⚠ Archived content — this site is no longer maintained.   Current WebKit documentation is at docs.webkit.org.

Timeline



Jun 5, 2006:

10:59 PM Changeset in webkit [14742] by ggaren
  • 2 edits
    1 add in trunk/JavaScriptCore

Reviewed and landed by Geoff.

  • fix a regression in ecma_3/String/regress-104375.js
  • kjs/string_object.cpp: (substituteBackreferences): If a 2-digit back reference is out of range, parse it as a 1-digit reference (followed by the other digit). This matches Firefox's behavior.
8:52 PM Changeset in webkit [14741] by sfalken
  • 1 edit in trunk/WebCore/platform/win/TemporaryLinkStubs.cpp

2006-06-06 Steve Falkenburg <​sfalken@apple.com>

Fix build break

  • platform/win/TemporaryLinkStubs.cpp: (GraphicsContext::roundToDevicePixels):
8:13 PM Changeset in webkit [14740] by ddkilzer
  • 3 edits
    8 adds in trunk

LayoutTests:

Reviewed by mjs.

testcases for ​http://bugzilla.opendarwin.org/show_bug.cgi?id=5564
'font' shorthand parsing should be more tolerant in quirks mode

  • fast/css/font_property_compatibility-expected.checksum: Added.
  • fast/css/font_property_compatibility-expected.png: Added.
  • fast/css/font_property_compatibility-expected.txt: Added.
  • fast/css/font_property_compatibility.html: Added.
  • fast/css/font_property_normal-expected.checksum: Added.
  • fast/css/font_property_normal-expected.png: Added.
  • fast/css/font_property_normal-expected.txt: Added.
  • fast/css/font_property_normal.html: Added.

WebCore:

Reviewed by mjs.

​http://bugzilla.opendarwin.org/show_bug.cgi?id=5564
'font' shorthand parsing should be more tolerant in quirks mode

When not in strict mode accept font property specifications
lacking a font family part, like WinIE does.

  • css/cssparser.cpp: (WebCore::CSSParser::parseFont):
7:27 PM Changeset in webkit [14739] by bdakin
  • 10 edits in trunk/WebCore

Reviewed by Darin.

Fix for <rdar://problem/4567520> Pixel cracks in weather widget at
1.83 scaling

To prevent pixel cracks at non-integral scaling factors, before we
call into CG to draw an image, we have to convert the rect to
device space, round the origin and size to integers in device
space, and convert back to user space.


No test cases added since this only affects non-1.0 resolution
scale factors.

  • bindings/js/JSCanvasRenderingContext2DCustom.cpp: (WebCore::JSCanvasRenderingContext2D::drawImage): drawImage() now takes FloatRects.
  • html/CanvasPattern.cpp: (WebCore::patternCallback): Call roundToDevicePixels()
  • html/CanvasRenderingContext2D.cpp: (WebCore::CanvasRenderingContext2D::drawImage): drawImage() now takes FloatRects and call roundToDevicePixels()
  • html/CanvasRenderingContext2D.h: drawImage() now takes FloatRects.
  • html/HTMLCanvasElement.cpp: (WebCore::HTMLCanvasElement::paint): Call roundToDevicePixels()
  • kcanvas/device/quartz/QuartzSupport.mm: (WebCore::debugDumpCGImageToFile): Same as above.
  • platform/GraphicsContext.h:
  • platform/cg/GraphicsContextCG.cpp: (WebCore::GraphicsContext::roundToDevicePixels): Takes care of converting between coordinate spaces and rounding. (WebCore::GraphicsContext::drawLineForText):
  • platform/mac/ImageMac.mm: Call roundToDevicePixels() (WebCore::Image::draw): Same as above. (WebCore::drawPattern): Same as above.
6:51 PM Changeset in webkit [14738] by ddkilzer
  • 2 edits in trunk/LayoutTests

LayoutTests:

Reviewed by ggaren.

  • fast/dom/global-constructors-expected.txt: Fixed test results.
5:23 PM Changeset in webkit [14737] by hyatt
  • 1 edit in trunk/WebKit/COM/WebView.cpp

Fix for perf regression with single-processor windows machines on ibench.

3:12 PM Changeset in webkit [14736] by ggaren
  • 3 edits in trunk/JavaScriptCore

Reviewed By Maciej.
Darin already reviewed this change on the branch. See <rdar://problem/4317701>.


  • Fixed <rdar://problem/4291345> PCRE overflow in Safari JavaScriptCore

No test case because there's no behavior change.


  • pcre/pcre_compile.c: (read_repeat_counts): Check for integer overflow / out of bounds
10:15 AM Changeset in webkit [14735] by ggaren
  • 12 edits
    2 adds
    2 deletes in trunk

LayoutTests:

  • Merged test-prototypes.html into global-constructors.html
  • Added prototypes.html, which tests for some missing prototypes in DOM objects


  • fast/dom/global-constructors-expected.txt:
  • fast/dom/global-constructors.html:
  • fast/dom/prototypes-expected.txt: Added.
  • fast/dom/prototypes.html: Added.
  • fast/dom/test-prototypes-expected.txt: Removed.
  • fast/dom/test-prototypes.html: Removed.
  • fast/dom/undetectable-document-all-expected.txt: Updated to reflect change from "HTMLCollection" to "Collection"

WebCore:

Reviewed by Darin.

  • Added the built-in object prototype to DOM objects that were missing it so that primitive operations like == work on them.


  • Removed dead EventConstructor class (it's autogenerated now)


  • Changed HTML collections to describe themselves as "[object Collection]", to match IE.


  • Made DOMCSSRule constructor private because JSCSSRule is the class to use.


  • bindings/js/JSHTMLOptionElementConstructor.cpp: (WebCore::JSHTMLOptionElementConstructor::JSHTMLOptionElementConstructor):
  • bindings/js/kjs_css.cpp: (KJS::DOMStyleSheet::DOMStyleSheet): (KJS::DOMStyleSheetList::DOMStyleSheetList): (KJS::DOMCSSRule::DOMCSSRule): (KJS::DOMCSSValue::DOMCSSValue): (KJS::toJS): (KJS::DOMRGBColor::DOMRGBColor): (KJS::getDOMRGBColor): (KJS::DOMRect::DOMRect):
  • bindings/js/kjs_css.h:
  • bindings/js/kjs_dom.cpp: (KJS::DOMNodeList::DOMNodeList): (KJS::DOMExceptionConstructor::DOMExceptionConstructor): (KJS::DOMExceptionConstructor::getOwnPropertySlot): (KJS::): (KJS::DOMNamedNodesCollection::DOMNamedNodesCollection):
  • bindings/js/kjs_dom.h: (KJS::DOMNamedNodesCollection::classInfo):
  • bindings/js/kjs_events.h:
  • bindings/js/kjs_html.cpp: (KJS::):
9:37 AM Changeset in webkit [14734] by ggaren
  • 4 edits in trunk/JavaScriptCore

Reviewed by aliu.

  • Changed CString length from int to size_t. We sould probably do this for UString, too. (Darin, if you're reading this: Maciej said so.)


  • kjs/function.cpp: (KJS::encode):
  • kjs/ustring.cpp: (KJS::CString::CString): (KJS::operator==):
  • kjs/ustring.h: (KJS::CString::size):

Jun 4, 2006:

10:20 PM Changeset in webkit [14733] by darin
  • 2 edits in trunk/WebCore

Reviewed by Maciej.

  • bindings/js/kjs_window.cpp: Added include of <wtf/MathExtras.h>. (WebCore::floatFeature): Remove WIN32 ifdef around isnan. (WebCore::toJS): Add null checks.
9:15 PM Changeset in webkit [14732] by darin
  • 4 edits
    2 adds in trunk

LayoutTests:

  • fast/loader/link-no-URL.html: Added.
  • fast/loader/link-no-URL-expected.txt: Added.

WebCore:

Reviewed by Maciej.

Test: fast/loader/link-no-URL.html

  • loader/Cache.h:
  • loader/Cache.cpp: (WebCore::Cache::updateCacheStatus): Removed the URL parameter. Instead get the URL from the cached object itself. Callers were passing in the wrong URL. When it was a null string, it caused the crash mentioned above. But in other cases it may have caused some other minor problems as well. Added an assert to catch cases where the URL is null (should never happen). (WebCore::Cache::requestImage): Changed call site to not pass URL. (WebCore::Cache::requestStyleSheet): Ditto. (WebCore::Cache::requestScript): Ditto. (WebCore::Cache::requestXSLStyleSheet): Ditto. (WebCore::Cache::requestXBLDocument): Ditto.
8:09 PM Changeset in webkit [14731] by ggaren
  • 2 edits in trunk/JavaScriptCore

Reviewed by Maciej.


  • kjs/value.h: Removed redundant declarations
4:05 PM Changeset in webkit [14730] by darin
  • 2 edits in trunk/WebCore

Reviewed by Adele.

  • platform/mac/WebCoreTextField.mm: (inputElement): Added. Helper function to get the Objective-C wrapper for an input element associated with a QTextEdit. Returns nil if the element is not an input element. The bug was caused by passing a non-input element to the delegate methods. (-[KWQTextFieldController controlTextDidBeginEditing:]): Use inputElement and don't make the delegate call at all if there is none. This happens only with isindex elements; it's not important to have auto-fill work with those. (-[KWQTextFieldController controlTextDidEndEditing:]): Ditto. (-[KWQTextFieldController controlTextDidChange:]): Ditto. (-[KWQTextFieldController control:textShouldEndEditing:]): Ditto. (-[KWQTextFieldController control:textView:doCommandBySelector:]): Ditto. (-[KWQTextFieldController textView:shouldHandleEvent:]): Ditto.
4:03 PM Changeset in webkit [14729] by darin
  • 36 edits in trunk/WebCore

Reviewed by Anders.

  • fix ​http://bugzilla.opendarwin.org/show_bug.cgi?id=8995 memory leak in XPath code -- seen in layout tests
  • fixed malloc error on exit due to SVGDOMImplementation destructor
  • attempted to fix the no-XPATH_SUPPORT build by adding an #if
  • changed around includes and header-file formatting in XPath code
  • bindings/scripts/CodeGeneratorJS.pm: Include PlatformString.h when generating the bindings for XPathNSResolver.
  • dom/DOMImplementation.h: Added a virtual destructor for the benefit of SVGDOMImplementation.
  • dom/DOMImplementation.cpp: (WebCore::DOMImplementation::~DOMImplementation): Added.
  • ksvg2/svg/SVGDOMImplementation.h: Reformatted a bit, removed unneeded declarations.
  • ksvg2/svg/SVGDOMImplementation.cpp: (SVGDOMImplementation::~SVGDOMImplementation): Remove commented-out code. (SVGDOMImplementation::createDocument): Remove a reference to KDOMView.
  • xpath/XPathExpression.cpp: (WebCore::XPathExpression::createExpression): Called release to avoid a tiny bit of reference count churn. (WebCore::XPathExpression::evaluate): Set exception code explicitly, because the code relies on it being 0 when the function succeeds.
  • xpath/XPathResult.cpp: (WebCore::InvalidatingEventListener::InvalidatingEventListener): Removed unneeded target node data member. (WebCore::XPathResult::XPathResult): Updated for enum name changes and InvalidatingEventListener name change. Added an assertion.
  • xpath/impl/XPathFunctions.h: Removed the FunctionLibrary class. Changed the parameter for createFunction to be a String instead of char*.
  • xpath/impl/XPathFunctions.cpp: Got rid of local implementation of "round", and used <wtf/MathExtras.h> instead. Also removed unused Interval::asString. Marked all the Interval functions inline. (WebCore::XPath::FunSubstring::doEvaluate): Use lround() instead of (long)round(). (WebCore::XPath::FunLang::doEvaluate): Removed some unneeded temporary String variables. Use equalIgnoringCase() instead of calling lower() on two strings. (WebCore::XPath::createFunctionMap): Changed how this works so that it's all local to this function. Also removed a duplicate entry for the "last" function. (WebCore::XPath::createFunction): Changed from a member function to a separate function. Also changed so that if the args are not used in the function they are deleted. This was one of the sources of the storage leaks.
  • xpath/impl/XPathGrammar.y: Put XPATH_SUPPORT ifdef in. Replaced <num> with values of specific types (numop, eqop). Updated for name change from AxisType to Axis. Changed tokens that were declared as <str> but weren't actually using the string to be declared with no type (PLUS, MINUS, OR, AND, DOTDOT, SLASHSLASH). Replaced use of unregisterString, unregisterPredicateVector, and unregisterExpressionVector to call delete functions instead that both unregister and delete. Change call sites for createFunction to just use the String, removing the deprecatedString().latin1() call that is no longer needed. Added a call to unregisterParseNode that was missing in the "FilterExpr DescendantOrSelf RelativeLocationPath" production.
  • xpath/impl/XPathParser.h: Changed Token to hold a String, Step::Axis, NumericOp::Opcode, and EqTestOp::Opcode. The token ID number tells us which to use. Rearranged to put the private part of the Parser class last. Moved a number of private things inside the cpp file.
  • xpath/impl/XPathParser.cpp: (WebCore::XPath::charCat): Made this a separate function, instead of a static member. (WebCore::XPath::isAxisName): Ditto. Also put the global axis map inside the function, allowing use of a map instead of a map pointer. (WebCore::XPath::isNodeTypeName): Ditto. (WebCore::XPath::Parser::isOperatorContext): Made this const. (WebCore::XPath::Parser::makeTokenAndAdvance): Got rid of the "int" version of this. By using the actual types, we can safely use overload instead. (WebCore::XPath::Parser::lexString): Use substring() instead of deprecatedString().mid(). (WebCore::XPath::Parser::lexNumber): Ditto. (WebCore::XPath::Parser::lexNCName): Changed to return a bool and a String instead of a Token. This avoids the hack of using a token value of "ERROR + 1" which I found quite confusing and is also a bit more explicit. (WebCore::XPath::Parser::lexQName): Ditto. (WebCore::XPath::Parser::nextTokenInternal): Changed to adapt to the changes above and to remove some unneeded else statements. (WebCore::XPath::Parser::lex): Changed the logic that sets up yylval to use the token ID to decide what the type is. This matches what the grammar expects, and removes the need for the token object to track which part of the value is good and the need to reserve 0 to have a special meaning. (WebCore::XPath::Parser::parseStatement): Added code to delete the predicates and the expressions in the predicate and expression vectors. Before, we were only deleting the vectors. This was one of the sources of the storage leaks. (WebCore::XPath::Parser::deletePredicateVector): Renamed, and changed to delete the vector. However this does not delete the predicates in the vector. (WebCore::XPath::Parser::deleteExpressionVector): Renamed, and changed to delete the vector. However this does not delete the expressions in the vector. (WebCore::XPath::Parser::deleteString): Renamed, and changed to delete the string.
  • xpath/impl/XPathPredicate.h:
  • xpath/impl/XPathPredicate.cpp: (WebCore::XPath::NumericOp::NumericOp): Changed opcode from an int to an enum, and renamed to "opcode" instead of "opCode". (WebCore::XPath::NumericOp::doEvaluate): More of the same. (WebCore::XPath::EqTestOp::EqTestOp): Ditto. (WebCore::XPath::EqTestOp::doEvaluate): Ditto. (WebCore::XPath::LogicalOp::LogicalOp): Ditto. (WebCore::XPath::LogicalOp::shortCircuitOn): Ditto. (WebCore::XPath::Predicate::evaluate): Removed an unneeded "new".
  • xpath/impl/XPathStep.h:
  • xpath/impl/XPathStep.cpp: Renamed AxisType to Axis. Removed unused axisAsString function and default constructor.
  • xpath/impl/XPathValue.h: Renamed the type constants to have a Value suffix, instead of the underscore suffix used before on NodeVector and String.
  • xpath/impl/XPathValue.cpp: Updated for name change. Added asserts.
  • xpath/impl/XPathVariableReference.cpp: (WebCore::XPath::VariableReference::doEvaluate): Use a reference instead of a pointer.
  • xpath/XPathEvaluator.cpp:
  • xpath/XPathEvaluator.h:
  • xpath/XPathExpression.h:
  • xpath/XPathNSResolver.cpp:
  • xpath/XPathNSResolver.h:
  • xpath/XPathNamespace.cpp:
  • xpath/XPathNamespace.h:
  • xpath/XPathResult.h:
  • xpath/impl/XPathExpressionNode.cpp:
  • xpath/impl/XPathExpressionNode.h:
  • xpath/impl/XPathPath.cpp:
  • xpath/impl/XPathPath.h:
  • xpath/impl/XPathUtil.cpp:
  • xpath/impl/XPathUtil.h:
  • xpath/impl/XPathVariableReference.h: Reformatted and changed includes around; no substantive changes.
3:51 PM Changeset in webkit [14728] by darin
  • 5 edits in trunk/JavaScriptCore

Reviewed by Anders.

  • changed deleteAllValues so it can work on "const" collections Deleting the values affects the values, not the pointers in the collection, so it's legitimate to do it to a const collection, and a case of that actually came up in the XPath code.
  • wtf/HashMap.h: (WTF::deleteAllPairSeconds): Use const iterators. (WTF::deleteAllValues): Take const HashMap reference as a parameter.
  • wtf/HashSet.h: (WTF::deleteAllValues): Take const HashSet reference as a parameter, and use const iterators.
  • wtf/Vector.h: (WTF::deleteAllValues): Take const Vector reference as a parameter.
  • added more functions that are present in <math.h> on some platforms, but not on others; moved here from various files in WebCore
  • wtf/MathExtras.h: (isinf): Added. (isnan): Added. (lround): Added. (lroundf): Tweaked. (round): Added. (roundf): Tweaked. (signbit): Added.
3:41 PM Changeset in webkit [14727] by ddkilzer
  • 4 edits
    32 adds in trunk

LayoutTests:

Reviewed by hyatt.

  • fast/block/positioning/absolute-in-inline-ltr-2.html: Added.
  • fast/block/positioning/absolute-in-inline-ltr-3.html: Added.
  • fast/block/positioning/absolute-in-inline-ltr.html: Added.
  • fast/block/positioning/absolute-in-inline-rtl-2.html: Added.
  • fast/block/positioning/absolute-in-inline-rtl-3.html: Added.
  • fast/block/positioning/absolute-in-inline-rtl.html: Added.
  • fast/block/positioning/absolute-in-inline-short-ltr.html: Added.
  • fast/block/positioning/absolute-in-inline-short-rtl.html: Added.

WebCore:

Reviewed by hyatt.

  • rendering/RenderBox.cpp: (WebCore::RenderBox::containingBlockWidthForPositioned): (WebCore::RenderBox::containingBlockHeightForPositioned): (WebCore::RenderBox::calcAbsoluteHorizontal): (WebCore::RenderBox::calcAbsoluteHorizontalValues): (WebCore::RenderBox::calcAbsoluteVertical): (WebCore::RenderBox::calcAbsoluteHorizontalReplaced): (WebCore::RenderBox::calcAbsoluteVerticalReplaced):
  • rendering/RenderBox.h:
2:41 PM Changeset in webkit [14726] by brmorris
  • 3 edits in S60/trunk

2006-06-04 Bradley <​bradley.morrison@nokia.com>

Reviewed by nobody (oops!)


Webkit.rsg gets processed before webkit.mmp. Fix for nonfatal:


No rule to make target `\EPOC32\INCLUDE\Webkit.rsg'

2:07 PM Changeset in webkit [14725] by brmorris
  • 22 edits
    17 adds in S60/trunk

2006-06-04 Zalan <​zalan.bujtas@nokia.com>

Reviewed by Sachin/Bradley


This commit introduces S60 3.1 support to S60WebKit's
WebKit / BrowserCore (part 3 of 3).

2:04 PM Changeset in webkit [14724] by brmorris
  • 19 edits
    15 adds in S60/trunk

2006-06-04 Zalan <​zalan.bujtas@nokia.com>

Reviewed by Sachin/Bradley


This commit introduces S60 3.1 support to S60WebKit's
WebKit / Plugin and BrowserView (part 2.8 of 3).

1:58 PM Changeset in webkit [14723] by brmorris
  • 46 edits
    11 adds in S60/trunk

2006-06-04 Zalan <​zalan.bujtas@nokia.com>

Reviewed by Sachin/Bradley


This commit introduces S60 3.1 support to S60WebKit's
WebKit / BrowserControl (part 2.6 of 3).

1:54 PM Changeset in webkit [14722] by brmorris
  • 11 edits
    1 add in S60/trunk

2006-06-04 Zalan <​zalan.bujtas@nokia.com>

Reviewed by Sachin/Bradley


This commit introduces S60 3.1 support to S60WebKit's
WebKit / BrowserControl (part 2.2 of 3).

1:37 PM Changeset in webkit [14721] by ddkilzer
  • 3 edits in trunk/WebKitSite

WebKitSite:

Reviewed by ggaren.

​http://bugzilla.opendarwin.org/show_bug.cgi?id=9303
JavaScript error when loading WebKit blog page

  • blog/wp-content/themes/webkit/sidebar.php: Removed unused JavaScript. Synced with nav.inc.
  • nav.inc: Removed evil tabs--whitespace change only.
1:12 PM Changeset in webkit [14720] by brmorris
  • 77 edits in S60/trunk

2006-06-04 Zalan <​zalan.bujtas@nokia.com>

Reviewed by Sachin/Bradley


This commit introduces S60 3.1 support to S60WebKit's
Reindeer UI and WebCore (part 2 of 3).

1:05 PM Changeset in webkit [14719] by brmorris
  • 19 edits
    2 adds in S60/trunk

2006-06-04 Zalan <​zalan.bujtas@nokia.com>

Reviewed by Sachin/Bradley


This commit introduces S60 3.1 support to S60WebKit's
JavaScriptCore, MemoryManager and S60Internals (part 1 of 3).

11:52 AM Changeset in webkit [14718] by ddkilzer
  • 4 edits in trunk/WebKitTools

WebKitTools:

Reviewed by darin.

​http://bugzilla.opendarwin.org/show_bug.cgi?id=9299
Teach svn-create-patch and friends to work with binary files

  • Scripts/svn-apply: Updated to use base64-encoded text for binary files when applying patches.
  • Scripts/svn-create-patch: Updated to include binary file content as base64-encoded text in patches.
  • Scripts/svn-unapply: Updated to recognize binary files when unapplying patches.
1:08 AM Changeset in webkit [14717] by andersca
  • 11 edits
    3 adds in trunk

WebCore:

2006-06-04 Anders Carlsson <​acarlsson@apple.com>

Reviewed by Darin.

​http://bugzilla.opendarwin.org/show_bug.cgi?id=9035
document.evaluate and createNSResolver miss on namespaced elements


  • xpath/XPathExpression.cpp: (WebCore::XPathExpression::createExpression): Pass the resolver to the parser.


  • xpath/XPathResult.cpp: (WebCore::XPathResult::iterateNext): Fix off-by-one error.


  • xpath/impl/XPathExpressionNode.h: (WebCore::XPath::EvaluationContext::EvaluationContext): Get rid of the resolver in the evaluation context.


  • xpath/impl/XPathGrammar.y: Fetch the resolver from the parser. Only return the local name.


  • xpath/impl/XPathParser.cpp: (WebCore::XPath::Parser::parseStatement):
  • xpath/impl/XPathParser.h: (WebCore::XPath::Parser::resolver): Update to hold the resolver.


  • xpath/impl/XPathStep.cpp: (WebCore::XPath::Step::Step): If the parser has stored a namespace URI, set it.


(WebCore::XPath::Step::nodeTestMatches):
Check namespace URIs.


  • xpath/impl/XPathStep.h: Add m_namespaceURI.

LayoutTests:

2006-06-04 Anders Carlsson <​acarlsson@apple.com>

Reviewed by Darin.

​http://bugzilla.opendarwin.org/show_bug.cgi?id=9035
document.evaluate and createNSResolver miss on namespaced elements

  • fast/xpath: Added.
  • fast/xpath/xpath-namespaces-expected.txt: Added.
  • fast/xpath/xpath-namespaces.html: Added.
Note: See TracTimeline for information about the timeline view.