14 years ago2008-06-07 Cameron Zwarich <cwzwarich@uwaterloo.ca>
cwzwarich@webkit.org [Sat, 7 Jun 2008 07:58:38 +0000 (07:58 +0000)]
2008-06-07  Cameron Zwarich  <cwzwarich@uwaterloo.ca>

        Reviewed by Oliver.

        Bug 17663: Wrong sequencing of argument list evaluation

        SquirrelFish fixed this bug, but we should add a test.


        * fast/js/order-of-operations-expected.txt:
        * fast/js/resources/order-of-operations.js:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34413 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by Darin.
ap@webkit.org [Sat, 7 Jun 2008 06:03:24 +0000 (06:03 +0000)]
    Reviewed by Darin.

        Combine per-thread objects into one, to make it easier to support legacy clients (for
        which they shouldn't be really per-thread).

        No change on SunSpider total.

        * JavaScriptCore.xcodeproj/project.pbxproj: Added JSGlobalData.{h,cpp}

        * kjs/JSGlobalData.cpp: Added.
        * kjs/JSGlobalData.h: Added.
        This class encapsulates all data that should be per-thread (or shared between legacy clients).
        It will also keep a Heap pointer, but right now, Heap (Collector) methods are all static.

        * kjs/identifier.h:
        Added a constructor explicitly taking JSGlobalData to access IdentifierTable. Actually,
        all of them should, but this will be a separate patch.

        * kjs/identifier.cpp:
        Combined IdentifierTable and LiteralIdentifierTable into a single class for simplicity.

        * kjs/grammar.y: kjsyyparse now takes JSGlobalData, not just a Lexer.

        * kjs/nodes.cpp:
        Changed to access Lexer and Parser via JSGlobalData::threadInstance(). This is also a
        temporary measure, they will need to use JSGlobalData explicitly.

        * VM/CodeGenerator.cpp:
        * VM/CodeGenerator.h:
        * VM/Machine.cpp:
        * kjs/CommonIdentifiers.cpp:
        * kjs/CommonIdentifiers.h:
        * kjs/DebuggerCallFrame.cpp:
        * kjs/ExecState.cpp:
        * kjs/ExecState.h:
        * kjs/InitializeThreading.cpp:
        * kjs/JSGlobalObject.cpp:
        * kjs/JSGlobalObject.h:
        * kjs/Parser.cpp:
        * kjs/Parser.h:
        * kjs/function.cpp:
        * kjs/function_object.cpp:
        * kjs/interpreter.cpp:
        * kjs/lexer.cpp:
        * kjs/lexer.h:
        * kjs/testkjs.cpp:
        Updated for the above changes. Most of threadInstance uses here will need to be replaced with
        explicitly passed pointers to support legacy JSC clients.

        * JavaScriptCore.exp: Removed KJS::parser().

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34412 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-06 Cameron Zwarich <cwzwarich@uwaterloo.ca>
cwzwarich@webkit.org [Sat, 7 Jun 2008 04:51:02 +0000 (04:51 +0000)]
2008-06-06  Cameron Zwarich  <cwzwarich@uwaterloo.ca>

        Reviewed by Oliver.

        Bug 19424: Add support for logging opcode pair counts


        * VM/Machine.cpp:
        * VM/Opcode.cpp:
        * VM/Opcode.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34411 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
jchaffraix@webkit.org [Sat, 7 Jun 2008 02:11:30 +0000 (02:11 +0000)]

2008-06-06  Julien Chaffraix  <jchaffraix@webkit.org>

        Reviewed by Eric.

        Part of bug 19200: *.in files should embed more information

        - Switched *.in files to XML format.

        - Added (simple) parsing facilities in make_names.pl using the XML::Tiny Perl parser.

        - Updated the build systems to include bindings/scripts when executing make_names.pl
        (for XMLTiny dependency).

        - Cleaned-up make_names.pl a little.

        * DerivedSources.make:
        * GNUmakefile.am:
        * WebCore.pro:
        * bindings/scripts/XMLTiny.pm: Added.
        * dom/make_names.pl:
        * html/HTMLAttributeNames.in:
        * html/HTMLTagNames.in:
        * svg/svgattrs.in:
        * svg/svgtags.in:
        * svg/xlinkattrs.in:
        * xml/xmlattrs.in:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34410 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-06 Robert Blaut <webkit@blaut.biz>
mrowe@apple.com [Fri, 6 Jun 2008 22:51:19 +0000 (22:51 +0000)]
2008-06-06  Robert Blaut  <webkit@blaut.biz>

        Reviewed by Darin Adler.

        Fix for https://bugs.webkit.org/show_bug.cgi?id=19407
        CSS3 Selector Test: *= ~= ^= $= should all accept the empty string, but match nothing

        Test: fast/css/begin-end-contain-selector-empty-value.html

        * css/CSSStyleSelector.cpp:

2008-06-06  Robert Blaut  <webkit@blaut.biz>

        Reviewed by Darin Adler.

        Test for https://bugs.webkit.org/show_bug.cgi?id=19407
        CSS3 Selector Test: *= ~= ^= $= should all accept the empty string, but match nothing.

        * fast/css/begin-end-contain-selector-empty-value.html: Added.
        * platform/mac/fast/css/begin-end-contain-selector-empty-value-expected.txt: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34409 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by mitz.
eric@webkit.org [Fri, 6 Jun 2008 22:09:40 +0000 (22:09 +0000)]
    Reviewed by mitz.

        More cleanup of CSSParser.
        Rename CSSParser::numParsedProperties and maxNumParsedProperites to m_*

        * css/CSSGrammar.y:
        * css/CSSParser.cpp:
        * css/CSSParser.h:
        * css/SVGCSSParser.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34408 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by mitz.
eric@webkit.org [Fri, 6 Jun 2008 22:03:41 +0000 (22:03 +0000)]
    Reviewed by mitz.

        More cleanup of CSSParser.
        Rename CSSParser::parsedProperties to m_parsedProperties.

        * css/CSSParser.cpp:
        * css/CSSParser.h:
        * css/SVGCSSParser.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34407 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by mitz.
eric@webkit.org [Fri, 6 Jun 2008 21:49:16 +0000 (21:49 +0000)]
    Reviewed by mitz.

        Start the cleanup of CSSParser.
        Rename CSSParser::styleElement to CSSParser::m_styleElement

        * css/CSSGrammar.y:
        * css/CSSParser.cpp:
        * css/CSSParser.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34406 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by adele. Landed by eseidel.
eric@webkit.org [Fri, 6 Jun 2008 21:07:06 +0000 (21:07 +0000)]
    Reviewed by adele.  Landed by eseidel.


        Calls listBoxOnChange() in typeAheadFind() when the <select> tag is a listBox.

        * html/HTMLSelectElement.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34405 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
justin.garcia@apple.com [Fri, 6 Jun 2008 20:39:49 +0000 (20:39 +0000)]

2008-06-05  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by John.

        <rdar://problem/5890684> REGRESSION(r23969): Deleting a line break changes the style of text after the break

        Serialization of font-family:-webkit-monospace when it came from a
        CSSMutableStyleDeclaration would wrap the value in single quotes but
        serialization of the same property value pair from a CSSComputedStyleDeclaration
        wouldn't, which kept code that checked for and removed non-redundant styles from
        style spans from working correctly.

        A redundant style left on a style span would not normally create a rendering
        difference, but it did in this case because an element with
        font-family:-webkit-monospace does not appear to inherit font-size.
        Added a FIXME about how we might need to keep font-size on a style span if its
        non-redundant styles include font-family:-webkit-monospace, although I couldn't
        create a test case to produce this scenario.

        * css/FontFamilyValue.cpp:
        (WebCore::isValidCSSIdentifier): Don't return false if the String starts
        with a '-'.  This function now appears identical to isCSSTokenizerIdentifier,
        so perhaps we should use a single function.
        * editing/ReplaceSelectionCommand.cpp:
        (WebCore::ReplaceSelectionCommand::handleStyleSpans): Added FIXME.  Removed the
        creation of two Strings that were never used.


2008-06-06  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by John.

        <rdar://problem/5890684> REGRESSION(r23969): Deleting a line break changes the style of text after the break

        * editing/deleting/5890684-expected.txt: Added.
        * editing/deleting/5890684.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34404 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
justin.garcia@apple.com [Fri, 6 Jun 2008 17:06:57 +0000 (17:06 +0000)]

2008-06-06  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Darin.

        Editing crash in IndentOutdentCommand::prepareBlockquoteLevelForInsertion, ASSERT in debug mode

        * editing/DeleteSelectionCommand.cpp:
        (WebCore::DeleteSelectionCommand::calculateTypingStyleAfterDelete): applyStyle
        can destroy insertedPlaceholder if it needs to move it, but it will set an
        endingSelection() at [movedPlaceholder, 0] if it does so.  Re-fetch insertedPlaceholder
        if that happens.  The long term solution for these crashes is to make moveParagraphs
        not copy/destroy/move content, and to make VisiblePositions update themselves on DOM
        * editing/IndentOutdentCommand.cpp:
        (WebCore::IndentOutdentCommand::indentRegion): Early return to avoid crashes like this
        in the future, but ASSERT to catch them in debug builds.


2008-06-06  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Darin.

        Editing crash in IndentOutdentCommand::prepareBlockquoteLevelForInsertion, ASSERT in debug mode

        * editing/execCommand/19087-expected.txt: Added.
        * editing/execCommand/19087.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34403 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-06 Kevin McCullough <kmccullough@apple.com>
kmccullough@apple.com [Fri, 6 Jun 2008 16:48:54 +0000 (16:48 +0000)]
2008-06-06  Kevin McCullough  <kmccullough@apple.com>

        Reviewed by Adam.

        <rdar://problem/5969992> JSProfiler: Remove the recursion limit in the
        - Change the remaining functions that do not take arguments, from using
        recursion to using iteration.

        * JavaScriptCore.exp:
        * profiler/Profile.cpp:
        (KJS::Profile::stopProfiling): Use foreach instead of recursion.
        (KJS::Profile::restoreAll): Ditto.
        * profiler/Profile.h:
        * profiler/ProfileNode.cpp: Remove recursion.
        * profiler/ProfileNode.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34402 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoLayoutTests:
jchaffraix@webkit.org [Fri, 6 Jun 2008 14:48:59 +0000 (14:48 +0000)]

2008-06-06  Julien Chaffraix  <jchaffraix@webkit.org>

        Reviewed by Alexey.

        xmlhttprequest-onProgress-open-should-zero-length.html supposes that the onprogress event
        is dispatched 3 times to test that the position is reset when XMLHttpRequest.open() is
        The test was failing because CoreFoundation keeps the data if they are not big enough
        (which led to only 1 progress event dispatched).

        We now generate a big response instead of a small one.

        Warning: this test is not guaranteed to succeed as we cannot guarantee that the server
                 will send responses' sizes that are close for 2 different requests to the same

        * http/tests/xmlhttprequest/resources/big-response.php: Added.
        * http/tests/xmlhttprequest/resources/five-tiny-chunk-one-second-delay.php: Removed.

        * http/tests/xmlhttprequest/xmlhttprequest-onProgress-open-should-zero-length-expected.txt: Added a
        warning about the possible failure of the test.
        * http/tests/xmlhttprequest/xmlhttprequest-onProgress-open-should-zero-length.html: Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34401 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoFix Greater and GreaterEq nodes to emit code in the correct order.
oliver@apple.com [Fri, 6 Jun 2008 05:09:27 +0000 (05:09 +0000)]
Fix Greater and GreaterEq nodes to emit code in the correct order.

Reviewed by Alexey.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34400 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
justin.garcia@apple.com [Fri, 6 Jun 2008 04:44:08 +0000 (04:44 +0000)]

2008-06-05  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Eric.

        ASSERT in InsertIntoTextNodeCommand::InsertIntoTextNodeCommand during RemoveFormat call

        * editing/RemoveFormatCommand.cpp:
        (WebCore::RemoveFormatCommand::doApply): If the selection was all
        formatting (like an empty list) the format-less text will be empty.
        Early return to avoid the ASSERT that fires if input(...) is called
        with an empty String.


2008-06-05  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Eric.

        ASSERT in InsertIntoTextNodeCommand::InsertIntoTextNodeCommand during RemoveFormat call

        * editing/execCommand/19403-expected.txt: Added.
        * editing/execCommand/19403.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34399 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-05 Alp Toker <alp@nuanti.com>
alp@webkit.org [Fri, 6 Jun 2008 01:43:37 +0000 (01:43 +0000)]
2008-06-05  Alp Toker  <alp@nuanti.com>

        Add 'Since' documentation for API versioning functions.

        * webkit/webkitversion.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34398 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-05 Alp Toker <alp@nuanti.com>
alp@webkit.org [Fri, 6 Jun 2008 01:09:31 +0000 (01:09 +0000)]
2008-06-05  Alp Toker  <alp@nuanti.com>

        Build fix for r34387.

        * GNUmakefile.am:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34397 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-05 Antti Koivisto <antti@apple.com>
antti@apple.com [Fri, 6 Jun 2008 01:00:16 +0000 (01:00 +0000)]
2008-06-05  Antti Koivisto  <antti@apple.com>

        Reviewed by Alp Toker.

        Fix whitespaces.

        * kjs/collector.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34396 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-05 Antti Koivisto <antti@apple.com>
antti@apple.com [Fri, 6 Jun 2008 00:50:29 +0000 (00:50 +0000)]
2008-06-05  Antti Koivisto  <antti@apple.com>

        Reviewed by Darin.

        Support compiling JavaScriptCore for ARM.

        * kjs/collector.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34395 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoOops, accidentally commited ChangeLog only.
antti@apple.com [Fri, 6 Jun 2008 00:49:37 +0000 (00:49 +0000)]
Oops, accidentally commited ChangeLog only.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34394 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-05 Antti Koivisto <antti@apple.com>
antti@apple.com [Fri, 6 Jun 2008 00:47:26 +0000 (00:47 +0000)]
2008-06-05  Antti Koivisto  <antti@apple.com>

        Reviewed by Darin.

        Support compiling JavaScriptCore for ARM.

        * kjs/collector.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34393 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-05 Kevin McCullough <kmccullough@apple.com>
kmccullough@apple.com [Fri, 6 Jun 2008 00:33:28 +0000 (00:33 +0000)]
2008-06-05  Kevin McCullough  <kmccullough@apple.com>

        Reviewed by Jon.

        - Name changes.

        * JavaScriptCore.exp:
        * profiler/Profile.cpp:
        * profiler/Profile.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34392 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-05 Geoffrey Garen <ggaren@apple.com>
ggaren@apple.com [Fri, 6 Jun 2008 00:28:57 +0000 (00:28 +0000)]
2008-06-05  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Stephanie Lewis.

        Added the -fno-move-loop-invariants flag to the pcre_exec.cpp build, to
        tell GCC not to perform loop invariant motion, since GCC's loop
        invariant motion doesn't do very well with computed goto code.

        SunSpider reports no change.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34391 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
justin.garcia@apple.com [Fri, 6 Jun 2008 00:20:36 +0000 (00:20 +0000)]

2008-06-05  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Eric.

        <https://bugs.webkit.org/show_bug.cgi?id=19089> Editing ASSERT in Range::compareBoundaryPoints

        ApplyStyleCommand::applyInlineStyle does some work and then checks to see
        if the start offset of the Range it's operating on is now past the caretMaxOffset
        of the start container.  If it is, applyInlineStyle updates the position to
        [startContainer->traverseNextNode(), 0] and then operates on that new start

        caretMaxOffset was incorrect for horizontal rules.  It should be 1 and it was 0.  So,
        if the horizontal rule was the last node in the document, applyInlineStyle would update
        its start position to [0, 0] and then try to operate on that position, assuming it was
        non-null, leading to the ASSERT.

        * rendering/RenderObject.cpp:


2008-06-05  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Eric.

        <https://bugs.webkit.org/show_bug.cgi?id=19089> Editing ASSERT in Range::compareBoundaryPoints

        * editing/execCommand/19089-expected.txt: Added.
        * editing/execCommand/19089.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34390 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago<rdar://problem/5947396> AccessibilityListBoxOptions should return their string as...
cfleizach@apple.com [Thu, 5 Jun 2008 23:19:55 +0000 (23:19 +0000)]
<rdar://problem/5947396> AccessibilityListBoxOptions should return their string as AXValue, not AXTitle

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34389 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-05 Geoffrey Garen <ggaren@apple.com>
ggaren@apple.com [Thu, 5 Jun 2008 23:13:24 +0000 (23:13 +0000)]
2008-06-05  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Stephanie Lewis.

        Added the -fno-tree-pre flag to the Machine.cpp build, to tell GCC not
        to perform Partial Redundancy Elimination (PRE) on trees in Machine.cpp,
        since GCC's PRE doesn't do very well with computed goto code.

        SunSpider reports a .7% speedup.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34388 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoPlease add a version to the Gtk port
christian@webkit.org [Thu, 5 Jun 2008 22:50:40 +0000 (22:50 +0000)]
Please add a version to the Gtk port

Reviewed by Alp Toker.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34387 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-05 Geoffrey Garen <ggaren@apple.com>
ggaren@apple.com [Thu, 5 Jun 2008 22:40:23 +0000 (22:40 +0000)]
2008-06-05  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Stephanie Lewis (or maybe the other way around).

        Minor change to PCRE to help out certain compilers.

        SunSpider reports no change, maybe a small speedup.

        * pcre/pcre_exec.cpp:
        (match): Use instructionPtr++ a little less, to avoid confusing the

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34386 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
justin.garcia@apple.com [Thu, 5 Jun 2008 22:34:19 +0000 (22:34 +0000)]

2008-06-05  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Harrison.

        <rdar://problem/5658933> GoogleDocs: Crash indenting a particular selection twice
        <rdar://problem/5831310> CrashTracer: [USER] 4 crashes in Mail at WebCore::IndentOutdentCommand::prepareBlockquoteLevelForInsertion
        <rdar://problem/5480141> GoogleDocs: Attempting to indent content in <td> closes <table> element

        A crash would occur when indenting a selection that started just before a table
        and ended inside that table.  With a selection like this, we'd indent the entire
        table, then try and indent paragraphs inside it, but the stored position for the
        next paragraph to indent was inside content that had been removed from the document
        during the first Indent.

        Fixed the crash adjusting the selection to perform the indent on to prepare it
        for paragraph iteration using selectionForParagraphIteration.  We do this adjustment
        for paragraph iteration in InsertList as well.

        Also, indenting paragraphs inside table cells would break table structure because
        we were splitting the DOM all the way up to the root editable element to insert
        a blockquote.  Fixed this by only splitting up to the enclosing table cell.  This
        matches FF.  Lastly, fixed a bug where indenting several paragraphs inside different
        table cells would move all the selected text into the same table cell. FF still has
        this bug.

        * editing/IndentOutdentCommand.cpp:


2008-06-05  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Harrison.

        <rdar://problem/5658933> GoogleDocs: Crash indenting a particular selection twice

        * editing/execCommand/5658933-1-expected.txt: Added.
        * editing/execCommand/5658933-1.html: Added.
        * editing/execCommand/5658933-2-expected.txt: Added.
        * editing/execCommand/5658933-2.html: Added.
        * editing/execCommand/5658933-3-expected.txt: Added.
        * editing/execCommand/5658933-3.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34385 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago<rdar://problem/5979998> AXList controls should have AXSelectedChildren and its child...
cfleizach@apple.com [Thu, 5 Jun 2008 22:17:56 +0000 (22:17 +0000)]
<rdar://problem/5979998> AXList controls should have AXSelectedChildren and its children's AXSelected attribute writable

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34384 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by mjs.
eric@webkit.org [Thu, 5 Jun 2008 22:01:34 +0000 (22:01 +0000)]
    Reviewed by mjs.

        Get rid of Uninitialized Memory Reads (UMRs) in the XPath::Value copy constructor.

        No functional changes, thus no tests.

        * xml/XPathValue.h:
        (WebCore::XPath::Value::Value): initialize all values in each constructor

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34383 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-04 David Hyatt <hyatt@apple.com>
hyatt@apple.com [Thu, 5 Jun 2008 20:51:32 +0000 (20:51 +0000)]
2008-06-04  David Hyatt  <hyatt@apple.com>

        Fix for <rdar://problem/5957994> Height doesn't animate

        Improve the behavior when transitions are dynamically changed.  Make sure to leave older transitions
        running and let them obsolete themselves when they finish.

        Reviewed by Dan Bernstein

        Added WebCore/manual-tests/transitions2.html (must be tested manually)

        * manual-tests/transitions.html:
        * manual-tests/transitions2.html: Added.
        * page/AnimationController.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34382 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by hyatt.
eric@webkit.org [Thu, 5 Jun 2008 20:33:42 +0000 (20:33 +0000)]
    Reviewed by hyatt.

        No functional changes, thus no tests.

        * platform/graphics/Color.h: remove unnecessary bitfield use

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34381 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Re-landing an independent part of a previously rolled out threading patch.
ap@webkit.org [Thu, 5 Jun 2008 20:03:43 +0000 (20:03 +0000)]
    Re-landing an independent part of a previously rolled out threading patch.

        * wtf/ThreadSpecific.h: Make sure to initialize POD thread-specific varaibles, too
        (replaced "new T" with "new T()").

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34380 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
antti@apple.com [Thu, 5 Jun 2008 19:56:44 +0000 (19:56 +0000)]

2008-06-05  Antti Koivisto  <antti@apple.com>

        Reviewed by Dave Hyatt.

        Make hasFeature() return true for the animation stuff.

        * dom/DOMImplementation.cpp:


2008-06-05  Antti Koivisto  <antti@apple.com>

        Reviewed by Dave Hyatt.

        Make hasFeature() return true for the animation stuff.

        * svg/custom/svg-features-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34379 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-05 Maciej Stachowiak <mjs@apple.com>
mjs@apple.com [Thu, 5 Jun 2008 17:54:59 +0000 (17:54 +0000)]
2008-06-05  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Hyatt.

        - force inlining of a template function that only has one call site per specialization
        1.3% speedup on SunSpider

        * kjs/collector.cpp:
        (KJS::Collector::heapAllocate): This template function is only
        called from allocate() and allocateNumber() (once per
        specialization) and the extra call overhead for GC allocation
        shows up, so force inlining.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34378 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
mitz@apple.com [Thu, 5 Jun 2008 16:55:13 +0000 (16:55 +0000)]

        Reviewed by Darin Adler.

        - fix https://bugs.webkit.org/show_bug.cgi?id=19397
          REGRESSION (3.1-TOT): Box-shadow renders weird when rotating a div with border radius

        Test: fast/transforms/shadows.html

        * platform/graphics/cg/GraphicsContextCG.cpp:
        (WebCore::GraphicsContext::setPlatformShadow): In Core Graphics, shadow
        dimensions are always in device space. Account for that, and apply the
        current context transform to shadow offset, and multiply the blur radius
        by the inverse of the operator norm of the inverse transform (which is
        the radius of the circle inscribed in the transformed unit circle).
        * rendering/RenderObject.cpp:
        (WebCore::RenderObject::paintBoxShadow): Added 1px separation between
        the edge of the filled shape and the edge of the clipped-out shape,
        because in a transformed context, due to antialiasing, the fill can
        bleed into the clip if they touch.


        Reviewed by Darin Adler.

        - pixel test for https://bugs.webkit.org/show_bug.cgi?id=19397
          REGRESSION (3.1-TOT): Box-shadow renders weird when rotating a div with border radius

        * fast/transforms/shadows.html: Added.
        * platform/mac/fast/transforms/shadows-expected.checksum: Added.
        * platform/mac/fast/transforms/shadows-expected.png: Added.
        * platform/mac/fast/transforms/shadows-expected.txt: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34377 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-05 Tor Arne Vestbø <tavestbo@trolltech.com>
hausmann@webkit.org [Thu, 5 Jun 2008 11:45:46 +0000 (11:45 +0000)]
2008-06-05  Tor Arne Vestbø  <tavestbo@trolltech.com>

        Reviewed by Simon

        Update layout tests on Mac based on the changes to html4.css
        in r34299. The missing backgroundClip and outlineClip in the
        new results is strange, but visually it looks the same.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34376 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-05 Tor Arne Vestbø <tavestbo@trolltech.com>
hausmann@webkit.org [Thu, 5 Jun 2008 10:36:03 +0000 (10:36 +0000)]
2008-06-05  Tor Arne Vestbø  <tavestbo@trolltech.com>

        Reviewed by Simon.

        Prevent cycles when converting from DOM objects to
        QVariant, due to cycles in the DOM tree.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34375 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-05 Maciej Stachowiak <mjs@apple.com>
mjs@apple.com [Thu, 5 Jun 2008 10:30:06 +0000 (10:30 +0000)]
2008-06-05  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Alexey and Oliver.

        - remove profiler fetch hack
        I measure an 0.5% progression from this, others show a wash. It seems not needed any more.

        * VM/Machine.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34374 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-05 Cameron Zwarich <cwzwarich@uwaterloo.ca>
cwzwarich@webkit.org [Thu, 5 Jun 2008 09:19:48 +0000 (09:19 +0000)]
2008-06-05  Cameron Zwarich  <cwzwarich@uwaterloo.ca>

        Reviewed by Maciej.

        Bug 19400: subscript operator does not protect base when necessary

        Use a temporary for the base in BracketAccessorNode if the subscript
        might possibly modify it.


        * kjs/grammar.y:
        * kjs/nodes.cpp:
        * kjs/nodes.h:


        * fast/js/codegen-temporaries-expected.txt:
        * fast/js/resources/codegen-temporaries.js:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34373 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-04 Sam Weinig <sam@webkit.org>
weinig@apple.com [Thu, 5 Jun 2008 05:36:55 +0000 (05:36 +0000)]
2008-06-04  Sam Weinig  <sam@webkit.org>

        Reviewed by Maciej Stachowiak.

        Big cleanup of formatting and whitespace.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34372 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-04 Cameron Zwarich <cwzwarich@uwaterloo.ca>
cwzwarich@webkit.org [Thu, 5 Jun 2008 04:39:09 +0000 (04:39 +0000)]
2008-06-04  Cameron Zwarich  <cwzwarich@uwaterloo.ca>

        Reviewed by Oliver.

        Add an option to dump statistics on executed instructions.


        * VM/Machine.cpp:
        * VM/Opcode.cpp:
        * VM/Opcode.h:


        * Scripts/check-for-global-initializers:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34371 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-04 Anders Carlsson <andersca@apple.com>
andersca@apple.com [Thu, 5 Jun 2008 01:03:35 +0000 (01:03 +0000)]
2008-06-04  Anders Carlsson  <andersca@apple.com>

        Reviewed by Mitz.

        Get rid of DeprecatedPtrList from StyleSheetList.

        * css/CSSStyleSelector.cpp:
        * css/StyleSheetList.cpp:
        * css/StyleSheetList.h:
        * dom/Document.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34370 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-04 Anders Carlsson <andersca@apple.com>
andersca@apple.com [Thu, 5 Jun 2008 00:31:18 +0000 (00:31 +0000)]
2008-06-04  Anders Carlsson  <andersca@apple.com>

        Reviewed by Darin.

        Convert more classes over to start out with a ref count of 1.

        * css/CSSImageGeneratorValue.cpp:
        * css/CSSImageValue.cpp:
        * css/CSSMediaRule.cpp:
        * css/CSSParser.cpp:
        * css/CSSRuleList.cpp:
        * css/CSSRuleList.h:
        * css/CSSStyleSelector.cpp:
        * css/CSSStyleSelector.h:
        * css/CSSStyleSheet.cpp:
        * css/CSSStyleSheet.h:
        * css/Pair.h:
        * html/HTMLMediaElement.cpp:
        * html/TimeRanges.cpp:
        * html/TimeRanges.h:
        * rendering/style/RenderStyle.cpp:
        * rendering/style/RenderStyle.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34369 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago<rdar://problem/5983804> Editable web areas do not properly report the AXFocused...
cfleizach@apple.com [Wed, 4 Jun 2008 23:45:57 +0000 (23:45 +0000)]
<rdar://problem/5983804> Editable web areas do not properly report the AXFocused attribute

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34368 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-03 Beth Dakin <bdakin@apple.com>
bdakin@apple.com [Wed, 4 Jun 2008 22:04:32 +0000 (22:04 +0000)]
2008-06-03  Beth Dakin  <bdakin@apple.com>

        Reviewed by Adele.

        Fix for https://bugs.webkit.org/show_bug.cgi?id=19381 WebKit
        Crashing Constantly on Gmail with accessibility turned on

        There was a missing null-check!

        * page/AccessibilityRenderObject.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34367 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-04 Anders Carlsson <andersca@apple.com>
andersca@apple.com [Wed, 4 Jun 2008 21:12:16 +0000 (21:12 +0000)]
2008-06-04  Anders Carlsson  <andersca@apple.com>

        Reviewed by Sam.

        Convert more classes to start with a ref count of 1.

        * bindings/js/JSCustomVoidCallback.cpp:
        * bindings/js/JSCustomVoidCallback.h:
        * css/StyleSheetList.cpp:
        * css/StyleSheetList.h:
        * dom/Document.cpp:
        * dom/NodeIterator.h:
        * dom/Traversal.cpp:
        * dom/Traversal.h:
        * dom/TreeWalker.h:
        * history/CachedPage.cpp:
        * html/CanvasGradient.cpp:
        * html/CanvasGradient.h:
        * html/CanvasRenderingContext2D.cpp:
        * html/CanvasRenderingContext2D.h:
        * html/HTMLCanvasElement.cpp:
        * html/VoidCallback.h:
        * loader/FormState.cpp:
        * platform/text/RegularExpression.cpp:
        * storage/SQLResultSet.cpp:
        * storage/SQLResultSetRowList.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34366 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
mitz@apple.com [Wed, 4 Jun 2008 20:47:24 +0000 (20:47 +0000)]

        Reviewed by Dave Hyatt.

        - fix <rdar://problem/5962118> Crash in RenderBlock::calcColumnWidth()

        Test: fast/multicol/gap-non-negative.html

        * css/CSSParser.cpp:
        (WebCore::CSSParser::parseValue): Changed to not allow negative
        values for column-gap, per


        Reviewed by Dave Hyatt.

        - test for <rdar://problem/5962118> Crash in RenderBlock::calcColumnWidth()

        * fast/multicol/gap-non-negative-expected.txt: Added.
        * fast/multicol/gap-non-negative.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34365 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
mitz@apple.com [Wed, 4 Jun 2008 20:45:46 +0000 (20:45 +0000)]

        Reviewed by Dave Hyatt.

        - fix <rdar://problem/5962270> Crash in WebCore::RenderTableCell::computeAbsoluteRepaintRect()

        Test: fast/reflections/table-cell.html

        * rendering/RenderTableCell.cpp:
        (WebCore::RenderTableCell::computeAbsoluteRepaintRect): Null-check
        (WebCore::RenderTableCell::absolutePosition): Ditto.


        Reviewed by Dave Hyatt.

        - test for <rdar://problem/5962270> Crash in WebCore::RenderTableCell::computeAbsoluteRepaintRect()

        * fast/reflections/table-cell.html: Added.
        * platform/mac/fast/reflections/table-cell-expected.checksum: Added.
        * platform/mac/fast/reflections/table-cell-expected.png: Added.
        * platform/mac/fast/reflections/table-cell-expected.txt: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34364 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoAdded index.html files to specification directories so
dino@apple.com [Wed, 4 Jun 2008 20:00:48 +0000 (20:00 +0000)]
Added index.html files to specification directories so
things are easier to find.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34363 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-04 Kevin McCullough <kmccullough@apple.com>
kmccullough@apple.com [Wed, 4 Jun 2008 18:53:10 +0000 (18:53 +0000)]
2008-06-04  Kevin McCullough  <kmccullough@apple.com>

        Reviewed by Geoff.

        <rdar://problem/5969992> JSProfiler: Remove the recursion limit in the
        - This patch removes the use of recursion for the sort functions.

        * JavaScriptCore.exp: Change the signatures of the functions being
        * profiler/Profile.cpp:
        (KJS::Profile::sort): This generic function will accept any of the
        static sort functions and apply them to the whole tree.
        * profiler/Profile.h: All of the sorting functions now call the new
        sort() function.
        * profiler/ProfileNode.cpp:
        (KJS::ProfileNode::ProfileNode): m_head used to point to the head node
        if this was the head node.  It now points to null to make iteration easy
        (KJS::ProfileNode::willExecute): Now must check if m_head is null, this
        check used to happend in the constructor.
        (KJS::ProfileNode::stopProfiling): Again the check is slightly different
        to determine if this is the head.
        (KJS::ProfileNode::traverseNextNode): This function returns the next
        node in post order.
        (KJS::ProfileNode::sort): This generic function will sort according to
        the comparator passed in, then reset the children pointers to macth the
        new order.
        * profiler/ProfileNode.h: The sorting function were removed from the
        definition file and instead use the new generic sort() function
        (KJS::ProfileNode::totalPercent): because the head can now be empty we
        need to check here too for the head node.
        (KJS::ProfileNode::selfPercent): Ditto
        (KJS::ProfileNode::firstChild): This function is necessary for the
        iterative algorithm in Profile.cpp.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34362 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by Darin.
ap@webkit.org [Wed, 4 Jun 2008 18:10:15 +0000 (18:10 +0000)]
    Reviewed by Darin.

        Fix JSClassCreate to work with old JSCore API threading model.

        No change on SunSpider.

        * API/JSClassRef.cpp: (OpaqueJSClass::OpaqueJSClass): Since JSClass is constructed without
        a context, there is no way for it to create Identifiers.
        Also, added initializeThreading(), just for good measure.

        * API/JSCallbackObjectFunctions.h: (KJS::::getPropertyNames): Make an Identifier out of the
        string here, because propertyNames.add() needs that.

        * kjs/identifier.cpp:
        * kjs/identifier.h:
        * kjs/ustring.cpp:
        Moved equal() from identifier.h to ustring.h, because it's not really about Identifiers,
        and to make it possible to use it from StrHash.
        Include StrHash.h from ustring.h to avoid having the behavior depend on headers that happen
        to be included.

        * wtf/StrHash.h: Removed.
        * kjs/ustring.h: Made RefPtr<UString::Rep> use the same default hash as UString::Rep* (it
        used to default to pointer equality). Moved the whole StrHash header into ustring.h.

        * JavaScriptCore.exp: Export equal() for WebCore use (this StrHash is used in c_class.cpp,
        jni_class.cpp, and npruntime.cpp).

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34361 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Rubber-stamped by Darin.
ap@webkit.org [Wed, 4 Jun 2008 16:29:49 +0000 (16:29 +0000)]
    Rubber-stamped by Darin.

        Fix spacing in collector.{h,cpp}.

        * kjs/collector.cpp:
        * kjs/collector.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34360 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-04 Tor Arne Vestbø <tavestbo@trolltech.com>
hausmann@webkit.org [Wed, 4 Jun 2008 13:40:54 +0000 (13:40 +0000)]
2008-06-04  Tor Arne Vestbø  <tavestbo@trolltech.com>

        Reviewed by Simon.

        Fix a failing assertion when calling QWebFrame::evaluateJavaScript.

        The starting line number has to be 1 instead of 0.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34359 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
antti@apple.com [Wed, 4 Jun 2008 02:57:46 +0000 (02:57 +0000)]

2008-06-03  Antti Koivisto  <antti@apple.com>

        Reviewed by Darin.

        Implement path morphing for SVG animation

        You can animate smoothly between paths that have equal number of control points of matching type.

        Test: svg/custom/animate-path-morphing.svg

        * svg/SVGAnimateElement.cpp:
        * svg/SVGAnimateElement.h:
        * svg/SVGPathSegList.cpp:
        * svg/SVGPathSegList.h:


2008-06-03  Antti Koivisto  <antti@apple.com>

        Reviewed by Darin.


        Test for path morphing.

        * platform/mac/svg/custom/animate-path-morphing-expected.txt: Added.
        * svg/custom/animate-path-morphing.svg: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34358 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-03 Darin Adler <darin@apple.com>
darin@apple.com [Wed, 4 Jun 2008 02:09:09 +0000 (02:09 +0000)]
2008-06-03  Darin Adler  <darin@apple.com>

        - try to fix the Qt build

        * bridge/qt/qt_runtime.cpp:
        (KJS::Bindings::QtRuntimeMetaMethod::lengthGetter): Remove the originalObject argument.
        (KJS::Bindings::QtRuntimeMetaMethod::connectGetter): Ditto.
        (KJS::Bindings::QtRuntimeMetaMethod::disconnectGetter): Ditto.
        (KJS::Bindings::QtRuntimeConnectionMethod::lengthGetter): Ditto.
        (KJS::Bindings::QtConnectionObject::execute): Use callAsFunction.
        * bridge/qt/qt_runtime.h: More of the same.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34357 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-03 Cameron Zwarich <cwzwarich@uwaterloo.ca>
cwzwarich@webkit.org [Wed, 4 Jun 2008 01:03:37 +0000 (01:03 +0000)]
2008-06-03  Cameron Zwarich  <cwzwarich@uwaterloo.ca>

        Reviewed by Maciej.

        Build fix. The cleanup in r34355 missed a method.

        * kjs/nodes.cpp:
        * kjs/nodes.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34356 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoJavaScriptCore:
darin@apple.com [Wed, 4 Jun 2008 00:40:47 +0000 (00:40 +0000)]

2008-06-03  Darin Adler  <darin@apple.com>

        Reviewed by Geoff.

        - https://bugs.webkit.org/show_bug.cgi?id=19269
          speed up SunSpider by eliminating the toObject call for most get/put/delete

        Makes standalone SunSpider 1.025x as fast as before.

        The getOwnPropertySlot virtual function now takes care of the toObject call
        for get. Similarly, the put function (and later deleteProperty) does the
        same for those operations. To do this, the virtual functions were moved from
        the JSObject class to the JSCell class. Also, since the caller no longer knows
        the identity of the "original object", which is used by JavaScript-function
        based getters, changed the PropertySlot class so the original object is
        already stored in the slot when getOwnPropertySlot is called, if the caller
        intends to call getValue.

        This affected the old interpreter code enough that the easiest thing for me
        was to just delete it. While I am not certain the mysterious slowdown is not
        still occurring, the net change is definitely a significant speedup.

        * JavaScriptCore.exp: Updated.

        * VM/Machine.cpp: Moved the UNLIKELY macro into AlwaysInline.h.
        (KJS::resolve): Set up the originalObject in the PropertySlot before
        calling getPropertySlot. Also removed the originalObject argument from
        (KJS::resolve_skip): Ditto.
        (KJS::resolveBaseAndProperty): Ditto.
        (KJS::resolveBaseAndFunc): Ditto.
        (KJS::Machine::privateExecute): Removed the toObject calls from the get and
        put functions where possible, instead calling directly with JSValue and letting
        the JSValue and JSCell calls handle toObject. Same for toThisObject.

        * kjs/ExecState.h: Removed OldInterpreterExecState.

        * API/JSBase.cpp: Updated includes.

        * kjs/LocalStorageEntry.h: Removed contents. Later we can remove the file too.

        * kjs/array_instance.cpp:
        (KJS::ArrayInstance::lengthGetter): Removed originalObject argumet.
        (KJS::ArrayInstance::inlineGetOwnPropertySlot): Don't pass a base value to
        setValueSlot. Also use UNLIKELY around the "getting elements past the end of
        the array" code path; less common than successfully getting an element.

        * kjs/array_object.cpp:
        (KJS::getProperty): Initialize the PropertySlot with the original object.
        Don't pass the original object to the get function.
        (KJS::arrayProtoFuncFilter): Ditto.
        (KJS::arrayProtoFuncMap): Ditto.
        (KJS::arrayProtoFuncEvery): Ditto.
        (KJS::arrayProtoFuncForEach): Ditto.
        (KJS::arrayProtoFuncSome): Ditto.

        * kjs/function_object.cpp:
        (KJS::FunctionObjectImp::construct): Removed an obsolete comment.

        * kjs/grammar.y: Eliminated support for some of the node types that were
        used to optimize executing from the syntax tree.

        * kjs/internal.cpp:
        (KJS::StringImp::toThisObject): Added. Same as toObject.
        (KJS::NumberImp::toThisObject): Ditto.
        (KJS::GetterSetterImp::getOwnPropertySlot): Added. Not reached.
        (KJS::GetterSetterImp::put): Ditto.
        (KJS::GetterSetterImp::toThisObject): Ditto.

        * kjs/internal.h: Added toThisObject to NumberImp for speed.

        * kjs/lexer.cpp:
        (KJS::Lexer::shift): Changed shift to just do a single character, to unroll
        the loop and especially to make the one character case faster.
        (KJS::Lexer::setCode): Call shift multiple times instead of passing a number.
        (KJS::Lexer::lex): Ditto.
        (KJS::Lexer::matchPunctuator): Ditto. Also removed unneeded elses after returns.
        (KJS::Lexer::scanRegExp): Ditto.
        * kjs/lexer.h: Removed the count argument from shift.

        * kjs/math_object.cpp:
        (KJS::mathProtoFuncPow): Call jsNaN instead of jsNumber(NaN).

        * kjs/nodes.cpp: Removed some of the things needed only for the pre-SquirrelFish
        execution model.
        (KJS::ForNode::emitCode): Handle cases where some expressions are missing by
        not emitting any code at all. The old way was to emit code for "true", but
        this is an unnecessary remnant of the old way of doing things.

        * kjs/nodes.h: Removed some of the things needed only for the pre-SquirrelFish
        execution model.

        * kjs/object.cpp:
        (KJS::JSObject::fillGetterPropertySlot): Changed to only pass in the getter
        function. The old code passed in a base, but it was never used when
        actually getting the property; the toThisObject call was pointless. Also
        changed to not pass a base for setUndefined.

        * kjs/object.h: Added the new JSCell operations to GetterSetterImp.
        Never called.
        (KJS::JSObject::get): Initialize the object in the PropertySlot and don't
        pass it in getValue.
        (KJS::JSObject::getOwnPropertySlotForWrite): Removed the base argument
        in calls to setValueSlot.
        (KJS::JSObject::getOwnPropertySlot): Ditto.
        (KJS::JSValue::get): Added. Here because it calls through to JSObject.
        A version of JSObject::get that also handles the other types of JSValue
        by creating the appropriate wrapper. Saves the virtual call to toObject.
        (KJS::JSValue::put): Ditto.
        (KJS::JSValue::deleteProperty): Ditto.

        * kjs/property_slot.cpp:
        (KJS::PropertySlot::undefinedGetter): Removed the originalObject argument.
        (KJS::PropertySlot::ungettableGetter): Ditto.
        (KJS::PropertySlot::functionGetter): Ditto. Use the value in the base
        as the "this" object, which will be set to the original object by the new
        PropertySlot initialization code. Also call toThisObject. The old code did
        not do this, but needed to so we can properly handle the activation object
        like the other similar code paths.

        * kjs/property_slot.h:
        (KJS::PropertySlot::PropertySlot): Added a constructor that takes a base
        object. In debug builds, set the base to 0 if you don't pass one.
        (KJS::PropertySlot::getValue): Don't take or pass the originalObject.
        (KJS::PropertySlot::setValueSlot): Don't take a base object, and clear the
        base object in debug builds.
        (KJS::PropertySlot::setGetterSlot): Ditto.
        (KJS::PropertySlot::setUndefined): Ditto.
        (KJS::PropertySlot::setUngettable): Ditto.
        (KJS::PropertySlot::slotBase): Assert that a base object is present.
        This will fire if someone actually calls the get function without having
        passed in a base object and the getter needs it.
        (KJS::PropertySlot::setBase): Added. Used by the code that implements
        toObject so it can supply the original object after the fact.
        (KJS::PropertySlot::clearBase): Added. Clears the base, but is debug-only
        code because it's an error to fetch the base if you don't have a guarantee
        it was set.

        * API/JSCallbackObject.h:
        * API/JSCallbackObjectFunctions.h:
        * kjs/JSActivation.cpp:
        * kjs/JSActivation.h:
        * kjs/JSVariableObject.h:
        * kjs/array_instance.h:
        * kjs/function.cpp:
        * kjs/function.h:
        * kjs/lookup.h:
        * kjs/string_object.cpp:
        * kjs/string_object.h:
        Removed originalObject arguments from getters. Don't pass base values to
        the various PropertySlot functions that no longer take them.

        * kjs/value.cpp:
        (KJS::JSCell::getOwnPropertySlot): Added. Calls toObject and then sets the slot.
        This function has to always return true, because the caller can't walk the prototype
        chain. Because of that, we do a getPropertySlot, not getOwnPropertySlot, which works
        for the caller. This is private, only called by getOwnPropertySlotInternal.
        (KJS::JSCell::put): Added. Calls toObject and then put.
        (KJS::JSCell::toThisObject): Added. Calls toObject.

        * kjs/value.h: Added get, put, and toThisObject to both JSValue
        and JSCell. These take care of the toObject operation without an additional virtual
        function call, and so make the common "already an object" case faster.

        * wtf/AlwaysInline.h: Moved the UNLIKELY macro here for now. Maybe we can find a
        better place later, or rename this header.


2008-06-03  Darin Adler  <darin@apple.com>

        - update for JavaScriptCore changes for https://bugs.webkit.org/show_bug.cgi?id=19269
          speed up SunSpider by eliminating the toObject call for most get/put/delete

        * UserObjectImp.cpp:
        (UserObjectImp::userObjectGetter): Removed originalObject argument.
        * UserObjectImp.h: Ditto.


2008-06-03  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by John.

        <rdar://problem/5763082> GMail: Hang when removing indent from nested list
        <rdar://problem/5775449> In Gmail and GoogleDocs, a hang occurs when I attempt to apply a list style to a large selection of text
        <rdar://problem/5937624> 9D32: Hang in Safari. Using 100% of processor

        * editing/InsertListCommand.cpp:
        (WebCore::InsertListCommand::modifyRange): doApply() may operate on and remove
        the last paragraph of the selection from the document if it's in the same list
        item as startOfCurrentParagraph.  Return early to avoid an infinite loop and
        because there is no more work to be done.  Added a FIXME (<rdar://problem/5983974>)
        about the incorrect endingSelection()s.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34355 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
justin.garcia@apple.com [Wed, 4 Jun 2008 00:36:28 +0000 (00:36 +0000)]

2008-06-03  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by John.

        <rdar://problem/5763082> GMail: Hang when removing indent from nested list
        <rdar://problem/5775449> In Gmail and GoogleDocs, a hang occurs when I attempt to apply a list style to a large selection of text
        <rdar://problem/5937624> 9D32: Hang in Safari. Using 100% of processor

        * editing/InsertListCommand.cpp:
        (WebCore::InsertListCommand::modifyRange): doApply() may operate on and remove
        the last paragraph of the selection from the document if it's in the same list
        item as startOfCurrentParagraph.  Return early to avoid an infinite loop and
        because there is no more work to be done.  Added a FIXME (<rdar://problem/5983974>)
        about the incorrect endingSelection()s.


2008-06-03  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by John.

        <rdar://problem/5763082> GMail: Hang when removing indent from nested list

        * editing/execCommand/5763082-expected.txt: Added.
        * editing/execCommand/5763082.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34354 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoBug 12983: Web Inspector break on the debugger keyword
oliver@apple.com [Tue, 3 Jun 2008 22:48:52 +0000 (22:48 +0000)]
Bug 12983: Web Inspector break on the debugger keyword

Reviewed by Tim

Added a DebuggerStatementNode to handle codegen, and added a new
DidReachBreakPoint debug event (which will hopefully be useful
if we ever move breakpoint management into JSC proper).  Also
added didReachBreakpoint to Debugger to allow us to actually respond
to this event.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34351 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-03 Darin Adler <darin@apple.com>
darin@apple.com [Tue, 3 Jun 2008 21:43:44 +0000 (21:43 +0000)]
2008-06-03  Darin Adler  <darin@apple.com>

        Fix build with GCC.

        * page/AccessibilityRenderObject.cpp:
        Put a default case back in the switch statement so we don't get warned about
        all the unhandled cases with GCC. Remove the unreachable code that was causing
        a warning in Visual Studio.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34349 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-03 Jonathon Jongsma <jonathon.jongsma@collabora.co.uk>
darin@apple.com [Tue, 3 Jun 2008 21:01:22 +0000 (21:01 +0000)]
2008-06-03  Jonathon Jongsma  <jonathon.jongsma@collabora.co.uk>

        Reviewed by Darin.

        * Api/qwebhistoryinterface.cpp: Remove WebCore::historyContains().  This
        function is not used anywhere internally and is only a convenience
        function that can still be accomplished using

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34348 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-03 Anders Carlsson <andersca@apple.com>
andersca@apple.com [Tue, 3 Jun 2008 19:25:59 +0000 (19:25 +0000)]
2008-06-03  Anders Carlsson  <andersca@apple.com>

        Reviewed by Darin.


        In 64-bit Web Kit, converting between float and double, can cause rounding errors which in turn causes
        newBottom to be larger than oldBottom which is illegal.

        * WebView/WebHTMLView.mm:
        (-[WebHTMLView adjustPageHeightNew:top:bottom:limit:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34347 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-30 Steve Falkenburg <sfalken@apple.com>
sfalken@apple.com [Tue, 3 Jun 2008 18:46:16 +0000 (18:46 +0000)]
2008-05-30  Steve Falkenburg  <sfalken@apple.com>

        Registry-free COM.

        WebKit no longer requires COM registration.
        Registry-free COM information is read from the application manifest.

        Only 1 set of CLSIDs are now used, since we don't need to worry about registry collisions.
        The second set remains, but only temporarily so nightlies continue to work.

        This is supported for XPSP2 and later.  Earlier systems require an installer
        to write the required registry keys.  Nightlies and developer builds require
        XPSP2, Server 2003, or later.

        Reviewed by Adam.

        * ForEachCoClass.cpp: Moved COM registration code here from WebKitDLL.cpp. This is ONLY used in the nightly case.
        (substituteGUID): Moved here from WebKitDLL.cpp.
        (registerWebKit): Moved here from WebKitDLL.cpp.
        (setUseOpenSourceWebKit): Call registerWebKit instead of swapping progids. There is only 1 set of progids now - for the nightly.
        * ForEachCoClass.h: Removed production progids.
        * Interfaces/WebKit.idl: Added prototype for shutDownWebKit.
        * WebKit.vcproj/Interfaces.vcproj: Changed path to type library to move it out of project directory.
        * WebKit.vcproj/WebKit.vcproj: Changed path to type library to move it out of project directory.
        * WebKitDLL.cpp:
        (DllUnregisterServer): Does nothing. Entry point is still present for backward compatibility.
        (DllRegisterServer): Does nothing. Entry point is still present for backward compatibility.
        (RunAsLocalServer): Does nothing. Entry point is still present for backward compatibility.
        (shutDownWebKit): Moved from ForEachCoClass.
        * WebKitDLL.h:

2008-05-30  Steve Falkenburg  <sfalken@apple.com>

        Generate an isolated COM manifest for registry free COM.

        * DumpRenderTree/win/DumpRenderTree.vcproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34346 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoFix build.
sfalken@apple.com [Tue, 3 Jun 2008 18:29:03 +0000 (18:29 +0000)]
Fix build.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34345 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-03 Justin Garcia <justin.garcia@apple.com>
justin.garcia@apple.com [Tue, 3 Jun 2008 17:12:37 +0000 (17:12 +0000)]
2008-06-03  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by John Sullivan.

        <rdar://problem/5943515> SelectionController()->modify Backward/Forward with LineGranularity sometimes don't work

        next/previousLinePosition didn't work if the input position was the last line in its
        block.  The code for handling such a position assumed that the input position was
        editable, and that the position to be returned needed to also be editable.  Changed
        this code to just maintain the same editability.

        No test case for now because we haven't enabled selection extension operations for
        non-editable selections.  I enabled them temporarily to test the fix, however.

        * dom/Node.cpp: Removed the now unused nextEditable(int).
        * dom/Node.h:
        * editing/visible_units.cpp:
        (WebCore::previousLeafWithSameEditability): Added.  Like previousEditable but just returns
        a node of the same editability.
        (WebCore::previousLinePosition): Call previousLeaf instead of previousEditable.
        (WebCore::nextLeafWithSameEditability): Added.
        (WebCore::nextLinePosition): Same as above.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34344 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
andersca@apple.com [Tue, 3 Jun 2008 17:07:58 +0000 (17:07 +0000)]

2008-06-02  Anders Carlsson  <andersca@apple.com>

        Reviewed by John and Sam.

        Assertion failure in ApplicationCacheGroup::addEntry when reloading a page with a cache manifest (19182)

        * loader/appcache/ApplicationCacheGroup.h:
        Add new accessor.

        * loader/appcache/ApplicationCacheStorage.cpp:
        Don't call newestCache since it will always be null. Instead, call hasNewestCache which returns the saved newest cache pointer.


2008-06-03  Anders Carlsson  <andersca@apple.com>

        Reviewed by John and Sam.

        Assertion failure in ApplicationCacheGroup::addEntry when reloading a page with a cache manifest (19182)

        * http/tests/appcache/crash-when-navigating-away-then-back-expected.txt: Added.
        * http/tests/appcache/crash-when-navigating-away-then-back.html: Added.
        * http/tests/appcache/resources/crash-when-navigating-away-then-back.html: Added.
        * http/tests/appcache/resources/crash-when-navigating-away-then-back.manifest: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34343 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Rubber-stamped by Mark Rowe.
ap@webkit.org [Tue, 3 Jun 2008 15:56:01 +0000 (15:56 +0000)]
    Rubber-stamped by Mark Rowe.

        Roll out r31215 - with Mac OS X 10.5.3, there is no need to pause update daemon any more.

        * sunspider:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34342 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
mitz@apple.com [Tue, 3 Jun 2008 15:29:14 +0000 (15:29 +0000)]

        Reviewed by Dave Hyatt.

        - fix <rdar://problem/5965410>
          In fixed table layout, the width of cells with auto width should be
          proportional to their column span, to match Firefox and IE.

        Test: fast/table/fixed-with-auto-with-colspan.html

        * rendering/FixedTableLayout.cpp:


        Reviewed by Dave Hyatt.

        - test and updated results for <rdar://problem/5965410>
          In fixed table layout, the width of cells with auto width should be
          proportional to their column span, to match Firefox and IE.

        * fast/table/fixed-with-auto-with-colspan.html: Added.
        * platform/mac/fast/table/fixed-with-auto-with-colspan-expected.checksum: Added.
        * platform/mac/fast/table/fixed-with-auto-with-colspan-expected.png: Added.
        * platform/mac/fast/table/fixed-with-auto-with-colspan-expected.txt: Added.
        * platform/mac/tables/mozilla_expected_failures/bugs/bug220653-expected.checksum:
        * platform/mac/tables/mozilla_expected_failures/bugs/bug220653-expected.png:
        * platform/mac/tables/mozilla_expected_failures/bugs/bug220653-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34341 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-03 Alp Toker <alp@nuanti.com>
alp@webkit.org [Tue, 3 Jun 2008 09:01:58 +0000 (09:01 +0000)]
2008-06-03  Alp Toker  <alp@nuanti.com>

        Rubber-stamped by Alexey.

        Remove GenerateNativeConverter added in r34331 to fix JS regressions.
        (JSTextCustom.cpp added in r34141 is now unused in any code path.)

        * dom/Text.idl:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34340 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-03 Maciej Stachowiak <mjs@apple.com>
mjs@apple.com [Tue, 3 Jun 2008 08:47:23 +0000 (08:47 +0000)]
2008-06-03  Maciej Stachowiak  <mjs@apple.com>

        Add a snapshot of the squirrelfish bytecode docs.

        * specs/squirrelfish-bytecode.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34339 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-03 Maciej Stachowiak <mjs@apple.com>
mjs@apple.com [Tue, 3 Jun 2008 08:44:55 +0000 (08:44 +0000)]
2008-06-03  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Oliver.

        - document remaining opcodes.

        * VM/Machine.cpp:
        (KJS::Machine::privateExecute): Document call, call_eval,
        construct, ret and end opcodes.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34338 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago <rdar://problem/5621979> editing/selection/5241148.html crashes on Windows
ap@webkit.org [Tue, 3 Jun 2008 08:42:42 +0000 (08:42 +0000)]
    <rdar://problem/5621979> editing/selection/5241148.html crashes on Windows

        Try removing it from the Skipped list, as this is not reproducible with current closed
        source libraries.

        * platform/win/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34337 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-03 Alexander Vassilev <avasilev@voipgate.com>
ap@webkit.org [Tue, 3 Jun 2008 07:56:36 +0000 (07:56 +0000)]
2008-06-03  Alexander Vassilev <avasilev@voipgate.com>

        Reviewed, tweaked and landed by Alexey.

        Compiler error in Entity.h

        * dom/Entity.h:
        (WebCore::Entity::Entity): Add a private constructor to shut down the errors.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34336 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-03 Maciej Stachowiak <mjs@apple.com>
mjs@apple.com [Tue, 3 Jun 2008 07:46:41 +0000 (07:46 +0000)]
2008-06-03  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Oliver.

        * VM/Machine.cpp:
        (KJS::Machine::privateExecute): Document throw and catch opcodes.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34335 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoJavaScriptCore:
ggaren@apple.com [Tue, 3 Jun 2008 06:26:54 +0000 (06:26 +0000)]

2008-06-02  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Alexey Proskuryakov.

        Removed JSObject::call, since it just called JSObject::callAsFunction.

        SunSpider reports no change.


2008-06-02  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Alexey Proskuryakov.

        Removed JSObject::call, since it just called JSObject::callAsFunction.

        SunSpider reports no change.


2008-06-02  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Alexey Proskuryakov.

        Removed JSObject::call, since it just called JSObject::callAsFunction.

        SunSpider reports no change.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34334 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 Maciej Stachowiak <mjs@apple.com>
mjs@apple.com [Tue, 3 Jun 2008 03:48:18 +0000 (03:48 +0000)]
2008-06-02  Maciej Stachowiak  <mjs@apple.com>

        * config.ini: Correct URL for just-added blog.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34333 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 Maciej Stachowiak <mjs@apple.com>
mjs@apple.com [Tue, 3 Jun 2008 03:15:11 +0000 (03:15 +0000)]
2008-06-02  Maciej Stachowiak  <mjs@apple.com>

        Rubber stamped by Mark.

        * config.ini: Add Camerom's Summer of JavaScriptCore blog.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34332 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 Alp Toker <alp@nuanti.com>
alp@webkit.org [Tue, 3 Jun 2008 02:17:15 +0000 (02:17 +0000)]
2008-06-02  Alp Toker  <alp@nuanti.com>

        Reviewed by Maciej.

        Two of the toJSNewlyCreated() optimisations in r34141 were never
        reached due to a generator bug and missing IDL attribute.

        Fix header generation to provide the necessary toJSNewlyCreated()

        Patch includes a related build fix by Jan Michael Alonzo.

        * GNUmakefile.am:
        * bindings/scripts/CodeGeneratorJS.pm:
        * dom/Text.idl:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34331 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 Geoffrey Garen <ggaren@apple.com>
ggaren@apple.com [Tue, 3 Jun 2008 01:05:12 +0000 (01:05 +0000)]
2008-06-02  Geoffrey Garen  <ggaren@apple.com>

        Transparent background!

        * blog-files/squirrelfish.png:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34330 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 Alp Toker <alp@nuanti.com>
alp@webkit.org [Tue, 3 Jun 2008 00:30:19 +0000 (00:30 +0000)]
2008-06-02  Alp Toker  <alp@nuanti.com>

        GTK+/autotools Windows build system fixes.

        * GNUmakefile.am:
        * configure.ac:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34329 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 Alice Liu <alice.liu@apple.com>
alice.liu@apple.com [Mon, 2 Jun 2008 23:04:13 +0000 (23:04 +0000)]
2008-06-02  Alice Liu  <alice.liu@apple.com>

        Reviewed by Beth Dakin.

        Some ARIA roles are designated to have presentational children, which
        means that their descendants aren't exposed to assistive technologies
        on an individual basis.  What this means in webcore is that descendants
        of such roles need to be ignored in terms of accessibility.

        * page/AccessibilityObject.cpp:
        * page/AccessibilityObject.h:
        * page/AccessibilityRenderObject.cpp:
        * page/AccessibilityRenderObject.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34328 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 Geoffrey Garen <ggaren@apple.com>
ggaren@apple.com [Mon, 2 Jun 2008 22:28:15 +0000 (22:28 +0000)]
2008-06-02  Geoffrey Garen  <ggaren@apple.com>


        * blog-files/squirrelfish-webkit-graph.png:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34327 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 Geoffrey Garen <ggaren@apple.com>
ggaren@apple.com [Mon, 2 Jun 2008 22:07:13 +0000 (22:07 +0000)]
2008-06-02  Geoffrey Garen  <ggaren@apple.com>

        Renamed this file quiet down the peanut gallery.

        * blog-files/squirrelfish-webkit-graph.png: Copied from blog-files/webkit-graph.png.
        * blog-files/webkit-graph.png: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34326 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 Geoffrey Garen <ggaren@apple.com>
ggaren@apple.com [Mon, 2 Jun 2008 22:02:51 +0000 (22:02 +0000)]
2008-06-02  Geoffrey Garen  <ggaren@apple.com>

        Added a graph.

        * blog-files/webkit-graph.png: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34325 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 David Hyatt <hyatt@apple.com>
hyatt@apple.com [Mon, 2 Jun 2008 21:54:23 +0000 (21:54 +0000)]
2008-06-02  David Hyatt  <hyatt@apple.com>

        Fix for https://bugs.webkit.org/show_bug.cgi?id=19303

        When the right slice value is omitted in border-image it should use the value of the left slice and not
        the top slice.

        Reviewed by darin

        Added fast/borders/border-image-omit-right-slice.html

        * css/CSSParser.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34324 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
andersca@apple.com [Mon, 2 Jun 2008 21:46:17 +0000 (21:46 +0000)]

2008-06-02  Anders Carlsson  <andersca@apple.com>

        Reviewed by David Hyatt and Mitz.

        repro crash in WebCore::RenderPart::setWidget (plugin-related?)

        Test: platform/mac/plugins/update-widget-from-style-recalc.html

        Make sure to update the document's rendering before calling updateWidget.

        * html/HTMLEmbedElement.cpp:
        * html/HTMLObjectElement.cpp:


2008-06-02  Anders Carlsson  <andersca@apple.com>

        Reviewed by David Hyatt and Mitz.

        repro crash in WebCore::RenderPart::setWidget (plugin-related?)

        Call -[WebView display] in the "plug-in" failed delegate method, simulating
        the sheet that Safari puts up.

        * DumpRenderTree/mac/ResourceLoadDelegate.mm:
        (-[ResourceLoadDelegate webView:plugInFailedWithError:dataSource:]):


2008-06-02  Anders Carlsson  <andersca@apple.com>

        Reviewed by David Hyatt and Mitz.

        repro crash in WebCore::RenderPart::setWidget (plugin-related?)

        * platform/mac/plugins/update-widget-from-style-recalc-expected.txt: Added.
        * platform/mac/plugins/update-widget-from-style-recalc.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34323 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 Geoffrey Garen <ggaren@apple.com>
ggaren@apple.com [Mon, 2 Jun 2008 21:45:59 +0000 (21:45 +0000)]
2008-06-02  Geoffrey Garen  <ggaren@apple.com>

        Moved the squirrelfish mascot to the right place.

        * blog-files/squirrelfish.png: Copied from images/squirrelfish.png.
        * images/squirrelfish.png: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34322 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 Geoffrey Garen <ggaren@apple.com>
ggaren@apple.com [Mon, 2 Jun 2008 21:41:10 +0000 (21:41 +0000)]
2008-06-02  Geoffrey Garen  <ggaren@apple.com>

        Uploaded the squirrelfish mascot.

        * images/squirrelfish.png: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34321 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoLayoutTests:
jchaffraix@webkit.org [Mon, 2 Jun 2008 20:47:54 +0000 (20:47 +0000)]

2008-06-02  Julien Chaffraix  <jchaffraix@webkit.org>

        Reviewed by Darin.

        Bug 18655: [XHR] OnProgress needs more test case

        * http/tests/xmlhttprequest/resources/five-tiny-chunk-one-second-delay.php: Added.
        * http/tests/xmlhttprequest/resources/noContentLength.asis: Added.
        * http/tests/xmlhttprequest/xmlhttprequest-LSProgressEvent-ProgressEvent-should-match-expected.txt: Added.
        * http/tests/xmlhttprequest/xmlhttprequest-LSProgressEvent-ProgressEvent-should-match.html: Added.
        * http/tests/xmlhttprequest/xmlhttprequest-addEventListener-onProgress-expected.txt: Added.
        * http/tests/xmlhttprequest/xmlhttprequest-addEventListener-onProgress.html: Added.
        * http/tests/xmlhttprequest/xmlhttprequest-no-content-length-onProgress-expected.txt: Added.
        * http/tests/xmlhttprequest/xmlhttprequest-no-content-length-onProgress.html: Added.
        * http/tests/xmlhttprequest/xmlhttprequest-onProgress-open-should-zero-length-expected.txt: Added.
        * http/tests/xmlhttprequest/xmlhttprequest-onProgress-open-should-zero-length.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34320 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 Geoffrey Garen <ggaren@apple.com>
ggaren@apple.com [Mon, 2 Jun 2008 20:45:13 +0000 (20:45 +0000)]
2008-06-02  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Darin Adler.

        A little cleanup in the CodeGenerator.

        * VM/CodeGenerator.cpp: A few changes here.

        (1) Removed remaining cases of the old hack of putting "this" into the
        symbol table; replaced with explicit tracking of m_thisRegister.

        (2) Made m_thisRegister behave the same for function, eval, and program
        code, removing the static programCodeThis() function.

        (3) Added a feature to nix a ScopeNode's declaration stacks when done
        compiling, to save memory.

        (4) Removed code that copied eval declarations into special vectors: we
        just use the originals in the ScopeNode now.

        * VM/CodeGenerator.h: Removed unneded parameters from the CodeGenerator
        constructor: we just use get that data from the ScopeNode now.

        * VM/Machine.cpp:
        (KJS::Machine::execute): When executing an eval node, don't iterate a
        special copy of its declarations; iterate the originals, instead.

        * kjs/nodes.cpp: Moved responsibility for knowing what AST data to throw
        away into the CodeGenerator. Nodes no longer call shrinkCapacity on
        their data directly.

        * kjs/nodes.h: Changed FunctionStack to ref its contents, so declaration
        data stays around even after we've thrown away the AST, unless we explicitly
        throw away the declaration data, too. This is useful for eval code, which
        needs to reference its declaration data at execution time. (Soon, it will
        be useful for program code, too, since program code should do the same.)

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34319 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 Kevin McCullough <kmccullough@apple.com>
kmccullough@apple.com [Mon, 2 Jun 2008 20:15:16 +0000 (20:15 +0000)]
2008-06-02  Kevin McCullough  <kmccullough@apple.com>

        Reviewed by Adam.

        <rdar://problem/5954226> JSProfiler: Move the calls from
        JavaScriptProfileNode to JavaScriptProfile in WebCore (19231)
        - This Makes the profiler now call sorting function s on a per-profile
        basis instead of on a node in the profile.
        - This is also needed to remove the recursion limit in the profiler
        <rdar://problem/5969992> JSProfiler: Remove the recursion limit in the
        since once the limit is removed calling sort on a node will only sort
        that node's children not the entire sub-tree of that node.

        * page/JavaScriptProfile.cpp: All of these functions were moved from
        * page/JavaScriptProfileNode.cpp:
        * page/inspector/ProfileView.js: Change uses of JavaScriptProfileNode
        for sorting to JavaScriptProfile.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34318 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
mitz@apple.com [Mon, 2 Jun 2008 20:00:34 +0000 (20:00 +0000)]

        Reviewed by Oliver Hunt.

        - fix more cases of https://bugs.webkit.org/show_bug.cgi?id=19344
          <rdar://problem/5977300> Regression: Shadow offsets seem to have changed in 10.5.3 in canvas

        Tests: fast/canvas/shadow-offset-1.html

        * html/CanvasRenderingContext2D.cpp:
        (WebCore::adjustedShadowSize): Added this helper function.
        (WebCore::CanvasRenderingContext2D::setShadow): Changed to call
        (WebCore::CanvasRenderingContext2D::applyShadow): Ditto.


        Reviewed by Oliver Hunt.

        - additional tests for https://bugs.webkit.org/show_bug.cgi?id=19344
          <rdar://problem/5977300> Regression: Shadow offsets seem to have changed in 10.5.3 in canvas

        * fast/canvas/resources/shadow-offset.js: Added.
        * fast/canvas/shadow-offset-1.html: Copied from LayoutTests/fast/canvas/shadow-offset.html.
        * fast/canvas/shadow-offset-2.html: Added.
        * fast/canvas/shadow-offset-3.html: Added.
        * fast/canvas/shadow-offset-4.html: Added.
        * fast/canvas/shadow-offset-5.html: Added.
        * fast/canvas/shadow-offset-6.html: Added.
        * fast/canvas/shadow-offset-7.html: Added.
        * fast/canvas/shadow-offset.html: Removed.
        * platform/mac/fast/canvas/shadow-offset-1-expected.checksum: Added.
        * platform/mac/fast/canvas/shadow-offset-1-expected.png: Added.
        * platform/mac/fast/canvas/shadow-offset-1-expected.txt: Added.
        * platform/mac/fast/canvas/shadow-offset-2-expected.checksum: Added.
        * platform/mac/fast/canvas/shadow-offset-2-expected.png: Added.
        * platform/mac/fast/canvas/shadow-offset-2-expected.txt: Added.
        * platform/mac/fast/canvas/shadow-offset-3-expected.checksum: Added.
        * platform/mac/fast/canvas/shadow-offset-3-expected.png: Added.
        * platform/mac/fast/canvas/shadow-offset-3-expected.txt: Added.
        * platform/mac/fast/canvas/shadow-offset-4-expected.checksum: Added.
        * platform/mac/fast/canvas/shadow-offset-4-expected.png: Added.
        * platform/mac/fast/canvas/shadow-offset-4-expected.txt: Added.
        * platform/mac/fast/canvas/shadow-offset-5-expected.checksum: Added.
        * platform/mac/fast/canvas/shadow-offset-5-expected.png: Added.
        * platform/mac/fast/canvas/shadow-offset-5-expected.txt: Added.
        * platform/mac/fast/canvas/shadow-offset-6-expected.checksum: Added.
        * platform/mac/fast/canvas/shadow-offset-6-expected.png: Added.
        * platform/mac/fast/canvas/shadow-offset-6-expected.txt: Added.
        * platform/mac/fast/canvas/shadow-offset-7-expected.checksum: Added.
        * platform/mac/fast/canvas/shadow-offset-7-expected.png: Added.
        * platform/mac/fast/canvas/shadow-offset-7-expected.txt: Added.
        * platform/mac/fast/canvas/shadow-offset-expected.checksum: Removed.
        * platform/mac/fast/canvas/shadow-offset-expected.png: Removed.
        * platform/mac/fast/canvas/shadow-offset-expected.txt: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34317 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Build fix for non-AllInOne builds
aroben@apple.com [Mon, 2 Jun 2008 19:58:02 +0000 (19:58 +0000)]
    Build fix for non-AllInOne builds

        * kjs/array_object.cpp: Added a missing #include.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34316 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
jchaffraix@webkit.org [Mon, 2 Jun 2008 19:51:32 +0000 (19:51 +0000)]

2008-06-02  Julien Chaffraix  <jchaffraix@webkit.org>

        Reviewed by Darin.

        Bug 18066: REGRESSION: createAttribute throws NAMESPACE_ERR exception

        Test: fast/dom/createAttribute-exception.html

        * dom/Document.cpp:
        * dom/Document.h: Add a bool parameter shouldIgnoreNamespaceChecks
        to createNamespaceNS() used by createAttribute() to bypass namespace


2008-06-02  Julien Chaffraix  <jchaffraix@webkit.org>

        Reviewed by Darin.

        Bug 18066: REGRESSION: createAttribute throws NAMESPACE_ERR exception

        * fast/dom/createAttribute-exception-expected.txt: Added.
        * fast/dom/createAttribute-exception.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34315 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 Anders Carlsson <andersca@apple.com>
andersca@apple.com [Mon, 2 Jun 2008 18:13:34 +0000 (18:13 +0000)]
2008-06-02  Anders Carlsson  <andersca@apple.com>

        Reviewed by Mitz.

        Speculative fix for <rdar://problem/5661112>
        CrashTracer: [USER] 49 crashes in DashboardClient at com.apple.WebCore: WebCore::RenderPart::setWidget + 62

        Defer loads while calling NPP_New. Some plug-ins start a run-loop inside NPP_New and finished loads could cause
        layouts to be triggered.

        * Plugins/WebBaseNetscapePluginView.mm:
        (-[WebBaseNetscapePluginView _createPlugin]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34314 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-06-02 Kevin McCullough <kmccullough@apple.com>
kmccullough@apple.com [Mon, 2 Jun 2008 17:32:19 +0000 (17:32 +0000)]
2008-06-02  Kevin McCullough  <kmccullough@apple.com>

        Took out accidental confilct lines I checked in.

        * ChangeLog:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34313 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
darin@apple.com [Mon, 2 Jun 2008 17:30:48 +0000 (17:30 +0000)]

2008-06-02  Darin Adler  <darin@apple.com>

        Reviewed by John Sullivan.

        - fix https://bugs.webkit.org/show_bug.cgi?id=18799
          apl.aip.org menu does not work (property names should be case sensitive)

        Test: fast/dom/CSSStyleDeclaration/css-properties-case-sensitive.html

        * bindings/js/JSCSSStyleDeclarationCustom.cpp:
        (WebCore::hasCSSPropertyNamePrefix): Tweak implementation a tiny bit for
        improved performance when the first character does not match.
        (WebCore::cssPropertyName): Reject property names with a leading uppercase


2008-06-02  Darin Adler  <darin@apple.com>

        Reviewed by John Sullivan.

        - test for https://bugs.webkit.org/show_bug.cgi?id=18799
          apl.aip.org menu does not work (property names should be case sensitive)

        * fast/dom/CSSStyleDeclaration: Added.

        * fast/dom/CSSStyleDeclaration/css-properties-case-sensitive-expected.txt: Added.
        * fast/dom/CSSStyleDeclaration/css-properties-case-sensitive.html: Added.
        * fast/dom/CSSStyleDeclaration/resources: Added.
        * fast/dom/CSSStyleDeclaration/resources/TEMPLATE.html: Copied from LayoutTests/fast/dom/Element/resources/TEMPLATE.html.
        * fast/dom/CSSStyleDeclaration/resources/css-properties-case-sensitive.js: Added.

        * fast/dom/CSSStyleDeclaration-empty-string-property-expected.txt: Removed.
        * fast/dom/CSSStyleDeclaration-empty-string-property.html: Removed.
        * fast/dom/CSSStyleDeclaration/empty-string-property-expected.txt: Copied from LayoutTests/fast/dom/CSSStyleDeclaration-empty-string-property-expected.txt.
        * fast/dom/CSSStyleDeclaration/empty-string-property.html: Copied from LayoutTests/fast/dom/CSSStyleDeclaration-empty-string-property.html.

        * fast/js/duplicate-param-crash.html: Allowed the make-js-test-wrappers script to update this file.
        * fast/js/duplicate-param-gc-crash.html: Ditto.
        * fast/js/resources/tostring-exception-in-property-access.js: Ditto.
        * fast/js/var-shadows-arg-crash.html: Ditto.
        * fast/js/var-shadows-arg-gc-crash.html: Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34312 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
darin@apple.com [Mon, 2 Jun 2008 17:26:04 +0000 (17:26 +0000)]

2008-06-02  Darin Adler  <darin@apple.com>

        Reviewed by Mitz.

        - fix https://bugs.webkit.org/show_bug.cgi?id=19358
          <rdar://problem/5951986> first line break in <pre> elements ignored
          (chess boards at http://rankzero.de/)

        Test: fast/parser/pre-first-line-break.html

        * editing/markup.cpp:
        (WebCore::appendStartMarkup): Fix handling of <listing>. Not part of the
        bug mentioned above, but uncovered by the regression test.

        * html/HTMLTokenizer.cpp:
        (WebCore::HTMLTokenizer::write): Clear the "discardLF" flag when
        processing a new tag.


2008-06-02  Darin Adler  <darin@apple.com>

        Reviewed by Mitz.

        - test for https://bugs.webkit.org/show_bug.cgi?id=19358
          <rdar://problem/5951986> first line break in <pre> elements ignored
          (chess boards at http://rankzero.de/)

        * fast/parser/pre-first-line-break-expected.txt: Added.
        * fast/parser/pre-first-line-break.html: Added.
        * fast/parser/resources/TEMPLATE.html: Copied from fast/dom/resources/TEMPLATE.html.
        * fast/parser/resources/pre-first-line-break.js: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@34311 268f45cc-cd09-0410-ab3c-d52691b4dbfc