GitLab

In CSSProperties.json5, to deal with type_names that are defined in
other files, we use the "field_type_path" parameter. For example, if we
wanted to use "Length" as the type_name, we can specify
"field_type_path" as "platform/Length" and the generator will
automatically include "platform/Length.h". We would also deduce that
the type_name is Length. However, this does not cater for type_names
like "Vector<String>", where multiple files need to be included.

This patch generalises "field_type_path" to "include_paths", which is
simply a list of files containing definitions for the types in
type_name. Since "include_paths" can contain multiple files, we can no
longer deduce the type_name, so we have to explicitly specify the
type_name.

BUG=628043

Review-Url: https://codereview.chromium.org/2904483003
Cr-Commit-Position: refs/heads/master@{#474928}

To avoid duplicate reports, this CL adds an ExpiringCache, keyed by
request hostname+port, to not send repeated Expect-CT reports.

As noted in a comment, it's possible that the same host+port could generate
legitimately different Expect-CT reports and it could be useful for the server
operator to receive both of them. However, this seems unlikely to happen in
practice. In a future refactor in which we move the Expect-CT reporting code
into //net, it would be easier to use more of the report contents as the cache
key (so that meaningfully different reports would not share a cache key), but
keying by host+port should suffice for now.

BUG=679012

Review-Url: https://codereview.chromium.org/2901183002
Cr-Commit-Position: refs/heads/master@{#474927}

This patch makes EImageRendering an enum class for better type safety.

BUG=684966
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Review-Url: https://codereview.chromium.org/2899633002
Cr-Commit-Position: refs/heads/master@{#474926}

This breaks most of our layout tests, since most things
depend on numbers and lengths, but it's better to be in
a single half state than have two half states hanging
around.

BUG=545318

Review-Url: https://codereview.chromium.org/2867883003
Cr-Commit-Position: refs/heads/master@{#474925}

Currently, there are a few *DataEquivalent functions in ComputedStyle
that take a pointer. All these functions assume that the pointers are
not null, so we can simply replace them with references.

BUG=628043

Review-Url: https://codereview.chromium.org/2905793002
Cr-Commit-Position: refs/heads/master@{#474924}

This patch makes RubyPosition an enum class for better type safety.

BUG=684966

Review-Url: https://codereview.chromium.org/2897883002
Cr-Commit-Position: refs/heads/master@{#474923}

Effected existing histograms:
Net.HttpContentLength.Http
Net.HttpContentLength.Https
Net.HttpContentLength.Video
Net.HttpContentLength.Http.Video
Net.HttpContentLength.Https.Video

New Histograms:

Net.HttpContentLength.Http.Direct
Net.HttpContentLength.Http.ViaDRP
Net.HttpContentLength.Http.BypassedDRP
Net.HttpContentLength.Http.Other
Net.HttpContentLength.Https.Direct
Net.HttpContentLength.Https.ViaDRP
Net.HttpContentLength.Https.BypassedDRP
Net.HttpContentLength.Https.Other

Net.HttpContentLength.Http.Direct.Video
Net.HttpContentLength.Http.ViaDRP.Video
Net.HttpContentLength.Http.BypassedDRP.Video
Net.HttpContentLength.Http.Other.Video
Net.HttpContentLength.Https.Direct.Video
Net.HttpContentLength.Https.ViaDRP.Video
Net.HttpContentLength.Https.BypassedDRP.Video
Net.HttpContentLength.Https.Other.Video

BUG=689659

Review-Url: https://codereview.chromium.org/2888653003
Cr-Commit-Position: refs/heads/master@{#474922}

These features being developed behind --enable-color-correct-rendering
and --enable-experimental-canvas-features.

We're turning on  --enable-color-correct-rendering for all layout tests,
some of which also use --enable-experimental-canvas-features, but do
not want to opt in to new canvas behaviors (at least yet).

To avoid this, use the Blink feature flag ColorCanvasExtensions. While
we're in the neighborhood, pull out the uses of the flag
--enable-color-correct-rendering-default-mode and related code.

BUG=724733

Review-Url: https://codereview.chromium.org/2893243003
Cr-Commit-Position: refs/heads/master@{#474921}

This will later be used to set proxy connection timeouts.

BUG=704339

Review-Url: https://codereview.chromium.org/2911473002
Cr-Commit-Position: refs/heads/master@{#474920}

InsertStyleSheet() returns sequence id. And RemoveInsertedStyleSheet()
removes the inserted style sheet with its id.

TEST=webkit_unit_tests --gtest_fiter=StyleEngineTest.AnalyzedInject, WebDocumentTest.InsertAndRemoveStyleSheet
BUG=608854

Review-Url: https://codereview.chromium.org/2835183002
Cr-Commit-Position: refs/heads/master@{#474919}

This test ensures that back-to-safety navigates backward for the main
frame case. This functionality was landed in crrev/474038.

BUG=684710

Change-Id: I2ff0223bdd6212d65afe6848894ce2d74cfae7ae
Reviewed-on: https://chromium-review.googlesource.com/516508Reviewed-by: Bo Liu <boliu@chromium.org>
Commit-Queue: Nate Fischer <ntfschr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#474918}

Add a timeout after user initiating voice interaction
session. All context requests can only be done with
pre-defined number of times during this time frame.

BUG=b/62065175

Review-Url: https://codereview.chromium.org/2898173006
Cr-Commit-Position: refs/heads/master@{#474917}

BUG=627945

Review-Url: https://codereview.chromium.org/2906013002
Cr-Commit-Position: refs/heads/master@{#474916}

This CL has the following changes
- Adds a showFeedback method to HelpAndFeedback which starts Feedback
  without the "Help" portion.
- FeedbackCollector can be initialized to not take screenshots (screenshot
  makes less sense for the VR feedback usecase).
- FeedbackCollector can set a CategoryTag to route the reports to a specific bucket.

BUG=722972

Review-Url: https://codereview.chromium.org/2897323002
Cr-Commit-Position: refs/heads/master@{#474915}

Adding this helper method allows us to simplify the implementation of
DocumentMarkerController::SetTextMatchMarkersActive(), which was previously
depending on details of the implementation of TextMatchMarkerListImpl (whether
or not it stores its markers in sorted order).

BUG=707867

Review-Url: https://codereview.chromium.org/2904973003
Cr-Commit-Position: refs/heads/master@{#474914}

These had been automatically disabled on Mac 10.10, 10.11 and 10.12. But
they are failing on ALL platforms, including Mac 10.9 which wasn't
explicitly mentioned. Updated to just mention "Mac" without specific
versions.

TBR=mgiuca@chromium.org

Bug: 726571
Change-Id: Ia1c33e13181a5b75f9e20f81f203c2d473c67811
Reviewed-on: https://chromium-review.googlesource.com/516863Reviewed-by: Matt Giuca <mgiuca@chromium.org>
Commit-Queue: Matt Giuca <mgiuca@chromium.org>
Cr-Commit-Position: refs/heads/master@{#474913}

After crrev.com/2721813002 tapping an insertion handle triggered a contextmenu event
which led to bug 717337. This CL fixes this bug by keeping track that the insertion
handle tap contextmenu event came from touch.

BUG=717337

Review-Url: https://codereview.chromium.org/2864833002
Cr-Commit-Position: refs/heads/master@{#474912}

Replace raw pointer with a stack variable.

BUG=chromium:690521, chromium:726579
TEST=passed trybots.

Review-Url: https://codereview.chromium.org/2902373003
Cr-Commit-Position: refs/heads/master@{#474911}

I accidentally introduced a bug in DocumentMarkerController::ShowMarkers() in
https://codereview.chromium.org/2820633002 causing it to crash if we have a node
with a DocumentMarkerList for at least one MarkerType but not all MarkerTypes
(this is very common). I hadn't noticed it until just now because this is a
method only used to assist with debugging. This CL fixes the bug.

BUG=

Review-Url: https://codereview.chromium.org/2908593003
Cr-Commit-Position: refs/heads/master@{#474910}

SelectionPaintRange has members of |selection_start_|,|selection_end_,
 |selection_start_pos_| and |selection_end_pos_| which LayoutSelection
 had directly.

BUG=708453
TEST=No change in behavior

Review-Url: https://codereview.chromium.org/2901263002
Cr-Commit-Position: refs/heads/master@{#474909}

When I moved mus_gpu_unittests into service_unittests, I didn't copy the
ozone initialization into the new test runner.

BUG=722546

Review-Url: https://codereview.chromium.org/2907663003
Cr-Commit-Position: refs/heads/master@{#474908}

This CL stores surface layer ids on LayerTreeHost which is pushed onto
LayerTreeImpl on commit and activation. These ids are used make the
compositor frame metadata instead of LayerTreeImpl::surface_layers and
so this patch also deletes LayerTreeImpl::surface_layers.

This CL is required to stop pushing hidden subtrees at commit.
(See comments in crrev.com/2846653002)

BUG=595843
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel

Review-Url: https://codereview.chromium.org/2905533002
Cr-Commit-Position: refs/heads/master@{#474907}

Allows the overlay to be any PersistentPrefStore, thereby allowing us to
use a PersistentPrefStoreClient to connect to a remote in-memory pref
store in the Mojo pref service instead of using a local one.

Bug: 719770
Change-Id: I992c416503ab6f115cbf3d98d658bc0911c4c0c7
Reviewed-on: https://chromium-review.googlesource.com/514842Reviewed-by: Sam McNally <sammc@chromium.org>
Reviewed-by: Bernhard Bauer <bauerb@chromium.org>
Commit-Queue: Johan Tibell <tibell@chromium.org>
Cr-Commit-Position: refs/heads/master@{#474906}

Enables developers to apply command-line flags on user builds of Android
by setting the Debug App setting appropriately in Developer Options.
Matches the usage in ChromeApplication.

/data/local/tmp/castshell-command-line on a public, user build of
Android (7.1.2 NHG67N on Pixel) iff ADB is enabled and DEBUG_APP is set
to "com.google.android.apps.mediashell".

Bug: internal b/62042278
Test: Verified that command-line flags were properly loaded from
Change-Id: Iabb6ed29bee8812bd8a9eb9a9815ac8f68192187
Reviewed-on: https://chromium-review.googlesource.com/516243Reviewed-by: Yaron Friedman <yfriedman@chromium.org>
Reviewed-by: Le-Chun Wu <lcwu@chromium.org>
Commit-Queue: Bryan Henry <bryanhenry@google.com>
Cr-Commit-Position: refs/heads/master@{#474905}

This patch makes TextIndentLine an enum class for better type safety.

BUG=684966

Review-Url: https://codereview.chromium.org/2905203002
Cr-Commit-Position: refs/heads/master@{#474904}

This patch
- Removed use of RefPtr::Release where the result is not used. Replaced
  this with setting the RefPtr to nullptr.
- Added WARN_UNUSED_RESULT to RefPtr::Release() to prevent it from
  being used in this way in future.

BUG=494719

Review-Url: https://codereview.chromium.org/2907543002
Cr-Commit-Position: refs/heads/master@{#474903}

This test is identical to the version available in the Web Platform
Tests project and may therefore be removed without impacting coverage.

BUG=688116
R=falken@chromium.org

Review-Url: https://codereview.chromium.org/2897313003
Cr-Commit-Position: refs/heads/master@{#474902}

Currently, the text-security property is a 'storage_only' field,
so it has no generated public getters/setters. This patch changes it to
a 'keyword' field so that its getters/setters can be generated as well.
We also generate the EWordBreak enum and use the generated
CSSValueID <-> EWordBreak mappings.

Diff of generated files:
https://gist.github.com/darrnshn/2811407420cd06be91eca37e52c974d6/revisions

BUG=628043

Review-Url: https://codereview.chromium.org/2911583002
Cr-Commit-Position: refs/heads/master@{#474901}

This patch makes TextEmphasisPosition an enum class for better type
safety.

BUG=684966
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Review-Url: https://codereview.chromium.org/2895883002
Cr-Commit-Position: refs/heads/master@{#474900}

Certain tabs API tests expect errors for passing invalid arguments.
Update the tests to accept the errors from the native bindings system.

BUG=653596
TEST=ExtensionApiTabTest.TabAudible --native-crx-bindings=1,
     ExtensionApiTabTest.TabMove --native-crx-bindings=1

Review-Url: https://codereview.chromium.org/2901403003
Cr-Commit-Position: refs/heads/master@{#474899}

This fixes a regression where we can't scroll the message center when
the mouse is over ARC notifications.

BUG=726574

Review-Url: https://codereview.chromium.org/2904933004
Cr-Commit-Position: refs/heads/master@{#474898}

CSSParserLocalContext represents local context for each property.
Currently it only has one boolean field (use_alias_parsing), but it
will be later extended to contain other info, such as whether the
property is a longhand of a shorthand.

BUG=668012

Review-Url: https://codereview.chromium.org/2901393002
Cr-Commit-Position: refs/heads/master@{#474897}

Currently the jinja template generates the properties getters/setters
in the order they appear in the group. This means changing groups will
change the location of the getter/setter making the diff generated hard
to read.
See diff here for example: https://codereview.chromium.org/2889323002

This CL sorts fields and subgroups deterministically in alphabetical
order.

Diff: https://gist.github.com/7b50d7f151021057eded58e5a405e724/revisions

BUG=710938

Review-Url: https://codereview.chromium.org/2905803002
Cr-Commit-Position: refs/heads/master@{#474896}

With browser-side navigation enabled, navigations in new foreground tabs are
considered to have started in the background. This is a change in behavior.
This causes page load metrics to incorrectly consider these page loads to have
started in the background, which skews all of Chrome's page load metrics
(PageLoad.* in UMA).

clamy explains:
"""
the background/foreground issue seems to be due to the ordering of calls in
chrome/browser/ui/browser_navigator.cc when we create a new tab.
What happens is:
- we first create a WebContents and ask it to navigate
(https://cs.chromium.org/chromium/src/chrome/browser/ui/browser_navigator.cc?type=cs&l=557).
This result in the creation of a NavigationHandle, so we fire DidStartNavigation.
In the current architecture, we fire DidStartNavigation after receiving
 DidStartProvisionalLoad.
- then we insert the WebContents in the tab strip
(https://cs.chromium.org/chromium/src/chrome/browser/ui/browser_navigator.cc?type=cs&l=589).
This fires WebContentsImpl::WasShown.
"""

This patch implements Nasko's proposed fix to expose the WebContents
CreateParams to interested WebContentsObservers, page load metrics's
MetricsWebContentsObserver being the first consumer.

We don't have the CreateParams in all places where a WebContents is passed to
TabHelpers, so we wrap the CreateParams in a base::Optional<> and provide
CreateParams in cases where they are available.

BUG=725347

Review-Url: https://codereview.chromium.org/2894973002
Cr-Commit-Position: refs/heads/master@{#474895}

Remove most of CronetTestFramework.  Cronet tests haven't passing options
on the command line for a long time so remove the command-line-args
abstraction layer.  Saves 250+ lines of code.

BUG=547160

CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.android:android_cronet_tester

Review-Url: https://codereview.chromium.org/2892013002
Cr-Commit-Position: refs/heads/master@{#474894}

This patch makes TextDecorationSkip an enum class for better type
safety.

BUG=684966
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_layout_tests_slimming_paint_v2

Review-Url: https://codereview.chromium.org/2910513003
Cr-Commit-Position: refs/heads/master@{#474893}

BUG=None

Review-Url: https://codereview.chromium.org/2901413002
Cr-Commit-Position: refs/heads/master@{#474892}

Currently, pref service connections are queued until all pref store
connections and the persistent pref store are ready. This is
unnecessarily restrictive and complicates changes necessary to support
sharing of default pref values. This CL changes the initialization of
pref connections to be more incremental.

Bug: 719770
Change-Id: Ie94f037b3ec9262ac12c048a2f31be20e0a3291c
Reviewed-on: https://chromium-review.googlesource.com/513644
Commit-Queue: Sam McNally <sammc@chromium.org>
Reviewed-by: Johan Tibell <tibell@chromium.org>
Cr-Commit-Position: refs/heads/master@{#474891}

When a user clears a chunk of history, the recency rule (previews aren't
shown for ~5 minutes after an opt out). Should be reset if the chunk of
history is longer than 5 minutes.

BUG=722436

Review-Url: https://codereview.chromium.org/2878363004
Cr-Commit-Position: refs/heads/master@{#474890}

The new login screen for multiple user pods involve creating small pods
and place them on a scroll bar.
 
The ‘small pod’ is implemented as an ‘add-on’ to the existing user pod
template, instead of being a new prototype. Pods can easily transform
itself among the ‘large’, ‘small’ and ‘extra small’ states, by toggling
the visibility of corresponding parts.
 
1) Pod placement works as follows:
Upon receiving the user list from backend, create a new pod for each user
(in rebuildPods()), append them to the pow row as children, and call
initialize(). This part remains the same with the old implementation, and
at this point there is no need to determine what size the pod should be
shown.

rebuildPods() calls placePods(). placePods() first determines the ‘main
pod’ (the one displayed as the large pod), and append all other pods to a
‘small pods container’, because they should be displayed as small pods.
This is done in appendPodsToParents().

The actual placement process starts here:
it’s done by: placeSinglePod_(), placePodsOnPodRow_(),
placePodsOnContainer_() or placePodsOnScrollableContainer_().

The placePods() function decides which one to call based on the number of
users we have. There’re lots of details in the placement process but the
general logic is simple, because we do not have to determine the best
numbers of rows and columns and can just follow the spec instead.

One special handling is: when scrollable container is shown it should
occupy the full screen, and a gradient mask is applied to avoid blocking
the header bar. The z-indexes should be well selected.
 
2) User removal works as follows:
A pod should always be switched to the main pod before being removed
(except the trivial 2-user case). Therefore after removing it from its
parent (the pod row), the main pod is set to null.

We call placePods(), and the following works exactly the same with the 1)
scenario. This is because placePods() will determine the main pod and re-
append the pods to different parents if necessary.
 
3) Window resizing works as follows:
Once there is a window resizing event, call placePods() directly. The
difference with the above two scenarios is that the main pod is not null
when window resizing happens. Therefore, placePods() will skip the main
pod selection and re-appending children process, and start the actual
placement directly.
 
4) Click event:
The only click event on small pods is to trigger the switch between the
small pod and the large pod. In the click event handler, we first find
out if it’s for a small pod, and if it is, simply switch it with the main
pod by switching the parents as well as positions / other relevant styles
(handled by changeMainPod()).

Please note:
We do not call placePods() because it may result in reordering the pods.
It’s OK that the pods are not in LRU order temporarily.

BUG=718159

Review-Url: https://codereview.chromium.org/2898283002
Cr-Commit-Position: refs/heads/master@{#474889}