Timeline
Oct 27, 2013:
- 10:42 PM Changeset in webkit [158101] by
-
- 5 edits in trunk/Source/WebKit2
WebIconDatabase can miss private browsing state changes.
<rdar://problem/15322318> and https://bugs.webkit.org/show_bug.cgi?id=123375
Reviewed by Alexey Proskuryakov.
- UIProcess/WebContext.cpp:
(WebKit::WebContext::willStartUsingPrivateBrowsing): Call setAnyPageGroupMightHavePrivateBrowsingEnabled(true) on each context.
(WebKit::WebContext::willStopUsingPrivateBrowsing): Call setAnyPageGroupMightHavePrivateBrowsingEnabled(false) on each context.
(WebKit::WebContext::setPrivateBrowsingEnabled): In addition to notifying other processes about private browsing
sessions, notify the context’s WebIconDatabase about the change in value.
- UIProcess/WebContext.h:
- UIProcess/WebIconDatabase.cpp:
(WebKit::WebIconDatabase::setDatabasePath): Prime the IconDatabase with an initial private browsing value.
(WebKit::WebIconDatabase::setAnyPageGroupMightHavePrivateBrowsingEnabled):
- UIProcess/WebIconDatabase.h:
- 6:59 PM Changeset in webkit [158100] by
-
- 2 edits in trunk/Source/WebCore
Fix 4 asserting SVG tests after r158097.
RenderElement::setStyle() is, quite surprisingly, a virtual function
with a single override in RenderSVGBlock.
To match the old behavior, we have to rewrite the display type from
any inline type to block instead.
- rendering/RenderElement.cpp:
(WebCore::RenderElement::initializeStyle):
- 5:17 PM Changeset in webkit [158099] by
-
- 2 edits in trunk/Source/WebCore
Tone down overzealous assertion from r158097.
RenderElement::initializeStyle() really only cares that there are no
text renderers that we should be calling styleDidChange() on.
Tweak the code to only check that there are no text children.
- rendering/RenderElement.cpp:
(WebCore::RenderElement::initializeStyle):
- 4:06 PM Changeset in webkit [158098] by
-
- 4 edits in trunk/Source/WebCore
Enable center and right text alignment for simple lines
https://bugs.webkit.org/show_bug.cgi?id=123398
Reviewed by Andreas Kling.
Support text-align:center and text-align:right on simple line layout path.
- rendering/SimpleLineLayout.cpp:
(WebCore::SimpleLineLayout::canUseFor):
text-align:justify is still not supported.
(WebCore::SimpleLineLayout::computeLineLeft):
(WebCore::SimpleLineLayout::createLines):
Do a rounding dance that matches the line boxes.
- rendering/SimpleLineLayout.h:
Add left position to lines.
- rendering/SimpleLineLayoutResolver.h:
(WebCore::SimpleLineLayout::Resolver::Line::rect):
We now do rounding during layout.
(WebCore::SimpleLineLayout::Resolver::Line::baseline):
- 3:54 PM Changeset in webkit [158097] by
-
- 346 edits in trunk/Source/WebCore
Renderers should receive their style at construction.
<https://webkit.org/b/123396>
Pass the RenderStyle to all non-text renderer constructors.
After construction, initializeStyle() must be called (as a stopgap
measure) until we are able to do style-derived initialization
without virtual function calls.
With this change, RenderElement::m_style is never null. Subsequent
patches will add enforcement for this and also make style() return
a RenderStyle&.
I'm adding three FIXME's in this patch:
- createRendererIfNeeded() calls AnimationController to set up the initial style manually instead of asking RenderElement's setAnimatedStyle() to do it. This can probably be done in a nicer way, but it's not clear yet how.
- ImageContentData::createRenderer() does a bit of unnecessary work. This should be easy to clean up but got too distracting to be part of this patch.
- Document::createRenderTree() creates the RenderView with an initial dummy RenderStyle. I've done this because resolving the document style assumes we already have a RenderView.
For styleWillChange() implementations to detect that they are
reacting to the initial style, I've added a hasInitializedStyle()
function on RenderElement. This will return false until you've
called initializeStyle() on the renderer. This should go away
along with initializeStyle() eventually.
Reviewed by Antti Koivisto.
Oct 26, 2013:
- 7:36 PM Changeset in webkit [158096] by
-
- 8 edits in trunk/Source
[mac] Remove WebTiledLayer
https://bugs.webkit.org/show_bug.cgi?id=123395
Reviewed by Anders Carlsson.
Source/WebCore:
Mac doesn't use CATiledLayer at all anymore. We have to keep
LayerTypeTiledLayer around for Windows, for now, but we can
get rid of WebTiledLayer and some related Mac-specific code.
No new tests, just removing dead code.
- WebCore.xcodeproj/project.pbxproj:
Remove WebTiledLayer.*
- platform/graphics/ca/PlatformCALayer.h:
- platform/graphics/ca/mac/PlatformCALayerMac.h:
- platform/graphics/ca/mac/PlatformCALayerMac.mm:
(PlatformCALayerMac::PlatformCALayerMac):
Remove synchronouslyDisplayTilesInRect, which was only used for WebTiledLayer on Mac.
Remove WebTiledLayer instantiation and setup code.
- platform/graphics/mac/WebTiledLayer.h: Removed.
- platform/graphics/mac/WebTiledLayer.mm: Removed.
Source/WebKit2:
- WebProcess/WebPage/mac/PlatformCALayerRemote.cpp:
- WebProcess/WebPage/mac/PlatformCALayerRemote.h:
Remove synchronouslyDisplayTilesInRect.
- 5:48 PM Changeset in webkit [158095] by
-
- 4 edits in trunk/Websites/test-results
Make new bug link in flakiness dashboard configurable
https://bugs.webkit.org/show_bug.cgi?id=123386
Reviewed by Alexey Proskuryakov.
- config.json: Added Bugzilla as the default destination for new bugs.
- public/api/manifest.php: Include newBugLinks in the manifest.
- public/index.html:
(TestResultsView): Initialize _newBugLinks.
(TestResultsView.setNewBugLinks): Added.
(TestResultsView._populateTestPane):
(TestResultsView._createTestResultRow): Replaced the hard-coded Bugzilla URL by the code
to generate hyper-links based on _newBugLinks. Also added a nullity check while formatting
revision checks so that we don't blow up when some build doesn't contain all revision info.
(TestResultsView._populateBuilderPane):
- 5:48 PM Changeset in webkit [158094] by
-
- 3 edits in trunk/Websites/test-results
New flakiness dashboard should support substring matching
https://bugs.webkit.org/show_bug.cgi?id=123393
Reviewed by Alexey Proskuryakov.
Addressed the use cases by
- Always showing the candidate even when there is exactly one test matching the current value.
- Adding all tests that match the current value upon the user pressing enter key.
- public/index.html:
(fetchManifest): Add all tests that match the current value. Confirm whether the user really
want to add all the tests when there are more than 15 tests to add.
- public/js/autocompleter.js:
(Autocompleter.prototype.filterCandidates): Extracted from _updateCandidates.
(Autocompleter.prototype._updateCandidates): Show the candidate window even when there is
exactly one test that matches the criteria so that the user can select this test.
- 5:47 PM Changeset in webkit [158093] by
-
- 2 edits in trunk/Websites/test-results
New flakiness dashboard shouldn't treat tests with right expectations as failing
https://bugs.webkit.org/show_bug.cgi?id=123385
Reviewed by Alexey Proskuryakov.
We define failing tests to be tests with wrong expectations whose actual results are not PASS
since tests with TEXT, IMAGE, etc... failures do not turn the bots red as long as the expectation
of the same type is specified in TestExpectation files.
- public/include/test-results.php:
(FailingResultsJSONWriter): Inherit from WrongExpectationsResultsJSONWriter.
- 1:42 PM Changeset in webkit [158092] by
-
- 2 edits in trunk/LayoutTests
Crashy WebGL test is crashy.
- platform/mac/TestExpectations:
- 1:23 PM Changeset in webkit [158091] by
-
- 2 edits in trunk/Source/WebKit2
[Cocoa] -[WKBackForwardList forwardItem] returns the back item
https://bugs.webkit.org/show_bug.cgi?id=123391
Reviewed by Mark Rowe.
- UIProcess/Cocoa/WKBackForwardList.mm:
(-[WKBackForwardList forwardItem]): Changed back to forward.
- 11:59 AM Changeset in webkit [158090] by
-
- 5 edits in trunk/Source/WebCore
CTTE: RenderImageResourceStyleImage always has a StyleImage.
<https://webkit.org/b/123390>
Codify the fact that RenderImageResourceStyleImage always wraps an
existing StyleImage object.
Reviewed by Antti Koivisto.
- 11:45 AM Changeset in webkit [158089] by
-
- 2 edits in trunk/LayoutTests
Skip two WebGL tests that either crash or fail.
- platform/mac/TestExpectations:
- 11:39 AM Changeset in webkit [158088] by
-
- 4 edits in trunk/Source/WebCore
Revert some accidental changes.
- css/CSSFontFaceSource.cpp:
(WebCore::CSSFontFaceSource::getFontData):
- css/CSSFontSelector.cpp:
- css/CSSFontSelector.h:
- 10:09 AM Changeset in webkit [158087] by
-
- 5 edits in trunk/Source/WebKit2
Unreviewed, rolling out r158075.
http://trac.webkit.org/changeset/158075
https://bugs.webkit.org/show_bug.cgi?id=123389
Broke WebKit2.PrivateBrowsingPushStateNoHistoryCallback API
test (Requested by ap on #webkit).
- UIProcess/WebContext.cpp:
(WebKit::WebContext::willStartUsingPrivateBrowsing):
(WebKit::WebContext::willStopUsingPrivateBrowsing):
- UIProcess/WebContext.h:
- UIProcess/WebIconDatabase.cpp:
- UIProcess/WebIconDatabase.h:
- 9:11 AM Changeset in webkit [158086] by
-
- 2 edits in trunk/Source/WebCore
Gardening: fixed broken build.
https://bugs.webkit.org/show_bug.cgi?id=123354.
- css/CSSFontSelector.h:
- 8:44 AM Changeset in webkit [158085] by
-
- 7 edits in trunk
Source/WebCore: fast/frames/seamless/seamless-nested-crash.html asserts on wk2 only
https://bugs.webkit.org/show_bug.cgi?id=123354
Reviewed by Andreas Kling.
- rendering/SimpleLineLayout.cpp:
(WebCore::SimpleLineLayout::canUseFor):
Don't enable simple line layout if the primary font is loading. The code expects
to use the primary font metrics for all lines but those won't match the fallbacks
when font is not loaded.
LayoutTests: fast/frames/seamless/seamless-nested-crash.html asserts on wk2 only
https://bugs.webkit.org/show_bug.cgi?id=123354
Reviewed by Andreas Kling.
- TestExpectations: Remove failure expectation
- 8:17 AM Changeset in webkit [158084] by
-
- 2 edits in trunk/Source/WebKit2
Remove Coordinated Graphics bits from DrawingAreaProxyImpl
https://bugs.webkit.org/show_bug.cgi?id=123382
Reviewed by Anders Carlsson.
As of r156861, Coordinated Graphics does not use DrawingAreaProxyImpl.
So, Coordinated Graphics bits should be removed from DrawingAreaProxyImpl.
- UIProcess/DrawingAreaProxyImpl.h:
- 4:02 AM Changeset in webkit [158083] by
-
- 14 edits in trunk/Source/WebCore
[GTK] Deprecate public dispatch_event method in objects implementing EventTarget interface
https://bugs.webkit.org/show_bug.cgi?id=123261
Reviewed by Gustavo Noronha Silva.
The interface function should be used instead.
- bindings/scripts/CodeGeneratorGObject.pm:
(GetFunctionDeprecationInformation): Helper function to return the
version when the function was deprecated and the function
replacing the deprecated one.
(GenerateFunction): Check if the function is deprecated to mark it
as such in the header and API docs.
- bindings/scripts/gobject-generate-headers.pl: Replace the unused
WEBKIT_OBSOLETE_API macro with new macros to mark function as
deprecated, using the glib macros so that we don't need checks for
the platform.
- bindings/scripts/test/GObject/WebKitDOMTestActiveDOMObject.h:
- bindings/scripts/test/GObject/WebKitDOMTestCallback.h:
- bindings/scripts/test/GObject/WebKitDOMTestCustomNamedGetter.h:
- bindings/scripts/test/GObject/WebKitDOMTestEventConstructor.h:
- bindings/scripts/test/GObject/WebKitDOMTestEventTarget.h:
- bindings/scripts/test/GObject/WebKitDOMTestException.h:
- bindings/scripts/test/GObject/WebKitDOMTestInterface.h:
- bindings/scripts/test/GObject/WebKitDOMTestObj.h:
- bindings/scripts/test/GObject/WebKitDOMTestSerializedScriptValueInterface.h:
- bindings/scripts/test/GObject/WebKitDOMTestTypedefs.h:
- bindings/scripts/test/GObject/WebKitDOMattribute.h:
- 12:16 AM Changeset in webkit [158082] by
-
- 2 edits in trunk/Source/WebKit2
[Cocoa] clang static analyzer warns about object being autoreleased too many times in +[WKNSArray web_arrayWithImmutableArray:]
https://bugs.webkit.org/show_bug.cgi?id=123384
Reviewed by Mark Rowe.
- Shared/Cocoa/WKNSArray.h: Annotated -web_initWithImmutableArray: as an init method.
