GitLab

When building Chrome with VS 2015 Update 2 RC there is a compiler error:

1>  texture_definition.cc
1>base/memory/ref_counted.h(414): error C2027: use of undefined type 'gpu::gles2::GLStreamTextureImage'
1>  gpu/command_buffer/service/texture_manager.h(31): note: see declaration of 'gpu::gles2::GLStreamTextureImage'
1>  base/memory/ref_counted.h(413): note: while compiling class template member function 'void scoped_refptr<gpu::gles2::GLStreamTextureImage>::AddRef(T *)'

This is caused by a compiler bug, together with a type trait used in
std::vector:
https://twitter.com/StephanTLavavej/status/710191543840219136

A VS bug has been filed at:
https://connect.microsoft.com/VisualStudio/feedback/details/2475971

Bug 595189 has a more minimal repro.

BUG=440500,595189
CQ_INCLUDE_TRYBOTS=tryserver.chromium.win:win_optional_gpu_tests_rel;tryserver.chromium.mac:mac_optional_gpu_tests_rel

Review URL: https://codereview.chromium.org/1806743002

Cr-Commit-Position: refs/heads/master@{#381640}

The GLHelperHolder in RenderWidgetHostViewAndroid wants to use a
command-buffer-aware context, which means it has to create a
WebGraphicsContext3DImpl. However we can hide that behind the
ContextProviderCommandBuffer API, to help us deprecate and
delete the WebGraphicsContext3D class.

We make a few cleanups to the class overall while we're touching it.

R=sievers@chromium.org
BUG=584497

Review URL: https://codereview.chromium.org/1805343005

Cr-Commit-Position: refs/heads/master@{#381639}

This change is a subset of https://codereview.chromium.org/1698183005/

A Task is a unit of work in the task scheduler. It has a closure,
a sequenced time, TaskTraits and other metadata inherited from
base::PendingTask.

A Sequence holds Tasks that must run in order. It is ref-counted
and has thread-safe Push, Pop and Peek operations. Priority queues,
worker threads and task runners will have references on Sequences.

BUG=553459

Review URL: https://codereview.chromium.org/1705253002

Cr-Commit-Position: refs/heads/master@{#381638}

Review URL: https://codereview.chromium.org/1813613002

Cr-Commit-Position: refs/heads/master@{#381637}

Reduced official build size from 7.9MB to 6.4MB. Can reenable WebRTC by changing the BUILD.gn file.

BUG=595038

Review URL: https://codereview.chromium.org/1806963002

Cr-Commit-Position: refs/heads/master@{#381636}

TBR=bruening@chromium.org, oshima@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1808023003

Cr-Commit-Position: refs/heads/master@{#381635}

BUG=595435

Review URL: https://codereview.chromium.org/1805913004

Cr-Commit-Position: refs/heads/master@{#381634}

Previously PairingAuthenticator classes were responsible for handling
client ID. This meant that in case there are multiple pairing
authentication methods the client would have to resend the client_id
if the host chooses different authentication method. Now the client
ID is handled by the negotiating layer, so client_id needs to be sent
only once.

Also now NegotiatingAuthenticators create PairingAuthenticators even
when falling back to PIN. This ensures that <pairing-failed> tag is
always present when falling back to PIN.

BUG=593123

Review URL: https://codereview.chromium.org/1781173005

Cr-Commit-Position: refs/heads/master@{#381633}

https://chromium.googlesource.com/native_client/src/native_client.git/+log/d26dd9f685f4..fc8541e88a9b

$ git log d26dd9f68..fc8541e88 --date=short --no-merges --format='%ad %ae %s'
2016-03-16 jpp PNaCl. Subzero. ARM32. Miscellaneous fixes.

CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_chromium_dbg_32_ng,linux_nacl_sdk_build
TBR=mseaborn@chromium.org

Review URL: https://codereview.chromium.org/1814673002

Cr-Commit-Position: refs/heads/master@{#381632}

Remove problematic dependencies on //ipc from //ui/events when targetting
iOS (Chrome on iOS cannot depends on //ipc nor //content).

BUG=459705

Review URL: https://codereview.chromium.org/1802333002

Cr-Commit-Position: refs/heads/master@{#381631}

Examples:
- load a page that takes some time to load (e.g. nytimes.com)
- begin navigating immediately

result:
wrap sound

expected:
to have something focused

- navigate to nfb.org
- navigate to the skip link
- continue navigating

result:
wrap sound

expected:
focus lands somewhere (it appears focusing some elements causes them to get invalidated).

m

BUG=

Review URL: https://codereview.chromium.org/1814633002

Cr-Commit-Position: refs/heads/master@{#381630}

According to the spec [1] "CSS-wide keywords are not valid <custom-ident>s",
so move isCSSWideKeyword to CSSPropertyParserHelper and use it
in consumeCustomIdent to exclude them.

BUG=584999

[1] https://drafts.csswg.org/css-values-3/#identifier-value

Review URL: https://codereview.chromium.org/1808753002

Cr-Commit-Position: refs/heads/master@{#381629}

BUG=577505

Review URL: https://codereview.chromium.org/1797323003

Cr-Commit-Position: refs/heads/master@{#381628}

https://chromium.googlesource.com/chromium/src/+/a7a987e9e

BUG=581974
TBR=kojii@chromium.org

Review URL: https://codereview.chromium.org/1806193002 .

Cr-Commit-Position: refs/heads/master@{#381627}

This was never implemented outside NSS sockets and things aren't using it
anywhere. Token Binding is now using key exporters, so we can remove this.

BUG=none

Review URL: https://codereview.chromium.org/1807063002

Cr-Commit-Position: refs/heads/master@{#381626}

https://chromium.googlesource.com/external/github.com/catapult-project/catapult.git/+log/468628c224cf..c4384a34d3b6

$ git log 468628c22..c4384a34d --date=short --no-merges --format='%ad %ae %s'

CQ_INCLUDE_TRYBOTS=
TBR=catapult-sheriff@chromium.org

Review URL: https://codereview.chromium.org/1812683002

Cr-Commit-Position: refs/heads/master@{#381625}

It seems that when the SeekDone comes from the remote side, we crash if there is no pipeline.
When we're remoting, we should probably get the pause_time_ from the remote side, so let's
avoid the crash by doing that.

BUG=586225

Review URL: https://codereview.chromium.org/1807053002

Cr-Commit-Position: refs/heads/master@{#381624}

There are two reasons to rename the worker:
- The formatting worker formats all kinds of formats.
- There's a confustion between JavaScriptFormatter.js and
  ScriptFormatterWorker.js

The patch also splits out FormattedContentBuilder into its own file.

BUG=none
R=dgozman, pfeldman

Review URL: https://codereview.chromium.org/1805343003

Cr-Commit-Position: refs/heads/master@{#381623}

R=dalecurtis@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1812713002

Cr-Commit-Position: refs/heads/master@{#381622}

cc::AnimationPlayer is an RC object, so it may outsurvive blink::CompositorAnimationPlayer if attached to cc::AnimationTimeline.

We must clean-up animation delegate in cc::AnimationPlayer on blink::CompositorAnimationPlayer deletion.

This is tests for this CL: https://codereview.chromium.org/1747283004/

BUG=590803

Review URL: https://codereview.chromium.org/1810833002

Cr-Commit-Position: refs/heads/master@{#381621}

The String.prototype.lineEndings pretended to cache the lineEndings value
on the instance of String object. However, it turned out to be working
only for string objects, and not for the string primitives.

For example, consider the following code snippet:

    var a = "foo";
    a.lineEndings(); // caching failed

    var b = new String(a);
    b.lineEndings(); // caching successful

    var c = b.substring(0);
    c.lineEndings(); // caching failed

The patch removes the caching from String.prototype.lineEndings(). Instead, it introduces
WI.Text object, which wraps String, caches lineEndings
properly, and provides some other handy methods.

BUG=none
R=dgozman, pfeldman

Review URL: https://codereview.chromium.org/1809533003

Cr-Commit-Position: refs/heads/master@{#381620}

Before this patch, <custom-ident> serializing used
quoteCSSStringIfNeeded to serialize, which can end up
inserting quotes.
However <custom-ident> are identifiers [1] and should
serialize as such, meaning no quotes [2].
This patches fixes that by calling serializeIdentifier.

Since the font family property relied on old CSSCustomIdentValue
behavior, this patch adds a CSSFontFamilyValue class to
keep said behavior, but with the change that previously we used
single quotes, now double quotes [3]. Most of the expected test
changes are because of this.

In order to fix parsing-css-string-characters.html and
string-quote-binary.html completely, the U+007F handling
mentioned in [3] is implemented.

Behavior matches Firefox.

BUG=584999

[1] https://drafts.csswg.org/css-values-3/#custom-idents
[2] https://drafts.csswg.org/cssom/#serialize-an-identifier
[3] https://drafts.csswg.org/cssom/#serialize-a-string

Review URL: https://codereview.chromium.org/1778743003

Cr-Commit-Position: refs/heads/master@{#381619}

The former method is deprecated, the latter is not. They have the same
functionality.
https://developer.apple.com/library/mac/releasenotes/General/MacOSXLionAPIDiffs/IOBluetooth.html

BUG=592663

Review URL: https://codereview.chromium.org/1813533002

Cr-Commit-Position: refs/heads/master@{#381618}

For reference, this color is overlaid atop the frame/toolbar when outlining tabs
and for outlines/shadows when drawing the new tab button.  It's either white or
black with a varying (but usually 0x40) alpha value.

Computing this is surprisingly complicated, beause we're trying to contrast with
two different colors simultaneously (tab and frame), and we also want to set our
magic numbers so as to achieve the colors from the design specs in the default
theme.

After quite a bit of thought, I elected to use a mechanism that defaults to
moving the frame away from the tab luminance; that is, if the tab is "brighter"
than the frame, we make the separator darken the frame, and if the tab is
darker, we try to lighten the frame.  If the frame is already so dark or light
that the result has too low of contrast with the frame color, we reverse
direction.  ("Too low" here is a lot lower than I'd like, but any higher and
we'd end up using light separators for the default theme in incognito mode,
which I think looks good but Sebastien dislikes.)  In the case where we reversed
direction, we have to worry that the result of all this won't contrast enough
with the tab; in that case, we push up the alpha value so the result contrasts
enough with the tab as well.  This last computation is the most expensive
because of how I chose to make it behave, but I think the behavior I selected
(too complicated to explain here, see code/comments for details) will feel more
consistent than any of the simpler methods I considered.

Because computing the separator color can be expensive (every call to
GetRelativeLuminance() can involve floating-point exponentiation among other
things, and in the worst case the color computation computes the desired color
via a 7-iteration loop), I elected to cache the computed value in a map.  This
might be a case of premature optimization, but in debugging it looked like this
color could be requested frequently, and I didn't want to risk performance
problems.

Even this choice presented options.  I used a simple map that I never prune
entries from; at worst, we'll add up to 2 entries per distinct theme the user
switches to while running, which didn't seem too bad.  I considered instead
using base::MRUCache, which would let me cap the size.  I also considered just
keeping a couple member structs containing the relevant information for the
normal and incognito color schemes, but this generally seemed like it ended up
as more code than the other routes.  None of these options seems wildly better
or worse than the others; I'm willing to change course in the face of violent
opinion :)

Finally, because the alpha value of the separator can now vary, I converted the
code in tab_strip.cc that set it to fixed values to instead use scaling
multipliers.  These will compute the same values as before when the separator
has its default (0x40) alpha, but in the case where we've computed some larger
alpha, scaling proportionally seemed like the best thing to do.  I used a
saturated_cast in one place where I wasn't sure the result was guaranteed to
stay <= 255.

BUG=585470
TEST=See bug comment 0

Review URL: https://codereview.chromium.org/1785613004

Cr-Commit-Position: refs/heads/master@{#381617}

These interfaces on LayoutObject will be used to more efficiently
represent object locations in the accessibility tree, rather than storing
everything using absolute coordinates.

BUG=551601

Review URL: https://codereview.chromium.org/1777613002

Cr-Commit-Position: refs/heads/master@{#381616}

This change adds the ability to transfer mach ports over Mojo. Mach ports can
either be wrapped using CreatePlatformHandleWrapper() or a Mojo shared buffer
can be created using CreateSharedBufferWrapper(). For now, Mojo shared buffers
created using MojoCreateSharedBuffer() (or MojoCreateDataPipe()) will still
use posix shared memory.

BUG=582468

Review URL: https://codereview.chromium.org/1712143002

Cr-Commit-Position: refs/heads/master@{#381615}

BUG=568830
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel

Review URL: https://codereview.chromium.org/1805343006

Cr-Commit-Position: refs/heads/master@{#381614}

This patch ensures that we process active low res pending high res
tilings, since they can have required for activation tiles. Previously,
this only applied to non ideal tiles, but clearly it's possible to hit
the rare case of low res on active while high res on pending.

R=enne, chrishtr
BUG=595209
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel

Review URL: https://codereview.chromium.org/1808043002

Cr-Commit-Position: refs/heads/master@{#381613}

As a part of https://codereview.chromium.org/1774323003, it became clear
that it needed to be possible to call SetClientReady repeatedly, which
some sources didn't support.

Instead of just making that part of the contract, instead fold the two
places where SetClientReady mattered into AddObserver (the first time
where a begin frame could actually be delivered) and remove
SetClientReady from the API.

R=skyostil@chromium.org,sunnyps@chomium.org
CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel

Review URL: https://codereview.chromium.org/1778853003

Cr-Commit-Position: refs/heads/master@{#381612}

https://chromium.googlesource.com/skia.git/+log/2732e186a113..bf680c30a280

$ git log 2732e186a..bf680c30a --date=short --no-merges --format='%ad %ae %s'
2016-03-16 senorblanco Image filters: fix the zero-sigma fast path in SkBlurImageFilter.

CQ_INCLUDE_TRYBOTS=tryserver.blink:linux_blink_rel
TBR=bungeman@google.com

Review URL: https://codereview.chromium.org/1808683003

Cr-Commit-Position: refs/heads/master@{#381611}

Summary of changes available at:
https://chromium.googlesource.com/v8/v8/+log/e27a07b4..eaa2e514

Please follow these instructions for assigning/CC'ing issues:
https://github.com/v8/v8/wiki/Triaging%20issues

Please close rolling in case of a roll revert:
https://v8-roll.appspot.com/
This only works with a Google account.

TBR=hablich@chromium.org,machenbach@chromium.org,yangguo@chromium.org,vogelheim@chromium.org

Review URL: https://codereview.chromium.org/1806963003

Cr-Commit-Position: refs/heads/master@{#381610}

Migrate to use SetContentSettingDefaultScope() which takes GURLs only in plugins.
For special test or specific patterns, remain to use SetContentSetting().

BUG=551747

Review URL: https://codereview.chromium.org/1800913002

Cr-Commit-Position: refs/heads/master@{#381609}

DesktopNotificationProfileUtil::ClearSetting() is changed to take GURLs directly
which internally call HostContentSettingsMap::SetContentSettingDefaultScope().

For case in MessageCenterSettingsController, since the patterns are from user
input, SetContentSetting() is used instead of ClearSetting() to take patterns
directly.

BUG=551747

Review URL: https://codereview.chromium.org/1694923004

Cr-Commit-Position: refs/heads/master@{#381608}

Previously, ToImageSkia, ToUIImage and ToNSImage would assume the
scoped_ptr they pass to AddRepresentation would remain valid after that
method returns. This *should* be the case, but it is poor form to assume
it. Now AddRepresentation returns a valid pointer and the callers return
that, rather than assuming the scoped_ptr stays alive.

BUG=590882

Review URL: https://codereview.chromium.org/1769433002

Cr-Commit-Position: refs/heads/master@{#381607}

This updates from revision
3594dcfbd8cf1cb3ba2c57bd56b761147ea31fca (2016-01-04)

to revision
fa6e6592cd76eb0cf6d442dbebe24f0a22787c6e (2016-03-01)

R=pkasting@chromium.org
BUG=37436

Review URL: https://codereview.chromium.org/1750853002

Cr-Commit-Position: refs/heads/master@{#381606}

PluginObject creation can fail if there isn't a valid V8 context. In this
case we need to take care to free the ppp_class_data.

BUG=594926

Review URL: https://codereview.chromium.org/1805303002

Cr-Commit-Position: refs/heads/master@{#381605}

This means that localization is performed before data is sent to the
WebUI client. This CL removes almost all existing i18n-template usage.
We still have a small number of more complex strings which are created
in JS which are more difficult to convert to the new system.

BUG=563884

Review URL: https://codereview.chromium.org/1609683002

Cr-Commit-Position: refs/heads/master@{#381604}

Move most of the logic for extension action execution into the
ExtensionActionRunner (instead of the ExtensionActionAPI), which
eliminates some of the back-and-forth and makes a more common place
for execution logic and adds slightly finer-grained control for
what to execute. This also becomes more important with upcoming
changes for webRequest/document_start handling.

BUG=595087

Review URL: https://codereview.chromium.org/1804123003

Cr-Commit-Position: refs/heads/master@{#381603}

BUG=595490
TBR=zhaoqin@chromium.org

Review URL: https://codereview.chromium.org/1809883002

Cr-Commit-Position: refs/heads/master@{#381602}

Since swapped out RenderFrameHosts are no longer used, its swapped out
state is no longer necessary. This CL removes it along with a public
API that exposed it outside of content/.

BUG=357747
CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_site_isolation

Review URL: https://codereview.chromium.org/1799163002

Cr-Commit-Position: refs/heads/master@{#381601}