GitLab

Summary of changes available at:
https://chromium.googlesource.com/chromium/blink/+log/97a4012..e49d160

TBR=dcheng@chromium.org,johnme@chromium.org

Review URL: https://codereview.chromium.org/1049983004

Cr-Commit-Position: refs/heads/master@{#325546}

This is because GCC also uses the -B search path at link time to find
"as", and our bundled "as" can only target x86. See also the discussion at
https://codereview.chromium.org/1084133002/

(I also tried adding this condition around line 858, but this won't work
because of scoping issues ("clang" is not defined here)).

R=thakis@chromium.org,fdegans@chromium.org
BUG=469376

Review URL: https://codereview.chromium.org/1096553002

Cr-Commit-Position: refs/heads/master@{#325545}

Previously if a target depended on a source set and that source set had
data deps, building the target would not force the data dep target to be
built.

This patch hooks up the dependencies such that this happens as expected.

BUG=477104

Review URL: https://codereview.chromium.org/1095673002

Cr-Commit-Position: refs/heads/master@{#325544}

Lean on ICU for month names and abbreviations rather than hardcoding
English ones.

BUG=none

Review URL: https://codereview.chromium.org/1080883002

Cr-Commit-Position: refs/heads/master@{#325543}

Original cl
https://codereview.chromium.org/1035983003/

In order for ChromeVox to queue up utterances, we can separate each attribute or rule into its own spannable. Each spannable is then sent to tts (queueing as needed).

BUG=477492
TBR=dtseng@chromium.org

Review URL: https://codereview.chromium.org/1094463003

Cr-Commit-Position: refs/heads/master@{#325542}

This matches the gn name and is less confusing with use_openssl and
use_openssl_certs both existing. The preprocessor flag will follow
in another CL. For now, the old name is still defined, but not used
within Chromium.

A follow-up CL will remove deprecated use_nss and USE_NSS #defines
together which will then cause downstream churn. Though from a grep
of known downstreams, the churn seems to be fairly minimal. The
removal is split from this CL so that, if we need to revert, the
CL to revert is small.

BUG=462040
TBR=wez@chromium.org

Review URL: https://codereview.chromium.org/1085923002

Cr-Commit-Position: refs/heads/master@{#325541}

TBR=phajdan.jr@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1095743002

Cr-Commit-Position: refs/heads/master@{#325540}

I can't reproduce a MemoryError locally or on the try bots, and the original bug only shows MemoryErrors on the top_25 pageset.

Successfult tryjobs:
https://codereview.chromium.org/1086243002/

BUG=463111

Review URL: https://codereview.chromium.org/1097573002

Cr-Commit-Position: refs/heads/master@{#325539}

Currently fixed-size views dialogs on Mac don't initialize properly.
(And, e.g., WebDialogBrowserTest.SizeWindow fails because of it).
Desktop Linux has a similar issue.

On desktop Linux, NativeWidgetPrivate::OnRootViewLayout triggers a call
to UpdateMinAndMaxSize. OnRootViewLayout happens during Widget::Init()
but the X11 window is not yet mapped, so this was resulting in a no-op.
For Linux, fix by explicitly calling UpdateMinAndMaxSize after the
window is mapped.

Mac needs to do a similar thing during Init. However, Mac was calling
OnSizeConstraintsChanged() before Widget::Init() had set the non-client
view, so Widget::GetMinimumSize() returned invalid sizes. For Mac, move
the OnSizeConstraintsChanged() call to OnRootViewLayout(), similar to
Linux.

Then, add a cross-platform test. Mac was not clamping to the size
constraints in SetBounds(), so do that.

For the test itself, collapse the testing WidgetDelegates in
widget_unittest.cc into one helper class rather than adding yet another
WidgetDelegate. Then, provide a way to query the OS for the minimum
window size to fill the gaps required for testing the propagation of
size constraints to the window server.

BUG=454698, 447086

Review URL: https://codereview.chromium.org/1059853007

Cr-Commit-Position: refs/heads/master@{#325538}

BUG=475715
BUG=477060

Review URL: https://codereview.chromium.org/1061063005

Cr-Commit-Position: refs/heads/master@{#325537}

Revert of Send resize event when the screen info changes. (patchset #5 id:80001 of https://codereview.chromium.org/1078123002/)

Reason for revert:
Suspected as cause of flaky failures for
WebViewInteractiveTest.PopupPositioningBasic  WebViewInteractiveTest.PopupPositioningMoved.  For details, see https://code.google.com/p/chromium/issues/detail?id=477783

Original issue's description:
> Send resize event when the screen info changes.
>
> This is a regression from https://codereview.chromium.org/953233002
>
> BUG=475939
>
> Committed: https://crrev.com/6e3582ad1a058af30aa44ca7fae2113b050aa437
> Cr-Commit-Position: refs/heads/master@{#325494}

TBR=danakj@chromium.org,jochen@chromium.org,piman@chromium.org,sadrul@chromium.org,sievers@chromium.org,mlamouri@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=475939

Review URL: https://codereview.chromium.org/1093713002

Cr-Commit-Position: refs/heads/master@{#325536}

to ensure that the callback function is set before ConnectChannel()
is called. This prevents a potential race condition that could occur
if messages were pending when ipc_adapter->ConnectChannel() is called.

TEST=git cl try, ARC still works, PNaCl demo still works
BUG=nativeclient:3802

Review URL: https://codereview.chromium.org/1090043002

Cr-Commit-Position: refs/heads/master@{#325535}

TBR=vabr@chromium.org
BUG=477812

Review URL: https://codereview.chromium.org/1096483003

Cr-Commit-Position: refs/heads/master@{#325534}

As of https://codereview.chromium.org/1008873006 it no longer does
anything.

BUG=426442,318217

Review URL: https://codereview.chromium.org/1045593002

Cr-Commit-Position: refs/heads/master@{#325533}

TBR=phajdan.jr@chromium.org
BUG=454413

Review URL: https://codereview.chromium.org/1062823006

Cr-Commit-Position: refs/heads/master@{#325532}

Convert the BadHTTPS and URLFetcher reuse tests to use the shared test
fixture.  As a bonus, make sure the reuse test actually updates the
response body between requests.

BUG=471069

Review URL: https://codereview.chromium.org/1091073003

Cr-Commit-Position: refs/heads/master@{#325531}

This allows an overloaded version of the function to be added to Blink.

BUG=460945

Review URL: https://codereview.chromium.org/1090863002

Cr-Commit-Position: refs/heads/master@{#325530}

unrecoverable_error_set and one of the InitialSyncEndedForType overloads
are only called by Directory so they can be made private.

Remove name accessor as it's never called.

BUG=475557

Review URL: https://codereview.chromium.org/1096543002

Cr-Commit-Position: refs/heads/master@{#325529}

- SessionRestore now exposes IsLoadingActiveTabsOnly
- Fixed 2 browser tests that were failing with new behavior
- One last telemetry test still failing, will be addressed in a separate CL

Notes:
- First patchset inverts the default behavior to show tests succeeding
- Second patchset restores default behavior

BUG=

Review URL: https://codereview.chromium.org/1087063003

Cr-Commit-Position: refs/heads/master@{#325528}

The hacks are:  HTTP/0.9 support, allowing bonus characters before
HTTP responses, and accepting a truncated set of HTTP headers.

Also rename the confusingly named "ParseResponseHeaders" and
"DoParseResponseHeaders" methods.

BUG=472762

Review URL: https://codereview.chromium.org/1074263003

Cr-Commit-Position: refs/heads/master@{#325527}

Revert of Created new URLRequestContext for secure proxy check (patchset #20 id:380001 of https://codereview.chromium.org/981633002/)

Reason for revert:
This is breaking all the tests again:
http://build.chromium.org/p/chromium.linux/builders/Android%20Tests%20%28dbg%29/builds/27288/

Original issue's description:
> Created new URLRequestContext for secure proxy check.
>
> The new URLRequestContext disables alternate protocols
> to ensure that the Data Reduction Proxy secure proxy check
> request goes out over vanilla HTTP so that it can be
> intercepted by middleboxes.
>
> Also, secure proxy check is now done on IO thread.
>
> BUG=437080
>
> Committed: https://crrev.com/652eabf1141c00594aa6e9ed81beb980aec89198
> Cr-Commit-Position: refs/heads/master@{#325302}
>
> Committed: https://crrev.com/afc53166cb6a0ed561f416403ab68261eae4427b
> Cr-Commit-Position: refs/heads/master@{#325464}

TBR=sclittle@chromium.org,rch@chromium.org,bengr@chromium.org,mmenke@chromium.org,sgurun@chromium.org,jeremyim@chromium.org,tbansal@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=437080

Review URL: https://codereview.chromium.org/1089443003

Cr-Commit-Position: refs/heads/master@{#325526}

BUG=chrome-os-partner:37102

Review URL: https://codereview.chromium.org/1085973002

Cr-Commit-Position: refs/heads/master@{#325525}

Previously for <input> with autocomplete="off", we send a synthesized keyCode.
This CL disables this feature to send 229 instead (however, backspace still
generates 8).

BUG=476497

Review URL: https://codereview.chromium.org/1082823002

Cr-Commit-Position: refs/heads/master@{#325524}

BUG=476592
TBR=sullivan@chromium.org

Review URL: https://codereview.chromium.org/1089843003

Cr-Commit-Position: refs/heads/master@{#325523}

TBR=dfalcantara@chromium.org

Review URL: https://codereview.chromium.org/1087363002

Cr-Commit-Position: refs/heads/master@{#325522}

Review URL: https://codereview.chromium.org/1095683002

Cr-Commit-Position: refs/heads/master@{#325521}

BUG=473250

Review URL: https://codereview.chromium.org/1091853002

Cr-Commit-Position: refs/heads/master@{#325520}

BUG=470391

Review URL: https://codereview.chromium.org/1090033002

Cr-Commit-Position: refs/heads/master@{#325519}

Rename the OWNER file to OWNERS so that presubmit reads it.

BUG=

Review URL: https://codereview.chromium.org/1093683002

Cr-Commit-Position: refs/heads/master@{#325518}

This reverts https://codereview.chromium.org/914363002 completely.
Some of the changes have already been reverted in
https://codereview.chromium.org/1008683002.

R=dalecurtis@chromium.org
BUG=457280

Review URL: https://codereview.chromium.org/1098503002

Cr-Commit-Position: refs/heads/master@{#325517}

Review URL: https://codereview.chromium.org/1086183003

Cr-Commit-Position: refs/heads/master@{#325516}

This CL maps an  ONC property for setting per-network roam threshold.

BUG=chrome-os-partner:38681
TEST=Expanded test ONC and JSON test sets

Review URL: https://codereview.chromium.org/1087353002

Cr-Commit-Position: refs/heads/master@{#325515}

In GLX, most of the drivers can cope with both PBuffer and Surfaceless modes
for offscreen rendering. In EGL though, Intel driver only works with
Surfaceless.

This CL changes all hardware platforms to use Surfaceless by forcing a zero
size GL surface initialization. Ozone-GBM wasn't previously working because of
that but now works fine.

BUG=423481
TEST=gpu_perftests on GBM

Review URL: https://codereview.chromium.org/1086703007

Cr-Commit-Position: refs/heads/master@{#325514}

BUG=467287
R=dcheng@chromium.org

Review URL: https://codereview.chromium.org/1095653002

Cr-Commit-Position: refs/heads/master@{#325513}

base/strings: Remove an unnecessary UTF8 check from one of the trim functions

gtest's message often use leading whitespace in a way that's visually meaningful. Some examples:

======== EXPECT_EQ(this, this+1);
my_test.cc(2048): error: Value of: this+1
  Actual: 24F8B8D4
Expected: this
Which is: 24F8B8A0
========
======== EXPECT_EQ("AA\nBB\nDD", std::string("AA\nXX\nDD"));
my_test.cc(2049): error: Value of: std::string("AA\nXX\nDD")
  Actual: "AA\nXX\nDD"
Expected: "AA\nBB\nDD"
With diff:
@@ -1,3 +1,3 @@
 AA
-BB
+XX
 DD
========

Today leading whitespaces are stripped out by the test launcher, leading to less readable test stdout on the bots (and headaches, especially, if the 'AA' value in the example above had leading whitespace itself -- which is how I noticed this).

The current whitespace trimming seems to be an accidental effect of [https://codereview.chromium.org/324893004], so it should be safe to change back.

BUG=475265

Review URL: https://codereview.chromium.org/1081493002

Cr-Commit-Position: refs/heads/master@{#325512}

create_nmf should determine the root of the SDK relative
to its own location.

The exception to this is when its run as part of the
chrome build, when it is not yet installed.  In this
case we pass --no-default-libpath and --objdump to
create_nmf to remove any use of the SDK root.

This is the second attempt at landing this change.
The first one was reverted:
https://codereview.chromium.org/737653003

CQ_EXTRA_TRYBOTS=tryserver.chromium.linux:linux_nacl_sdk;tryserver.chromium.mac:mac_nacl_sdk
BUG=none

Review URL: https://codereview.chromium.org/1066823003

Cr-Commit-Position: refs/heads/master@{#325511}

BUG=477556

Review URL: https://codereview.chromium.org/1081073004

Cr-Commit-Position: refs/heads/master@{#325510}

This make the API more consistent with RefCountedMemory and
other discardable memory interfaces such as SkDiscardableMemory.

Also improves readability by replacing memory->Memory() calls
with memory->data() calls.

A handy method is also added to simplify calling data() with a
reinterpret_cast. This is also consistent with RefCountedMemory.

BUG=

Review URL: https://codereview.chromium.org/1094433002

Cr-Commit-Position: refs/heads/master@{#325509}

In the FrameTree unittests, illustrate the liveness of frames in the ASCII
dumps. Update some tests so that children aren't born to dead parents.

BUG=474231

Review URL: https://codereview.chromium.org/1070553005

Cr-Commit-Position: refs/heads/master@{#325508}

1. Increasing hard-coded estimates to fix stalls in low fps YouTube videos
2. Moving video to use max duration to avoid stalls.
3. Adjusting estimates in SourceBufferRange upon future appends.

*More on hard-coded estimates*
The hard-coded video duration estimate was changed from 42ms (24fps) to 63ms (16fps). While 42ms is more common than 63ms, both are just wild guesses and 42ms was causing low framerate videos (12fps and lower) to stall by incorrectly triggering gaps in the MSE buffered range. 63ms will prevent these gaps for video as low as 8fps. To the extent that 63ms favors over-estimation, this is mitigated by the adjustments in SourceBufferRange.

*What about audio*
This CL lays the foundation for a similar change to audio's estimation. I'm splitting them up to keep the size reasonable.

Internal Google bug: b/18090240

Review URL: https://codereview.chromium.org/1018373003

Cr-Commit-Position: refs/heads/master@{#325507}