WebCore:

2008-07-21  Anatoli Papirovski  <apapirovski@mac.com>

        Reviewed by Dave Hyatt.

        Fix for https://bugs.webkit.org/show_bug.cgi?id=14346
        Modified the css parser:
        - not to fail when closing braces are not found for
          a declaration at the end of the file
        - not to accept "!important fail" as valid
        - to keep accepting @import when it comes after invalid @ rules
        - not to drop the whole @media block when there's an error before the
          closing brace
        + some other minor css parsing revisions.

        * css/CSSGrammar.y:

LayoutTests:

2008-07-21  Anatoli Papirovski  <apapirovski@mac.com>

        Reviewed by Dave Hyatt.

        Updated layout CSS2.1 test suite tests related to css parsing:
        https://bugs.webkit.org/show_bug.cgi?id=14346

        * platform/mac/css2.1/t040105-import-01-b-expected.checksum:
        * platform/mac/css2.1/t040105-import-01-b-expected.png:
        * platform/mac/css2.1/t040105-import-01-b-expected.txt:
        * platform/mac/css2.1/t0402-syntax-05-f-expected.checksum:
        * platform/mac/css2.1/t0402-syntax-05-f-expected.png:
        * platform/mac/css2.1/t0402-syntax-05-f-expected.txt:
        * platform/mac/css2.1/t0402-syntax-06-f-expected.checksum:
        * platform/mac/css2.1/t0402-syntax-06-f-expected.png:
        * platform/mac/css2.1/t0402-syntax-06-f-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35261 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Touch file to force rebuild for vsprops change.

        * WebCorePrefix.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35260 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Build fix.

        * win/tools/vsprops/common.vsprops:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35259 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Whoops, forgot to commit expected output for this test.

RS=Dan Bernstein

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35258 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-20  Sam Weinig  <sam@webkit.org>

        Remove extra space Oliver Hunt left in.

        * page/Console.cpp:
        (WebCore::Console::reportException):
        (WebCore::Console::reportCurrentException):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35257 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Bug 19757: Crash when an ondragstart handler hides the element
<https://bugs.webkit.org/show_bug.cgi?id=19757>

Reviewed by Dan Bernstein.

The solution to this is problem is just to null check the renderer
immediately before launching the system drag, and terminate the
drag if the renderer is gone.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35256 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Oliver.

Fixes: https://bugs.webkit.org/show_bug.cgi?id=12171
Remove manual SVG property <-> XML attribute synchronization in SVGPolyElement.
Added svg/custom/poly-points-attribute-changes.svg (testcase from Rob)

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35255 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Oliver.
Cleanup JSSVGPODTypeWrapper code.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35254 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Bug 15979: Console logs in a database callback lose line number information
<https://bugs.webkit.org/show_bug.cgi?id=15979>

Reviewed by Mark Rowe

The problem was that there were multiple versions of the logic to
log an exception to the console, many of which were incorrect.  We
resolve this by making one single shared reportException method,
which handles the exceptions correctly.

This improves fidelity of callback errors, and ensures that
exceptions thrown in timer callbacks correctly report line numbers,
etc.

Test: fast/js/exceptions-thrown-in-callbacks.html

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35253 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Rubber stamped by Oliver.
Another attempt to fix the win build.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35252 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by David Kilzer.

        Fix high CPU usage on testmyiphone.com.

        - Made the logic for determining when to stop looking for meta charset more strainghtforward.
        Previously, this happened if a tag that's disallowed in HEAD was seen past the first 512
        bytes. Now, the algorithm bails out at the boundary if we are lo longer in HEAD (i.e, an
        offending tag was seen at any point before).

        - The above change made one of our regression tests fail, because it had its <meta>
        declaration past the 512 byte boundary. Fixed it by raising the boundary to 1024 bytes.

        - Made the algorithm bail out quickly if a comment that's not in HEAD crosses the boundary.

        - Moved a check for XML content type out of the loop.

        * loader/TextResourceDecoder.cpp:
        (WebCore::TextResourceDecoder::checkForHeadCharset):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35251 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Bug 20104: Exception in tables/mozilla_expected_failures/bugs/bug92868_1.html includes the equals operator in the quoted expression
<https://bugs.webkit.org/show_bug.cgi?id=20104>

Reviewed by Cameron Zwarich

To make this correct we make the dot and bracket assign nodes emit the information to indicate
the failure range is the dot/bracket accessor.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35250 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Attempt to fix windows build.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35249 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Oliver & parts by Eric.

Fixes: https://bugs.webkit.org/show_bug.cgi?id=20051
Rewrite animated property concept without heavy macro usage, replace by a templatified solution.
Fewer virtual function calls, no more usage of the tear-off's within internal code (synchronization needed it before.)

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35248 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Windows build fix.

        * kjs/JSGlobalObjectFunctions.cpp:
        (KJS::isStrWhiteSpace):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35247 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Windows build fix.

        * kjs/nodes.h:
        (KJS::ThrowableExpressionData::ThrowableExpressionData):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35246 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Bug 18774: SQUIRRELFISH: print meaningful error messages <https://bugs.webkit.org/show_bug.cgi?id=18774>
<rdar://problem/5769353> SQUIRRELFISH: JavaScript error messages are missing informative text

Reviewed by Cameron Zwarich

Add support for decent error messages in JavaScript.  This patch achieves this by providing
ensuring the common errors and exceptions have messages that provide the text of expression
that trigger the exception.  In addition it attaches a number of properties to the exception
object detailing where in the source the expression came from.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35245 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-18  Maxime Britto  <britto@apple.com>

        Reviewed by Adele.

        Fixed <rdar://problem/6049803>
        Prevent the autoscroll to trigger in WebClips when starting or hovering on an editable field.

        Test: fast/events/autoscroll-with-non-scrollable-parent.html

        * ChangeLog:
        * page/EventHandler.cpp: Edited
        (WebCore::EventHandler::handleMousePressEvent): changed the name of the funtion called to canBeProgramaticallyScrolled()
        (WebCore::EventHandler::handleMouseDraggedEvent): prevent the autoscroll to keep looking for a renderer when it's already triggered
        * rendering/RenderLayer.cpp:
        (WebCore::RenderLayer::scrollRectToVisible): verifies that the top layer can be programmatically scrolled before asking him to make the rect visible
        * rendering/RenderListBox.h:
        (WebCore::RenderListBox::canBeProgramaticallyScrolled):
        * rendering/RenderObject.cpp:
        (WebCore::RenderObject::canBeProgramaticallyScrolled): Edited : For the 3rd case we want document's renderer to have scrollbar as it's the top layer
        (WebCore::RenderObject::hasScrollableView): Verifies that the Object has a view with scrollBars
        * rendering/RenderObject.h: Renamed shouldAutosroll() for canBeProgramaticallyScrolled()
        * rendering/RenderTextControl.h:
        (WebCore::RenderTextControl::canBeProgramaticallyScrolled):

2008-07-18  Adele Peterson & Maxime Britto  <britto@apple.com>

        Reviewed by Adele.

        Test for <rdar://problem/6049803> Autoscroll triggered on no scrolling iframes

        * ChangeLog:
        * fast/events/autoscroll-with-non-scrollable-parent-expected.txt: Added.
        * fast/events/autoscroll-with-non-scrollable-parent.html: Added.
        * fast/events/resources/big-page-with-input.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35244 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-18  Sam Weinig  <sam@webkit.org>

        Reviewed by Anders Carlsson.

        <rdar://problem/6087283> Add support for uploading files via XMLHttpRequest

        - Overload XMLHttpRequests send() method to accept File tokens.

        * bindings/js/JSXMLHttpRequestCustom.cpp:
        (WebCore::JSXMLHttpRequest::send):
        * xml/XMLHttpRequest.cpp:
        (WebCore::XMLHttpRequest::send):
        * xml/XMLHttpRequest.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35243 268f45cc-cd09-0410-ab3c-d52691b4dbfc

JavaScriptCore:

2008-07-18  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Cameron Zwarich.

        Three renames:

        "CallTypeNative" => "CallTypeHost"
        "code" => "byteCode"
        "generatedCode" => "generatedByteCode"

JavaScriptGlue:

2008-07-18  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Cameron Zwarich.

        Three renames:

        "CallTypeNative" => "CallTypeHost"
        "code" => "byteCode"
        "generatedCode" => "generatedByteCode"

        * ChangeLog:
        * JSObject.cpp:
        (JSUserObject::getCallData):

WebCore:

2008-07-18  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Cameron Zwarich.

        Three renames:

        "CallTypeNative" => "CallTypeHost"
        "code" => "byteCode"
        "generatedCode" => "generatedByteCode"

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35242 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-18  Adele Peterson  <adele@apple.com>

        Reviewed by Dan Bernstein.

        I recently made a change to avoid fading media controls in and out if a video element actually only contains audio,
        which broke the code that did the same thing for audio elements that contain video (but don't actually display that video).
        This change will now check both the media element and the player to decide whether or not to make the controls persistent.

        * rendering/RenderMedia.cpp: (WebCore::RenderMedia::updateControlVisibility):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35241 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-18  Simon Fraser  <simon.fraser@apple.com>

        Reviewed by mitz

        Remove braces around single-line conditional.

        * css/CSSComputedStyleDeclaration.cpp:
        (WebCore::computedTransform):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35240 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-18  Simon Fraser  <simon.fraser@apple.com>

        Reviewed by Dave Hyatt

        getComputedStyle() for -webkit-transform should return
        'none' for elements with no renderer, or those with no
        transform.

        https://bugs.webkit.org/show_bug.cgi?id=20008

        Testcase: LayoutTests/fast/css/computed-style-without-renderer
                  LayoutTests/fast/css/computed-style-expected.txt

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35239 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-18  Simon Fraser  <simon.fraser@apple.com>

        Reviewed by Dave Hyatt

        Fix assertion about creating Length values with percentage types
        when blending transforms.
        https://bugs.webkit.org/show_bug.cgi?id=20086

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35238 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-18  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Oliver Hunt.

        Optimized <= for immediate number cases.

        SunSpider reports no overall change, but a 10% speedup on access-nsieve.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35237 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix some casts added in a previous build fix to match the style used
throughout WebKit.

Rubber-stamped by Sam Weinig.

* VM/Machine.cpp:
(KJS::Machine::initializeCallFrame):
* VM/Register.h:
(KJS::Register::Register):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35236 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Update expected results for a recently added test to what is seen when running the
        full test suite.

        * fast/forms/form-and-frame-interaction-retains-values-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35235 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Bug 19975: [OpenBSD] Patches to enable build of WebKit

JavaScriptCore:

2008-07-18  Landry Breuil  <landry@openbsd.org>

        Bug 19975: [OpenBSD] Patches to enable build of WebKit

        <https://bugs.webkit.org/show_bug.cgi?id=19975>

        Reviewed by David Kilzer.

        Support for OpenBSD, mostly threading and libm tweaks.

        * kjs/collector.cpp: #include <pthread.h>
        (KJS::currentThreadStackBase): use pthread_stackseg_np() to get stack base
        * kjs/config.h: OpenBSD also provides <pthread_np.h>
        * wtf/MathExtras.h: #include <sys/types.h> and <machine/ieee.h>
        (isfinite), (signbit): as long as we don't have those functions provide fallback implementations
        * wtf/Platform.h: Add support for PLATFORM(OPENBSD) and PLATFORM(SPARC64) macro

WebKitTools:

2008-07-18  Landry Breuil  <landry@openbsd.org>

        Bug 19975: [OpenBSD] Patches to enable build of WebKit

        <https://bugs.webkit.org/show_bug.cgi?id=19975>

        Reviewed by David Kilzer.

        * DumpRenderTree/DumpRenderTree.h: OpenBSD doesn't support wide characters.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35234 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Fixed indentation of example code for C++ pointers/references!

        * coding/coding-style.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35233 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-18  Robert Blaut  <webkit@blaut.biz>

        Reviewed by David Kilzer.

        Fix for <https://bugs.webkit.org/show_bug.cgi?id=12313>
        Added code style guidelines for multi-line boolean expressions

        * coding/coding-style.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35232 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-17  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Oliver Hunt.

        Next step toward putting doubles in registers: Store constant pool
        entries as registers, not JSValue*s.

        SunSpider reports no change.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35231 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-17  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by John Sullivan and Oliver Hunt.

        A tiny bit of tidying in function call register allocation.

        This patch saves one register when invoking a function expression and/or
        a new expression that is stored in a temporary.

        Since it's just one register, I can't make a testcase for it.

        * VM/CodeGenerator.cpp:
        (KJS::CodeGenerator::emitCall): No need to ref the function we're calling
        or its base. We'd like the call frame to overlap with them, if possible.
        op_call will read the function and its base before writing the call frame,
        so this is safe.

        * kjs/nodes.cpp:
        (KJS::NewExprNode::emitCode): No need to ref the function we're new-ing,
        for the same reasons stated above.

        (KJS::FunctionCallValueNode::emitCode): ditto

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35230 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-17  Steve Falkenburg  <sfalken@apple.com>

        Build fix.

        * kjs/InternalFunction.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35229 268f45cc-cd09-0410-ab3c-d52691b4dbfc

JavaScriptCore:

2008-07-17  Sam Weinig  <sam@webkit.org>

        Roll out r35199 as it is causing failures on the PPC build.

LayoutTests:

2008-07-17  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by David Kilzer.

        Test for https://bugs.webkit.org/show_bug.cgi?id=20067
        Support function.name (Firefox extension)

        * fast/js/function-name-expected.txt: Added.
        * fast/js/function-name.html: Added.
        * fast/js/resources/function-name.js: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35228 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-17  Sam Weinig  <sam@webkit.org>

        Roll out r35199 as it is causing failures on the PPC build.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35227 268f45cc-cd09-0410-ab3c-d52691b4dbfc

JavaScriptCore:

2008-07-17  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Oliver Hunt.

        Fixed <rdar://problem/6081636> Functions calls use more temporary
        registers than necessary

        Holding a reference to the last statement result register caused each
        successive statement to output its result to an even higher register.

        Happily, statements don't actually need to return a result register
        at all. I hope to make this clearer in a future cleanup patch,
        but this change will fix the major bug for now.

        * kjs/nodes.cpp:
        (KJS::statementListEmitCode):

LayoutTests:

2008-07-17  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Oliver Hunt.

        Test for <rdar://problem/6081636> Functions calls use more temporary
        registers than necessary

        * fast/js/function-call-register-allocation.html: Added.
        * fast/js/function-call-register-allocation-expected.txt: Added.
        * fast/js/resources/function-call-register-allocation.js: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35226 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by rwlbuis@gmail.com

        https://bugs.webkit.org/show_bug.cgi?id=19965

        - Added "@"{ident} rule (below other @-rules) to use flex for longest match
          (if an earlier rule - e.g. @media - also matches then flex chooses that one
           however, if a longer match - e.g. @mediaall matches flex will choose that)
        - Updated grammar to defined ATKEYWORD token and to use that instead of '@'
          in the error recovery grammar.

        Test: css2.1/atrule_longest_match.html

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35225 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-17  Gavin Barraclough  <barraclough@apple.com>

        Reviewed by Sam Weinig.

        Merge pre&post dot nodes to simplify the parse tree.
        Sunspider results show 0.6% progression (no performance change expected).

        * kjs/grammar.y:
        * kjs/nodes.cpp:
        * kjs/nodes.h:
        * kjs/nodes2string.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35224 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-17  David Hyatt  <hyatt@apple.com>

        Make sure the check to see if a frame/iframe is being rendered inside a transparency layer is
        recursive (and checks for transparency layers all the way up the ancestor document chain back to
        the top-level frame).

        Reviewed by Dan

        * rendering/RenderView.cpp:
        (WebCore::RenderView::paintBoxDecorations):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35223 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-17  Gavin Barraclough  <barraclough@apple.com>

        Reviewed by Cameron Zwarich.

        Merge pre&post resolve nodes to simplify the parse tree.
        Sunspider results show no performance change.

        * kjs/grammar.y:
        * kjs/nodes.cpp:
        * kjs/nodes.h:
        * kjs/nodes2string.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35222 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-17  Gavin Barraclough  <barraclough@apple.com>

        Reviewed by Cameron Zwarich.

        Merge logical nodes to simplify the parse tree.
        Sunspider results show 0.6% progression (no performance change expected).

        * kjs/grammar.y:
        * kjs/nodes.cpp:
        * kjs/nodes.h:
        * kjs/nodes2string.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35221 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-17  Ariya Hidayat  <ariya.hidayat@trolltech.com>

        Reviewed by Simon.

        Fix MinGW build (broken in r35198) and simplify getLocalTime().

        * kjs/DateMath.cpp:
        (KJS::getLocalTime):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35220 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[GTK] WebKit crashes on invalid settings notify callback

https://bugs.webkit.org/show_bug.cgi?id=19813

Disconnect in webkit_web_view_finalize

Reviewed by Holger Freyther.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35219 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-17  Gavin Barraclough  <barraclough@apple.com>

        Merge pre&post bracket nodes to simplify the parse tree.
        Sunspider results show no performance change.

        Reviewed by Sam Weinig.

        * kjs/grammar.y:
        * kjs/nodes.cpp:
        * kjs/nodes.h:
        * kjs/nodes2string.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35218 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-17  Ariya Hidayat  <ariya.hidayat@trolltech.com>

        Reviewed by Simon.

        Fix the 32-bit gcc builds, conversion from "long int" to Register is
        ambiguous. Explicitly choose the intptr_t constructor.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35217 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by mitz.

        Attempt to make initializeATSUStyle human-readable by
        splitting it out into better-named static inline functions.

        No functional changes, thus no tests.

        * platform/graphics/mac/FontMac.mm:
        (WebCore::fontHasMirroringInfo):
        (WebCore::disableLigatures):
        (WebCore::initializeATSUStyle):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35216 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    No review, build fix only.

        Attempt to fix Mac build

        * WebCore.xcodeproj/project.pbxproj: Make TextRun.h a private header.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35215 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-16  Jon Honeycutt  <jhoneycutt@apple.com>

        CygwinDownloader fails to install necessary packages
        https://bugs.webkit.org/show_bug.cgi?id=20075

        Reviewed by Mark Rowe.

        * CygwinDownloader/cygwin-downloader.zip: Rebuilt after two broken
        mirrors removed in r34116. Set MIME type to application/octet-stream.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35214 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-16  Jon Honeycutt  <jhoneycutt@apple.com>

        CygwinDownloader's make-zip.sh script fails
        https://bugs.webkit.org/show_bug.cgi?id=20074

        Reviewed by Mark Rowe.

        * CygwinDownloader/cygwin-downloader.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35213 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-16  Jon Honeycutt  <jhoneycutt@apple.com>

        REGRESSION: Can't create windowless plug-in with Flash 9
        https://bugs.webkit.org/show_bug.cgi?id=20070

        Reviewed by Mark Rowe.

        * plugins/win/PluginPackageWin.cpp:
        (WebCore::PluginPackage::determineQuirks): Reverse argument order;
        PlatformModuleVersion constructor takes leastSig, mostSig.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35212 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by mitz.

        Make ATSULayoutParameters a real C++ class
        (Give it a destructor to clean up after itself)
        (Also use OwnArrayPtr instead of manual member cleanup)

        * platform/graphics/mac/FontMac.mm:
        (WebCore::ATSULayoutParameters::ATSULayoutParameters):
        (WebCore::ATSULayoutParameters::~ATSULayoutParameters):
        (WebCore::overrideLayoutOperation):
        (WebCore::ATSULayoutParameters::initialize):
        (WebCore::Font::selectionRectForComplexText):
        (WebCore::Font::drawComplexText):
        (WebCore::Font::floatWidthForComplexText):
        (WebCore::Font::offsetForPositionForComplexText):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35211 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by mitz.

        Make ownership of copied UChar buffer clearer
        (and more leak-proof) by using a OwnArrayPtr.

        No functional changes, thus no tests.

        * platform/graphics/mac/FontMac.mm:
        (WebCore::copyRunForDirectionalOverrideIfNecessary):
        (WebCore::Font::selectionRectForComplexText):
        (WebCore::Font::drawComplexText):
        (WebCore::Font::offsetForPositionForComplexText):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35210 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix JavaScript in 64-bit by using a pointer-sized integer
type in the Register union.  Also includes a rename of
the intType constant to IntType.

Rubber-stamped by Geoff Garen.

* VM/Machine.cpp:
(KJS::Machine::initializeCallFrame):
* VM/Register.h:
(KJS::Register::):
(KJS::Register::Register):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35209 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Sam.

        Split out TextRun into its own header file.
        Re-order members to place all bools together
        (to allow compilers to better pack the struct)

        No functional changes, thus no tests.

        * WebCore.vcproj/WebCore.vcproj:
        * WebCore.xcodeproj/project.pbxproj:
        * platform/graphics/Font.h: Moved TextRun into its own file.
        * platform/graphics/TextRun.h: Split out from Font.h

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35208 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-16  Simon Fraser  <simon.fraser@apple.com>

        Reviewed by Dave Hyatt

        AnimationController needs to adjust z-index in the blended
        style in case opacity or transform were changed by blending.

        https://bugs.webkit.org/show_bug.cgi?id=20047

        Test: transitions/opacity-transition-zindex.html

        * page/AnimationController.cpp:
        (WebCore::AnimationController::updateImplicitAnimations):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35207 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Sam.

Fixes: https://bugs.webkit.org/show_bug.cgi?id=20052
Export the tag & attribute names as string literals for SVGNames/XLinkNames and one property in HTMLNames.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35206 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Rubber stamped by Eric.
Update pixel test results for two tests that have been changed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35205 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Eric.

Fixes: https://bugs.webkit.org/show_bug.cgi?id=10745 (SVGLength object needs some weight loss)
Don't store a context pointer in SVGLength, saving memory.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35204 268f45cc-cd09-0410-ab3c-d52691b4dbfc

JavaScriptCore:

2008-07-16  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Oliver Hunt.

        First step toward putting doubles in registers: Turned Register into a
        proper abstraction layer. It is no longer possible to cast a Register
        to a JSValue*, or a Register& to a JSValue*&, or to access the union
        inside a Register directly.

        SunSpider reports no change.

        In support of this change, I had to make the following mechanical changes
        in a lot of places:

        1. Clients now use explicit accessors to read data out of Registers, and
        implicit copy constructors to write data into registers.

        So, assignment that used to look like

            x.u.jsValue = y;

        now looks like

            x = y;

        And access that used to look like

            x = y.u.jsValue;

        now looks like

            x = y.jsValue();

        2. I made generic flow control specific in opcodes that made their flow
        control generic by treating a Register& as a JSValue*&. This had the
        added benefit of removing some exception checking branches from immediate
        number code.

        3. I beefed up PropertySlot to support storing a Register* in a property
        slot. For now, only JSVariableObject's symbolTableGet and symbolTablePut
        use this functionality, but I expect more clients to use it in the future.

        4. I changed ArgList to be a buffer of Registers, not JSValue*'s, and I
        changed ArgList iterator clients to iterate Registers, not JSValue*'s.

WebCore:

2008-07-16  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Oliver Hunt.

        Support for JavaScriptCore's first step toward putting doubles in
        registers: Treat ArgList iterators as Register*'s, not JSValue*'s.

        * bindings/js/ScheduledAction.cpp:
        (WebCore::ScheduledAction::ScheduledAction):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35203 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Fixed build.

        * kjs/JSGlobalObject.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35202 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Anders Carlsson.
Add missing layout test results.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35201 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-16  Kevin McCullough  <kmccullough@apple.com>

        Reviewed by Sam and Geoff.

        <rdar://problem/5958840> Navigating to another page while profiler is
        attached results in slow JavaScript for all time.

        - The UNLIKELY keeps this from being a sunspider performance regression.

        * kjs/JSGlobalObject.cpp:
        (KJS::JSGlobalObject::~JSGlobalObject): Stop the profiler associated
        with this exec state.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35200 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-16  Sam Weinig  <sam@webkit.org>

        Reviewed by Steve Falkenburg.

        Replace adopting UString constructor in favor of explicit
        static adopt method.

        * API/JSStringRefCF.cpp:
        (JSStringCreateWithCFString):
        * kjs/StringConstructor.cpp:
        (KJS::stringFromCharCode):
        * kjs/StringPrototype.cpp:
        (KJS::stringProtoFuncToLowerCase):
        (KJS::stringProtoFuncToUpperCase):
        (KJS::stringProtoFuncToLocaleLowerCase):
        (KJS::stringProtoFuncToLocaleUpperCase):
        * kjs/ustring.cpp:
        (KJS::UString::adopt):
        * kjs/ustring.h:
        (KJS::UString::UString):
        (KJS::UString::~UString):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35199 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-16  Ariya Hidayat  <ariya.hidayat@trolltech.com>

        Reviewed by Simon.

        http://trolltech.com/developer/task-tracker/index_html?method=entry&id=216179
        Fix potential crash (on Qt for Windows port) when performing JavaScript date
        conversion.

        * kjs/DateMath.cpp:
        (KJS::getLocalTime): For the Qt port, prefer to use Windows code, i.e.
        localtime_s() instead of localtime() since the latter might crash (on Windows)
        given a non-sensible, e.g. NaN, argument.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35198 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix a typo in ChangeLog.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35197 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Anders and Geoff.

        https://bugs.webkit.org/show_bug.cgi?id=20023
        Failed assertion in PropertyNameArray.cpp

        This is already tested by testapi.

        * API/JSObjectRef.cpp: (JSPropertyNameAccumulatorAddName): Add the string to identifier
        table to appeaase PropertyNameArray.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35196 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Geoff.

        Dereference identifiers when deleting a hash table (fixes leaks with private JSGlobalData
        objects).

        * kjs/JSGlobalData.cpp: (KJS::JSGlobalData::~JSGlobalData):
        * kjs/lookup.cpp: (KJS::HashTable::deleteTable):
        * kjs/lookup.h:
        * kjs/lexer.cpp: (KJS::Lexer::~Lexer)
        HashTable cannot have a destructor, because check-for-global-initializers complains about
        having a global constructor then.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35195 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Geoff.

        Check pthread_key_create return value.

        This check was helpful when debugging a crash in run-webkit-tests --threaded that happened
        because JSGlobalData objects were not deleted, and we were running out of pthread keys soon.
        It also looks useful for production builds.

        * wtf/ThreadSpecific.h: (WTF::::ThreadSpecific):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35194 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebKitTools:

2008-07-15  Sam Weinig  <sam@webkit.org>

        Reviewed by Anders Carlsson.

        Refactor accessibility testing code.

        We now have:
        - AccessibilityController
          Controller which has access to the WebView and can provide the focused element, root element
          and in the future, elementAtPoint.

        - AccessibilityUIElement
          Object which provides access to the data provided by the Accessibility APIs.

        * DumpRenderTree/AccessibilityController.cpp:
        (getFocusedElementCallback):
        (getRootElementCallback):
        (AccessibilityController::getJSClass):
        * DumpRenderTree/AccessibilityController.h:
        * DumpRenderTree/AccessibilityUIElement.cpp: Added.
        (allAttributesCallback):
        (attributesOfLinkedUIElementsCallback):
        (attributesOfChildrenCallback):
        (getRoleCallback):
        (getTitleCallback):
        (getDescriptionCallback):
        (getWidthCallback):
        (getHeightCallback):
        (getIntValueCallback):
        (getMinValueCallback):
        (getMaxValueCallback):
        (finalize):
        (AccessibilityUIElement::makeJSAccessibilityUIElement):
        (AccessibilityUIElement::getJSClass):
        * DumpRenderTree/AccessibilityUIElement.h: Added.
        (AccessibilityUIElement::platformUIElement):
        * DumpRenderTree/DumpRenderTree.xcodeproj/project.pbxproj:
        * DumpRenderTree/mac/AccessibilityControllerMac.mm:
        (AccessibilityController::AccessibilityController):
        (AccessibilityController::~AccessibilityController):
        (AccessibilityController::focusedElement):
        (AccessibilityController::rootElement):
        * DumpRenderTree/mac/AccessibilityUIElementMac.mm: Added.
        (AccessibilityUIElement::AccessibilityUIElement):
        (descriptionOfValue):
        (attributesOfElement):
        (nsStringToJSStringRef):
        (concatenateAttributeAndValue):
        (AccessibilityUIElement::getLinkedUIElements):
        (AccessibilityUIElement::getChildren):
        (AccessibilityUIElement::attributesOfLinkedUIElements):
        (AccessibilityUIElement::attributesOfChildren):
        (AccessibilityUIElement::allAttributes):
        (AccessibilityUIElement::role):
        (AccessibilityUIElement::title):
        (AccessibilityUIElement::description):
        (AccessibilityUIElement::width):
        (AccessibilityUIElement::height):
        (AccessibilityUIElement::intValue):
        (AccessibilityUIElement::minValue):
        (AccessibilityUIElement::maxValue):

LayoutTests:

2008-07-15  Sam Weinig  <sam@webkit.org>

        Reviewed by Anders Carlsson.

        Update for refactored accessibility testing code.

        * accessibility/aria-describedby-on-input.html:
        * accessibility/aria-labelledby-on-input.html:
        * accessibility/aria-range-value.html:
        * accessibility/aria-range.html:
        * accessibility/aria-roles.html:
        * accessibility/aria-slider.html:
        * accessibility/aria-spinbutton.html:
        * accessibility/document-attributes.html:
        * accessibility/image-map1.html:
        * accessibility/image-map2.html:
        * accessibility/internal-link-anchors.html:
        * accessibility/radio-button-checkbox-size.html:
        * accessibility/radio-button-group-members.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35193 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Versioning.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35192 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-15  Maxime Britto  <britto@apple.com>

        Reviewed by Eric.

        https://bugs.webkit.org/show_bug.cgi?id=17589
        <rdar://problem/5770893>
        Retrieve the wheel sensitivity setting from windows system settings for the horizontal and the vertical scrolling.
        Allows the page scrolling with the wheel (if activated in windows settings) on both horizontal and vertical ways
        For the other platforms, nothing is changed but if someone want to activate this feature
        he/she has to retrieve the settings and fill out the m_lineScrollSensitivity and/or m_charScrollSensitivity in his/her PlatformWheelEvent constructor.

        * ChangeLog:
        * page/EventHandler.cpp:
        (WebCore::scrollAndAcceptEvent): static inline function to avoid copy/paste in the member functions
        (WebCore::EventHandler::handleWheelEvent): Call to the above mentionned static inline function with the right parameters
        * platform/PlatformWheelEvent.h:
        (WebCore::PlatformWheelEvent::deltaX): Now returns the delta multiplied by the user settings for the number of lines to scroll
        (WebCore::PlatformWheelEvent::deltaY): Now returns the delta multiplied by the user settings for the number of chars to scroll
        (WebCore::PlatformWheelEvent::isPageXScrollModeEnabled):
        (WebCore::PlatformWheelEvent::isPageYScrollModeEnabled):
        * platform/gtk/WheelEventGtk.cpp: Add the values to avoid build break but the factors are initialized to 1 not to change the previous behavior
        (WebCore::PlatformWheelEvent::PlatformWheelEvent):
        * platform/mac/WheelEventMac.mm: Add the values to avoid build break but the factors are initialized to 1 not to change the previous behavior
        (WebCore::PlatformWheelEvent::PlatformWheelEvent):
        * platform/qt/WheelEventQt.cpp: Add the values to avoid build break but the factors are initialized to 1 not to change the previous behavior
        (WebCore::PlatformWheelEvent::PlatformWheelEvent):
        * platform/win/ScrollViewWin.cpp: Added the page scroll (horiz/vertical) and reduced the number of pixels to scroll per line to keep the standard behavior with the factor
        (WebCore::adjustDeltaForPageScrollMode): static inline function to avoid copy/paste in the member functions
        (WebCore::ScrollView::updateScrollbars):
        (WebCore::ScrollView::wheelEvent):
        * platform/win/WheelEventWin.cpp: Retrieves the new values by using the win32 API.
        (WebCore::PlatformWheelEvent::PlatformWheelEvent):
        * platform/wx/MouseWheelEventWx.cpp: Add the values to avoid build break but the factors are initialized to 1 not to change the previous behavior
        (WebCore::PlatformWheelEvent::PlatformWheelEvent):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35190 268f45cc-cd09-0410-ab3c-d52691b4dbfc

wx build fix - adding AccessibilityImageMapLink.cpp to the wx build.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35189 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

2008-07-15  Kevin McCullough  <kmccullough@apple.com>

        Reviewed by Dan.

        <rdar://problem/5620273> REGRESSION: Apparent caching between form
        submits twice (on Maconomy) (18401)

        Frames did not set the createdByParser flag even when they were.

        * html/HTMLElementFactory.cpp:
        (WebCore::frameConstructor):
        (WebCore::iframeConstructor):

LayoutTests:

2008-07-15  Kevin McCullough  <kmccullough@apple.com>

        Reviewed by Dan.

        <rdar://problem/5620273> REGRESSION: Apparent caching between form
        submits twice (on Maconomy) (18401)

        * fast/forms/form-and-frame-interaction-retains-values-expected.txt: Added.
        * fast/forms/form-and-frame-interaction-retains-values.html: Added.
        * fast/forms/resources/form-and-frame-interaction-retains-values-main.html: Added.
        * fast/forms/resources/form-and-frame-interaction-retains-values-submit.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35188 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Rubber-stamped by Sam Weinig.

        - update results for the addition of the compareDocumentPosition-related constants

        * fast/dom/domListEnumeration-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35187 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

        Reviewed by Dave Hyatt.

        - fix https://bugs.webkit.org/show_bug.cgi?id=19525
          <rdar://problem/5961768> -webkit-box-reflect in hyperlink causes webkit to crash

        Test: fast/reflections/inline-crash.html

        * rendering/RenderInline.cpp:
        (WebCore::RenderInline::setStyle): Added setHasReflection(false).
        Inline flows never have reflections.

LayoutTests:

        Reviewed by Dave Hyatt.

        - test for https://bugs.webkit.org/show_bug.cgi?id=19525
          <rdar://problem/5961768> -webkit-box-reflect in hyperlink causes webkit to crash

        * fast/reflections/inline-crash.html: Added.
        * platform/mac/fast/reflections/inline-crash-expected.checksum: Added.
        * platform/mac/fast/reflections/inline-crash-expected.png: Added.
        * platform/mac/fast/reflections/inline-crash-expected.txt: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35186 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-14  David Hyatt  <hyatt@apple.com>

        Allow <style> and <link> pretty much anywhere.  They will still be moved to the <head> if
        no <body> exists yet though.

        Reviewed by Sam

        Added fast/css/style-parsed-outside-of-head.html

        * html/HTMLElement.cpp:
        (WebCore::inlineTagList):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35185 268f45cc-cd09-0410-ab3c-d52691b4dbfc

JavaScriptCore:

2008-07-15  Kevin McCullough  <kmccullough@apple.com>

        Reviewed by Geoff.

        Rename pageGroupIdentifier to profileGroup to keep mention of a
        pageGroup out of JavaScriptCore.

        * kjs/JSGlobalObject.cpp:
        (KJS::JSGlobalObject::init):
        * kjs/JSGlobalObject.h:
        (KJS::JSGlobalObject::setProfileGroup):
        (KJS::JSGlobalObject::profileGroup):
        * profiler/ProfileGenerator.cpp:
        (KJS::ProfileGenerator::create):
        (KJS::ProfileGenerator::ProfileGenerator):
        * profiler/ProfileGenerator.h:
        (KJS::ProfileGenerator::profileGroup):
        * profiler/Profiler.cpp:
        (KJS::Profiler::startProfiling):
        (KJS::dispatchFunctionToProfiles):
        (KJS::Profiler::willExecute):
        (KJS::Profiler::didExecute):

WebCore:

2008-07-15  Kevin McCullough  <kmccullough@apple.com>

        Reviewed by Geoff.

        Rename pageGroupIdentifier to profileGroup to keep mention of a
        pageGroup out of JavaScriptCore.

        * bindings/js/JSQuarantinedObjectWrapper.cpp:
        (WebCore::JSQuarantinedObjectWrapper::allowsUnwrappedAccessFrom):
        * bindings/js/ScriptController.cpp:
        (WebCore::ScriptController::clear):
        (WebCore::ScriptController::initScript):
        * history/CachedPage.cpp:
        (WebCore::CachedPage::restore):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35184 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-15  Adele Peterson  <adele@apple.com>

        Updating results for my fix to <rdar://problem/5784890>.

        * http/tests/misc/xhtml-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35183 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Add WebKit[Set]ShouldUseFontSmoothing functions

WebCore:

        WebCore part of adding WebKit[Set]ShouldUseFontSmoothing functions

        <rdar://6059127>

        Reviewed by John Sullivan.

        * platform/graphics/win/FontCGWin.cpp:
        (WebCore::Font::drawGlyphs): Pass the result of
        WebCoreShouldUseFontSmoothing() into wkSetFontSmoothingStyle.
        * platform/win/WebCoreTextRenderer.cpp:
        (WebCore::WebCoreSetShouldUseFontSmoothing):
        (WebCore::WebCoreShouldUseFontSmoothing):
        Added.
        * platform/win/WebCoreTextRenderer.h:

WebKit/win:

        Add WebKit[Set]ShouldUseFontSmoothing functions

        <rdar://6059127>

        Reviewed by John Sullivan.

        * WebKit.vcproj/WebKit.def:
        * WebKit.vcproj/WebKit_debug.def:
        Export the new functions.
        * WebKitGraphics.cpp:
        (WebKitSetShouldUseFontSmoothing):
        (WebKitShouldUseFontSmoothing):
        Call through to the relevant WebCore functions.
        * WebKitGraphics.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35182 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Rubber-stamped by Anders Carlsson.

        - updated results for https://bugs.webkit.org/show_bug.cgi?id=18088
          <rdar://problem/6036232> white-space:nowrap; float:left; causing the misalignment issue

        * platform/mac/tables/mozilla/bugs/bug26553-expected.checksum:
        * platform/mac/tables/mozilla/bugs/bug26553-expected.png:
        * platform/mac/tables/mozilla/bugs/bug26553-expected.txt:
        * platform/mac/tables/mozilla/bugs/bug57828-2-expected.checksum:
        * platform/mac/tables/mozilla/bugs/bug57828-2-expected.png:
        * platform/mac/tables/mozilla/bugs/bug57828-2-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35181 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Windows build fix

        * WebCore.vcproj/WebCore.vcproj: Add AccessibilityImageMapLink.{cpp,h}
        to the project.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35180 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-15  Holger Hans Peter Freyther  <zecke@selfish.org>

        Build fix for the Gtk+ and Qt platform.

        * GNUmakefile.am: Add AccessibilityImageMapLink.cpp to the build
        * WebCore.pro: Add AccessibilityImageMapLink.cpp to the build

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35179 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    - Tiger build fix

        * platform/mac/WidgetMac.mm:
        (WebCore::Widget::paint):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35178 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix https://bugs.webkit.org/show_bug.cgi?id=20037
Bug 20037: GCC 4.2 build broken due to strict aliasing violation.

Reviewed by Sam Weinig.

* kjs/ustring.cpp:
(KJS::UString::Rep::computeHash): Add a version of computeHash that takes a char* and explicit length.
* kjs/ustring.h:
* profiler/CallIdentifier.h:
(WTF::): Use new version of computeHash that takes a char* and explicit length to avoid unsafe aliasing.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35177 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

        Reviewed by John Sullivan.

        - WebCore part of fixing <rdar://problem/6071850> Subviews not drawn correctly when using -cacheDisplayInRect:toBitmapImageRep:

        * platform/mac/WidgetMac.mm:
        (WebCore::Widget::paint): Changed to account for the case of drawing
        into a bitmap context that is not a window's backing store. In that
        case, -displayRectIgnoringOpacity:inContext: is used to redirect the
        drawing to the correct context, after setting up the right transform
        on it. For subframes, additional code ensures that the scroll view will
        not paint the background behind a transparent frame.

WebKit/mac:

        Reviewed by John Sullivan.

        - WebKit part of fixing <rdar://problem/6071850> Subviews not drawn correctly when using -cacheDisplayInRect:toBitmapImageRep:

        * Plugins/WebBaseNetscapePluginView.mm:
        (-[WebBaseNetscapePluginView saveAndSetNewPortStateForUpdate:]): Changed
        to account for the case of being drawn into a bitmap context that is not
        a window's backing store. In that case, there are no valid "rects being
        drawn" to clip to.
        * WebView/WebHTMLView.mm:
        (-[WebHTMLView _recursive:displayRectIgnoringOpacity:inContext:topView:]):
        Added an override of this NSView method which is used in for
        -cacheDisplayInRect:toBitmapImageRep:. Like two existing NSView drawing
        machinery overrides, it sets subviews aside before invoking the
        superclass implementation. On Tiger, it also updates the layout.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35176 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Don't include autoversion.h in WebKitDLL.cpp.
        autoversion.h changes for every build, so we don't want to include it unless necessary.

        Rubber stamped by Ada Chan.

        * WebKitDLL.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35175 268f45cc-cd09-0410-ab3c-d52691b4dbfc

<rdar://problem/6038106> AXChildren returned for this web area is null

The accessibility of image maps needed to be brought inline with the new AX architecture

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35174 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-14  Sam Weinig  <sam@webkit.org>

        Rubber-stamped by David Hyatt.

        Remove unused ExceptionCode parameter from compareDocumentPosition.

        * dom/Document.cpp:
        (WebCore::Document::addStyleSheetCandidateNode):
        * dom/Node.cpp:
        (WebCore::Node::compareDocumentPosition):
        * dom/Node.h:
        * dom/Node.idl:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35173 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-14  David Hyatt  <hyatt@apple.com>

        Fix for bugzilla bug 5476, link and style elements should be allowed outside the <head>.

        Reviewed by Sam

        Added fast/css/style-outside-head.html
        Added fast/css/link-outside-head.html

        * dom/Document.cpp:
        (WebCore::Document::addStyleSheetCandidateNode):
        (WebCore::Document::removeStyleSheetCandidateNode):
        (WebCore::Document::recalcStyleSelector):
        * dom/Document.h:
        * dom/Node.cpp:
        (WebCore::Node::compareDocumentPosition):
        * dom/ProcessingInstruction.cpp:
        (WebCore::ProcessingInstruction::checkStyleSheet):
        (WebCore::ProcessingInstruction::insertedIntoDocument):
        (WebCore::ProcessingInstruction::removedFromDocument):
        (WebCore::ProcessingInstruction::finishParsingChildren):
        * dom/ProcessingInstruction.h:
        (WebCore::ProcessingInstruction::setCreatedByParser):
        * dom/StyleElement.cpp:
        (WebCore::StyleElement::removedFromDocument):
        * dom/XMLTokenizer.cpp:
        (WebCore::XMLTokenizer::startElementNs):
        (WebCore::XMLTokenizer::processingInstruction):
        (WebCore::):
        * html/HTMLElementFactory.cpp:
        (WebCore::linkConstructor):
        * html/HTMLLinkElement.cpp:
        (WebCore::HTMLLinkElement::HTMLLinkElement):
        (WebCore::HTMLLinkElement::insertedIntoDocument):
        (WebCore::HTMLLinkElement::removedFromDocument):
        (WebCore::HTMLLinkElement::finishParsingChildren):
        * html/HTMLLinkElement.h:
        (WebCore::HTMLLinkElement::setCreatedByParser):
        * html/HTMLStyleElement.cpp:
        (WebCore::HTMLStyleElement::insertedIntoDocument):
        (WebCore::HTMLStyleElement::removedFromDocument):
        * svg/SVGStyleElement.cpp:
        (WebCore::SVGStyleElement::insertedIntoDocument):
        (WebCore::SVGStyleElement::removedFromDocument):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35172 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-14  Steve Falkenburg  <sfalken@apple.com>

        Build fix.

        * DumpRenderTree/win/DumpRenderTree.vcproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35171 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Attempted Windows build fix

        * DumpRenderTree/win/TestNetscapePlugin/TestNetscapePlugin.vcproj: Add
        CoreFoundation/OSXCompatibilityHeaders[/GNUCompatibility] to the
        include path.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35170 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-14  Sam Weinig  <sam@webkit.org>

        Reviewed by Mark Rowe.

        Sort interface extended attributes to appease *the* Mark Rowe.

        * page/DOMWindow.idl:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35169 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

2008-07-14  Anders Carlsson  <andersca@apple.com>

        Reviewed by Geoff.

        <rdar://problem/6073974>
        https://bugs.webkit.org/show_bug.cgi?id=18106
        The "onscroll" event bubbles, which is inconsistent with Firefox and IE

        Don't bubble the scroll event.

        * rendering/RenderLayer.cpp:
        (WebCore::RenderLayer::scrollToOffset):
        * rendering/RenderListBox.cpp:
        (WebCore::RenderListBox::valueChanged):

LayoutTests:

2008-07-14  Anders Carlsson  <andersca@apple.com>

        Reviewed by Geoff.

        <rdar://problem/6073974>
        https://bugs.webkit.org/show_bug.cgi?id=18106
        The "onscroll" event bubbles, which is inconsistent with Firefox and IE

        Add test.

        * fast/events/scroll-event-does-not-bubble-expected.txt: Added.
        * fast/events/scroll-event-does-not-bubble.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35168 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-14  Adele Peterson  <adele@apple.com>

        Reviewed by Geoff.

        Fix for <rdar://problem/5784890> REGRESSION (Safari 3): Can't edit SpeedStream 4100 DSL modem settings

        The web server on this particular piece of hardware doesn't handle "text/xml" in the
        Accept header if the headers are sent in a certain order.  Safari 2 used to send "*/*" in
        the Accept header.  Firefox 2 sent "text/xml" but in an acceptable order for the web
        server.  And Firefox 3 doesn't send "text/xml" at all in the Accept header since it
        is being deprecated in favor of "application/xml".  We decided that the best solution is
        to match Firefox 3 and stop sending "text/xml" in the Accept header.

        No test.  There appears to be no way to get the raw headers in the original order in perl/php.

        * loader/FrameLoader.cpp: (WebCore::FrameLoader::addExtraFieldsToRequest):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35167 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

2008-07-14  Sam Weinig  <sam@webkit.org>

        Reviewed by Adele Peterson.

        Fix for <rdar://problem/5769819>

        Test: http/tests/security/cross-frame-access-object-prototype.html

        * bindings/js/JSDOMWindowCustom.cpp:
        (WebCore::JSDOMWindow::getPropertyAttributes): Perform security check.
        (WebCore::JSDOMWindow::defineGetter): Ditto.
        (WebCore::JSDOMWindow::defineSetter): Ditto.
        (WebCore::JSDOMWindow::lookupGetter): Ditto.
        (WebCore::JSDOMWindow::lookupSetter): Ditto.
        * bindings/scripts/CodeGeneratorJS.pm: Add support for custom versions
        of all the core JSObject functionality.
        * page/DOMWindow.idl: Override remaining core JSObject functionality,
        to inject security checks.

LayoutTests:

2008-07-14  Sam Weinig  <sam@webkit.org>

        Reviewed by Adele Peterson.

        Test for <rdar://problem/5769819>

        * http/tests/security/cross-frame-access-object-prototype-expected.txt: Added.
        * http/tests/security/cross-frame-access-object-prototype.html: Added.
        * http/tests/security/resources/cross-frame-iframe-for-object-prototype-test.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35166 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-14  David Hyatt  <hyatt@apple.com>

        Fix a crashing bug in ListHashSet's -- operator.  Make sure that end() can be -- by special-casing the null
        position.

        Reviewed by Maciej

        * wtf/ListHashSet.h:
        (WTF::ListHashSetConstIterator::operator--):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35165 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-14  David Hyatt  <hyatt@apple.com>

        Buidl fix.  Make sure the second insertBefore method returns a value.

        * wtf/ListHashSet.h:
        (WTF::::insertBefore):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35164 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Windows build fixes

JavaScriptCore:

        Windows build fix

        * JavaScriptCore.vcproj/jsc/jsc.vcproj: Added include/pthreads to the
        include path.

WebCore:

        Windows build fixes

        * WebCore.vcproj/WebCore.vcproj: Add include/pthreads to the include
        path.
        * bindings/scripts/CodeGeneratorCOM.pm: Touched this to force COM
        bindings to regenerate.
        * loader/FTPDirectoryDocument.cpp:
        * loader/FTPDirectoryParser.cpp:
        Fixed some duplicate macro definitions now that we're including
        pthread.h again.

WebKit/win:

        Windows build fix

        * WebJavaScriptCollector.cpp:
        (WebJavaScriptCollector::objectCount): Updated to match
        -[WebCoreStatistics javaScriptsObjectCount].

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35163 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Kevin McCullough.

        Make JSGlobalData refcounted in preparation to adding a way to create contexts that share
        global data.

        * JavaScriptCore.exp:
        * kjs/JSGlobalData.cpp:
        (KJS::JSGlobalData::create):
        * kjs/JSGlobalData.h:
        Made contructor private, and added a static create() method. Made the class inherit from
        RefCounted.

        * kjs/JSGlobalObject.h:
        (KJS::JSGlobalObject::globalData):
        JSGlobalData is now owned by JSGlobalObject (except for the shared one, and the common
        WebCore one, which are never deleted).

        * kjs/Shell.cpp: (main): Create JSGlobalData with create() method.

WebCore:
        * bindings/js/JSDOMWindowBase.cpp:
        (WebCore::JSDOMWindowBase::commonJSGlobalData): Create JSGlobalData with create() method.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35162 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-07-14  Simon Hausmann  <hausmann@webkit.org>

        Reviewed by Holger.

        Make listDirectory() in the Qt port work with empty namefilters.
        Don't include . and .. to avoid scanning the parent directory.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@35161 268f45cc-cd09-0410-ab3c-d52691b4dbfc