GitLab

BUG=348911
TBR=eae

Review URL: https://codereview.chromium.org/183743017

git-svn-id: svn://svn.chromium.org/blink/trunk@168372 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=258421

Review URL: https://codereview.chromium.org/18132024

git-svn-id: svn://svn.chromium.org/blink/trunk@168371 bbb929c8-8fbe-4397-9dbb-9b2b20218538

R=lushnikov@chromium.org, lushnikov, pfeldman

Review URL: https://codereview.chromium.org/183663014

git-svn-id: svn://svn.chromium.org/blink/trunk@168370 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Revert of Ensure DOMWrapperWorld always exists in all webkit_unit_tests (https://codereview.chromium.org/180743013/)

Reason for revert:
A few inspector tests are crashing:
WebCore::DOMWrapperWorld::ensureIsolatedWorld(int, int) [DOMWrapperWorld.cpp : 124 + 0x33]

Original issue's description:
> Ensure DOMWrapperWorld always exists in all webkit_unit_tests
> 
> Currently DOMWrapperWorld can be 0 in some tests. This is problematic since I want to make DOMWrapperWorld::current() callable from anywhere. Thus this CL adds DOMWrapperWorld to all webkit_unit_tests.
> 
> - It's verbose to initialize v8::HandleScope, v8::Context, v8::Context::Scope, V8PerContextData and DOMWrapperWorld in each test. So this CL adds a helper class (V8BindingTestScope) that does all the initialization work.
> 
> - This CL also fixes indentation of V8Binding.h.
> 
> BUG=341032
> NOTRY=true
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=168355

TBR=jochen@chromium.org,dcarney@chromium.org,haraken@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=341032

Review URL: https://codereview.chromium.org/183803017

git-svn-id: svn://svn.chromium.org/blink/trunk@168369 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This moves 8200 lines out of BisonCSSParser leaving only 1700 lines left. :)

When I split CSSPropertyParser from BisonCSSParser I left
all the code in BisonCSSParser-in.cpp to make the diff
small and to avoid conflicting with others changes.

Now I'm moving (almost) all of CSSPropertyParser into the
correct .cpp file.

The only bit left is some transform parsing code which uses
the -in.cpp preprocessor bits of BisonCSSParser. Its
the only code which uses this fancy SWITCH/CASE preprocessor
and should just be moved out into its own file.

BUG=330389

Review URL: https://codereview.chromium.org/185293006

git-svn-id: svn://svn.chromium.org/blink/trunk@168368 bbb929c8-8fbe-4397-9dbb-9b2b20218538

1) Selected window doesn't slide while recording. Why: there used to be a call for _updateWindow from _update method.
2) No overview for Events mode. Why: raw records may have no valid endTime.
3) The window selection is dropping by recording timeline. Why: earlier overview pane saved the selected time range at _onWindowChanged.

BUG=
R=pfeldman@chromium.org

Review URL: https://codereview.chromium.org/185543012

git-svn-id: svn://svn.chromium.org/blink/trunk@168367 bbb929c8-8fbe-4397-9dbb-9b2b20218538

R=erik.corry@gmail.com, haraken@chromium.org
BUG=

Review URL: https://codereview.chromium.org/183683012

git-svn-id: svn://svn.chromium.org/blink/trunk@168366 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This should have been removed in https://codereview.chromium.org/184693003/

Review URL: https://codereview.chromium.org/178543008

git-svn-id: svn://svn.chromium.org/blink/trunk@168365 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=347562
R=pfeldman
NOTRY=true

Review URL: https://codereview.chromium.org/178473018

git-svn-id: svn://svn.chromium.org/blink/trunk@168364 bbb929c8-8fbe-4397-9dbb-9b2b20218538

These strings need to be copied along with the event. We only used to
support static strings so far.

BUG=

Review URL: https://codereview.chromium.org/184303002

git-svn-id: svn://svn.chromium.org/blink/trunk@168363 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Before this CL, it was possible that DatabaseThread held the last
references of DatabaseBackend objects and it was released in the
database thread.
We must destruct objects in the originator thread in Oilpan.

Implementation:
DatabaseThread::recordDatabaseClosed doesn't remove a RefPtr<
DatabaseBackend>, and we release RefPtr<DatabaseBackend> objects
in ~DatabaseThread, which is executed in the main/worker thread.
Note that there are no ways to close a database explicitly, and
DatabaseBackend objects are never destructed before their
DatabaseContext is stopped. DatabaseThread::recordDatabaseClosed
is actually unnecessary.

DatabaseThread doesn't know DatabaseBackend open status any longer.
We remove DatabaseThread::isDatabaseOpen, and its existing
callsites use DatabaseBackend::open() instead.

BUG=347902
TEST=none; no visible behavior changes hopefully.
R=michaeln@chromium.org

Review URL: https://codereview.chromium.org/183883005

git-svn-id: svn://svn.chromium.org/blink/trunk@168362 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=
R=pfeldman@chromium.org

Review URL: https://codereview.chromium.org/185533004

git-svn-id: svn://svn.chromium.org/blink/trunk@168361 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This CL moves the early out to the top of the function, which lets us merge the
two stacking node iterations.

R=ojan@chromium.org

Review URL: https://codereview.chromium.org/183973017

git-svn-id: svn://svn.chromium.org/blink/trunk@168360 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Prior to this CL, we were incorrectlyd dirtying
RenderLayer::hasVisibleContent() when removing a RenderLayer from the layer
tree. This caused us to incorrectly believe that the parent layer lacked
visible content.

R=ojan@chromium.org

Review URL: https://codereview.chromium.org/185653011

git-svn-id: svn://svn.chromium.org/blink/trunk@168359 bbb929c8-8fbe-4397-9dbb-9b2b20218538

These tests that compositing updates work even when the render tree changes in
a minor way. These tests would have failed when introducing dirty bits for
CompositedLayerMapping in https://codereview.chromium.org/183763016.

R=chrishtr@chromium.org

Review URL: https://codereview.chromium.org/183833016

git-svn-id: svn://svn.chromium.org/blink/trunk@168358 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch fixes a bug where 'removetrack' events were being fired
when all the inband tracks were being removed. This behavior was trying
to implement the "forget the media element's media-resource-specific
tracks" algorithm, but it wasn't correct.

This patch includes the following changes:
- Renames removeAllInbandTracks() to forgetResourceSpecificTracks()
- Reorganized logic in prepareForLoad() to match the spec order. The
  shuffling should not have any user visible behaviour change.
- Added forgetResourceSpecificTracks() calls to other places that
  the spec indicates it should run.
- Added code to make sure forgetResourceSpecificTracks() doesn't
  trigger 'removetrack' events.
- Converted the existing media/track/track-remove-track.html test
  to W3C test & added a new test case that verifies that the event
  doesn't fire.

BUG=249427
TEST=Existing LayoutTests still pass. Added a new case to media/track/track-remove-track.html

Review URL: https://codereview.chromium.org/177243018

git-svn-id: svn://svn.chromium.org/blink/trunk@168357 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Currently DOMWrapperWorld can be 0 in some tests. This is problematic since I want to make DOMWrapperWorld::current() callable from anywhere. Thus this CL adds DOMWrapperWorld to all webkit_unit_tests.

- It's verbose to initialize v8::HandleScope, v8::Context, v8::Context::Scope, V8PerContextData and DOMWrapperWorld in each test. So this CL adds a helper class (V8BindingTestScope) that does all the initialization work.

- This CL also fixes indentation of V8Binding.h.

BUG=341032
NOTRY=true

Review URL: https://codereview.chromium.org/180743013

git-svn-id: svn://svn.chromium.org/blink/trunk@168355 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Moving Event to oilpan's heap caused a lot of crashes/failures. So this CL reverts Event back to RefCounted.

This CL is a revert for https://codereview.chromium.org/182063004/. However, this CL does not revert trace() methods in the original CL because there is no harm in having trace() methods in the code base. (If I revert trace() methods, I have to resolve a bunch of conflicts manually, which will be more dangerous than leaving trace() methods in the code base.)

BUG=340522
NOTRY=true

Review URL: https://codereview.chromium.org/185463005

git-svn-id: svn://svn.chromium.org/blink/trunk@168354 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://src.chromium.org/viewvc/blink?view=revision&revision=168336

BUG=346132
TBR=tsepez@chromium.org

Review URL: https://codereview.chromium.org/183893017

git-svn-id: svn://svn.chromium.org/blink/trunk@168353 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch fixes a crash on empty documents where a cluster would not
get created and we would crash during inflation. I've changed
beginLayout to only call inflate when a block has children, and
refactored endLayout similarly.

BUG=348458
NOTRY=true

Review URL: https://codereview.chromium.org/182443004

git-svn-id: svn://svn.chromium.org/blink/trunk@168352 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Before this change, stylesheets in imports are ordered based on
<link>s that "own" these imports. The ownership however is an
implementation detail and shouldn't be observable.

With this change, it takes only the tree order of <link>s into account.
The ownership no longer plays any role.

BUG=346878
TEST=import-style-tree-order-dedup.html
R=esprehn@chromium.org, dglazkov@chromium.org

Review URL: https://codereview.chromium.org/179873020

git-svn-id: svn://svn.chromium.org/blink/trunk@168351 bbb929c8-8fbe-4397-9dbb-9b2b20218538

- This CL removes V8PerIsolateData::m_domDataStoreList and related methods.

- The only customer of these methods is SerializedScriptValue::neuterBinding(). This CL replaces it to use DOMWrapperWorld::allWorldsInMainThread() instead.

BUG=341032
NOTRY=true

Review URL: https://codereview.chromium.org/182513003

git-svn-id: svn://svn.chromium.org/blink/trunk@168347 bbb929c8-8fbe-4397-9dbb-9b2b20218538

After haraken's change, Event is a heap-allocated object. Events should not be dispatched while document is being disposed.

BUG=None

Review URL: https://codereview.chromium.org/182383005

git-svn-id: svn://svn.chromium.org/blink/trunk@168346 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=None

Review URL: https://codereview.chromium.org/184103012

git-svn-id: svn://svn.chromium.org/blink/trunk@168345 bbb929c8-8fbe-4397-9dbb-9b2b20218538

MeasureTextData::processRenderer is only used within
SVGTextMetricsBuilder::measureTextRenderer, and is always set before a
call to that method. This makes a good candidate for removal from the
traversal state struct (MeasureTextData) and being passed as an argument
to measureTextRenderer instead.

There's no need to track both the last character and whether there was
a previous character. It's sufficient to track if the the last character
was a whitespace or not. (The initial case which previously had
hasLastCharacter == false can be represented as having a been a whitespace
character.) This allows replacing the two fields hasLastCharacter and
lastCharacter with a single field lastCharacterWasWhiteSpace.

MeasureTextData::skippedCharacters is reset to zero (0) at the end of
SVGTextMetricsBuilder::measureTextRenderer, and is only read or written
if MeasureTextData::allCharactersMap is non-null.
This means it can be kept on the stack of measureTextRenderer instead.

Also change the type of surrogatePairCharacters from int to unsigned,
since it's a simple counter, that should never be negative (and is used
in arithmetic with unsigned typed variables.)

Review URL: https://codereview.chromium.org/183203003

git-svn-id: svn://svn.chromium.org/blink/trunk@168342 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This does some cleanup to GYP;
only change to build is consolidating two identical gyp-time files into one.

* Move temporary list file variables to the top, with other variables
  (keeps all these lists together, removes duplicate variables, and removes
   duplicate comments)
* Add comment lines between actions (GYP files are hard to read,
  and we're going to add more actions for caching lexer/parser)
* Fix indentation so consistent
(.gyp file formatting isn't terribly consistent or linted,
but this makes reading and edits easier)

BUG=341748
R=haraken

Review URL: https://codereview.chromium.org/181063013

git-svn-id: svn://svn.chromium.org/blink/trunk@168341 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Refactoring CL, inspired by and preliminary to Terry's "speed up r-b-t" CL.
This adds a thin wrapper class, IdlCompiler, which has two methods:
* Initialization (both IdlReader and CodeGeneratorV8)
* .compile()

The resulting main() is v. simple (clear initialization/compile);
this will also make Terry's CL considerably simpler and easier to review,
since he can just instantiate this class in r-b-t and call it repeatedly.

Terry's:
Faster PYTHON IDL dependency computation and parsing.
https://codereview.chromium.org/169743005/

R=haraken

Review URL: https://codereview.chromium.org/178003010

git-svn-id: svn://svn.chromium.org/blink/trunk@168340 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Set the initial directionality to Neutral in
RenderText::computePreferredLogicalWidths. This ensures that leading and
trailing SCRIPT_COMMON characters gets the correct directionality and
thus the correct width.

TEST=fast/text/international/rtl-space-in-ltr-element.html
BUG=333004
R=leviw@chromium.org

Review URL: https://codereview.chromium.org/185123002

git-svn-id: svn://svn.chromium.org/blink/trunk@168339 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Currently gin::IsolateHolder of the main thread is owned by V8PerIsolateData, and gin::IsolateHolder of a worker thread is owned by WorkerScriptController. This is inconsistent. gin::IsolateHolder of worker threads should be owned by V8PerIsolateData as well.

(c.f., gin::ContextHolder is owned by V8PerContextData.)

BUG=341032
NOTRY=true

Review URL: https://codereview.chromium.org/185653002

git-svn-id: svn://svn.chromium.org/blink/trunk@168338 bbb929c8-8fbe-4397-9dbb-9b2b20218538

R=tonyg@chromium.org
BUG=348655

Review URL: https://codereview.chromium.org/177983010

git-svn-id: svn://svn.chromium.org/blink/trunk@168337 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=346132
TBR=cbiesinger@chromium.org

Review URL: https://codereview.chromium.org/183893014

git-svn-id: svn://svn.chromium.org/blink/trunk@168336 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=
CC=apavlov@chromium.org,dongseong.hwang@intel.com
BUG=

Review URL: https://codereview.chromium.org/185293013

git-svn-id: svn://svn.chromium.org/blink/trunk@168335 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://src.chromium.org/viewvc/blink?view=revision&revision=168326

BUG=258420
TBR=a.suchit@samsung.com

Review URL: https://codereview.chromium.org/181573003

git-svn-id: svn://svn.chromium.org/blink/trunk@168334 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Prior to https://src.chromium.org/viewvc/blink?revision=166428&view=revision
scroll, scale and form state was lazily serialized at the moment of
sending up the NavigationEntry message to the browser.  However, this
read-time mutation sometimes inappropriately clobbered the scroll offset
so it was deleted in the above patch.  But it's important to frequently
reserialize this state to defend against unexpected process deaths.

This patch brings a more principled approach to history save/restore:
- scroll save now happens immediately whenever one of the inputs changes.
- for scroll restore, we now attempt to do it every contents size
  change, which is similar to the policy that's proven optimal for page
  scale initialization.  This should improve usability by converging to
  the scroll position more quickly (instead of waiting until the
  progress spinner completes).
- for form fields specifically, I brought back the previous behavior of
  serializing on currentHistoryItem() calls (but added a "dirty" flag to
  do it only when needed).  This was a batching optimization dating from
  2009 (https://bugs.webkit.org/show_bug.cgi?id=31276), I'm guessing to
  avoid O(n) cost per letter when typing into huge textboxes.

NOTRY=true
BUG=344072

Review URL: https://codereview.chromium.org/172893007

git-svn-id: svn://svn.chromium.org/blink/trunk@168333 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This change is to match Blink's HTML Imports implementation
match the spec change [1] which says that all imported document
should follow CSP of the master document, instead of its own.

The implementation is straightforward. It:

- Makes ContentSecurityPolicy object ref-counted and passes
  one of the master to its imported document.
- Disables <meta> directive for CSP when the document is an import.

Tests have to be changed much. As one CSP constraints all the scripts in each test,
things like shouldBe() is no longer usable in many test. Also, many imports
are now need to be servced via CORS so the change turns some of them from HTML to CGI.

Removed tests are ones that is no longer relevant: if script-src is unsafe-inline,
we cannot load any imports thus cannot test scripts in such imports.

[1] https://www.w3.org/Bugs/Public/show_bug.cgi?id=24268

BUG=334210
TEST=http/tests/htmlimports/
R=abarth@chromium.org, dglazkov@chromium.org

Review URL: https://codereview.chromium.org/183883004

git-svn-id: svn://svn.chromium.org/blink/trunk@168331 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The reported crash can't be reproduced anymore.

BUG=232903

Review URL: https://codereview.chromium.org/180213005

git-svn-id: svn://svn.chromium.org/blink/trunk@168330 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=pfeldman

Review URL: https://codereview.chromium.org/178463005

git-svn-id: svn://svn.chromium.org/blink/trunk@168328 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://src.chromium.org/viewvc/blink?view=revision&revision=168320

BUG=114658
TBR=cbiesinger@chromium.org

Review URL: https://codereview.chromium.org/176843013

git-svn-id: svn://svn.chromium.org/blink/trunk@168327 bbb929c8-8fbe-4397-9dbb-9b2b20218538

When all rows in spanning cell are having only row spanning cells with empty
cell than None of the rows in spanning cell was not getting any height.

Total height of the rows in spanning cell should have at least spanning cell
height so apply spanning cell height in the last row of the spanning cells
if all rows in spanning cell are have zero height.
The total height of the rows in a spanning cell should be at least the height of
the spanning cell's itself because content should fit in the cell. This change
adds the remaining height to the last row because every row is belong to this cell only.

R=jchaffraix@chromium.org
BUG=258420,330564,341366,346150

Review URL: https://codereview.chromium.org/47923009

git-svn-id: svn://svn.chromium.org/blink/trunk@168326 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Change to use same logic for verifying keySystem and contentType values when creating MediaKeys and MediaKeySessions as is used for isTypeSupported(). Removes unfinished stub code in ContentDecryptionModule.

BUG=342107
TEST=manual

Review URL: https://codereview.chromium.org/183943003

git-svn-id: svn://svn.chromium.org/blink/trunk@168321 bbb929c8-8fbe-4397-9dbb-9b2b20218538