn-id: svn://svn.chromium.org/blink/trunk@171391 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This CL removes HTMLImportRoot abstract class and merge it
into HTMLImportsController.

TEST=none
BUG=none
R=dglazkov@chromium.org, dominicc@chromium.org

Review URL: https://codereview.chromium.org/234063004

git-svn-id: svn://svn.chromium.org/blink/trunk@171390 bbb929c8-8fbe-4397-9dbb-9b2b20218538

We're hitting an assert in requiresCompositingForScrollableFrame,
because it depends on layout information. With
http://src.chromium.org/viewvc/blink?revision=171366&view=revision,
we now try to update this state when setting forceCompositingMode,
at which point, layout is not necessarily up to date.

We can just early return here because this method will be called
again after we've done the layout. This does potentially expose
us to bugs from chicken-egg issues that depend on the
m_forceCompositingMode bit for subframes.

Review URL: https://codereview.chromium.org/236193002

git-svn-id: svn://svn.chromium.org/blink/trunk@171389 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://src.chromium.org/viewvc/blink?view=revision&revision=171367

BUG=362078
TBR=enne@chromium.org

Review URL: https://codereview.chromium.org/236243002

git-svn-id: svn://svn.chromium.org/blink/trunk@171388 bbb929c8-8fbe-4397-9dbb-9b2b20218538

In the internal WebCrypto API, WebCryptoDigestor can act as a superset of
digestSynchronous. It will simlify things to not have digestSynchronous anymore,
so as a first step, this removes its one use.

Review URL: https://codereview.chromium.org/235893003

git-svn-id: svn://svn.chromium.org/blink/trunk@171387 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=361828

Review URL: https://codereview.chromium.org/225053005

git-svn-id: svn://svn.chromium.org/blink/trunk@171386 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This function does a hash lookup and is needlessly indirect. Instead, we can
just keep a bit on RenderLayerScrollableArea to remember whether we actually
scroll. I was hoping this would make the tweetus demo from famo.us faster, but
it didn't seem to have much of an effect. It's worth doing just for code
clarity, however.

Review URL: https://codereview.chromium.org/235903003

git-svn-id: svn://svn.chromium.org/blink/trunk@171385 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://src.chromium.org/viewvc/blink?view=revision&revision=171367

BUG=362078
TBR=enne@chromium.org

Review URL: https://codereview.chromium.org/235863004

git-svn-id: svn://svn.chromium.org/blink/trunk@171384 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Found via the W3C web-platform-tests. Once the back-end is
sending the version to the front-end (through existing plumbing),
create and populate the correct kind of event, per the spec.

Blink side must land first, then Chromium side, then Blink tests.

BUG=362251

Review URL: https://codereview.chromium.org/235493003

git-svn-id: svn://svn.chromium.org/blink/trunk@171382 bbb929c8-8fbe-4397-9dbb-9b2b20218538

See spec bug: https://www.w3.org/Bugs/Public/show_bug.cgi?id=21275

R=abarth@chromium.org, morrita@chromium.org
BUG=336698

Review URL: https://codereview.chromium.org/178883002

git-svn-id: svn://svn.chromium.org/blink/trunk@171381 bbb929c8-8fbe-4397-9dbb-9b2b20218538

There was a "fast path" that just compared the strings and then another
that split and then compared. The two paths were slightly different for
no obvious reasons which doesn't really follow the spec where the tokens
can be all mixed with each other (except for "shortcut icon" apparently).

Allocating the separate strings for "alternate stylesheet" should not be
an issue.

I also removed the extra constructor, we can just use a default argument
value.

The only behavior change is that now we allow things like "alternate import"
(or dns-prefetch) and just ignore the alternate part when deciding what
to do with the link.

Review URL: https://codereview.chromium.org/235063003

git-svn-id: svn://svn.chromium.org/blink/trunk@171380 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Review URL: https://codereview.chromium.org/235093005

git-svn-id: svn://svn.chromium.org/blink/trunk@171379 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Make shape-margin parsing accept percentages and properly compute the
length once the percentage is accepted.

This is a merge of http://trac.webkit.org/changeset/166787 by Bem Jones-Bey <bjonesbe@adobe.com>.

BUG=353357

Review URL: https://codereview.chromium.org/226373002

git-svn-id: svn://svn.chromium.org/blink/trunk@171378 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Unhandled case found via W3c web-platform-tests. If an indexing
operation is kicked off then the object store is deleted, an
ASSERT was hit since the cursor is invalid. Check (via the cursor)
that the store has not been deleted; if it has, stop indexing.

BUG=362711

Review URL: https://codereview.chromium.org/235883002

git-svn-id: svn://svn.chromium.org/blink/trunk@171377 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://src.chromium.org/viewvc/blink?view=revision&revision=171364

BUG=302005
TBR=skobes@google.com

Review URL: https://codereview.chromium.org/232463008

git-svn-id: svn://svn.chromium.org/blink/trunk@171376 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Traditionally, we have carefully handled Cache-Control headers that were set on a ResourceResponse, but have pretended Cache-Control doesn't exist on ResourceRequests. As of http://src.chromium.org/viewvc/blink?view=rev&rev=165603, we do the smarter naive thing, and never allow resources to be reused from MemoryCache if the ResourceRequest includes a Cache-Control header.

This patch moves ResourceResponse's Cache-Control parsing to HTTPParsers and handles Cache-Control headers consistently, whether the header appears on a ResourceRequest or a ResourceResponse.

BUG=

This relands r171178.

Review URL: https://codereview.chromium.org/232993004

git-svn-id: svn://svn.chromium.org/blink/trunk@171375 bbb929c8-8fbe-4397-9dbb-9b2b20218538

> We'll soon get rid of the v8 default isolate, so we'll need to manage
> our own.
>
> To test the platform implementation, run with
>   --js-flags="--noparallel_sweeping --concurrent-sweeping --job-based-sweeping"
>
> BUG=359977
> R=dcarney@chromium.org,haraken@chromium.org
>
> Original review https://codereview.chromium.org/227653002/

BUG=359977
TBR=dcarney@chromium.org

Review URL: https://codereview.chromium.org/235453004

git-svn-id: svn://svn.chromium.org/blink/trunk@171374 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This does not remove the code yet, only makes x-webkit-speech
test-only (only exposed during LayoutTests) which makes
this patch only a one-line change. We should follow up this
change with a full removal of this feature including
all tests.

x-webkit-speech is a Chrome-only feature which was
announced as deprecated Feb 2014:
https://groups.google.com/a/chromium.org/forum/#!topic/blink-dev/jlPdpx6jErE
Current usage is at about 0.487% of reported Chrome page views
which is (well) above our normal 0.06% removal threshold:
http://www.chromestatus.com/metrics/feature/popularity#PrefixedSpeechAttribute

However, hours ago we recieved report of cases where *requiring
user interaction* speech can be captured for up to 8 seconds
or until the user stops un-interrupted speaking without proper
appearance of the "Allow this site to use your microphone"
infobar.

Although this is only a medium-security issue (requires
user interaction and very limited duration), this is still
severe enough that we no longer believe we can support
this deprecated, webkit-only feature.

Removing this feature will not affect mobile users
which already have access to speech-input via various
speech-enabled keyboards on Android.  Desktop sites which
were using this feature can use the modern Web Speech APIs
instead, which are standardized and supported accross many
browsers: http://caniuse.com/web-speech

Without this feature pages will continue to work, they just
will not be able to get speech input via <input> controls
via these APIs and will either have to use the standard
Web Speech APIs or depend on the operating system to expose
speech input via other methods.

I've verified that disabling this does not affect the new
JS Web Speech API by testing with:
https://www.google.com/intl/en/chrome/demos/speech.html

BUG=360448

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=171187

Review URL: https://codereview.chromium.org/231323003

git-svn-id: svn://svn.chromium.org/blink/trunk@171373 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=358675

Review URL: https://codereview.chromium.org/227593006

git-svn-id: svn://svn.chromium.org/blink/trunk@171372 bbb929c8-8fbe-4397-9dbb-9b2b20218538

There's no reason for this function to return a bool. One caller
cares about it and the other doesn't. It's should be
the caller's job to track whether the bool changed.

Review URL: https://codereview.chromium.org/234423007

git-svn-id: svn://svn.chromium.org/blink/trunk@171371 bbb929c8-8fbe-4397-9dbb-9b2b20218538

These were marked as NeedsManualRebaseline in Blink r171349.  The change
to scrollbars was made in Chromium r263273.  Marking these now as
NeedsRebaseline.

TBR=abarth@chromium.org
BUG=362078

Review URL: https://codereview.chromium.org/235823002

git-svn-id: svn://svn.chromium.org/blink/trunk@171367 bbb929c8-8fbe-4397-9dbb-9b2b20218538

m_hasAcceleratedCompositing, m_showRepaintCounter and 
m_compositingTriggers are all just caches of values in Settings.
Instead of updating them after every style recalc and layout,
we should update them when the settings in question change.

Add an invalidation hook to Settings for updating compositing
triggers. Also, remove the hook that forces a style recalc
in acceleratedCompositingEnabled and showRepaintCounter.
As best I can tell, this call doesn't make sense anymore.
Compositing state can affect the results of style recalc,
but we're not actually setting needs recalc on any nodes,
so we wouldn't update them unless they already needed
recalc. I traced this back to
http://src.chromium.org/viewvc/blink?view=revision&revision=45199.

In addition to minimizing the amount of chicken-egg settings
update we have, this enables moving the accelerated fixed position
setting to being a trigger in CompositingReasonFinder
in https://codereview.chromium.org/231613002/. Without this patch,
we incorrectly get the wrong trigger value because we read
the compositing triggers inside of style recalc.

Unfortunately, we still need to set m_forceCompsitingMode in
updateCompositingLayersAfterLayout because layout can affect
whether we believe an iframe's content to be scrollable,
and thus whether it should be force composited.

Review URL: https://codereview.chromium.org/233323004

git-svn-id: svn://svn.chromium.org/blink/trunk@171366 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=362061

Review URL: https://codereview.chromium.org/234853002

git-svn-id: svn://svn.chromium.org/blink/trunk@171365 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Previously the multiplier was based on the first root touched by the current
layout pass (for table cells) or the deepest common ancestor of all the roots
(for other superclusters).

We now base the multiplier on the max known width of any root.  This is still
not great since not all widths are known, but it gives us consistent results
on hacker news.

BUG=361749
R=pdr@chromium.org

Review URL: https://codereview.chromium.org/231663006

git-svn-id: svn://svn.chromium.org/blink/trunk@171364 bbb929c8-8fbe-4397-9dbb-9b2b20218538

IntSize::area_safe was not named correctly for blink, and the old area
function should have been deprecated anyhow. This CL cleans this up.

Review URL: https://codereview.chromium.org/226523005

git-svn-id: svn://svn.chromium.org/blink/trunk@171363 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Reason for revert:
Under-repaint if body has percentage positioned or bottom aligned background-image when changing height.

Will work on bug 358460 first.

Original issue's description:
> Optimize repaint on FrameView resize
> 
> Previously we always repaint the full FrameView on resize.
> 
> Don't mark full repaint if the resize is height-only and writing mode
> is horizontal. Any descendants needing repaint because of FrameView
> resize will still schedule their own repaints.
> 
> BUG=258219
> TEST=WebFrameTest.sizeChangeRepaint
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=170427

TBR=ojan@chromium.org,dsinclair@chromium.org,esprehn@chromium.org,eseidel@chromium.org,jchaffraix@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=258219,362407

Review URL: https://codereview.chromium.org/235323005

git-svn-id: svn://svn.chromium.org/blink/trunk@171362 bbb929c8-8fbe-4397-9dbb-9b2b20218538

I don't see how we could have a null page here. I put an
assert(page) in CompositingReasonFinder::updateTriggers to
verify that no tests hit this.

Review URL: https://codereview.chromium.org/233963002

git-svn-id: svn://svn.chromium.org/blink/trunk@171361 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Don't use the same no-op filter for all url() inputs and outputs, or caching gets confused. NULL will do.

Covered by css3/filters/effect-reference-ordering-hw.html and others.

(This doesn't cause any test failures yet, but it will if https://codereview.chromium.org/230653005/ lands and rolls in.)

BUG=362561

Review URL: https://codereview.chromium.org/235303002

git-svn-id: svn://svn.chromium.org/blink/trunk@171360 bbb929c8-8fbe-4397-9dbb-9b2b20218538

I left the Persistents in the supplements to not rock the boat.
We should now be in position to move the supplements to the heap
and get them traced.

R=oilpan-reviews@chromium.org, zerny@chromium.org
BUG=

Review URL: https://codereview.chromium.org/230173004

git-svn-id: svn://svn.chromium.org/blink/trunk@171359 bbb929c8-8fbe-4397-9dbb-9b2b20218538