GitLab

Revert of [ServiceWorker] Don't allow registration of the ServiceWorker scope outside the script directory. (patchset #4 id:70001 of https://codereview.chromium.org/672383003/)

Reason for revert:
This patch seems to have broken the following two browser_tests on all platforms, as can be seen on the non-layout bots on the Blink waterfall. We won't be able to roll Blink in Chromium until this is resolved.

PushMessagingBrowserTest.RegisterFailureNoPermission
PushMessagingBrowserTest.RegisterSuccess

Example output:
http://build.chromium.org/p/chromium.webkit/builders/Linux%20Tests%20%28dbg%29/builds/4400/steps/browser_tests/logs/RegisterSuccess

[INFO:CONSOLE(7)] "SecurityError - The scope must be under the directory of the script URL.", source: https://127.0.0.1:39935/files/push_messaging/test.html (7)
../../chrome/browser/services/gcm/push_messaging_browsertest.cc:142: Failure
Value of: register_worker_result
  Actual: "SecurityError - The scope must be under the directory of the script URL."
Expected: "ok"

Original issue's description:
> [ServiceWorker] Don't allow registration of the ServiceWorker scope outside the script directory.
> 
> As per discussion on https://github.com/slightlyoff/ServiceWorker/issues/468
> 
> BUG=423983
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=184336

TBR=falken@chromium.org,horo@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=423983

Review URL: https://codereview.chromium.org/672353003

git-svn-id: svn://svn.chromium.org/blink/trunk@184364 bbb929c8-8fbe-4397-9dbb-9b2b20218538

We used to have garbage in resulting bitmap if we abort redraw before fill commands
in SkPicture are executed.

BUG=426100,426102

Review URL: https://codereview.chromium.org/654843010

git-svn-id: svn://svn.chromium.org/blink/trunk@184363 bbb929c8-8fbe-4397-9dbb-9b2b20218538

According to our object profiling in micro-benchmarks and real-world benchmarks (https://docs.google.com/a/google.com/spreadsheets/d/1EBYDfkvJ1aleIj0iSNmk0zEsq0AqVOsjEyI6XfOoLIU/edit#gid=562560234), the following objects are allocated very frequently.

- Node
- CSSValue
- RenderObject
- HTMLStackItem
- ElementRecord

Given the results, this CL introduces a dedicated heap for CSSValue and RenderObject. (Note: PartitionAlloc also has a dedicated heap for RenderObject.)

This improves various benchmarks a lot:

css/CSSPropertySetterGetter: +35%
css/ClassInvalidation: +32%
css/StyleSheetInsert: +21%
layout/floats-*: ~+20%
layout/large-table-with-*: ~+10%

BUG=420515

Review URL: https://codereview.chromium.org/667053009

git-svn-id: svn://svn.chromium.org/blink/trunk@184362 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This makes the main thread block and wait for the parser thread when all the
script data has arrived. The goal is to ensure that the script (which can now be
compiled) will get the main thread's attention as soon as possible.

This feature is temporary: The usefulness of blocking will be evaluated with
Finch, and this option will be removed (blocking will be always enabled or never
enabled, based on which option turns out to be better).

Previous version: https://codereview.chromium.org/651163002/

R=haraken@chromium.org
BUG=

Review URL: https://codereview.chromium.org/674133002

git-svn-id: svn://svn.chromium.org/blink/trunk@184361 bbb929c8-8fbe-4397-9dbb-9b2b20218538

R=pfeldman
BUG=426815

Review URL: https://codereview.chromium.org/640953004

git-svn-id: svn://svn.chromium.org/blink/trunk@184360 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=caseq

Review URL: https://codereview.chromium.org/675173002

git-svn-id: svn://svn.chromium.org/blink/trunk@184359 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This is required for the validator to work on Windows, as otherwise
the command line exceeds the maximum length there.

R=sergeyv@chromium.org, sergeyv

Review URL: https://codereview.chromium.org/669123005

git-svn-id: svn://svn.chromium.org/blink/trunk@184358 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Design doc:
https://docs.google.com/a/chromium.org/document/d/16f_RIhZa47uEK_OdtTgzWdRU0RFMTQWMpEWyWXIpXUo/edit#heading=h.srz53flt1rrp

BUG=391005

Review URL: https://codereview.chromium.org/656323005

git-svn-id: svn://svn.chromium.org/blink/trunk@184357 bbb929c8-8fbe-4397-9dbb-9b2b20218538

We used to change cursor upon drag on body only, but this now doesn't
have effect since we explicitly set cursor: default in .component-root.
So set the cursor explicitly both on the event target and body (the latter
is needed so that cursor change also has effect on the glass pane).

BUG=

Review URL: https://codereview.chromium.org/658723006

git-svn-id: svn://svn.chromium.org/blink/trunk@184356 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=none

Review URL: https://codereview.chromium.org/675753002

git-svn-id: svn://svn.chromium.org/blink/trunk@184355 bbb929c8-8fbe-4397-9dbb-9b2b20218538

We have many .*Policy in this file. Use a more specific name for
readability.

R=mkwst
BUG=none

Review URL: https://codereview.chromium.org/666533005

git-svn-id: svn://svn.chromium.org/blink/trunk@184352 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=309710

Review URL: https://codereview.chromium.org/652813005

git-svn-id: svn://svn.chromium.org/blink/trunk@184351 bbb929c8-8fbe-4397-9dbb-9b2b20218538

WebIDL Dictionaries should access prototype chains when searching for properties. This change adds getPropertyNames to the v8 Dictionary class.

Web Animations Keyframes should make use of prototype chains when looking for CSS properties to animate. This change makes use of getPropertyNames to ensure this happens.

Review URL: https://codereview.chromium.org/661453004

git-svn-id: svn://svn.chromium.org/blink/trunk@184350 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This reverts r184277.

Reason: ASSERTS fail in debug mode because ScriptStreamer gets destroyed while
holding the mutex.

BUG=
TBR=marja@chromium.org, haraken@chromium.org
NOTRY=true

Review URL: https://codereview.chromium.org/680493002

git-svn-id: svn://svn.chromium.org/blink/trunk@184349 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Switched from using dumpResourceRequestCallbacks to using direct
memoryCache APIs in internals.

BUG=426041

Review URL: https://codereview.chromium.org/675853003

git-svn-id: svn://svn.chromium.org/blink/trunk@184348 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch fixes our handling of escaped newlines in the css-syntax
tokenizer. Since we don't perform preprocessing, the logic is slightly
trickier than in the spec. The preprocessing replaces \r, \r\n, \f with
\n.

Regarding escaped newlines in strings, the spec states:
"Otherwise, if the next input code point is a newline, consume it."
In this case, we may need to consume two tokens if we have \r\n.

The other cases don't require special handling, since \r\n starts with
\r (also a newline) and in these cases the following token is just going
to be a whitespace token.

BUG=424988

Review URL: https://codereview.chromium.org/656033010

git-svn-id: svn://svn.chromium.org/blink/trunk@184347 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The line-height property has lots of special handling in the style
resolver which we want to remove at some point. When we do, it's nice to
know that this still works correctly.

R=rune@opera.com
BUG=353932

Review URL: https://codereview.chromium.org/678483005

git-svn-id: svn://svn.chromium.org/blink/trunk@184346 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Use dataEquivalent instead of comparing pointers. Updating the base
RenderStyle in ActiveAnimations asserts that the cached base RenderStyle
is correct by comparing RenderStyle objects. It triggered an assert when
comparing two shape-outside values represented by different objects
representing the same value.

R=esprehn@chromium.org,dstockwell@chromium.org
BUG=425473

Review URL: https://codereview.chromium.org/669093002

git-svn-id: svn://svn.chromium.org/blink/trunk@184345 bbb929c8-8fbe-4397-9dbb-9b2b20218538

R=vsevik@chromium.org,aandrey@chromium.org,pfeldman@chromium.org

Review URL: https://codereview.chromium.org/649183010

git-svn-id: svn://svn.chromium.org/blink/trunk@184344 bbb929c8-8fbe-4397-9dbb-9b2b20218538

NOTRY=true
TBR=sigbjornf@opera.com

Review URL: https://codereview.chromium.org/674103002

git-svn-id: svn://svn.chromium.org/blink/trunk@184343 bbb929c8-8fbe-4397-9dbb-9b2b20218538

It's not necessary to do initialization of FontBuilder separately from
StyleResolverState::setStyle. It just adds noise.

Also, use a reference for Document in FontBuilder, since its existence can
be guaranteed in all cases.

R=dglazkov@chromium.org

Review URL: https://codereview.chromium.org/671613003

git-svn-id: svn://svn.chromium.org/blink/trunk@184342 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Because there is a "cue" entry in both pseudoTypeWithoutArgumentsMap and
pseudoTypeWithArgumentsMap, this path is never taken (verified with
ASSERT(0)) when running LayoutTests/media/.

It is taken for something like ::cuefoo, but that makes no sense.

This was initially the only code path:
http://trac.webkit.org/changeset/136991

Later commits made it obsolete:
http://trac.webkit.org/changeset/137955
https://src.chromium.org/viewvc/blink?revision=155834&view=revision

This was the last usage of PseudoUserAgentCustomElement, so remove that
too, together with its use counter:
https://src.chromium.org/viewvc/blink?revision=155928&view=revision
https://www.chromestatus.com/metrics/feature/timeline/popularity/103

At the time pseudos starting with "x-" were also
PseudoUserAgentCustomElement, but has since been removed:
https://src.chromium.org/viewvc/blink?revision=170210&view=revision

Review URL: https://codereview.chromium.org/646853007

git-svn-id: svn://svn.chromium.org/blink/trunk@184341 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=364627
TEST=virtual/stable/webexposed

Review URL: https://codereview.chromium.org/680473002

git-svn-id: svn://svn.chromium.org/blink/trunk@184340 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=424767

Review URL: https://codereview.chromium.org/640823005

git-svn-id: svn://svn.chromium.org/blink/trunk@184339 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The "stylsheets" field is not used at runtime,
so it is removed to reduce the footprint.

R=vsevik

Review URL: https://codereview.chromium.org/676993002

git-svn-id: svn://svn.chromium.org/blink/trunk@184338 bbb929c8-8fbe-4397-9dbb-9b2b20218538

It was reverted due to bug 426608; the fix was landed on the Chromium side.

Original review url: https://codereview.chromium.org/671003002

BUG=426608,426053
TBR=leviw

Review URL: https://codereview.chromium.org/663233004

git-svn-id: svn://svn.chromium.org/blink/trunk@184337 bbb929c8-8fbe-4397-9dbb-9b2b20218538

As per discussion on https://github.com/slightlyoff/ServiceWorker/issues/468

BUG=423983

Review URL: https://codereview.chromium.org/672383003

git-svn-id: svn://svn.chromium.org/blink/trunk@184336 bbb929c8-8fbe-4397-9dbb-9b2b20218538

When resizing the viewport, srcset re-evaluates its resources, and downloads
a new one if needed.

However, since srcset is by definition not related to art-direction,
there's no need to download a new resource is there is a higher density one
that's already in the memory cache.

This CL avoids that unneccessary download.

BUG=425701

Review URL: https://codereview.chromium.org/674923004

git-svn-id: svn://svn.chromium.org/blink/trunk@184335 bbb929c8-8fbe-4397-9dbb-9b2b20218538

It seems that TypeErrors can be thrown when we modify the content of
pickle files (e.g. InterfacesInfoCoreIndividual.pickle). We can
just overwrite pickle files if pickle.load() throws an exception.

Review URL: https://codereview.chromium.org/643153006

git-svn-id: svn://svn.chromium.org/blink/trunk@184334 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=426047

Review URL: https://codereview.chromium.org/672423003

git-svn-id: svn://svn.chromium.org/blink/trunk@184333 bbb929c8-8fbe-4397-9dbb-9b2b20218538

With WebLocalFrameImpl on the heap, this object no longer needs to use
persistent (or untraced) references.

R=haraken
BUG=340522

Review URL: https://codereview.chromium.org/674773003

git-svn-id: svn://svn.chromium.org/blink/trunk@184332 bbb929c8-8fbe-4397-9dbb-9b2b20218538

We added AXComputedObjectAttributeCache to WebKit because we wanted to
cache some expensive computations in the AX tree, but none of the other
WebKit ports wanted it.

A much simpler solution is to just store the cached value inside the
AXObject. To determine if the value is out of date, I added a
modification count that increments whenever there's a layout or DOM
mutation. We can reuse the same modification count for more cached
attributes.

BUG=425879

Review URL: https://codereview.chromium.org/670093002

git-svn-id: svn://svn.chromium.org/blink/trunk@184331 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch cleans up the reason strings to not have periods. Additionally,
a typo has been fixed in InspectorTraceEvents.

BUG=410701,424451

Review URL: https://codereview.chromium.org/650993004

git-svn-id: svn://svn.chromium.org/blink/trunk@184330 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Revert of Timeout expectation for serviceworker/fetch-mixed-content.html (patchset #1 id:1 of https://codereview.chromium.org/668773003/)

Reason for revert:
Fixed by https://codereview.chromium.org/664343005

Original issue's description:
> Timeout expectation for serviceworker/fetch-mixed-content.html
> 
> Timing out following chromium r300644, but this looks like a test-only
> failure due to invalid cert in the localhost HTTPS server.
> 
> BUG=426047
> TBR=aboxhall@chromium.org,horo@chromium.org
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=184199

TBR=aboxhall@chromium.org,jsbell@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=426047

Review URL: https://codereview.chromium.org/648333003

git-svn-id: svn://svn.chromium.org/blink/trunk@184329 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This change displays CSS animation keyframes alongside the associated
animation player in the Animations sidebar pane. No visible changes are
introduced except with the hidden 'Animation Inspection' experiment enabled.

BUG=419269

Review URL: https://codereview.chromium.org/664993002

git-svn-id: svn://svn.chromium.org/blink/trunk@184328 bbb929c8-8fbe-4397-9dbb-9b2b20218538

It is critical to display list canvases, due to the fact, that we fallback from
rendering on impl thread, when the canvas is not fully redrawn inside each
frame. On Browsermark 2 it is actually messed up due to floating point rounding.
The area in SkScalar is {0.00061, 0.00061, 1599.99997, 999.99998}, rounded down
to {1,1,1599,999} and the entire canvas area seems to be not redrawn.

BUG=386601

Review URL: https://codereview.chromium.org/666253003

git-svn-id: svn://svn.chromium.org/blink/trunk@184327 bbb929c8-8fbe-4397-9dbb-9b2b20218538

SVG font implementation is removed from Blink. Remove LeakExpectations
entries for SVG font tests as they don't leak Documents anymore.

BUG=366455,242735
TBR=pdr

Review URL: https://codereview.chromium.org/673093002

git-svn-id: svn://svn.chromium.org/blink/trunk@184326 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://src.chromium.org/viewvc/blink?view=revision&revision=184321

BUG=242735
TBR=pdr@chromium.org

Review URL: https://codereview.chromium.org/673883005

git-svn-id: svn://svn.chromium.org/blink/trunk@184325 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This is a minor style correction. We'd prefer the 'implicit false' rather
than 'len(array) > 0'.

Review URL: https://codereview.chromium.org/678453004

git-svn-id: svn://svn.chromium.org/blink/trunk@184324 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch adds comprehensive unit tests to the css-syntax based
tokenizer. I've tried to cover all the edge cases I could think of here.

A few cases don't work correctly, which I've left as commented out
tests:
- Idents starting with two dashes like --abc aren't parsed as idents
- Escape should be replaced with U+FFFD when they represent either
(U+0000), a surrogate codepoint (U+D800 - U+DFFF) or are larger than
the maximum allowed codepoint (U+110000).
- We don't correctly replace U+0000 with U+FFFD
- Our handling of newline escapes (i.e. "check if two code points are a
valid escape") is incorrect in some cases as we don't perform
preprocessing.

Note that there have been minor spec changes since the candidate rec and
the tests here are based on the latest editor's draft.

http://dev.w3.org/csswg/css-syntax

BUG=424988

Review URL: https://codereview.chromium.org/646893003

git-svn-id: svn://svn.chromium.org/blink/trunk@184322 bbb929c8-8fbe-4397-9dbb-9b2b20218538