GitLab

gpu code shouldn't know about surfaces/display-compositor (except for
gpu-main which is responsible for setting everything up). And surfaces
should not know about gpu code. The ws (or any other parts of mus) should
know about either gpu or surfaces.

BUG=none
TBR=sky@ for services/ui/DEPS

Review-Url: https://codereview.chromium.org/2546123002
Cr-Commit-Position: refs/heads/master@{#436165}

This patch is for code health, it only modifies the JSDocs to add
template types to fromTarget calls on the SDKModels, so the developer
does not need to explicitly cast them.

R=lushnikov
BUG=None

Review-Url: https://codereview.chromium.org/2543633009
Cr-Commit-Position: refs/heads/master@{#436164}

Cr-Commit-Position: refs/heads/master@{#436163}

Refactor Content-Range response header parsing into http_util so that
it's easier for other components to depend on it without needing to go
through HttpResponseHeaders.

BUG=670913

Review-Url: https://codereview.chromium.org/2545213002
Cr-Commit-Position: refs/heads/master@{#436162}

BUG=643746

Review-Url: https://codereview.chromium.org/2537423004
Cr-Commit-Position: refs/heads/master@{#436161}

Merge internal change: 139517817
Merge internal change: 139517817

BUG=

Review-Url: https://codereview.chromium.org/2551603002
Cr-Commit-Position: refs/heads/master@{#436160}

Ruthlessly borrowed from https://codereview.chromium.org/2547273002/ by
Jacob Dufault <jdufault@chromium.org> and simplified :). From that CL's
description:

Locally, this brings first paint time from 1000ms to 500ms.

R=jdufault@chromium.org,xiaoyinh@chromium.org
BUG=670914
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation

Review-Url: https://codereview.chromium.org/2529293015
Cr-Commit-Position: refs/heads/master@{#436159}

BUG=none

Review-Url: https://codereview.chromium.org/2550563002
Cr-Commit-Position: refs/heads/master@{#436158}

We query the autohide taskbar state when a window is maximized to ensure
that we leave just enough space at the edge so that the taskbar shows up
when the mouse is moved there.

Once we find the taskbar via the SHAppBarMessage API we also check if the
taskbar is a topmost window. While this works in most cases, I found cases
where in the taskbar loses the topmost style before the window is resized
away from fullscreen. This can presumably happen if there are other
activation changes going on at the same time, or if the machine is slow.

I think it is safe to remove the topmost style check for the taskbar. In
any case we also check the monitor where the taskbar is showing up to
ensure that they are the same. Also this code only executes when a window
is maximized. From my testing this works correctly in all cases.

Added code to register for shell fullscreen notifications via SHAppBarMessage in hwnd_message_handler.cc.
We call SendFrameChanged() if we are in maximized window and a window switched away from fullscreen

BUG=152241

Review-Url: https://codereview.chromium.org/2542533002
Cr-Commit-Position: refs/heads/master@{#436157}

BUG=669256
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation

Review-Url: https://codereview.chromium.org/2540023002
Cr-Commit-Position: refs/heads/master@{#436156}

The "Just a sec" page looks redundant if we already show
"This may take up to a minute or so..." loading page,
from users' perspective.
Then, skip to show it.

BUG=670606
BUG=b/33163416
TEST=Ran bots.
TEST=Manually run opt-in flow.
     Accept ToS, and made sure no "Just a sec" page before "SIGN IN" page.
TEST=Manually run opt-in flow.
     Accept ToS. While "SIGN IN" is shown, log-out from ChromeOS account.
     Re-login. Then "Just a sec" is shown shortly just before "SIGN IN" page.
TEST=Manually run opt-in flow.
     Accept ToS. While "SIGN IN" is shown, disable network.
     Click "SIGN IN" and show error page. Re-establish the network.
     Click "TRY AGAIN" and made sure "Just a sec" is shown shortly just before
     "SIGN IN" page is re-shown.
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation

Review-Url: https://codereview.chromium.org/2545793003
Cr-Commit-Position: refs/heads/master@{#436155}

This is a prereq to https://codereview.chromium.org/2464233002 which
will from now on reset BrowserThread globals associated with destroyed
BrowserThreadImpls and thus makes DCHECK_CURRENTLY_ON fail after it
was.

BUG=653916

Review-Url: https://codereview.chromium.org/2549623006
Cr-Commit-Position: refs/heads/master@{#436154}

[DevTools] Mark inspector/sources/debugger-pause/debugger-pause-in-internal.html as NeedsManualRebaseline.

BUG=109362

Review-Url: https://codereview.chromium.org/2548093002
Cr-Commit-Position: refs/heads/master@{#436153}

BUG=666732

Review-Url: https://codereview.chromium.org/2547943003
Cr-Commit-Position: refs/heads/master@{#436152}

Out-of-flow descendants need to know what their static position would
have been. This is because abspos spec defaults to static position if
TLBL are not specified.

Oof descendant's dimensions are not known during propagation, only
it's logical top/left offset. The offset is propagated as NGCorner,
which is a PhysicalOffset + vertex location (topLeft, bottomLeft, etc)

Next step will be positioning of oof descendants.

BUG=635619

Review-Url: https://codereview.chromium.org/2540653003
Cr-Commit-Position: refs/heads/master@{#436151}

The tasks aren't getting picked up because it defaults to 6 and the
swarming bots only have 1 available. (No clue why; I'll look at that later.)
This should at least get them running, albeit really slowly.

BUG=534815, 664190

Review-Url: https://codereview.chromium.org/2544223003
Cr-Commit-Position: refs/heads/master@{#436150}

When computing min-content and max-content of the content-sized tracks
we are using the marginIntrinsicLogicalWidthForChild function, which
uses the grid's writing-mode to determine wether to use the child's
margin width or height. This is not correct when the grid item is
orthogonal.

This patch changes how we compute the tracks width so we use always
the item's marginLogicalWidth, which depends only on its own writing
mode.

BUG=670642

Review-Url: https://codereview.chromium.org/2543193002
Cr-Commit-Position: refs/heads/master@{#436149}

We would like to have all Chrome clients use a single service-side
CompositorFrameSink implementation as an incremental step towards pulling the
display compositor into the gpu process. This CL extract non-mojo-specific
code from GpuCompositorFrameSink in services/ui to CompositorFrameSinkSupport
in a central place: cc/surfaces that can be used by Chrome today. Individual
clients can transition to mojo after we carefully measure the performance
impact of the transition.

BUG=668126, 657959
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel

Review-Url: https://codereview.chromium.org/2543373002
Cr-Commit-Position: refs/heads/master@{#436148}

Fetch requires [1] that "the value pairs to iterate over are the
return value of running sort and combine with the header list", which
is to say we don't simply iterate over the internal data of Headers
but make a sorted (and combined) copy [2]. This manifested in some
failing imported cache tests that were sensitive to the order.

This CL implements making a copy and sorting the list for iteration;
it does not implement combining[3]. So not perfect but an incremental
improvement.

[1] https://fetch.spec.whatwg.org/#headers-class
[2] https://fetch.spec.whatwg.org/#concept-header-list-sort-and-combine
[3] https://fetch.spec.whatwg.org/#concept-header-value-combined

BUG=667376,670439
R=yhirano@chromium.org

Review-Url: https://codereview.chromium.org/2542073005
Cr-Commit-Position: refs/heads/master@{#436147}

When the tab's RenderWidgetHostView is nullptr, we cannot and should not select any
RenderWidgetHosts for a drag destination. Therefore, in response to draggingEntered
we should return no operation.

This CL will fix that issue as well as making RenderWidgetHostInputEventRouter::GetRenderWidgetHostAtPoint()
handle nullptr |root_view| by returning nullptr.

BUG=670645

Review-Url: https://codereview.chromium.org/2547213002
Cr-Commit-Position: refs/heads/master@{#436146}

BUG=653350

Review-Url: https://codereview.chromium.org/2543403002
Cr-Commit-Position: refs/heads/master@{#436145}

- Also fixed source_buffer_state failed to find codec with vp9 in mp4 contents when VP9 in MP4 is enabled.

- And added a few playback/demuxer tests.

BUG=580623

Review-Url: https://codereview.chromium.org/2495193004
Cr-Commit-Position: refs/heads/master@{#436144}

- Adds ui.mojom.CompositionUnderline and ui.mojom.CompositionText.
- Adds corresponding struct traits + unittest for struct traits.
- Breaks ui.mojom.TextInputClient::OnCompositionEvent() into multiple
  functions, each corresponding to a function in ui::TextInputClient.

BUG=631524

Review-Url: https://codereview.chromium.org/2539453002
Cr-Commit-Position: refs/heads/master@{#436143}

BUG=666732

Review-Url: https://codereview.chromium.org/2545143003
Cr-Commit-Position: refs/heads/master@{#436142}

BUG=664308

Review-Url: https://codereview.chromium.org/2549863003
Cr-Commit-Position: refs/heads/master@{#436141}

BUG=

Review-Url: https://codereview.chromium.org/2524993003
Cr-Commit-Position: refs/heads/master@{#436140}

BUG=670397

Review-Url: https://codereview.chromium.org/2550883002
Cr-Commit-Position: refs/heads/master@{#436139}

BUG=666732

Review-Url: https://codereview.chromium.org/2544413002
Cr-Commit-Position: refs/heads/master@{#436138}

The LazySyncableAdapter is used to wrap LazySyncables of any changeset type into a LazySyncable with a string changeset by serializing the changesets into a string and parsing them from strings. That way, ObjectSyncState can just work with LazySyncables of type string.

BUG=663065

Review-Url: https://codereview.chromium.org/2522633004
Cr-Commit-Position: refs/heads/master@{#436137}

UnsafeResource is used in creating/showing SB interstitial, thus this is a preparation step of componentizing SB interstitial.

BUG=666100

Review-Url: https://codereview.chromium.org/2540563002
Cr-Commit-Position: refs/heads/master@{#436136}

Supported profiles will not be used. Support is tested by trying to
initialize the decoder.

BUG=660942
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel

Review-Url: https://codereview.chromium.org/2548923002
Cr-Commit-Position: refs/heads/master@{#436135}

Mappings need to stay the same across frames so textures aren't
unnecessarily reallocated, but they should be deleted once they're not
used to avoid leaking memory.
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel

Review-Url: https://codereview.chromium.org/2544203003
Cr-Commit-Position: refs/heads/master@{#436134}

Some types of phones (such as Google Pixel) report ClassCastException
error when it populates data for SearchEnginePreference.

The reason of this bug is because SearchEnginePreference manually sets
LayoutParams for the ListView in PreferenceFragment and some types of
phones (Nexus 6P) use LinearyLayout to hold the ListView while other
types of phones (Google Pixel) use FrameLayout to hold it.

To solve this problem, get LayoutParams of ListView and cast it to
MarginLayoutParams before setting margin values of it. Because both
LinearLayout.LayoutParams and FrameLayout.LayoutParms inherit from
MarginLayoutParams.

BUG=670297

Review-Url: https://codereview.chromium.org/2543023004
Cr-Commit-Position: refs/heads/master@{#436133}

With this patch new requestPresent calls that are made while the first one is
still pending will simply be queued up Blink-side and all resolved at once when
the original request succeeds or fails.

BUG=670441

Review-Url: https://codereview.chromium.org/2550863002
Cr-Commit-Position: refs/heads/master@{#436132}

This is dead code.

BUG=none
TBR=tsepez@chromium.org

Review-Url: https://codereview.chromium.org/2550833002
Cr-Commit-Position: refs/heads/master@{#436131}

BUG=666602
R=lushnikov@chromium.org

Review-Url: https://codereview.chromium.org/2544383002
Cr-Commit-Position: refs/heads/master@{#436130}

Before moving DisplayChangeObserver from src/ash/display to
src/ui/display we need to remove all ash dependencies.

1. DisplayManager and DisplayConfigurator are injected instead of
   getting them from ash::Shell.
2. DCO no longer acts as a ShellObserver for shutdown. Add
   ShutdownObserver to do this instead.
3. Move display strings from ash to ui.

BUG=667344

Review-Url: https://codereview.chromium.org/2523873002
Cr-Commit-Position: refs/heads/master@{#436129}

PerformanceMonitor: do not maintain a vector of frames in the monitor, single ExecutionContext member should suffice.

Review-Url: https://codereview.chromium.org/2539323002
Cr-Commit-Position: refs/heads/master@{#436128}

BUG=666732

Review-Url: https://codereview.chromium.org/2544153004
Cr-Commit-Position: refs/heads/master@{#436127}

These were added to the code in
https://codereview.chromium.org/1855423003

BUG=

Review-Url: https://codereview.chromium.org/2544883004
Cr-Commit-Position: refs/heads/master@{#436126}