GitLab

It traversed the ancestors to throw a INVALID_NODE_TYPE_ERR if one of them was a DocumentType Node. However, it is impossible for a DocumentType Node to have children.

Picked from webkit https://trac.webkit.org/changeset/190145

No new tests, covered by existing tests.

BUG=None

Review URL: https://codereview.chromium.org/1363833002

Cr-Commit-Position: refs/heads/master@{#351046}

This patch makes |makeSearchRange()| to use |EphemeralRange| instead of
|Range| object to avoid registering temporary |Range| object into |Document|
for Oilpan friendly and as a preparation of introducing full version of
|VisibleSelectionTemplate|.

This patch is a preparation of making selection to handle granularity for web
component, http://crrev.com/1363873004

BUG=388681, 513568
TEST=n/a; no behavior changes

Review URL: https://codereview.chromium.org/1375453002

Cr-Commit-Position: refs/heads/master@{#351045}

To handle workloads with high allocation rates, where the Oilpan heap
allocations only make up a small fraction compared to what's allocated
by either PartitionAlloc or V8, use V8's major GC notification as
a signal to check for the need to perform an "emergency" memory
pressure conservative GC. Otherwise there might not be enough Oilpan
allocations made to trigger out-of-line heap allocations which trigger
that same check. These GCs and workloads are rare, but reduces the
possibility of running into OOM conditions when Oilpan is faced with
those.

An example where it does show up is Dromaeo's dom-modify.html, which
has a subtest that heavily allocates Text nodes, each holding a longer
string. As each Text node is a small Oilpan heap object, not a lot of heap is
needed for the nodes, but PartitionAlloc's buffer partition size grows much
more sharply and risks signalling OOM. V8's RTS notices the memory pressure
and triggers extra GCs; reuse that signal in Oilpan.

Hence, bring back the check that https://codereview.chromium.org/1190513006/
added for checking this condition.

R=haraken
BUG=474470

Review URL: https://codereview.chromium.org/1369783002

Cr-Commit-Position: refs/heads/master@{#351044}

This causes flake during local test runs (for those who use braille when
developing) and since bots don't generally have braille displays
connected, autostarting in tests doesn't add any value.

R=dmazzoni@chromium.org
BUG=None

Review URL: https://codereview.chromium.org/1354143005

Cr-Commit-Position: refs/heads/master@{#351043}

Revert of Move status bar color setting to ChromeActivity. (patchset #5 id:70001 of https://codereview.chromium.org/1364043003/ )

Reason for revert:
Broke CustomTabActivityTest#testToolbarColor, see crbug.com/536698 for the tracking bug, and crrev.com/1370003002 for a potential fix.

Original issue's description:
> Move status bar color setting to ChromeActivity.
>
> BUG=535235
>
> Committed: https://crrev.com/4b74179f72ea6c4f0b1539a8c31277e8d29f7711
> Cr-Commit-Position: refs/heads/master@{#350961}

TBR=yusufo@chromium.org,tedchoc@chromium.org,aurimas@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=535235

Review URL: https://codereview.chromium.org/1372883002

Cr-Commit-Position: refs/heads/master@{#351042}

BUG=535460, 534993

Review URL: https://codereview.chromium.org/1371853003

Cr-Commit-Position: refs/heads/master@{#351041}

This CL makes the WebSocket server in net/server use the net/websockets
parser for parsing Sec-WebSocket-Extensions in the extension negotiation.

The new implementation validates the extension negotiation offer more
strictly than before. Specifically,

- Malformed Sec-WebSocket-Extensions header value causes connection failure.
 - Previously it was just ignored.
- Malformed permessage-deflate parameters are declined.
 - Previously part of such params were accepted partially.

BUG=523228

Review URL: https://codereview.chromium.org/1340523002

Cr-Commit-Position: refs/heads/master@{#351040}

This patch utilizes |VisibleSelectionTemplate| template class in
|SelectionController| class as replacement of |VisibleSelection::InDOMTree| and
|VisibleSelection::InComposedTree| for a preparation of full version of
|VisibleSelectionTemplate|.

This patch is a preparation of making selection to handle granularity for web
component, http://crrev.com/1363873004

BUG=513568
TEST=n/a; no behavior changes

Review URL: https://codereview.chromium.org/1371003002

Cr-Commit-Position: refs/heads/master@{#351039}

correcting my mistake from last july

R=dbeam@chromium.org

Review URL: https://codereview.chromium.org/1372783002

Cr-Commit-Position: refs/heads/master@{#351038}

The Bluetooth address of a connected device changes when the device
is paired while connected.

BUG=533449

Review URL: https://codereview.chromium.org/1351623006

Cr-Commit-Position: refs/heads/master@{#351037}

The iOS port needs to turn down its usage of ChromeStabilityMetricsProvider in
favor of using its own stability metrics provider. To minimize code duplication
between the two providers, this CL extracts the code that will be common
between the two into a shared helper class.

As a necessary dependence, this CL also componentizes
system_memory_stats_recorder*. Those files can be componentized as-is.

BUG=512422

Review URL: https://codereview.chromium.org/1366643002

Cr-Commit-Position: refs/heads/master@{#351036}

Passing std::string by reference can prevent extra copying of object.

BUG=367418
TEST=
R=rogerta@chromium.org,sdefresne@chromium.org

Review URL: https://codereview.chromium.org/1354373002

Cr-Commit-Position: refs/heads/master@{#351035}

Closing the local streams seems to solve the buffer thrashing that's
going on while gathering frames; this should stop the test from
timing out on Windows.

BUG=532918,webrtc:4986

Review URL: https://codereview.chromium.org/1367453002

Cr-Commit-Position: refs/heads/master@{#351034}

Chrome doesn't serialize string as is stated in the spec here:
https://drafts.csswg.org/cssom/#serialize-a-url. Each string passed to
a URL is supposed to be in quotes. As opposed to what we do currently.
Gecko on Firefox does this right.

BUG=534674

Review URL: https://codereview.chromium.org/1363233003

Cr-Commit-Position: refs/heads/master@{#351033}

BUG=535367
TEST=it all stays working

Review URL: https://codereview.chromium.org/1361223003

Cr-Commit-Position: refs/heads/master@{#351032}

Among a number of minor changes to padding and alignmnent, we also switch
viewer-zoom-toolbar to use CSS transitions rather than neon-animation, as
it's much easier to make RTL compatible and more concise code. We also
switch from number input in the page selector, as it does not have
suitable localization support.

Disabling presubmit since it doesn't like paper-tooltip

BUG=506035
NOPRESUBMIT=true

Review URL: https://codereview.chromium.org/1364163002

Cr-Commit-Position: refs/heads/master@{#351031}

BUG=500428

Review URL: https://codereview.chromium.org/1365733006

Cr-Commit-Position: refs/heads/master@{#351030}

If the item range being removed has no extent, no item is removed
from the vector. Return early, if so.

Style invalidation uses this Vector remove() operation upon restoring
state (RecursionCheckpoint), with the length being zero often.
Optimize its handling.

The performance of blink_perf.css.ClassInvalidation and
blink_perf.css.ClassDescendantSelector is improved 15-20% as a result
(without Oilpan, windows + linux64.)

R=haraken
BUG=

Review URL: https://codereview.chromium.org/1370933002

Cr-Commit-Position: refs/heads/master@{#351029}

This issue was caused because overaly canvas size and main canvas size
were different. Main canvas can show image with scaling.

BUG=535854
TEST=manually tested as described in issue.

Review URL: https://codereview.chromium.org/1370683003

Cr-Commit-Position: refs/heads/master@{#351028}

Revert of Upgrading binutils to 2.25 for -ggnu-pubnames (patchset #2 id:20001 of https://codereview.chromium.org/1368233002/ )

Reason for revert:
binutils is not deployed in goma backend, and this roll causes slow compile on linux

BUG=536663

Original issue's description:
> Upgrading binutils to 2.25 for -ggnu-pubnames
>
> To make static_library builds fast we need binutils 2.25 to enable
> -ggnu-pubnames support.
>
> The ehframe-race.patch is removed because it is now included upstream. Dates
> and expected release versions have been added for the remaining patches to make
> it easier to upgrade in the future.
>
> (Also use a precise chroot rather than lucid for building binutils.)
>
> BUG=374952
>
> Committed: https://crrev.com/1a4e8c12aa264a54432b38fa009b0aa4814caea1
> Cr-Commit-Position: refs/heads/master@{#351019}

TBR=thestig@chromium.org,thakis@chromium.org,mithro@mithis.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=374952

Review URL: https://codereview.chromium.org/1374563002

Cr-Commit-Position: refs/heads/master@{#351027}

Original CL:
> Files.app: add create directory operation to context menu of directory tree.
>
> Create directory operation is consisted with two operations,
> 1. Create new directory with default name.
> 2. Start rename operation of it.
>
> Since rename operation itself is already implemented with different CL,
> this CL focuses on to create new directory with default name and update
> the UI.
>
> BUG=502171
> TEST=browser_tests:DirectoryTreeContextMenu/FileManagerBrowserTest.Test/*
>
> Committed: https://crrev.com/57bd0537cd33f15039a9c6d929ce27f2515f8149
> Cr-Commit-Position: refs/heads/master@{#350522}

patch from issue 1353753002 at patchset 40001 (http://crrev.com/1353753002#ps40001)

Additional changes to the original CL:
* Waits an element before starting to input text in test case.

BUG=502171
TEST=run flaky test case 100 times; browser_tests:DirectoryTreeContextMenu/FileManagerBrowserTest.Test/18

Review URL: https://codereview.chromium.org/1373733002

Cr-Commit-Position: refs/heads/master@{#351026}

StyleResolver::computeFont does not rely on the computed style of any
elements in the document tree. The updateLayoutTree call was there to
avoid an assert due to an unnecessary composed tree traversal for a
parent node. That traversal was removed by [1], so the updateLayoutTree
call should no longer be needed.

This change gets rid of synchronous style recalcs when FontFaceSet.load()
is called.

[1] https://codereview.chromium.org/1367743002/

BUG=536625

Review URL: https://codereview.chromium.org/1372823002

Cr-Commit-Position: refs/heads/master@{#351025}

Revert of Replace RawPtr with RefPtr on StylePendingImage (patchset #2 id:20001 of https://codereview.chromium.org/1368613002/ )

Reason for revert:
Causing crashes in Canary.

BUG=536655

Original issue's description:
> Replace RawPtr with RefPtr on StylePendingImage
>
> This change ensures StylePendingImages set on ComputedStyles will keep
> their corresponding CSSValue alive.
> The PendingImagePropertyMap on ElementStyleResources was supposed to do
> this however it only keeps alive one per property while properties can
> have any number of images set on them (background-image).
>
> Prior to this change StylePendingImages held a RawPtr to their CSSValue
> to avoid a ref loop because the CSSValue would hold a RefPtr to the
> StylePendingImage as a cache.
>
> After this change StylePendingImages hold RefPtrs to their CSSValue
> while CSSValues no longer cache StylePendingImages. Having multiple
> StylePendingImages per CSSValue is equivalent to using cached
> StylePendingImages as no state changes on the StylePendingImage
> during the StyleResourceLoader pass.
>
> The old scenario is not currently a problem as all applied CSSValues
> have their refs held elsewhere (style rules or AnimatableImages).
> Future animation work intends to apply temporary CSSValue images and
> must ensure StylePendingImages keep a ref to avoid a use after free.
>
> This patch is a refactor towards memory management correctness and has
> no behavioural changes.
>
> BUG=437696
>
> Committed: https://crrev.com/1dd18b7ea8db7d3344024277e04c8a5365c03cf1
> Cr-Commit-Position: refs/heads/master@{#350808}

TBR=timloh@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=437696

Review URL: https://codereview.chromium.org/1369213002

Cr-Commit-Position: refs/heads/master@{#351024}

Passing std::string by reference can prevent extra copying of object.

BUG=367418
TEST=
R=blundell@chromium.org,jochen@chromium.org,sdefresne@chromium.org

Review URL: https://codereview.chromium.org/1355983002

Cr-Commit-Position: refs/heads/master@{#351023}

After linking to the 10.10 SDK, the text in the find bar (Cmd+F) loses
its subpixel AA when the textfield doesn't have focus.

https://developer.apple.com/library/prerelease/mac/releasenotes/AppKit/RN-AppKitOlderNotes/#10_8Layers
documents new logic for applications linked on 10.8 and higher.
Specifically, NSTextField now allows LCD font smoothing when the view is
layer-backed, but this will be disabled if AppKit can't find an opaque
ancestor view.

The findbar textfield cell isn't opaque (it has rounded corners), but
the background behind the text is always solid (even when a theme is
installed), so enabling AA is safe. In fact, this is true for all
StyledTextFieldCells, of which there are 2 (findbar and omnibox).

Share the "force-enable-AA" logic already in BlueLabelButton and
AutocompleteTextFieldCell that does this by adding
ui::ScopedCGContextSmoothFonts. Note AutocompleteTextFieldCell overrides
-[NSCell drawWithFrame:inView:] so doesn't inherit the fix from
StyledTextFieldCell.

BUG=535790
TBR=avi@chromium.org
TEST=On Mac (without a retina screen), press Cmd+F, type something, then
click the page to defocus. The text in the find bar should keep its font
smoothing.

Review URL: https://codereview.chromium.org/1368883002

Cr-Commit-Position: refs/heads/master@{#351022}

Instead of poking at the DOM, there is a gentler and more accurate
way to determine whether the document has a pending location
change or a redirect. Let's use it!

BUG=521166
R=jam

Review URL: https://codereview.chromium.org/1363803003

Cr-Commit-Position: refs/heads/master@{#351021}

This CL adds "issuedNonGetFetchFromScript" flag to WebPageImportanceSignals.

From this signal, we assume that the page state is not simply discardable,
as it has issued some dynamic resource request with side-effects.

BUG=520838

Review URL: https://codereview.chromium.org/1303833002

Cr-Commit-Position: refs/heads/master@{#351020}

To make static_library builds fast we need binutils 2.25 to enable
-ggnu-pubnames support.

The ehframe-race.patch is removed because it is now included upstream. Dates
and expected release versions have been added for the remaining patches to make
it easier to upgrade in the future.

(Also use a precise chroot rather than lucid for building binutils.)

BUG=374952

Review URL: https://codereview.chromium.org/1368233002

Cr-Commit-Position: refs/heads/master@{#351019}

In OOPIF based <webview>, BrowserPluginGuest isn't used much and the
plan is to remove its usage in --site-per-process. However, the
Web Speech API relies on BrowserPluginGuest to return its embedder
WebContents. I've made it so that we ask for the WebContentsImpl to
return its embedder or outer WebContents instead of going through
BrowserPluginGuest.

TBR=jam@chromium.org for chromium.fyi.json
BUG=533647
Test=Try WebSpeech API from a Chrome App <webview>, with --site-per-process.
This should work and not crash the browser/ process.

Review URL: https://codereview.chromium.org/1351643003

Cr-Commit-Position: refs/heads/master@{#351018}

The behaviour of dom-if changed in Polymer 1.0 to make it useful for
lazily stamping complex DOM subtrees. This CL adds dom-if to
viewer-bookmark, so that child bookmarks are only created when they are
made visible. On my workstation, this cuts the time to initially render
a complex bookmarks menu from 6 seconds to 0.2 seconds.

Also, make use of reflectToAttribute to remove some single-use CSS
classes.

BUG=110020

Review URL: https://codereview.chromium.org/1369163002

Cr-Commit-Position: refs/heads/master@{#351017}

Cr-Commit-Position: refs/heads/master@{#351016}

This should be a NOP from a build setup perspective, but aims to make the large common shared build configuration files easier to understand.

Review URL: https://codereview.chromium.org/1368263002

Cr-Commit-Position: refs/heads/master@{#351015}

This is split off from https://codereview.chromium.org/1350023003/ so that it can have another EDK in src/mojo/edk without conflicts. Also update header include guards as well so they don't clash with the ones in src/mojo/edk.

BUG=478251

Review URL: https://codereview.chromium.org/1367303002

Cr-Commit-Position: refs/heads/master@{#351014}

This flag is set by calling |Channel::WillShutdownSoon()|, but that may
not be called in all cases (i.e. on broken connection due to renderer
crash). crrev.com/95049185 relies on
this flag being set on shutdown to prevent a race with
|Channel::AttachAndRunEdpoint()|.

Review URL: https://codereview.chromium.org/1367383002

Cr-Commit-Position: refs/heads/master@{#351013}

Currently composited UI frames during a live-resize arrive out of sync,
and get queued up, resulting in wasted work and odd behavior where the
window contents lag behind the window size.

To fix, use ui::WindowResizeHelperMac to watch messages coming in from
the GPU process and block the UI thread until it's ready to display a
composited frame that matches the window.

For now, just fix resize, which hooks in via -[NSWindowDelegate
windowDidResize:]. In future, we may want to block for other cases (E.g.
showing a window-modal sheet should wait before starting the animation.)

BUG=531831

Review URL: https://codereview.chromium.org/1352743002

Cr-Commit-Position: refs/heads/master@{#351012}

Review URL: https://codereview.chromium.org/1367713002

Cr-Commit-Position: refs/heads/master@{#351011}

This is a follow-up fix to https://codereview.chromium.org/1306733003 with some
opportunistic refactoring to remove duplicate code.

When tying floats to the (current) last line-box in block-flow we may hit a float
in lines previously considered clean but which we have since decided to continue
laying out past (e.g.because one of our floats has changed size). If that happens
we need to add the floats to the linebox rather than bailing early. In
https://codereview.chromium.org/1306733003 we weren't catering for the possibility
that we may have given up on the plan to stop at clean lines.

In the refactoring I got rid of a superfluous bool and refactored some common code
into a function. Now, if we give up on staying within clean lines we immediately
delete them rather than relying on layout state to figure out if we need to do it
later.

lgtm'd pre-merge at https://codereview.chromium.org/1355003002

TBR=eae@chromium.org

BUG=532712

Review URL: https://codereview.chromium.org/1367363002

Cr-Commit-Position: refs/heads/master@{#351010}

Passing std::string by reference can prevent extra copying of object.

BUG=367418
TEST=
R=jochen@chromium.org,sky@chromium.org,thakis@chromium.org

Review URL: https://codereview.chromium.org/1358783002

Cr-Commit-Position: refs/heads/master@{#351009}

Summary of changes available at:
https://chromium.googlesource.com/skia/+log/97be9a0..17e3806

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

Commits in this roll:
17e3806 update-skps@skia.org Update SKP version

Review URL: https://codereview.chromium.org/1369943002

Cr-Commit-Position: refs/heads/master@{#351008}

Passing std::string by reference can prevent extra copying of object.

BUG=367418
TEST=
R=jam@chromium.org,jochen@chromium.org,kinuko@chromium.org

Review URL: https://codereview.chromium.org/1354363002

Cr-Commit-Position: refs/heads/master@{#351007}