14 years ago2008-05-31 Maciej Stachowiak <mjs@apple.com>
mjs@apple.com [Sat, 31 May 2008 22:33:46 +0000 (22:33 +0000)]
2008-05-31  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Sam.

        - avoid assertion in updateDocument() when running debug

        * bindings/js/kjs_proxy.cpp:
        (WebCore::KJSProxy::updateDocument): Exit early if the document is null. We
        don't need to update in this case.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34279 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-31 Alp Toker <alp@nuanti.com>
alp@webkit.org [Sat, 31 May 2008 20:46:53 +0000 (20:46 +0000)]
2008-05-31  Alp Toker  <alp@nuanti.com>

        Fix inconsistent zoom docs and enum name landed in r34249. Issue
        noticed by Marco Barisione.

        * webkit/webkitwebview.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34278 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
mitz@apple.com [Sat, 31 May 2008 06:28:21 +0000 (06:28 +0000)]

        Reviewed by Darin Adler.

        - fix https://bugs.webkit.org/show_bug.cgi?id=19278
          <rdar://problem/5968621> REGRESSION (r31114-r31132): image float disappears on refresh

        Test: fast/dynamic/float-in-trailing-whitespace-after-last-line-break.html

        * rendering/bidi.cpp:
        (WebCore::RenderBlock::layoutInlineChildren): Ensured that floats
        occurring in trailing whitespace after a line break are added to the
        last line's floats vector.


        Reviewed by Darin Adler.

        - test for https://bugs.webkit.org/show_bug.cgi?id=19278
          <rdar://problem/5968621> REGRESSION (r31114-r31132): image float disappears on refresh

        * fast/dynamic/float-in-trailing-whitespace-after-last-line-break.html: Added.
        * platform/mac/fast/dynamic/float-in-trailing-whitespace-after-last-line-break-expected.checksum: Added.
        * platform/mac/fast/dynamic/float-in-trailing-whitespace-after-last-line-break-expected.png: Added.
        * platform/mac/fast/dynamic/float-in-trailing-whitespace-after-last-line-break-expected.txt: Added.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34277 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Implements the back and forward buttons in the Scripts panel.
timothy@apple.com [Sat, 31 May 2008 06:06:11 +0000 (06:06 +0000)]
    Implements the back and forward buttons in the Scripts panel.

        Reviewed by Adam Roben.

        * page/inspector/ScriptsPanel.js:
        (WebInspector.ScriptsPanel): Add the back and forward button elements
        and remove the FIXMEs.
        (WebInspector.ScriptsPanel.prototype.reset): Reset _backForwardList
        and _currentBackForwardIndex. Then call _updateBackAndForwardButtons().
        Added a fromBackForwardAction argument. If fromBackForwardAction is false
        then update the _currentBackForwardIndex and _backForwardList.
        Update the disabled state of the buttons.
        (WebInspector.ScriptsPanel.prototype._goBack): Update the _currentBackForwardIndex
        and call _showScriptOrResource() and _updateBackAndForwardButtons().
        (WebInspector.ScriptsPanel.prototype._goForward): Ditto.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34276 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Made clicking a Console message URL originating from a JavaScript
timothy@apple.com [Sat, 31 May 2008 06:05:57 +0000 (06:05 +0000)]
    Made clicking a Console message URL originating from a JavaScript
        source show the Resource in the Scripts panel. This only happens
        when the debugger is attached.

        Reviewed by Adam Roben.

        * page/inspector/Console.js:
        Add a preferredPanel of "scripts".
        Add a preferredPanel of "scripts" when the source is JS.
        * page/inspector/ScriptView.js:
        (WebInspector.ScriptView.prototype.highlightLine): Added.
        Calls through to the sourceFrame. Matches SourceView.
        * page/inspector/ScriptsPanel.js:
        (WebInspector.ScriptsPanel.prototype.canShowResource): Added.
        Returns true if the debugger is attached and the resource has scripts.
        (WebInspector.ScriptsPanel.prototype._showScriptOrResource): Call
        highlightLine on the view.
        * page/inspector/inspector.js:
        (WebInspector.documentClick): Pass the preferredPanel to showResourceForURL.
        (WebInspector.showResourceForURL): Resolve the preferredPanel from the string
        passed in. Checks if the panel implements showResource and canShowResource.
        Reverts to the Resources panel if any of that is false or not implemented.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34275 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Fixed an "undefined type" exception that was being thrown when
timothy@apple.com [Sat, 31 May 2008 06:05:43 +0000 (06:05 +0000)]
    Fixed an "undefined type" exception that was being thrown when
        selecting a eval script from the file menu in the Scripts panel.
        This would also happen when stepping into an eval which would
        break other parts of the Inspector interface.

        Reviewed by Adam Roben.

        * page/inspector/ScriptsPanel.js:
        Use the scriptOrResource variable instead of the undefined script variable.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34274 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoMade the starting line number of scripts be 1-based throughout the engine.
timothy@apple.com [Sat, 31 May 2008 06:05:28 +0000 (06:05 +0000)]
Made the starting line number of scripts be 1-based throughout the engine.


2008-05-30  Timothy Hatcher  <timothy@apple.com>

        Made the starting line number of scripts be 1-based throughout the engine.
        This cleans up script line numbers so they are all consistent now and fixes
        some cases where script execution was shown as off by one line in the debugger.

        No change in SunSpider.

        Reviewed by Oliver Hunt.

        * API/minidom.c:
        (main): Pass a line number of 1 instead of 0 to parser().parse().
        * API/testapi.c:
        (main): Ditto. And removes a FIXME and changed an assertEqualsAsNumber
        to use 1 instead of 2 for the line number.
        * VM/Machine.cpp:
        (KJS::callEval): Pass a line number of 1 instead of 0.
        (KJS::Machine::debug): Use firstLine for WillExecuteProgram instead of
        lastLine. Use lastLine for DidExecuteProgram instead of firstLine.
        * kjs/DebuggerCallFrame.cpp:
        (KJS::DebuggerCallFrame::evaluate): Pass a line number of 1 instead of
        0 to parser().parse().
        * kjs/Parser.cpp:
        (KJS::Parser::parse): ASSERT startingLineNumber is greatter than 0. Change
        the startingLineNumber to be 1 if it was less than or equal to 0. This is needed
        for release builds to maintain compatibility with the JavaScriptCore API.
        * kjs/function.cpp:
        (KJS::globalFuncEval): Pass a line number of 1 instead of 0 to parser().parse().
        * kjs/function_object.cpp:
        (FunctionObjectImp::construct): Pass a line number of 1 instead of 0 to construct().
        * kjs/lexer.cpp:
        (Lexer::setCode): Made yylineno = startingLineNumber instead of adding 1.
        * kjs/testkjs.cpp:
        (functionRun): Pass a line number of 1 instead of 0 to Interpreter::evaluate().
        (functionLoad): Ditto.
        (prettyPrintScript): Ditto.
        (runWithScripts): Ditto.
        * profiler/Profiler.cpp:
        (WebCore::createCallIdentifier): Removed a plus 1 of startingLineNumber.


2008-05-30  Timothy Hatcher  <timothy@apple.com>

        Made the starting line number of scripts be 1-based throughout the engine.
        This cleans up script line numbers so they are all consistent now and fixes
        some cases where script execution was shown as off by one line in the debugger.

        Doing this also exposed a bug where JSLazyEventListener created in XHML or SVG
        documents would always have a line number of 0. So this change fixed that bug
        to pass all the SVG and XHTML tests.

        All layout tests pass.

        Reviewed by Oliver Hunt.

        * bindings/js/kjs_events.cpp:
        (WebCore::JSLazyEventListener::JSLazyEventListener): Set the line number to 1
        if it was passed in as 0. This can happen when listeners are created with
        a setAttribute call from JavaScript.
        (WebCore::JSLazyEventListener::parseCode): Add a FIXME about the URL being
        incorrect when listeners are created with a setAttribute call from JavaScript.
        * bindings/js/kjs_events.h: Remove the default value for lineNumber, since no
        callers need it.
        * bindings/objc/WebScriptObject.mm:
        (-[WebScriptObject evaluateWebScript:]): Pass a line number of 1 instead of 0
        to Interpreter::evaluate().
        * bridge/NP_jsobject.cpp:
        (_NPN_Evaluate): Ditto.
        * bridge/jni/jni_jsobject.mm:
        (JavaJSObject::eval): Ditto.
        * dom/XMLTokenizer.cpp:
        (WebCore::XMLTokenizer::startElementNs): Call KJSProxy::setEventHandlerLineno()
        around the call to handleElementAttributes, so any JSLazyEventListener created
        from those attributes have line numbers.
        (WebCore::XMLTokenizer::endElementNs): Remove a minus 1 of the line number.
        (WebCore::XMLTokenizer::notifyFinished): Pass a line number of 1 instead of 0.
        (WebCore::XMLTokenizer::parseEndElement): Remove a minus 1 of the line number.
        * html/HTMLScriptElement.cpp:
        (WebCore::HTMLScriptElement::evaluateScript): Add a FIXME about the starting
        line number being incorrect in some cases when this function is called.
        * html/HTMLTokenizer.cpp:
        (WebCore::HTMLTokenizer::parseSpecial): Add a plus 1 to the line number when
        setting scriptStartLineno so it is 1-based. Same for calling setEventHandlerLineno().
        (WebCore::HTMLTokenizer::processToken): Ditto.
        * html/HTMLTokenizer.h: Change the default line number on scriptExecution() to 1 from 0.
        * loader/FrameLoader.cpp:
        (FrameLoader::executeIfJavaScriptURL): Pass a line number of 1 instead of 0 to executeScript().


2008-05-30  Timothy Hatcher  <timothy@apple.com>

        Made the starting line number of scripts be 1-based throughout the engine.
        This cleans up script line numbers so they are all consistent now.

        Reviewed by Oliver Hunt.

        * DumpRenderTree/mac/ObjCController.m:
        (runJavaScriptThread): Pass a line number of 1 instead of 0 to JSEvaluateScript.
        * DumpRenderTree/pthreads/JavaScriptThreadingPthreads.cpp:
        (runJavaScriptThread): Ditto.
        * DumpRenderTree/win/DumpRenderTree.cpp:
        (runJavaScriptThread): Ditto.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34273 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Fixes the regression where image resources don't have a preview icon.
timothy@apple.com [Sat, 31 May 2008 06:05:04 +0000 (06:05 +0000)]
    Fixes the regression where image resources don't have a preview icon.


        Reviewed by Adam Roben.

        * page/inspector/ResourcesPanel.js:
        (WebInspector.ResourceSidebarTreeElement): Call createIconElement before
        calling the superclass.
        Create an iconElement, if the category is images then make a div that
        contains a preview image. Otherwise just make an img element. If there
        was a previous iconElement, then replace it.
        (WebInspector.ResourceSidebarTreeElement.prototype.refresh): Call
        createIconElement if the category changed.
        * page/inspector/SidebarTreeElement.js:
        (WebInspector.SidebarTreeElement): Create an iconElement if one hasn't
        been made already by a subclass.
        (WebInspector.SidebarTreeElement.prototype.onattach): Append the iconElement
        instead of creating one each time.
        * page/inspector/inspector.css: New and updated style rules.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34272 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-30 Maciej Stachowiak <mjs@apple.com>
mjs@apple.com [Sat, 31 May 2008 05:37:03 +0000 (05:37 +0000)]
2008-05-30  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Oliver (earlier version reviewed by Alexey).

        - speculative fix for "REGRESSION(r34143?): Frequent crash while browsing"

        ("This Time for Sure" Edition)

        I'm pretty sure this fixes it but I have not been able to
        reproduce and am unsure if my theory of the bug is right.

        I belive the bug was because JSDOMWindowBase accessed
        JSDOMWindowShell in its destructor to remove itself from a
        hashtable, but GC destructor order is not guaranteed, so the
        hashtable may have been freed already. This patch changes things
        so that a non-GC object (the KJSProxy) does the tracking of live
        window objects for a frame. JSDOMWindowBase can null check the frame
        pointer to verify if it is still good.

        In addition, we must create a similar setup between DOMWindow and
        Frame; since the DOMWindow of a given frame can now change over
        time, we must ensure that the Frame disconnects every live
        DOMWindow when destroyed, not just the last.

        * bindings/js/JSDOMWindowBase.cpp:
        * bindings/js/JSDOMWindowShell.cpp:
        * bindings/js/JSDOMWindowShell.h:
        * bindings/js/kjs_proxy.cpp:
        * bindings/js/kjs_proxy.h:
        * page/DOMWindow.cpp:
        * page/Frame.cpp:
        * page/Frame.h:
        * page/FramePrivate.h:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34271 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
mitz@apple.com [Fri, 30 May 2008 21:34:30 +0000 (21:34 +0000)]

        Reviewed by Darin Adler and Dave Hyatt.

        - eliminate excessive repainting when an object's final position
          after layout is unchanged

        * rendering/RenderBlock.cpp:
        (WebCore::RenderBlock::layoutBlockChildren): Removed the full repaint
        for the case that the final position is unchanged but the position at
        which an intermediate layout occurred was different. Any repainting
        done during the intermediate layout would have happened at the object's
        initial (and also final) coordinates, since layoutDelta is factored
        into repaint rect calculations.


        Reviewed by Darin Adler.

        - updated pixel results for "eliminate excessive repainting when an
          object's final position after layout is unchanged"

        * platform/mac/fast/repaint/intermediate-layout-position-clip-expected.checksum:
        * platform/mac/fast/repaint/intermediate-layout-position-clip-expected.png:
        * platform/mac/fast/repaint/intermediate-layout-position-expected.checksum:
        * platform/mac/fast/repaint/intermediate-layout-position-expected.png:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34267 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-29 Justin Garcia <justin.garcia@apple.com>
justin.garcia@apple.com [Fri, 30 May 2008 21:30:39 +0000 (21:30 +0000)]
2008-05-29  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Darin.

        <rdar://problem/5949462> REGRESSION: Can't paste screen captures into Mail

        AppKit started putting PNG instead of PICT onto the pasteboard for screen
        captures.  Added support for PNG with kUTTypePNG.  Tiger doesn't support
        setting and retrieving pasteboard types with UTIs, but we don't know of any
        applications on Tiger that put only PNG on the pasteboard.

        * WebView/WebHTMLView.mm:
        (-[WebHTMLView _documentFragmentFromPasteboard:inContext:allowPlainText:]):
        (-[WebHTMLView _documentFragmentFromPasteboard:forType:inContext:subresources:]):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34266 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by Darin.
ap@webkit.org [Fri, 30 May 2008 20:10:38 +0000 (20:10 +0000)]
    Reviewed by Darin.

        speed up SunSpider by optimizing immediate number cases

        Also fixed a JavaScriptCore regression seen on PowerPC - we didn't clip left shift
        parameter to 0...31.

        0.5% improvement on SunSpider overall, although a 8.5 regression on bitops-3bit-bits-in-byte.

        * VM/Machine.cpp:
        Added fast paths for >>> and <<.

        * kjs/JSImmediate.h:
        (KJS::JSImmediate::toTruncatedUInt32): Added. Same as getTruncatedInt32, but casts the result
        to unsigned.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34265 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago<rdar://problem/5959478> r34079: AX: crash at stringForReplacedNode
cfleizach@apple.com [Fri, 30 May 2008 18:05:04 +0000 (18:05 +0000)]
<rdar://problem/5959478> r34079: AX: crash at stringForReplacedNode

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34264 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-30 Carlos Martín Nieto <carlos@cmartin.tk>
jmalonzo@webkit.org [Fri, 30 May 2008 14:15:35 +0000 (14:15 +0000)]
2008-05-30  Carlos Martín Nieto  <carlos@cmartin.tk>

        Reviewed by Alp Toker.

        [GTK] The "hovering-over-link" signal arguments aren't documented.

        * webkit/webkitwebview.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34263 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-30 Tor Arne Vestbø <tavestbo@trolltech.com>
hausmann@webkit.org [Fri, 30 May 2008 11:28:46 +0000 (11:28 +0000)]
2008-05-30  Tor Arne Vestbø  <tavestbo@trolltech.com>

        Reviewed by Simon.

        Fix the build dependencies for the Qt build. Don't generate
        SVGCSSPropertyNames.cpp for every build.

        * WebCore.pro: The output file was renamed, so the compiler rule needs
        to be adjusted in order for qmake to generate correct depdencies.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34262 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-30 Maciej Stachowiak <mjs@apple.com>
mjs@apple.com [Fri, 30 May 2008 10:40:19 +0000 (10:40 +0000)]
2008-05-30  Maciej Stachowiak  <mjs@apple.com>

        Revert fix for 19285, it just caused more crashes and I need time
        to fix it properly.

        * bindings/js/JSDOMWindowBase.cpp:
        * bindings/js/JSDOMWindowShell.cpp:
        * bindings/js/JSDOMWindowShell.h:
        * bindings/js/kjs_proxy.cpp:
        * bindings/js/kjs_proxy.h:
        * page/Frame.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34261 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-30 Adam Treat <treat@kde.org>
hausmann@webkit.org [Fri, 30 May 2008 10:28:25 +0000 (10:28 +0000)]
2008-05-30  Adam Treat  <treat@kde.org>

        Reviewed by Simon.

        Fix a huge memory leak by ensuring that on application shutdown
        the shared timer is fired one last time if it is active. This
        ensures that the GCController timer is fired at the end to free
        references to JavaScript objects.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34260 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by Darin.
ap@webkit.org [Fri, 30 May 2008 10:08:31 +0000 (10:08 +0000)]
    Reviewed by Darin.

        <rdar://problem/4657563> Use of Ctrl as access key modifier conflicts with Mac OS X
        emacs-style keybindings

        * page/EventHandler.cpp:
        (WebCore::EventHandler::handleAccessKey): Use Ctrl+Option for access keys on Mac OS X.

        * fast/events/access-key-self-destruct.html:
        * fast/forms/access-key.html:
        * fast/forms/focus-selection-input-expected.txt:
        * fast/forms/focus-selection-input.html:
        * fast/forms/legend-access-key.html:
        * fast/forms/select-accesskey.html:
        Updated tests to keep passing.

        * fast/forms/select-accesskey-expected.txt: Added.
        * fast/forms/select-accesskey.html:
        * platform/mac/fast/forms/select-accesskey-expected.checksum: Removed.
        * platform/mac/fast/forms/select-accesskey-expected.png: Removed.
        * platform/mac/fast/forms/select-accesskey-expected.txt: Removed.
        Updated to keep passing, made text-only.

        * fast/forms/focus-selection-textarea.html:
        * fast/forms/focus-selection-textarea-expected.txt:
        Added missing platform-specific logic.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34259 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by Oliver Hunt.
ap@webkit.org [Fri, 30 May 2008 09:10:42 +0000 (09:10 +0000)]
    Reviewed by Oliver Hunt.

        speed up SunSpider by optimizing immediate number cases

        Also fixed three JavaScriptCore regressions seen on PowerPC - we didn't clip right shift
        parameter to 0...31.

        1.6% improvement on SunSpider, without significant regressions on any tests.

        * VM/Machine.cpp:
        Added fast paths for >>, ==, ===, !=, !==. Changed order of memory accesses in many
        cases, making them less dependent on gcc's ability to properly assign registers. With this,
        I could move exception checks back into slow code paths, and saw less randomness in general.

        * kjs/JSImmediate.h:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34258 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-30 Maciej Stachowiak <mjs@apple.com>
mjs@apple.com [Fri, 30 May 2008 08:15:31 +0000 (08:15 +0000)]
2008-05-30  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Alexey.

        - speculative fix for "REGRESSION(r34143?): Frequent crash while browsing"

        I'm pretty sure this fixes it but I have not been able to
        reproduce and am unsure if my theory of the bug is right.

        I belive the bug was because JSDOMWindowBase accessed
        JSDOMWindowShell in its destructor to remove itself from a
        hashtable, but GC destructor order is not guaranteed, so the
        hashtable may have been freed already. This patch changes things
        so that a non-GC object (the KJSProxy) does the tracking of live
        window objects for a frame. JSDOMWindowBase can null check the frame
        pointer to verify if it is still good.

        * bindings/js/JSDOMWindowBase.cpp:
        * bindings/js/JSDOMWindowShell.cpp:
        * bindings/js/JSDOMWindowShell.h:
        * bindings/js/kjs_proxy.cpp:
        * bindings/js/kjs_proxy.h:
        * page/Frame.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34257 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago<rdar://problem/5091386> Seed: VO reads form labels twice in Safari
cfleizach@apple.com [Fri, 30 May 2008 06:05:44 +0000 (06:05 +0000)]
<rdar://problem/5091386> Seed: VO reads form labels twice in Safari
<rdar://problem/4783102> Radio button/checkbox embedded with <label> tag should combine the text and the label as a single element

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34254 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoJavaScriptCore:
mjs@apple.com [Fri, 30 May 2008 05:45:16 +0000 (05:45 +0000)]

2008-05-29  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Oliver.

        - fixed <rdar://problem/5972943> REGRESSION(r33979): Flash clips do not play on cnn.com

        Finally blocks could clobber registers that had to remain live
        until they returned. This patch takes a conservative approach and
        makes sure that finally blocks do not reuse any registers that
        were previously allocated for the function. In the future this
        could probably be tightened up to be less profligate with the
        register allocation.

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


2008-05-29  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Oliver. Test by Geoff Garen.

        - fixed <rdar://problem/5972943> REGRESSION(r33979): Flash clips do not play on cnn.com

        * fast/js/finally-codegen-failure-expected.txt: Added.
        * fast/js/finally-codegen-failure.html: Added.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34250 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-29 Marco Barisione <marco@collabora.co.uk>
alp@webkit.org [Fri, 30 May 2008 03:44:11 +0000 (03:44 +0000)]
2008-05-29  Marco Barisione  <marco@collabora.co.uk>

        Reviewed (and tweaked) by Alp Toker.

        [GTK] add functions to set/get the zoom level

        * webkit/webkitwebsettings.cpp:
        * webkit/webkitwebview.cpp: Add functions to get and set the zoom
        * webkit/webkitwebview.h: Ditto.
        * webkit/webkitprivate.h:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34249 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoVersioning.
mrowe@apple.com [Thu, 29 May 2008 23:51:11 +0000 (23:51 +0000)]

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34248 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Update PrettyPatch directory list
aroben@apple.com [Thu, 29 May 2008 22:06:42 +0000 (22:06 +0000)]
    Update PrettyPatch directory list

        * PrettyPatch/PrettyPatch.rb:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34214 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Removes a use of the internal _childrenListNode property by adding
timothy@apple.com [Thu, 29 May 2008 22:03:21 +0000 (22:03 +0000)]
    Removes a use of the internal _childrenListNode property by adding
        a getter/setter for smallChildren to SidebarSectionTreeElement.

        Reviewed by Adam Roben.

        * page/inspector/ResourcesPanel.js:
        Toggle smallChildren on the resourcesTreeElement instead of setting
        the style class directly.
        * page/inspector/SidebarTreeElement.js:
        Sets or removes the small class on the _childrenListNode.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34213 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Geoffrey Garen <ggaren@apple.com>
ggaren@apple.com [Thu, 29 May 2008 21:35:12 +0000 (21:35 +0000)]
  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Adam Roben.

        Fixed VCSUtils.pm to work with git repositories inside symlinks.

        * Scripts/VCSUtils.pm: Compute a relative path from the git repository
        root, instead of the root of the filesystem, to work around a bug in
        abs2rel when traversing symlinked home directories.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34212 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoFixes a bug where the className would be reset for Resource
timothy@apple.com [Thu, 29 May 2008 21:25:43 +0000 (21:25 +0000)]
Fixes a bug where the className would be reset for Resource
tree elements and the selected state would be lost. This could
happen when sorting the resources.


Reviewed by Adam Roben.

* page/inspector/ResourcesPanel.js:
(WebInspector.ResourceSidebarTreeElement.prototype.refresh): Call
removeMatchingStyleClasses to remove previous category classes instead
of setting the whole className directly.
* page/inspector/utilities.js:
(Element.prototype.removeStyleClass): Moved code to
removeMatchingStyleClasses and call removeMatchingStyleClasses.
(Element.prototype.removeMatchingStyleClasses): Added. Code moved from
removeStyleClass and don't escape the string.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34211 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-29 David Hyatt <hyatt@apple.com>
hyatt@apple.com [Thu, 29 May 2008 21:05:11 +0000 (21:05 +0000)]
2008-05-29  David Hyatt  <hyatt@apple.com>

        Improve the performance of the GUIMark benchmark by 2x in the CoreGraphics code path.

        Whenever a foreground image changes size rapidly, we will now dynamically shift into rendering it
        using low quality scaling.  Once the animation completes, the image will repaint at high quality.
        Scaled images will still render at high quality by default, only shifting into low quality if
        the scale factor is rapidly changing.  This change raises GUIMark from 21fps to 34fps on my machine.

        Rewrite the Image draw method to avoid the use of throwaway CG subimages.  Instead the entire image is
        always drawn (with the appropriate clip and scale set up to make sure the correct subimage portion shows up
        in the destination rect).  This change raises GUIMark from 34fps to 43fps on my machine.

        Reviewed by Darin

        * platform/graphics/GraphicsContext.cpp:
        * platform/graphics/cg/ImageCG.cpp:
        * rendering/RenderImage.cpp:
        * rendering/RenderImage.h:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34210 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
justin.garcia@apple.com [Thu, 29 May 2008 20:37:53 +0000 (20:37 +0000)]

2008-05-29  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Eric.

        <rdar://problem/5847330> REGRESSION CrashTracer: [USER] 536 crashes at WebCore::highestAncestor

        * editing/CompositeEditCommand.cpp:
        (WebCore::CompositeEditCommand::breakOutOfEmptyListItem): Don't break out of an empty
        list item if its parent is non-editable, since doing so will move into non-editable content.
        * editing/DeleteSelectionCommand.cpp:
        (WebCore::DeleteSelectionCommand::initializePositionData): We want to prevent merges from
        table cells even if those table cells are non-editable.
        * editing/htmlediting.cpp:
        (WebCore::enclosingNodeOfType): Can now return non-editable nodes, for cases where the caller
        isn't going to do editing in the returned node.
        * editing/htmlediting.h:


2008-05-28  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Eric.

        <rdar://problem/5847330> REGRESSION CrashTracer: [USER] 536 crashes at WebCore::highestAncestor

        * editing/deleting/5847330-1-expected.txt: Added.
        * editing/deleting/5847330-1.html: Added.
        * editing/deleting/5847330-2-expected.txt: Added.
        * editing/deleting/5847330-2.html: Added.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34209 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago../mac:
andersca@apple.com [Thu, 29 May 2008 19:22:28 +0000 (19:22 +0000)]

2008-05-29  Anders Carlsson  <andersca@apple.com>

        Reviewed by Brady.

        icon file specified for stand alone web app causes crash if the icon can't be found

        Handle the case where iconData is null.

        * Misc/WebIconFetcher.mm:


2008-05-29  Anders Carlsson  <andersca@apple.com>

        Reviewed by Brady.

        icon file specified for stand alone web app causes crash if the icon can't be found

        Handle the case where iconData is null.

        * WebIconFetcher.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34208 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
andersca@apple.com [Thu, 29 May 2008 18:57:00 +0000 (18:57 +0000)]

2008-05-29  Anders Carlsson  <andersca@apple.com>

        Reviewed by Mitz.

        Add version member to NPCocoaEvent

        * bridge/npapi.h:


2008-05-29  Anders Carlsson  <andersca@apple.com>

        Reviewed by Mitz.

        Add version member to NPCocoaEvent

        Initialize the version member to 0 for all events.

        * Plugins/WebNetscapePluginEventHandlerCocoa.mm:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34207 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoReviewed by Darin Adler.
kevino@webkit.org [Thu, 29 May 2008 18:02:57 +0000 (18:02 +0000)]
Reviewed by Darin Adler.

Restore original behavior of isOSX() referring to the Mac port, not the OS itself.


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34206 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-29 Steve Falkenburg <sfalken@apple.com>
sfalken@apple.com [Thu, 29 May 2008 17:31:43 +0000 (17:31 +0000)]
2008-05-29  Steve Falkenburg  <sfalken@apple.com>

        Build fix.

        * kjs/array_instance.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34205 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by Darin.
ap@webkit.org [Thu, 29 May 2008 17:14:02 +0000 (17:14 +0000)]
    Reviewed by Darin.

        <rdar://problem/5969062> A crash when iterating over a sparse array backwards.

        * kjs/array_instance.cpp: Turned sparseArrayCutoff into a macro, so that using max() on it
        doesn't cause a PIC branch.
        (KJS::ArrayInstance::increaseVectorLength): Added a comment about this function not
        preserving class invariants.
        (KJS::ArrayInstance::put): Update m_storage after reallocation. Move values that fit to
        the vector from the map in all code paths.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34204 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by Jessica Kahn.
mitz@apple.com [Thu, 29 May 2008 16:08:55 +0000 (16:08 +0000)]
    Reviewed by Jessica Kahn.

        - fix <rdar://problem/5965013> Page 2 does not print correctly

        When printing, _recursiveDisplayRectIfNeededIgnoringOpacity:... and
        _recursiveDisplayAllDirtyWithLockFocus:... can be invoked without
        -viewWillDraw being sent first, which could lead to painting without
        valid layout. The fix is to ensure up-to-date layout in those methods
        when printing.

        * WebView/WebHTMLView.mm:
        (-[WebHTMLView _recursiveDisplayRectIfNeededIgnoringOpacity:isVisibleRect:rectIsVisibleRectForView:topView:]):
        (-[WebHTMLView _recursiveDisplayAllDirtyWithLockFocus:visRect:]):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34203 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by Adam Roben.
ap@webkit.org [Thu, 29 May 2008 15:05:24 +0000 (15:05 +0000)]
    Reviewed by Adam Roben.

        <rdar://problem/5960682> REGRESSION(r30535): Crashes on iExploder in checkForHeadCharset().

        I don't see how to make a reliable test, but iExploder catches this quickly.

        * loader/TextResourceDecoder.cpp:
        (WebCore::TextResourceDecoder::checkForHeadCharset): Correct the length passed to

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34202 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by Adam Roben.
ap@webkit.org [Thu, 29 May 2008 13:42:29 +0000 (13:42 +0000)]
    Reviewed by Adam Roben.

        Fix run-iexploder-tests and run-mangleme-tests to work with updated shared scripts and
        configuration files.

        * Scripts/run-iexploder-tests:
        * Scripts/run-mangleme-tests:
        Renamed runSafari to not conflict with the one in webkitdirs.
        Added SSLCertificateFile option for httpd, as now needed.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34201 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-29 Peter Kasting <zerodpx@gmail.com>
alp@webkit.org [Thu, 29 May 2008 12:53:03 +0000 (12:53 +0000)]
2008-05-29  Peter Kasting  <zerodpx@gmail.com>

        Reviewed by Dave Hyatt.

        Handle looping GIFs correctly (at least in the open source GIF decoder) even when
        the loop count doesn't appear in the initial data packets.

        * platform/graphics/BitmapImage.cpp:
        * platform/image-decoders/gif/GIFImageDecoder.cpp:
        * platform/image-decoders/gif/GIFImageDecoder.h:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34200 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-29 Brent Fulgham <bfulgham@gmail.com>
alp@webkit.org [Thu, 29 May 2008 11:28:19 +0000 (11:28 +0000)]
2008-05-29  Brent Fulgham  <bfulgham@gmail.com>

        Reviewed by Alp Toker.

        Correct Windows (Cairo) Build Regressions

        Correct small Windows (Cairo) build regressions in recent updates.

        * platform/graphics/SimpleFontData.h:  Correct build regression due
          to r32781 (additional use of m_syntheticBoldOffset in GDI code).
        * platform/graphics/cairo/GraphicsContextPlatformPrivateCairo.h:
          Correct build regression due to r31830 (change of clip signature
          to use FloatRect rather than IntRect).
        * platform/graphics/win/SimpleFontDataCairoWin.cpp:
        (WebCore::SimpleFontData::platformInit):  Initialize
        * platform/image-decoders/jpeg/JPEGImageDecoder.cpp:
        (WebCore::JPEGImageReader::JPEGImageReader):  Work around build
          error due to warnings about _setjmp treated as errors.
        * platform/image-decoders/png/PNGImageDecoder.cpp:  Work around
          build error due to warnings about _setjmp treated as errors.
        * platform/network/curl/ResourceHandleCurl.cpp:  Correct build error
          due to different order (and content) of include files when building
          with CoreFoundation (but not CoreGraphics).
        * platform/network/curl/ResourceHandleManager.cpp:
        (WebCore::ResourceHandleManager::setupPOST):  Correct build error
          due to Visual Studio compiler bug.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34199 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-29 Thiago Macieira <tjmaciei@trolltech.com>
hausmann@webkit.org [Thu, 29 May 2008 10:16:18 +0000 (10:16 +0000)]
2008-05-29  Thiago Macieira  <tjmaciei@trolltech.com>

        Reviewed by Simon.

        Fix compilation in Solaris with Sun CC

        Lots of WebKit code uses C99 functions that, strict as it
        is, the Solaris system doesn't provide in C++. So we must define them
        for both GCC and the Sun CC.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34198 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by Darin.
ap@webkit.org [Thu, 29 May 2008 09:43:00 +0000 (09:43 +0000)]
    Reviewed by Darin.

        REGRESSION: transformToDocument fails when xsl includes &#160;

        Test: fast/xsl/nbsp-in-stylesheet.html

        * editing/markup.cpp:
        Only escape non-breaking spaces in HTML documents. In Firefox, this behavior is also limited
        to innerHTML, but we don't have a separate code path for it, and do not necessarily want to.

        * page/Frame.cpp:
        * dom/CDATASection.cpp:
        * dom/CDATASection.h:
        * dom/Comment.cpp:
        * dom/Comment.h:
        * dom/DocumentType.cpp:
        * dom/DocumentType.h:
        * dom/ProcessingInstruction.cpp:
        * dom/ProcessingInstruction.h:
        While at it, I also wanted to fix escaping for these nodes, but it turned out that FIXMEs were
        incorrect. So, I just moved their serialization to markup.cpp for consistency.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34197 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoKavindra Devi Palaraja <kdpalara@trolltech.com>
hausmann@webkit.org [Thu, 29 May 2008 09:37:49 +0000 (09:37 +0000)]
Kavindra Devi Palaraja <kdpalara@trolltech.com>

Doc: Mention the requirement of a QApplication with QtWebKit

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34196 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoDavid Boddie <dboddie@trolltech.com>
hausmann@webkit.org [Thu, 29 May 2008 09:37:35 +0000 (09:37 +0000)]
David Boddie <dboddie@trolltech.com>

Some clarifications for the documentation.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34195 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoFix ChangeLog typos.
mjs@apple.com [Thu, 29 May 2008 08:11:30 +0000 (08:11 +0000)]
Fix ChangeLog typos.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34194 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
mjs@apple.com [Thu, 29 May 2008 08:05:12 +0000 (08:05 +0000)]

2008-05-29  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Dave Hyatt.

        - speed up DHTML using lazy style resolution and rendere creation

        This change introduces the concept of "lazy attach" - when a node
        is lazy attached, then instead of resolving style and creating a
        renderer right away, we just mark it as needing a style recalc.

        The patch makes use of this mechanism when inserting nodes directly
        using DOM APIs from script. For now this is only done for the
        JavaScript language binding but could also be done for other
        bindings in the future.

        Lazy attach helps some common DHTML patterns - when a node is
        added to the DOM, and then subsequently changed in a
        style-affecting way, this causes an extra style recalc. This is a
        fairly common pattern so it is better to be lazy.

        * bindings/js/JSNodeCustom.cpp:
        (WebCore::JSNode::insertBefore): Request lazy attach.
        (WebCore::JSNode::replaceChild): ditto
        (WebCore::JSNode::appendChild): ditto
        * dom/ContainerNode.cpp:
        (WebCore::ContainerNode::insertBefore): Support lazy attach.
        (WebCore::ContainerNode::replaceChild): ditto
        (WebCore::ContainerNode::appendChild): ditto
        (WebCore::ContainerNode::detach): Clear "changed child" bit if still set.
        * dom/ContainerNode.h:
        * dom/Element.cpp:
        (WebCore::Element::recalcStyle): Adjusted to properly reattach a
        lazy-attached node.
        * dom/Node.cpp:
        (WebCore::Node::insertBefore): Extra parameter for lazy attach
        (still doesn't do anything).
        (WebCore::Node::replaceChild): ditto
        (WebCore::Node::appendChild): ditto
        (WebCore::Node::setChanged): Unrelated but obvious optimization -
        stop marking ancestor as having a changed child once we already reach
        an ancestor so marked.
        (WebCore::outermostLazyAttachedAncestor): Helper function for lazyAttach.
        (WebCore::Node::lazyAttach): Implement lazy attach.
        (WebCore::Node::canLazyAttach): Virtual method - true for most nodes.
        * dom/Node.h:
        * dom/Text.cpp:
        (WebCore::Text::recalcStyle): Properly handle the case of a reattached node.
        * html/HTMLEmbedElement.h:
        (WebCore::HTMLEmbedElement::canLazyAttach): Refuse lazy attach, since
        plugins and frames do important work at rederer creation time.
        * html/HTMLFrameElementBase.h:
        (WebCore::HTMLFrameElementBase::canLazyAttach): Refuse lazy attach, since
        plugins and frames do important work at rederer creation time.
        * html/HTMLFrameSetElement.cpp:
        (WebCore::HTMLFrameSetElement::recalcStyle): Change order so that
        reattach works properly.
        * html/HTMLObjectElement.h:
        (WebCore::HTMLObjectElement::canLazyAttach): Refuse lazy attach, since
        plugins and frames do important work at rederer creation time.
        * html/HTMLOptGroupElement.cpp:
        (WebCore::HTMLOptGroupElement::insertBefore): Pass along extra param.
        (WebCore::HTMLOptGroupElement::replaceChild): ditto
        (WebCore::HTMLOptGroupElement::appendChild): ditto
        * html/HTMLOptGroupElement.h:
        * html/HTMLSelectElement.cpp:
        (WebCore::HTMLSelectElement::insertBefore): Pass along extra param.
        (WebCore::HTMLSelectElement::replaceChild): ditto
        (WebCore::HTMLSelectElement::appendChild): ditto
        * html/HTMLSelectElement.h:
        * svg/SVGLocatable.cpp:
        (WebCore::SVGLocatable::getBBox): Add missing updateLayout call.
        * svg/SVGTextContentElement.cpp:
        (WebCore::SVGTextContentElement::getNumberOfChars): ditto
        (WebCore::SVGTextContentElement::getComputedTextLength): ditto
        (WebCore::SVGTextContentElement::getSubStringLength): ditto
        (WebCore::SVGTextContentElement::getStartPositionOfChar): ditto
        (WebCore::SVGTextContentElement::getEndPositionOfChar): ditto
        (WebCore::SVGTextContentElement::getExtentOfChar): ditto
        (WebCore::SVGTextContentElement::getRotationOfChar): ditto
        (WebCore::SVGTextContentElement::getCharNumAtPosition): ditto


2008-05-29  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Dave Hyatt.

        - Test cases for this change: "speed up DHTML using lazy style resolution and rendere creation"

        * http/tests/misc/acid3-expected.txt:
        * platform/mac/fast/dynamic/insert-before-table-part-in-continuation-expected.checksum:
        * platform/mac/fast/dynamic/insert-before-table-part-in-continuation-expected.png:
        * platform/mac/fast/dynamic/insert-before-table-part-in-continuation-expected.txt:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34193 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoFix codegen for assignment being used as a function.
oliver@apple.com [Thu, 29 May 2008 04:15:45 +0000 (04:15 +0000)]
Fix codegen for assignment being used as a function.

Reviewed by Anders.

FunctionCallValueNode::emitCode failed to account for the
potential of the function expression to allocate arbitrary

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34192 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-28 Geoffrey Garen <ggaren@apple.com>
bdakin@apple.com [Thu, 29 May 2008 02:11:41 +0000 (02:11 +0000)]
2008-05-28  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Oliver Hunt.

        Fixed total destruction in Software Update.app.

        Updated JavaScriptGlue to honor squirrelfish's new getCallData mechanism for
        indicating that an object can be called as a function.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34191 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago <rdar://problem/5957036> REGRESSION (r31960): 20-30% slowdown in i-Bench...
adachan@apple.com [Thu, 29 May 2008 00:51:06 +0000 (00:51 +0000)]
    <rdar://problem/5957036> REGRESSION (r31960): 20-30% slowdown in i-Bench JavaScript test on XP Home
        Conditionalize the timer latency change (r31960) only for vista since it does not
        help xp anyway.

        Reviewed by Steve Falkenburg.

        * platform/win/SharedTimerWin.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34190 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-28 Anders Carlsson <andersca@apple.com>
andersca@apple.com [Thu, 29 May 2008 00:44:03 +0000 (00:44 +0000)]
2008-05-28  Anders Carlsson  <andersca@apple.com>

        Reviewed by Jon.

        Remove workaround, this is no longer a problem.

        * WebView/WebView.mm:
        (-[WebView _removeObjectForIdentifier:]):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34189 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-28 Anders Carlsson <andersca@apple.com>
andersca@apple.com [Wed, 28 May 2008 23:20:17 +0000 (23:20 +0000)]
2008-05-28  Anders Carlsson  <andersca@apple.com>

        Build fix.

        * storage/LocalStorage.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34188 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-28 Anders Carlsson <andersca@apple.com>
andersca@apple.com [Wed, 28 May 2008 22:50:02 +0000 (22:50 +0000)]
2008-05-28  Anders Carlsson  <andersca@apple.com>

        Reviewed by Tim.

        CrashTracer: [USER] 4 crashes in Safari at libxml2.2.dylib: xmlGenericErrorDefaultFunc + 76

        Turns out that while libxml has two different error callbacks, they both use the same context, so we need to
        set both error callbacks to prevent one of the error callbacks from using the context as a FILE pointer.

        * xml/XSLStyleSheet.cpp:
        * xml/XSLTProcessor.cpp:
        * xml/XSLTProcessor.h:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34187 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-28 Brady Eidson <beidson@apple.com>
beidson@apple.com [Wed, 28 May 2008 22:45:13 +0000 (22:45 +0000)]
2008-05-28  Brady Eidson  <beidson@apple.com>

        Reviewed by Adam

        Fix hang I introduced to WebKit clients without their LocalStorage paths set
        Such as DRT...

        * storage/LocalStorage.cpp:
        (WebCore::LocalStorage::scheduleImport): Return a bool, namely to signify
          whether or not there is a LocalStorageThread to schedule the import on.
        * storage/LocalStorage.h:

        * storage/LocalStorageArea.cpp:
        (WebCore::LocalStorageArea::LocalStorageArea): If the scheduleImport failed,
          mark the imported flag as true so the LocalStorageArea doesn't wait around
          locking the main thread.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34186 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-28 Anthony Ricaud <rik24d@gmail.com>
pewtermoose@webkit.org [Wed, 28 May 2008 22:25:41 +0000 (22:25 +0000)]
2008-05-28  Anthony Ricaud  <rik24d@gmail.com>

        User stylesheets are now non-editable and displayed as user-stylesheets


        Reviewed by Tim Hatcher.

        * English.lproj/localizedStrings.js: Added a string
        * page/inspector/StylesSidebarPane.js:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34185 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-28 Anders Carlsson <andersca@apple.com>
andersca@apple.com [Wed, 28 May 2008 21:00:44 +0000 (21:00 +0000)]
2008-05-28  Anders Carlsson  <andersca@apple.com>

        Reviewed by Mitz.

        Crash in WebKit!KJS::Window::printErrorMessage quitting Safari after pop up window on uk.real.com was closed

        Add null check for Frame::settings().

        * bindings/js/JSDOMWindowBase.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34184 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoReviewed by Kevin Ollivier.
kevino@webkit.org [Wed, 28 May 2008 20:57:52 +0000 (20:57 +0000)]
Reviewed by Kevin Ollivier.

This patch adds a new wx event type and code to send it when a new title is set by the page being loaded.


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34183 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoJavaScriptCore:
ggaren@apple.com [Wed, 28 May 2008 20:47:13 +0000 (20:47 +0000)]

2008-05-27  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Tim Hatcher.

        Fixed https://bugs.webkit.org/show_bug.cgi?id=19183
        REGRESSION (r33979): Crash in DebuggerCallFrame::functionName when
        clicking button in returnEvent-crash.html

        Added two new debugger hooks, willExecuteProgram and didExecuteProgram,
        along with code to generate them, code to invoke them when unwinding
        due to an exception, and code to dump them.

        SunSpider reports no change.

        * VM/CodeBlock.cpp:
        (KJS::debugHookName): I had to mark this function NEVER_INLINE to avoid
        a .4% performance regression. The mind boggles.


2008-05-27  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Tim Hatcher.

        Fixed https://bugs.webkit.org/show_bug.cgi?id=19183
        REGRESSION (r33979): Crash in DebuggerCallFrame::functionName when
        clicking button in returnEvent-crash.html

        Added implementations for willExecuteProgram and didExecuteProgram. They
        take care to update our call frame when entering and exiting programs,
        preventing us from keeping around a stale global frame after executing
        a program.

        eval programs now show up as "anonymous function" in a new scope. This
        is slightly better than what they used to do -- overwriting the current
        scope -- but obviously we can do better.


2008-05-27  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Tim Hatcher.

        Fixed https://bugs.webkit.org/show_bug.cgi?id=19183
        REGRESSION (r33979): Crash in DebuggerCallFrame::functionName when
        clicking button in returnEvent-crash.html

        Added implementations for willExecuteProgram and didExecuteProgram, which
        take care of making sure we're not hanging on to stale data.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34182 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoReviewed by Kevin Ollivier.
kevino@webkit.org [Wed, 28 May 2008 20:28:01 +0000 (20:28 +0000)]
Reviewed by Kevin Ollivier.

This patch tweaks the wxWebView class to make it conform to normal wx patterns for widget classes.  It adds a default ctor and the Create method so it can use the 2-phase create pattern, adds wxRTTI macros which is important for wxPython and XRC, and fixes the LoadURL method to pass a wxString reference to save a copy.


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34181 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Fix JavaScriptCore tests on OS X
aroben@apple.com [Wed, 28 May 2008 20:13:53 +0000 (20:13 +0000)]
    Fix JavaScriptCore tests on OS X

        We were quoting the path to testkjs too late, after it had already
        been combined with spaces and other options.

        * tests/mozilla/jsDriver.pl:
        (top level): Move path quoting from here...
        (sub get_kjs_engine_command): ...to here.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34180 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-28 Brady Eidson <beidson@apple.com>
beidson@apple.com [Wed, 28 May 2008 18:55:32 +0000 (18:55 +0000)]
2008-05-28  Brady Eidson  <beidson@apple.com>

        Reviewed by Dan Bernstein

        <rdar://problem/5946834> - LocalStorage threads appearing in apps that don't want them.

        Decided to fix this bug from a few different avenues to make sure we only create LocalStorage resources
        when we *really* want them.

        * page/DOMWindow.cpp:
        (WebCore::DOMWindow::localStorage): Account for the possibility that the LocalStorage object might be null.

        * storage/LocalStorage.cpp:
        (WebCore::LocalStorage::LocalStorage): Don't create the thread if there is no path name - we know we'll never
          be needing it when there is no file backing us on disk.
        (WebCore::LocalStorage::scheduleImport): Null check the thread, as it might not exist.
        (WebCore::LocalStorage::scheduleSync): Ditto

        * storage/LocalStorageThread.cpp:
        (WebCore::LocalStorageThread::localStorageThread): Clear the threadID after termination - clear way to signify
          that this thread is truly done.
        (WebCore::LocalStorageThread::scheduleImport): ASSERT that the queue is live, and the threadID exists
        (WebCore::LocalStorageThread::scheduleSync): Ditto
        (WebCore::LocalStorageThread::terminate): Don't schedule termination and wait for it to complete if the thread
          isn't live.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34179 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoReviewed by Kevin Ollivier.
kevino@webkit.org [Wed, 28 May 2008 18:36:50 +0000 (18:36 +0000)]
Reviewed by Kevin Ollivier.

Rescale image to fit the destination rect so that we don't get a crash when drawing in a destination rect that's bigger than the image rect. Also adds an optimization for drawing an image that is a solid color.


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34178 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoJavaScriptCore:
andersca@apple.com [Wed, 28 May 2008 18:13:37 +0000 (18:13 +0000)]

2008-05-28  Anders Carlsson  <andersca@apple.com>

        Reviewed by Oliver.

        <rdar://problem/5968071> "const f" crashes in JavaScriptCore

        Make sure to null check the initializer.

        * kjs/nodes.cpp:


2008-05-28  Anders Carlsson  <andersca@apple.com>

        Reviewed by Oliver.

        <rdar://problem/5968071> "const f" crashes in JavaScriptCore

        * fast/js/const-without-initializer-expected.txt: Added.
        * fast/js/const-without-initializer.html: Added.
        * fast/js/resources/const-without-initializer.js: Added.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34177 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-28 Alp Toker <alp@nuanti.com>
alp@webkit.org [Wed, 28 May 2008 18:04:20 +0000 (18:04 +0000)]
2008-05-28  Alp Toker  <alp@nuanti.com>

        Reviewed by Alexey Proskuryakov.

        [gtk] cannot type letters with dead keys in textarea

        [GDK] Support input methods

        Add support for input method composition.

        Remove some old hacks so we handle input method events more

        * WebCoreSupport/EditorClientGtk.cpp:
        * webkit/webkitwebview.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34176 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Make PrettyPatch able to handle diffs taken from outside the WebKit source...
aroben@apple.com [Wed, 28 May 2008 17:31:37 +0000 (17:31 +0000)]
    Make PrettyPatch able to handle diffs taken from outside the WebKit source tree

        Part of Bug 19290: More patches not handled by PrettyPatch.rb

        Reviewed by David Kilzer.

        * PrettyPatch/PrettyPatch.rb:
        (PrettyPatch.find_url_and_path): Added. Searches the file path from
        the bottom up looking for a directory that exists in the source tree.
        (PrettyPatch.linkifyFilename): Changed to call find_url_and_path.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34175 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Make run-javascriptcore-tests work with a space in the path to testkjs
aroben@apple.com [Wed, 28 May 2008 17:31:22 +0000 (17:31 +0000)]
    Make run-javascriptcore-tests work with a space in the path to testkjs

        Reviewed by Alexey Proskuryakov.

        * tests/mozilla/jsDriver.pl: Quote the path to the engine so that
        spaces will be interpreted correctly.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34174 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Print exceptions from PrettyPatch to stdout
aroben@apple.com [Wed, 28 May 2008 15:46:18 +0000 (15:46 +0000)]
    Print exceptions from PrettyPatch to stdout

        This will make it much easier to debug problems with PrettyPatch,
        since we'll be able to see the exceptions in the browser.

        Reviewed by Sam Weinig.

        * PrettyPatch/prettify.rb: Added a --html-exceptions option, which
        will print exceptions to stdout as HTML.
        * attachment.cgi: Changed to pass --html-exceptions to prettify.rb.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34173 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Fixed a misguiding comment - my measurement for negative numbers only include...
ap@webkit.org [Wed, 28 May 2008 12:36:31 +0000 (12:36 +0000)]
    Fixed a misguiding comment - my measurement for negative numbers only included cases
        where both operands were negative, which is not very interesting.

        * VM/Machine.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34172 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by Maciej.
ap@webkit.org [Wed, 28 May 2008 10:48:04 +0000 (10:48 +0000)]
    Reviewed by Maciej.

        Based on a patch by Oliver Hunt.

        speed up SunSpider by optimizing immediate number cases

        1.4% speedup on SunSpider.

        * VM/Machine.cpp:
        * kjs/JSImmediate.h:
        Added fast paths for ++ and --.

        (KJS::JSImmediate::canDoFastAdditiveOperations): Corrected a comment.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34171 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Reviewed by Darin.
ap@webkit.org [Wed, 28 May 2008 09:47:29 +0000 (09:47 +0000)]
    Reviewed by Darin.

        speed up SunSpider by optimizing immediate number cases

        2% speedup overall, maximum 10% on controlflow-recursive and bitops-3bit-bits-in-byte,
        but a 4% regression on bitops-bits-in-byte and bitops-bitwise-and.

        * kjs/JSImmediate.h:
        Added fast cases that work with positive values less than 2^30.

        * VM/Machine.cpp:
        (KJS::Machine::privateExecute): Use the above operations. Also updated SunSpider frequencies
        with my results (looks like tag values have changed, not sure what caused the minor variation
        in actual frequencies).

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34170 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-27 Tor Arne Vestbø <tavestbo@trolltech.com>
hausmann@webkit.org [Wed, 28 May 2008 07:44:44 +0000 (07:44 +0000)]
2008-05-27  Tor Arne Vestbø  <tavestbo@trolltech.com>

        Reviewed by Simon

        Close the video widget explicitly so we don't hang
        when exiting the application.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34169 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoWebCore:
justin.garcia@apple.com [Wed, 28 May 2008 04:20:36 +0000 (04:20 +0000)]

2008-05-27  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Oliver.

        <rdar://problem/5803706> Pressing return at the end of a header doesn't break out of header (17460)

        It wasn't enough to break out of the header, to match other browsers we
        also need to not preserve the typing style.

        * editing/InsertParagraphSeparatorCommand.cpp:
        * editing/InsertParagraphSeparatorCommand.h:


2008-05-27  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Oliver.

        <rdar://problem/5803706> Pressing return at the end of a header doesn't break out of header (17460)

        * editing/inserting/5803706-1-expected.txt: Adjust this already checked-in test to reflect the new
        expected behavior.
        * editing/inserting/5803706-1.html:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34166 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Windows build fix
aroben@apple.com [Wed, 28 May 2008 03:54:35 +0000 (03:54 +0000)]
    Windows build fix

        * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreGenerated.make:
        Remove code that appended Cygwin's /bin directory to PATH.
        * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreGenerated.vcproj:
        Prepend Cygwin's /bin directory to PATH. We prepend instead of append
        so that Cygwin's utilities will win out over Win32 versions of the
        same utilities (particularly perl). We do the prepend here instead of
        in the Makefile because nmake doesn't seem to like prepending to PATH
        inside the Makefile. This also matches the way WebCoreGenerated works.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34165 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Roll out r34163
aroben@apple.com [Wed, 28 May 2008 03:54:06 +0000 (03:54 +0000)]
    Roll out r34163

        A better fix is on the way.

        * DerivedSources.make:
        * JavaScriptCore.vcproj/JavaScriptCore/build-generated-files.sh:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34164 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Windows build fix
aroben@apple.com [Wed, 28 May 2008 03:31:44 +0000 (03:31 +0000)]
    Windows build fix

        * DerivedSources.make: Don't generate the bytecode docs if
        OMIT_BYTECODE_DOCS is set to 1.
        * JavaScriptCore.vcproj/JavaScriptCore/build-generated-files.sh: Set
        OMIT_BYTECODE_DOCS for production builds.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34163 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago<rdar://problem/5966961> REGRESSION: Font previews in the Web
timothy@apple.com [Wed, 28 May 2008 02:45:14 +0000 (02:45 +0000)]
<rdar://problem/5966961> REGRESSION: Font previews in the Web
Inspector always show in Lucida Grande (not the real font)

Reviewed by Adam Roben.

* page/inspector/FontView.js: Add a style element that loads the
@font-face rule that is excepted by the preview element.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34162 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoFixes a regression where the Image and Font preview views do not
timothy@apple.com [Wed, 28 May 2008 02:45:01 +0000 (02:45 +0000)]
Fixes a regression where the Image and Font preview views do not
size correctly.

<rdar://problem/5966723> REGRESSION (r33991): Image and font previews
are shrunken and can't be made larger (19281)

Reviewed by Adam Roben.

* page/inspector/FontView.js: Adjust the font size based only on the
container width minus padding.
* page/inspector/inspector.css: Updated styles for the image and
font views.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34161 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoJavaScriptCore:
andersca@apple.com [Wed, 28 May 2008 00:44:40 +0000 (00:44 +0000)]

2008-05-27  Anders Carlsson  <andersca@apple.com>

        Reviewed by Geoff and Maciej.

        Crash in KJS::JSObject::put after setting this.__proto__

        Set slotIsWriteable to false for __proto__, we want setting __proto__ to go through JSObject::put instead.

        * kjs/object.h:


2008-05-27  Anders Carlsson  <andersca@apple.com>

        Reviewed by Geoff and Maciej.

        Crash in KJS::JSObject::put after setting this.__proto__

        * fast/js/resources/this-non-object-proto.js: Added.
        * fast/js/this-non-object-proto-expected.txt: Added.
        * fast/js/this-non-object-proto.html: Added.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34160 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoFix typo in comment.
ddkilzer@apple.com [Tue, 27 May 2008 23:38:36 +0000 (23:38 +0000)]
Fix typo in comment.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34159 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago Fix Bug 19276: XHR resources are classified as "Other" until Inspector is...
aroben@apple.com [Tue, 27 May 2008 23:33:08 +0000 (23:33 +0000)]
    Fix Bug 19276: XHR resources are classified as "Other" until Inspector is closed and reopened


        Reviewed by Tim Hatcher.

        * page/InspectorController.cpp:
        (WebCore::InspectorController::updateScriptResourceResponse): Moved
        code that updates the resource's type from here...
        (WebCore::InspectorController::updateScriptResourceType): ...to here.
        Call updateScriptResourceType now that we know this resource was
        retrieved via an XHR.
        * page/InspectorController.h:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34158 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoRoll out r34156, which was accidentally checked in
aroben@apple.com [Tue, 27 May 2008 23:25:09 +0000 (23:25 +0000)]
Roll out r34156, which was accidentally checked in

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34157 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoadd printInstruction
aroben@apple.com [Tue, 27 May 2008 23:22:52 +0000 (23:22 +0000)]
add printInstruction

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34156 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoRemoves a redundant call to InspectorController::stopDebugging().
timothy@apple.com [Tue, 27 May 2008 21:50:02 +0000 (21:50 +0000)]
Removes a redundant call to InspectorController::stopDebugging().

Rubber-stamped by Adam Roben.

* page/InspectorController.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34155 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoFixes a bug where unplugging the monitor from a video card and
timothy@apple.com [Tue, 27 May 2008 20:44:16 +0000 (20:44 +0000)]
Fixes a bug where unplugging the monitor from a video card and
moving it to another video card would no longer show OpenGL plugins
until you relaunched Safari.

<rdar://problem/5790983> Add AllowOfflineDisplays pixel format
attribute to OpenGL contexts

Reviewed by Kevin Decker.

* Plugins/WebBaseNetscapePluginView.mm:
(-[WebBaseNetscapePluginView _createWindowedAGLContext]): Added
(-[WebBaseNetscapePluginView _createWindowlessAGLContext]): Ditto.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34154 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-27 Kevin McCullough <kmccullough@apple.com>
kmccullough@apple.com [Tue, 27 May 2008 18:46:57 +0000 (18:46 +0000)]
2008-05-27  Kevin McCullough  <kmccullough@apple.com>

        Reviewed by Tim.

        - Added and updated some manual test for the profiler.

        * manual-tests/inspector/profiler-test-anonymous-function-calls-eval.html:
        * manual-tests/inspector/profiler-test-apply.html:
        * manual-tests/inspector/profiler-test-call.html:
        * manual-tests/inspector/profiler-test-deep-recursion.html: Added.
        * manual-tests/inspector/profiler-test-simple-event-call.html: Added.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34153 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agowx build fix. Update the sample app after wxWebFrame->wxWebBrowserShell rename.
kevino@webkit.org [Tue, 27 May 2008 17:49:31 +0000 (17:49 +0000)]
wx build fix. Update the sample app after wxWebFrame->wxWebBrowserShell rename.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34152 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agowx build fix. Accidently left a couple fixes out of the previous commit.
kevino@webkit.org [Tue, 27 May 2008 16:29:07 +0000 (16:29 +0000)]
wx build fix. Accidently left a couple fixes out of the previous commit.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34151 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agowx build fixes after introduction of SquirrelFish and other updates.
kevino@webkit.org [Tue, 27 May 2008 16:12:45 +0000 (16:12 +0000)]
wx build fixes after introduction of SquirrelFish and other updates.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34150 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-27 Darin Adler <darin@apple.com>
darin@apple.com [Tue, 27 May 2008 14:43:36 +0000 (14:43 +0000)]
2008-05-27  Darin Adler  <darin@apple.com>

        Reviewed by Tim Hatcher.

        - https://bugs.webkit.org/show_bug.cgi?id=19180
          speed up SunSpider by optimizing immediate number cases

        Add immediate number cases for the &, |, and ^ operators.
        Makes standalone SunSpider 1.010x faster.

        * VM/Machine.cpp:
        (KJS::Machine::privateExecute): Add areBothImmediateNumbers special cases
        for the &, |, and ^ operators.

        * kjs/JSImmediate.h:
        (KJS::JSImmediate::xorImmediateNumbers): Added.
        (KJS::JSImmediate::orImmediateNumbers): Added.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34149 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-27 Ariya Hidayat <ariya.hidayat@trolltech.com>
hausmann@webkit.org [Tue, 27 May 2008 10:31:02 +0000 (10:31 +0000)]
2008-05-27  Ariya Hidayat  <ariya.hidayat@trolltech.com>

        Reviewed by Simon.

        Fix linking error (missing gdi32 and user32 libraries).

        * WebCore.pro:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34148 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-27 Ariya Hidayat <ariya.hidayat@trolltech.com>
hausmann@webkit.org [Tue, 27 May 2008 09:49:36 +0000 (09:49 +0000)]
2008-05-27  Ariya Hidayat  <ariya.hidayat@trolltech.com>

        Reviewed by Simon.

        Fix web inspector does not returns its state properly after its window is minimized.

        As hinted by Holger, InspectorClientView::hideEvent is not needed.
        This fixes bug https://bugs.webkit.org/show_bug.cgi?id=18967

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34147 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-27 Ariya Hidayat <ariya.hidayat@trolltech.com>
hausmann@webkit.org [Tue, 27 May 2008 09:02:16 +0000 (09:02 +0000)]
2008-05-27  Ariya Hidayat  <ariya.hidayat@trolltech.com>

        Reviewed by Simon.

        Update the WebKit.qrc so that Web Inspector is functional again.

        * page/inspector/WebKit.qrc:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34146 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-26 Stephanie Lewis <slewis@apple.com>
slewis@apple.com [Tue, 27 May 2008 05:04:48 +0000 (05:04 +0000)]
2008-05-26  Stephanie Lewis  <slewis@apple.com>

        Windows build fix.

        * kjs/testkjs.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34145 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-26 Alp Toker <alp@nuanti.com>
alp@webkit.org [Tue, 27 May 2008 02:48:14 +0000 (02:48 +0000)]
2008-05-26  Alp Toker  <alp@nuanti.com>

        Reviewed by Anders.

        [GTK] Accessibility support with ATK/AT-SPI

        Implement basic action, text and editable text accessibility.

        Provide utility functions to simplify const string return and casting
        and add some more stubs.

        It's now possible to navigate and manipulate forms and text using
        assistive technologies.

        * page/gtk/AccessibilityObjectWrapperAtk.cpp:
        * platform/gtk/LocalizedStringsGtk.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34144 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoJavaScriptCore:
mjs@apple.com [Tue, 27 May 2008 02:10:23 +0000 (02:10 +0000)]

2008-05-26  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Anders.

        - make addStaticGlobals protected instead of private so subclasses can use it

        * JavaScriptCore.exp:
        * kjs/JSGlobalObject.h:


2008-05-25  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Anders.

        - bind the window's "document" and "window" properties to static slots
        1.7% speedup on Acid3 test 26

        Only the "document" part matters for Acid3, but since I was doing
        this anyway, "window" also seemed like a good one to optimize.

        * bindings/js/JSDOMWindowBase.cpp:
        (WebCore::JSDOMWindowBase::JSDOMWindowBase): Allocate variable
        slots for document and window; but just put null in the document
        slot for now.
        (WebCore::JSDOMWindowBase::updateDocument): Fill in the document
        (WebCore::JSDOMWindowBase::~JSDOMWindowBase): Make our shell stop
        tracking us.
        * bindings/js/JSDOMWindowBase.h:
        * bindings/js/JSDOMWindowShell.cpp:
        (WebCore::JSDOMWindowShell::JSDOMWindowShell): Allocate HashSet
        for our still-not-dead inner windows.
        (WebCore::JSDOMWindowShell::updateDocument): Update all our
        inner windows ever (to make sure the update to the document
        property shows up everywhere).
        * bindings/js/JSDOMWindowShell.h:
        (WebCore::JSDOMWindowShell::setWindow): Update our hash.
        (WebCore::JSDOMWindowShell::clearFormerWindow): Ditto.
        * bindings/js/JSDocumentCustom.cpp:
        (WebCore::toJS): Don't set self as a property map property of
        the window, this is no longer necessary.
        * bindings/js/kjs_proxy.cpp:
        (WebCore::KJSProxy::initScript): Update document for the
        newly created window.
        * bindings/js/kjs_proxy.h:
        * page/Frame.cpp:
        (WebCore::Frame::setDocument): If the document is not null,
        update our window.


2008-05-26  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Anders.

        - update test result for document/window being allocated to var slots

        * http/tests/security/cross-frame-access-put-expected.txt:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34143 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-26 James Ren <jamesjren@gmail.com>
collinj@webkit.org [Mon, 26 May 2008 23:49:08 +0000 (23:49 +0000)]
2008-05-26  James Ren  <jamesjren@gmail.com>

        Reviewed by Sam Weinig, landed by Collin Jackson.


        Modified some http/security layout tests to allow them to run in Firefox.

        * http/tests/security/aboutBlank/xss-DENIED-set-opener.html:
        * http/tests/security/frameNavigation/xss-ALLOWED-targeted-subframe-navigation-change.html:
        * http/tests/security/resources/localhost-accesssor.html:
          Added a workaround for https://bugzilla.mozilla.org/show_bug.cgi?id=434912

        * http/tests/security/resources/cross-frame-access.js:
          Changed shouldBe to also gracefully handle exceptions

        * http/tests/security/resources/foreign-domain-iframe-for-selection-test.html:
          Added feature detection for setBaseAndExtent and selectAllChildren

        * http/tests/security/cross-frame-access-enumeration-expected.txt:
        * http/tests/security/cross-frame-access-enumeration.html:
          Fixed typo (missing '.html')

        * http/tests/security/cross-frame-access-call-expected.txt:
        * http/tests/security/cross-frame-access-call.html:
        * http/tests/security/cross-frame-access-first-time.html:
        * http/tests/security/cross-frame-access-put.html:
        * http/tests/security/cross-frame-access-selection.html:
          Caught exceptions that Firefox throws, to allow the test to run to completion.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34142 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-26 Anders Carlsson <andersca@apple.com>
andersca@apple.com [Mon, 26 May 2008 21:13:03 +0000 (21:13 +0000)]
2008-05-26  Anders Carlsson  <andersca@apple.com>

        Reviewed by Sam.

        This patch does two things:

        * It adds a toJSNewlyCreated function for creating JS wrappers for nodes where we know
        that the node doesn't have a wrapper. This avoids a hash lookup.

        * It adds toJSNewlyCreated methods for JSElement and JSText.

        All in all this is a 4.3% speedup on Acid 3 test 26.

        * WebCore.vcproj/WebCore.vcproj:
        * WebCore.xcodeproj/project.pbxproj:
        * bindings/js/JSElementCustom.cpp:
        * bindings/js/JSNodeCustom.cpp:
        * bindings/js/JSTextCustom.cpp: Added.
        * bindings/scripts/CodeGeneratorJS.pm:
        * dom/Document.idl:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34141 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years agoFixed date of ChangeLog entry.
darin@apple.com [Mon, 26 May 2008 20:53:32 +0000 (20:53 +0000)]
Fixed date of ChangeLog entry.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34140 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-26 Sam Weinig <sam@webkit.org>
weinig@apple.com [Mon, 26 May 2008 20:41:08 +0000 (20:41 +0000)]
2008-05-26  Sam Weinig  <sam@webkit.org>

        Reviewed by Anders Carlsson.

        Inline the getOwnPropertySlot for JSNode, JSEventTargetNode, JSElement,
        and JSDocument for a 1-2% speedup on Acide 3 test 26.

        * bindings/js/JSEventTargetNode.cpp:
        * bindings/js/JSEventTargetNode.h:
        * bindings/scripts/CodeGeneratorJS.pm:
        * dom/Document.idl:
        * dom/Element.idl:
        * dom/Node.idl:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34139 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-25 Darin Adler <darin@apple.com>
darin@apple.com [Mon, 26 May 2008 20:07:40 +0000 (20:07 +0000)]
2008-05-25  Darin Adler  <darin@apple.com>

        Reviewed by Maciej.

        - https://bugs.webkit.org/show_bug.cgi?id=19239

        Removed code that would only useful if if there was any way to create an Entity
        node or if we populated EntityReference nodes. It's likely that we'll remove
        these node types entirely in some future version of WebKit, depending in part
        on what happens with these in HTML 5.

        1.8% speedup of Acid3 test 26.

        The only read-only nodes we ever create are EntityReference nodes.
        These are only created by explicit calls to the Document.createEntityReference
        function. Since the created nodes are read-only, there can be no children of
        these nodes, so no other types of nodes can be read-only despite the rule that
        makes descendants of Entity and EntityReference nodes read-only.

        * bindings/objc/DOM.mm: Removed unneeded include.

        * dom/Attr.cpp:
        (WebCore::Attr::setValue): Removed unnecessary isReadOnlyNode check.
        (WebCore::Attr::setNodeValue): Ditto.
        * dom/CharacterData.cpp:
        (WebCore::CharacterData::setData): Removed unnecessary isReadOnlyNode check.
        (WebCore::CharacterData::substringData): Don't initialize the ec variable before
        calling checkCharDataOperation, since that function already does it.
        (WebCore::CharacterData::appendData): Removed unnecessary isReadOnlyNode check.
        (WebCore::CharacterData::insertData): Don't initialize ec.
        (WebCore::CharacterData::deleteData): Ditto.
        (WebCore::CharacterData::replaceData): Ditto.
        (WebCore::CharacterData::setNodeValue): Removed now-misleading comment.
        (WebCore::CharacterData::checkCharDataOperation): Removed unnecessary
        isReadOnlyNode check.
        * dom/CharacterData.h: Made checkCharDataOperation private and non-virtual
        since it's never used or defined in any other class.

        * dom/Element.cpp:
        (WebCore::Element::setAttribute): Removed unneeded isReadOnlyNode check.

        * dom/Entity.cpp: Removed the entire contents of this file.

        * dom/Entity.h: Replaced the class with a stub -- just enough to keep the
        public bindings (JavaScript and Objective-C compiling), but the class is
        abstract and can't be instantianted.

        * dom/EntityReference.cpp: Removed unused constructor.
        (WebCore::EntityReference::cloneNode): Removed unneeded code and comments
        to clone nonexistent child nodes.
        * dom/EntityReference.h: Removed unused constructor and unnecessary override
        of childTypeAllowed, since no children are allowed at all.

        * dom/NamedAttrMap.cpp:
        (WebCore::NamedAttrMap::setNamedItem): Removed unneeded isReadOnlyNode check.
        (WebCore::NamedAttrMap::removeNamedItem): Ditto.
        * dom/NamedAttrMap.h: Removed isReadOnlyNode function.

        * dom/Node.cpp: Removed the old isReadOnlyNode function. It made lots of
        virtual nodeType function calls.
        * dom/Node.h:
        (WebCore::Node::isReadOnlyNode): Replaced the old virtual isReadOnlyNode
        function with a new simple non-virtual one.

        * dom/ProcessingInstruction.cpp:
        (WebCore::ProcessingInstruction::setData): Removed unneeded isReadOnlyNode

        * dom/Text.cpp:
        (WebCore::Text::splitText): Removed unneeded isReadOnlyNode check.
        (WebCore::Text::replaceWholeText): Removed unnecessary isReadOnlyNode assertion.
        There is no reason for this function to go out of its way to mention the
        read-only node issue when none of the other DOM mutation functions do.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34138 268f45cc-cd09-0410-ab3c-d52691b4dbfc

14 years ago2008-05-26 Adam Barth <abarth@webkit.org>
abarth@webkit.org [Mon, 26 May 2008 19:38:14 +0000 (19:38 +0000)]
2008-05-26  Adam Barth  <abarth@webkit.org>

        Reviewed by Sam.

        Cleanup SecurityOrigin::canAccess.  Removed comment referring to an
        old version of the code.  Removed some useless code.

        * platform/SecurityOrigin.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@34137 268f45cc-cd09-0410-ab3c-d52691b4dbfc