GitLab

buildtools changes:
https://codereview.chromium.org/2977093002/

R=thakis@chromium.org
TBR=michaelpg@chromium.org

Bug: 593874
Change-Id: I825899b1f8d55e577f0bd061f53fd26b9b1a9dd8
Reviewed-on: https://chromium-review.googlesource.com/570743
Commit-Queue: Thomas Anderson <thomasanderson@chromium.org>
Reviewed-by: Nico Weber <thakis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486646}

This CL observes changes to extensions that may control the NTP and
sends that information to the OnStartup settings page.

Bug: 740779
Change-Id: Icdc539f0fcb64eeb6f0dc6f8eb9ade061d4e870e
Reviewed-on: https://chromium-review.googlesource.com/567419
Commit-Queue: Dave Schuyler <dschuyler@chromium.org>
Reviewed-by: Demetrios Papadopoulos <dpapad@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486645}

This CL sends a notification to the FeatureEngagementTracker when
the user opens Chrome both from a cold start and a warm start.

Warm starts include backgrounding the app and reopening it, launching
another app from Chrome and returning to it, and locking then unlocking
the screen when Chrome is in the foreground.

-[AppState applicationWillEnterForeground: ... ] handles the warm
start event because it is called whenever the app is taken out of
the background (except on the initial launch, which is a cold start).

The cold start event must occur in the startup flow, a complex process,
so finding a good place for the notification is tricky. It needs to
be called exactly once, and it needs a ChromeBrowserState object.
-[MainController startUpBrowserForegroundInitialization] was chosen
for a couple reasons.
 * It's called as part of the startup flow (necessary to make the
   chrome launched event accurate).
 * It's only called once. startUpBrowserToStage, which calls
   startUpBrowserForegroundInitialization, keeps track of what stage
   of initialization the app is in and makes sure it is not called
   twice.
 * This is where the ChromeBrowserState object is initialized. The
   FeatureEngagementTrackerFactory uses ChromeBrowserState to store
   and retrieve the FeatureEngagementTracker.

Locking then unlocking the screen with Chrome in the foreground is
considered a warm start because applicationWillEnterForeground is
called. Although this is not necessarily desired behavior, there's no
way to determine whether the app is returning from the lock screen.
This slightly increases the potential number of times the Chrome
Opened Event is fired, so in-product help promotions relying on this
event should increase the number of times this event must have fired
to compensate.

R=edchin@chromium.org, gchatz@chromium.org

Bug: 738164
Change-Id: I656297a800600672931c785fcac0fd6dcd4996aa
Reviewed-on: https://chromium-review.googlesource.com/558666
Commit-Queue: Cooper Knaak <cooperknaak@google.com>
Reviewed-by: David Trainor <dtrainor@chromium.org>
Reviewed-by: Mark Cogan <marq@chromium.org>
Reviewed-by: Ed Chin <edchin@chromium.org>
Reviewed-by: Gregory Chatzinoff <gchatz@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486644}

This is preparation to split LayoutSelection start/end calculation w/o 
canonicalization.

Bug: 739062
Change-Id: Ic318f07693a25e160060cb413f3c6edbb8f9e6b1
Reviewed-on: https://chromium-review.googlesource.com/569553Reviewed-by: Yoshifumi Inoue <yosin@chromium.org>
Commit-Queue: Yoshifumi Inoue <yosin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486643}

This CL modifies a cc::BlockingTaskRunner::PostTask's argument from
base::Callback to base::OnceCallback, so that it can take OnceCallback
in addition to base::Callback via Callback to OnceCallback conversion.

Change-Id: If45f17bf53ca56162cfa66cfa3cd9d02bbca7d42
Reviewed-on: https://chromium-review.googlesource.com/570079Reviewed-by: danakj <danakj@chromium.org>
Commit-Queue: Taiju Tsuiki <tzik@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486642}

Pressing enter key on the icon-button is expected to open the action-menu, but instead it incorrectly focuses the row-item.

BUG=740741
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation

Review-Url: https://codereview.chromium.org/2945433002
Cr-Commit-Position: refs/heads/master@{#486641}

Bug: 740075
Change-Id: I12f04bd64aa3d29bd72b75a54e7745741b284ce0
Reviewed-on: https://chromium-review.googlesource.com/568718
Commit-Queue: Justin Cohen <justincohen@chromium.org>
Reviewed-by: Rohit Rao <rohitrao@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486640}

This is a copy-paste of the "CFI Linux Full" configuration on
chromium.fyi.

Bug: 732652
Change-Id: I20cd4b9af27a99576111a40ede0582ed1f2a93be
Reviewed-on: https://chromium-review.googlesource.com/570833Reviewed-by: Dirk Pranke <dpranke@chromium.org>
Commit-Queue: Peter Collingbourne <pcc@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486639}

These calls are no longer needed after r485461
(https://crrev.com/2478803003) which has made the
domAutomationController.setAutomationId call a no-op.

BUG=662543
This CL was uploaded by git cl split.

R=michaelpg@chromium.org

Review-Url: https://codereview.chromium.org/2974323003
Cr-Commit-Position: refs/heads/master@{#486638}

This is a reland of f51e5d49
Original change's description:
> Add browser test and testing config for dynamic Expect-CT reporting
> 
> This CL adds a browser test to check that the ChromeExpectCTReporter is properly
> created and set up to receive reports, particularly for dynamic Expect-CT
> violations. It also enables the dynamic Expect-CT feature on the waterfall.
> 
> BUG=679012,642517
> 
> Change-Id: I3ea01207ea5852b9bd0bbb1d90799ee4555c854e
> Reviewed-on: https://chromium-review.googlesource.com/566572
> Reviewed-by: Jesse Doherty <jwd@chromium.org>
> Reviewed-by: Mustafa Emre Acer <meacer@chromium.org>
> Commit-Queue: Emily Stark <estark@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#486158}

TBR=jwd@chromium.org

Bug: 679012, 642517
Change-Id: If0a05ea0823f0427f04c455206ab8a7f7020fc89
Reviewed-on: https://chromium-review.googlesource.com/570561Reviewed-by: Emily Stark <estark@chromium.org>
Reviewed-by: Mustafa Emre Acer <meacer@chromium.org>
Commit-Queue: Emily Stark <estark@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486637}

Introduces a data item storage for data created on lock screen.

Data items are persisted on disk in per-extension value stores backed by
leveldb (same backend as chrome.storage.local API).
Value store is structured as follows:
* registered_items -> dictionary whose keys are IDs of registered item
  Dictionary values are currently empty dictionaries, but could be
  extended in future to contain data item metadata, if required.
* item ID -> base64 encoded encrypted data item content

Note that data item content is encrypted because the storage is kept
outside user directory. Encrypted data is base 64 encoded because value
store does not currently support binary data (the DB values are
preserved as JSON strings, and JSON writer does not handle binary
values well).

Additionally, this CL adds a dictionary to local state that maps extension
IDs to number of lock screen data items registered for the extension.
The purpose of this dictionary is to enable the lockScreen.data API to
determine set of apps to which it should set OnDataItemsAvailable event
when user session is activated without querying the values store
database for each installed extension.

BUG=715781

Review-Url: https://codereview.chromium.org/2934293003
Cr-Commit-Position: refs/heads/master@{#486636}

WebPluginContainerBase is no longer required as all of the references in the
code have been updated to use WebPluginContainerImpl.

Bug: 731490
Change-Id: Iad08ea5246f37db6609c20b8a4e0b8fd35811550
Reviewed-on: https://chromium-review.googlesource.com/571379Reviewed-by: Kentaro Hara <haraken@chromium.org>
Commit-Queue: Stuart Langley <slangley@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486635}

GLSurface::SetRelyOnImplicitSync was added to implement a workaround
on CrOS/Intel.

Initially that method was expected to be called only on GbmSurfaceless
that extends GLSurface.

https://chromium-review.googlesource.com/c/565842/ restricted the method
call only to platforms where the workaround is enabled.

Turns out that on Intel on CrOS we still allocate other types of
GLSurfaces. SurfacelessEGL is allocated as an offscreen surface.

In debug builds, where the target_os is chromeos and ozone is not used
we also end up allocating GLSurfaces that are not derived from
GbmSurfaceless, effectively calling GLSurface::SetRelyOnImplicitSync
and hitting the NOTIMPLEMENTED.

This CL removes NOTIMPLEMENTED from GLSurface::SetRelyOnImplicitSync
since it's fine in the scenarios described above to ignore the call
if the surface doesn't support it.

Bug: 740630 , 721463
Change-Id: Ib04bfff85bd90509d1f3028af93a2edc138fd129
Reviewed-on: https://chromium-review.googlesource.com/570740Reviewed-by: Mitsuru Oshima <oshima@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Daniele Castagna <dcastagna@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486634}

This CL deletes |result_map_| entry before send
completion signal, hence resolved a race condition.

Bug: 287726
Change-Id: I8d4e025eff4906bb657c3f28eb7c302293d176b5
Reviewed-on: https://chromium-review.googlesource.com/571060Reviewed-by: Selim Gurun <sgurun@chromium.org>
Commit-Queue: Shimi Zhang <ctzsm@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486633}

Reland "Remove unsupported perm, symlink calls for Fuchsia."
The previous CL was landed at the same time as net_unittests,
which depended on calls that were conditionally removed from this CL.

This CL adds !OS_FUCHSIA guards to the relevant ifdef blocks in
url_request_unittest and sql/connection.cc.

Removing these functions from Fuchsia builds will make it easier to
detect and prevent any future regressions, since they'll appear as
build breakages.

R: thakis@chromium.org
Bug: 706592
Change-Id: Iad6343a3cb6fcc90b742de04bb8052e77ba498c7
Reviewed-on: https://chromium-review.googlesource.com/569845
Commit-Queue: Kevin Marshall <kmarshall@chromium.org>
Reviewed-by: Victor Costan <pwnall@chromium.org>
Reviewed-by: David Benjamin <davidben@chromium.org>
Reviewed-by: Nico Weber <thakis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486632}

This CL moves the following classes:

SurfaceSequence
SurfaceReferenceFactory
SurfaceReferenceOwner
SurfaceSequenceGenerator

to components/viz/common/surfaces and updates includes, forward declarations,
and namespaces.

TBR=junov@chromium.org, tsepez@chromium.org

Bug: 722935
Change-Id: Ic429323a71c30430754335f003ed3f0f9ab300d3
Reviewed-on: https://chromium-review.googlesource.com/570511
Commit-Queue: Fady Samuel <fsamuel@chromium.org>
Reviewed-by: Fady Samuel <fsamuel@chromium.org>
Reviewed-by: Antoine Labour <piman@chromium.org>
Reviewed-by: Scott Violet <sky@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486631}

Make a copy of the GpuPreferences inside the ContextGroup to make it
easier for all callers -- including test harnesses -- to be correct.

BUG=742440

Change-Id: I839866e24ec6ebe7b88e7012b7165ac5f9e10bc0
Reviewed-on: https://chromium-review.googlesource.com/570675Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Reviewed-by: Antoine Labour <piman@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486630}

Tests in tools/idl_parser are broken as of [1] because the current IDL
parser discards comments but tests assumed that there are comment nodes
which contain test expectations. This CL is basically a revert of [1]
with following modifications:
- Introduce SpecialComments instead of Comments
- SpecialComments are Javadoc-style comment (/** ... */) and they are
  parsed to AST nodes
- Normal comments (C/C++ style; /* ... */ and // ...) are still ignored

The reason to introduce SpecialComments: If we treat normal comments as
tokens and parse them as AST nodes, we have to add a lot of custom rules
to the parser, which will mess up the parser.

[1] https://chromium-review.googlesource.com/c/544424

BUG=740524

Change-Id: I872b6c19e6b844859bbcaa48c828decda36cb300
Reviewed-on: https://chromium-review.googlesource.com/569940Reviewed-by: Hitoshi Yoshida <peria@chromium.org>
Reviewed-by: Kentaro Hara <haraken@chromium.org>
Reviewed-by: Yuki Shiino <yukishiino@chromium.org>
Commit-Queue: Kenichi Ishibashi <bashi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486629}

Bug=742672
TBR=kochi@chromium.org
NOTRY=true

Change-Id: I11b056184b38826439b995221d9a14b12bdd922e
Reviewed-on: https://chromium-review.googlesource.com/571341Reviewed-by: Kent Tamura <tkent@chromium.org>
Commit-Queue: Kent Tamura <tkent@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486628}

- Added an entry to chromium.chromiumos.json.
- Added missing data_deps to ui_test.pak.

Bug: 737433
Test: keyboard_unittests runs on the chromiumos bots.
Change-Id: I17bdf38c9012372221580c811cf28d4b8e63258e
Reviewed-on: https://chromium-review.googlesource.com/551436
Commit-Queue: Yuichiro Hanada <yhanada@chromium.org>
Reviewed-by: Dirk Pranke <dpranke@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486627}

The scripts we run in the integration tests via the
content::ExecuteScript* APIs are run in a shared environment. So when a
script calls domAutomationController.send() more than once, the result
of the subsequent calls would leak to the next ExecuteScript call,
causing all sorts of weird behavior. This patch makes sure send() is
called exactly once in all code paths in those scripts.

Bug: 742443
Change-Id: Idcf4df408d1e0d0a5f85860c47bf85e0088e881b
Reviewed-on: https://chromium-review.googlesource.com/570600
Commit-Queue: Derek Cheng <imcheng@chromium.org>
Reviewed-by: mark a. foltz <mfoltz@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486626}

BUG=689520

Review-Url: https://codereview.chromium.org/2975153002
Cr-Commit-Position: refs/heads/master@{#486625}

Adds a "using" for a cross-platform ScopedPlatformFile that can be used regardless of system.

Converts base::File to use the new typedef. Some other users will be migrated in a followup.

Change-Id: If26a829a315cb32c9bfbbee48dd005ebb567bb38
Reviewed-on: https://chromium-review.googlesource.com/570768
Commit-Queue: Brett Wilson <brettw@chromium.org>
Reviewed-by: Daniel Cheng <dcheng@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486624}

This reverts commit 28c8676d.

Reason for revert: Linux TSanTests found a ThreadSanitizer warning that
appears to likely be associated with this change.

TBR: mattcary
Bug: 742533
Change-Id: I5a9f462a89b11ec7c4d5106ae2407505ab017960
Reviewed-on: https://chromium-review.googlesource.com/571359Reviewed-by: Quinten Yearsley <qyearsley@chromium.org>
Commit-Queue: Quinten Yearsley <qyearsley@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486623}

Those buttons will eventually be used both in a Keyboard Accessory View (KAV)
(on iPhone) and as UITextInputAssistantItems (on iPad).

The toolbar_assistive_keyboard_views.* files are in their own directory.
A follow up CL will move all the code for the KAV in that directory.

Bug: None
Change-Id: I430791e16aaab98e1dce92f311e85f600370f420
Reviewed-on: https://chromium-review.googlesource.com/567509
Commit-Queue: Jean-François Geyelin <jif@chromium.org>
Reviewed-by: Justin Cohen <justincohen@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486622}

For testing, we want to validate that OriginTrials is actually the cause
of the feature being enabled, not that a feature might be enabled by
RuntimeEnabledFeatures (early out).

RuntimeEnabledFeatures::SetOriginTrialControlledFeaturesEnabled allows
you to turn off all the features that can be turned on by OriginTrials.

BUG=685788

Change-Id: I7d18673db24b5b90d139b171e3094b66566b693c
Reviewed-on: https://chromium-review.googlesource.com/568368Reviewed-by: Ian Clelland <iclelland@chromium.org>
Reviewed-by: Jason Chase <chasej@chromium.org>
Commit-Queue: Alex Vallee <avallee@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486621}

https://chromium.googlesource.com/webm/libvpx.git/+log/4e16f7070354..b578d59623dd

$ git log 4e16f7070..b578d5962 --date=short --no-merges --format='%ad %ae %s'
2017-07-11 johannkoenig sad4d neon: 64x[32,64]
2017-07-11 johannkoenig sad4d neon: 32x[16,32,64]
2017-07-11 johannkoenig sad4d neon: 16x[8,16,32]
2017-07-11 johannkoenig sad4d neon: 8x[4,8,16]
2017-07-11 johannkoenig sad4d neon: 4x4, 4x8
2017-07-07 jzern remove vp9_firstpass.c w/CONFIG_REALTIME_ONLY
2017-07-11 urvang Remove the token state array from greedy optimize_b.
2017-07-11 jimbankoski Reintroduce fix for max qindex calculation of a gf interval
2017-07-10 marpan vp9: Fix to SVC and denoising for fixed pattern case.
2017-07-10 jianj vp9: Move skinmap computation into multithreading loop.
2017-07-10 johannkoenig avg_neon: fix 4x4, update 8x8
2017-07-07 johannkoenig neon: consolidate horizontal adds
2017-07-07 johannkoenig remove vp9_full_sad_search
2017-06-30 jianj vp9: Remove alt-ref from denoiser.

Created with:
  roll-dep src/third_party/libvpx/source/libvpx
R=johannkoenig@google.com

Bug: 
Change-Id: I858f7fabcce1491f8930f609fec6ad1d9aea001c
Reviewed-on: https://chromium-review.googlesource.com/570188Reviewed-by: Johann Koenig <johannkoenig@google.com>
Commit-Queue: Marco Paniconi <marpan@google.com>
Cr-Commit-Position: refs/heads/master@{#486620}

Android Signin Manager already has the functionality to delete all data
when a managed user signs out.

This CL reuses that functionality to delete service worker caches when
any user signs out.

BUG=739982

Review-Url: https://codereview.chromium.org/2966763003
Cr-Commit-Position: refs/heads/master@{#486619}

Bug: 695939
Change-Id: I4f3ad7e06d44db7534a8750f3a344ad27bb4f460
Reviewed-on: https://chromium-review.googlesource.com/569328Reviewed-by: Takeshi Yoshino <tyoshino@chromium.org>
Commit-Queue: Yutaka Hirano <yhirano@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486618}

BUG=709244

Review-Url: https://codereview.chromium.org/2970463002
Cr-Original-Commit-Position: refs/heads/master@{#483877}
Committed: https://chromium.googlesource.com/chromium/src/+/30815d143e71eb731bea80ba5fbd236d628ee9f3
Review-Url: https://codereview.chromium.org/2970463002
Cr-Commit-Position: refs/heads/master@{#486617}

Bug: 739389, 739384
Change-Id: I8637ce1ad693f566f738058e7fef07d025109d72
Reviewed-on: https://chromium-review.googlesource.com/563884
Commit-Queue: Bryan McQuade <bmcquade@chromium.org>
Reviewed-by: Robert Kaplow <rkaplow@chromium.org>
Reviewed-by: Charlie Harrison <csharrison@chromium.org>
Reviewed-by: Matt Falkenhagen <falken@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486616}

This CL adds queuing capability to the download service.
DownloadServiceImpl would queue all the requests such as StartDownload,
Pause/Resume/CancelDownload requests from the client as well as
OnStartScheduledTask calls from the system. After the controller is
ready, the service will execute the tasks in the same order as received.
If init failed, it will drop all the client requests, but will let the
scheduled tasks run which will be handled in the Controller
appropriately i.e. the finished callback will be invoked to notify the
OS of completion/rescheduling of the task.

This CL also ensures that the task finished callbacks will be invoked
only after the task to run is complete.

BUG=736229

Change-Id: I139ab0e0345be49c5e0351eb7d0533767ad8ff65
Reviewed-on: https://chromium-review.googlesource.com/550516
Commit-Queue: Shakti Sahu <shaktisahu@chromium.org>
Reviewed-by: David Trainor <dtrainor@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486615}

Since the backends for downloads and offline pages load asynchrously,
the DownloadHistoryAdapter waits for items from both backends to be
ready. There was a bug where this was not happening and which resulted
in download items shifting down when offline pages are added.
This CL fixes the bug.

Bug: 741201
Change-Id: I45b58179721b79adc93bf2018dae3698babe2f17
Reviewed-on: https://chromium-review.googlesource.com/571187Reviewed-by: Theresa <twellington@chromium.org>
Commit-Queue: Shakti Sahu <shaktisahu@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486614}

cc must send the computed whitelisted touch action to the browser. A
new IPC Message (|InputHostMsg_SetWhiteListedTouchAction|) has been
added to complete this communication. The reasoning for this is to
keep |InputHostMsg_SetTouchAction| as the means of communicating
effective touch actions.

Bug: 727855
Cq-Include-Trybots: master.tryserver.blink:linux_trusty_blink_rel
Change-Id: Ida9e27b4fe222474292f786ea5ab36d799ec3bad
Reviewed-on: https://chromium-review.googlesource.com/562376
Commit-Queue: Xida Chen <xidachen@chromium.org>
Reviewed-by: Xida Chen <xidachen@chromium.org>
Reviewed-by: Ken Buchanan <kenrb@chromium.org>
Reviewed-by: Antoine Labour <piman@chromium.org>
Reviewed-by: Dave Tapuska <dtapuska@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486613}

We would stop drawing ChromeVR once we entered WebVR mode. In additon to this,
we would start the timer to show the transient URL bar when we entered WebVR
mode, but the splash screen would stay visible until WebVR frames would draw.
The transient url bar timer would time out and as a result the user would see it
for a very short period of time or not at all.

This CL disables WebVR rendering until submit frame is called in cases where
we expect auto-presentation.

Bug: 740262
Change-Id: I119d3e528d403735c93a757f2be061bb8999cf43
Reviewed-on: https://chromium-review.googlesource.com/568998
Commit-Queue: Yash Malik <ymalik@chromium.org>
Reviewed-by: Michael Thiessen <mthiesse@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486612}

Migrate ExecuteCodeFunction away from BrowserThread::PostTask to post a
task to the FILE thread and instead use base::PostTask with MayBlock
traits. Also avoid a double post and instead just use PostTask*AndReply.

BUG=689520

Review-Url: https://codereview.chromium.org/2978843002
Cr-Commit-Position: refs/heads/master@{#486611}

This CL removes the usage of MessageLoop and TestBrowserThread in
ExtensionUserScriptLoaderTest, both of which should not be used directly in
tests, and replaces them with a TestBrowserThreadBundle instance. This also
gives (implicitly) ExtensionUserScriptLoaderTest a TaskScheduler instance.

BUG=272091

Change-Id: I7fc45c0d6bf30b25aa4b5029abfec013f29e4c0b
Reviewed-on: https://chromium-review.googlesource.com/571182
Commit-Queue: Karan Bhatia <karandeepb@chromium.org>
Reviewed-by: Devlin <rdevlin.cronin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486610}

This ancient API has several Chrome dependencies, which have been moved
into a FileSystemDelegate class that performs the Chrome-specific work.

I also made a delegate for apps::SavedFilesService which is only
implemented in //apps, but allows //extensions to invoke this
apps-specific code without adding a hard dependency.

Since so much of the code was shuffled around anyway, I also ran
`git cl format` on everything.

Future work includes creating an AppShellFileSystemDelegate to make
certain functionality available in app_shell.

BUG=729713
R=benwells@chromium.org,rdevlin.cronin@chromium.org

Review-Url: https://codereview.chromium.org/2934143002
Cr-Commit-Position: refs/heads/master@{#486609}

GCC gets a bit confused by libc++'s map headers, and including <map>
after <set> can cause problems under certain conditions [1].

[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81347

BUG=webrtc:7922
R=pkasting@chromium.org,dpranke@chromium.org

Review-Url: https://codereview.chromium.org/2974603002
Cr-Commit-Position: refs/heads/master@{#486608}

The previous set of changes introduced a bug where clicking
on the 'old' (current) image would discard it since that is
the desired keyboard action for that item.

Bug: 730031
Change-Id: I8cf0b715b6183cd313e63005c8e2460181bad549
Reviewed-on: https://chromium-review.googlesource.com/569022
Commit-Queue: Steven Bennetts <stevenjb@chromium.org>
Reviewed-by: Tommy Li <tommycli@chromium.org>
Cr-Commit-Position: refs/heads/master@{#486607}