WebCore/ChangeLog

   1 2007-03-19  Mitz Pettel  <mitz@webkit.org>
   2
   3         Reviewed by Hyatt.
   4
   5         - fix http://bugs.webkit.org/show_bug.cgi?id=13117
   6         REGRESSION (r14658): Flickering text fields when updating via javascript
   7
   8         * page/Frame.cpp:
   9         (WebCore::Frame::forceLayout): Added a flag to allow partial relayout.
  10         * page/Frame.h:
  11         * page/mac/WebCoreFrameBridge.mm:
  12         (-[WebCoreFrameBridge forceLayoutAdjustingViewSize:]): Allow partial relayout
  13         if not adjusting view size.
  14         (-[WebCoreFrameBridge needsLayout]): Changed to check for pending subtree
  15         layout instead of just whether the root needs layout.
  16
  17 2007-03-18  Nikolas Zimmermann  <zimmermann@kde.org>
  18
  19         Reviewed by Maciej.
  20
  21         Make JS function calls on POD-types (ie. SVGLength) take effect.
  22         Calling ie. newValueInSpecifiedUnits on SVGLength works as expected.
  23
  24         This worked fine before we switched to SVGLength POD type, so this
  25         can be considered as P1 regression fix.
  26
  27         Partly fix svg/hixie/dynamic/005-broken.svg. We match Opera now though
  28         it's not yet completly fixed. Setting baseVal attribute on SVGAnimatedString
  29         doesn't work as expected - needs a seperated fix.
  30
  31         * bindings/scripts/CodeGeneratorJS.pm:
  32
  33 2007-03-19  Adele Peterson  <adele@apple.com>
  34
  35         Reviewed by Adam.
  36
  37         Adding comments and a little cleanup from my last checkin.
  38
  39         * html/HTMLSelectElement.cpp:
  40         (WebCore::HTMLSelectElement::setSelectedIndex): Removed commented out assert.  Added comment about how we use onChange.
  41         (WebCore::HTMLSelectElement::selectAll): Added comment about how we use saveLastSelection and onChange.
  42         (WebCore::HTMLSelectElement::dispatchFocusEvent): ditto.
  43         (WebCore::HTMLSelectElement::dispatchBlurEvent): ditto.
  44         (WebCore::HTMLSelectElement::menuListDefaultEventHandler): ditto.
  45         (WebCore::HTMLSelectElement::listBoxDefaultEventHandler): ditto.
  46         (WebCore::HTMLSelectElement::menuListOnChange): Added assert that usesMenuList() is true.
  47         (WebCore::HTMLSelectElement::listBoxOnChange): Added assert that usesMenuList() is false.
  48         (WebCore::HTMLSelectElement::saveLastSelection): Added early return for menu lists.
  49
  50 2007-03-19  Mitz Pettel  <mitz@webkit.org>
  51
  52         Reviewed by Tim Hatcher.
  53
  54         - fix http://bugs.webkit.org/show_bug.cgi?id=13091
  55           REGRESSION (r20075): Wrong text style and pixel break in Apple Store Locator HUD
  56
  57         Test: fast/innerHTML/additional-inline-style.html
  58
  59         * editing/markup.cpp:
  60         (WebCore::startMarkup): Changed to add inline style based on CSS rules only
  61         in AnnotateForInterchange mode.
  62
  63 2007-03-19  Zack Rusin  <zrusin@trolltech.com>
  64
  65         Compile fix.
  66
  67         * platform/qt/ScrollViewQt.cpp:
  68         (WebCore::ScrollView::update):
  69
  70 2007-03-19  Adam Roben  <aroben@apple.com>
  71
  72         Reviewed by Hyatt and Maciej.
  73
  74         Renamed WebCoreJavaScript to JavaScriptStatistics and made it C++ as
  75         part of <rdar://problem/5071653>.
  76
  77         All layout tests pass.
  78
  79         * WebCore.exp: Updated.
  80         * WebCore.xcodeproj/project.pbxproj: Updated.
  81         * bridge/JavaScriptStatistics.cpp: Moved from
  82         bridge/mac/WebCoreJavaScript.mm, and converted to C++.
  83         (WebCore::collect):
  84         (WebCore::JavaScriptStatistics::objectCount):
  85         (WebCore::JavaScriptStatistics::interpreterCount):
  86         (WebCore::JavaScriptStatistics::protectedObjectCount):
  87         (WebCore::JavaScriptStatistics::rootObjectTypeCounts):
  88         (WebCore::JavaScriptStatistics::garbageCollect):
  89         (WebCore::JavaScriptStatistics::garbageCollectOnAlternateThread):
  90         (WebCore::JavaScriptStatistics::shouldPrintExceptions):
  91         (WebCore::JavaScriptStatistics::setShouldPrintExceptions):
  92         * bridge/JavaScriptStatistics.h: Added.
  93         * bridge/mac/WebCoreJavaScript.h: Removed.
  94         * bridge/mac/WebCoreJavaScript.mm: Removed.
  95
  96 2007-03-18  Maciej Stachowiak  <mjs@apple.com>
  97
  98         Reviewed by Mark.
  99
 100         - avoid static construction (and global variable access) in a smarter, more portable way,
 101         to later enable MUTLI_THREAD mode to work on other platforms and compilers.
 102
 103         * bindings/js/JSCustomXPathNSResolver.cpp:
 104         (WebCore::JSCustomXPathNSResolver::lookupNamespaceURI):
 105         * bindings/js/JSHTMLInputElementBase.cpp:
 106         (WebCore::JSHTMLInputElementBaseFunction::JSHTMLInputElementBaseFunction):
 107         * bindings/js/JSHTMLOptionElementConstructor.cpp:
 108         (WebCore::JSHTMLOptionElementConstructor::JSHTMLOptionElementConstructor):
 109         * bindings/js/JSXMLHttpRequest.cpp:
 110         (KJS::JSXMLHttpRequestConstructorImp::JSXMLHttpRequestConstructorImp):
 111         * bindings/js/JSXSLTProcessor.cpp:
 112         (KJS::XSLTProcessorConstructorImp::XSLTProcessorConstructorImp):
 113         * bindings/js/kjs_dom.cpp:
 114         (KJS::DOMNamedNodeMap::getOwnPropertySlot):
 115         (KJS::DOMNamedNodesCollection::getOwnPropertySlot):
 116         * bindings/js/kjs_events.cpp:
 117         (KJS::JSAbstractEventListener::handleEvent):
 118         * bindings/js/kjs_html.cpp:
 119         (KJS::HTMLElementFunction::HTMLElementFunction):
 120         (KJS::JSHTMLCollection::getOwnPropertySlot):
 121         * bindings/js/kjs_window.cpp:
 122         (KJS::ScheduledAction::execute):
 123         * bindings/objc/WebScriptObject.mm:
 124         * bindings/scripts/CodeGeneratorJS.pm:
 125
 126 2007-03-18  Andrew Wellington  <proton@wiretapped.net>
 127
 128         Reviewed by Mark Rowe
 129
 130         Set Xcode editor to use 4 space indentation (http://webkit.org/coding/coding-style.html)
 131
 132         * WebCore.xcodeproj/project.pbxproj:
 133
 134 2007-03-18  David Hyatt  <hyatt@apple.com>
 135
 136         Fix for REGRESSION bug 13108, frame borders being painted when they shouldn't be.  Rework both frame borders
 137         and resizing to be more like other browsers.
 138
 139         Reviewed by ggaren
 140
 141         fast/frames/no-frame-borders.html
 142
 143         * html/HTMLFrameElement.cpp:
 144         (WebCore::HTMLFrameElement::HTMLFrameElement):
 145         (WebCore::HTMLFrameElement::attach):
 146         (WebCore::HTMLFrameElement::parseMappedAttribute):
 147         * html/HTMLFrameElement.h:
 148         (WebCore::HTMLFrameElement::hasFrameBorder):
 149         * html/HTMLFrameElementBase.cpp:
 150         (WebCore::HTMLFrameElementBase::HTMLFrameElementBase):
 151         (WebCore::HTMLFrameElementBase::parseMappedAttribute):
 152         * html/HTMLFrameElementBase.h:
 153         * html/HTMLFrameSetElement.cpp:
 154         (WebCore::HTMLFrameSetElement::attach):
 155         * html/HTMLFrameSetElement.h:
 156         (WebCore::HTMLFrameSetElement::hasFrameBorder):
 157         * html/HTMLIFrameElement.cpp:
 158         (WebCore::HTMLIFrameElement::HTMLIFrameElement):
 159         * rendering/RenderFrame.cpp:
 160         (WebCore::RenderFrame::edgeInfo):
 161         * rendering/RenderFrame.h:
 162         * rendering/RenderFrameSet.cpp:
 163         (WebCore::RenderFrameSet::paint):
 164         (WebCore::RenderFrameSet::GridAxis::resize):
 165         (WebCore::RenderFrameSet::fillFromEdgeInfo):
 166         (WebCore::RenderFrameSet::computeEdgeInfo):
 167         (WebCore::RenderFrameSet::edgeInfo):
 168         (WebCore::RenderFrameSet::layout):
 169         (WebCore::RenderFrameSet::startResizing):
 170         (WebCore::RenderFrameSet::continueResizing):
 171         (WebCore::RenderFrameSet::canResizeRow):
 172         (WebCore::RenderFrameSet::canResizeColumn):
 173         (WebCore::RenderFrameSet::splitPosition):
 174         (WebCore::RenderFrameSet::hitTestSplit):
 175         (WebCore::RenderFrameSet::dump):
 176         * rendering/RenderFrameSet.h:
 177         (WebCore::):
 178         (WebCore::FrameEdgeInfo::FrameEdgeInfo):
 179         (WebCore::FrameEdgeInfo::preventResize):
 180         (WebCore::FrameEdgeInfo::allowBorder):
 181         (WebCore::FrameEdgeInfo::setPreventResize):
 182         (WebCore::FrameEdgeInfo::setAllowBorder):
 183
 184 2007-03-19  Mark Rowe  <mrowe@apple.com>
 185
 186         Rubber-stamped by Brady.
 187
 188         Update references to bugzilla.opendarwin.org with bugs.webkit.org.
 189
 190         * ChangeLog:
 191         * WebCore.vcproj/WebCore/build-generated-files.sh:
 192         * manual-tests/ATSU-bad-layout.html:
 193         * manual-tests/accidental-strict-mode.html:
 194         * manual-tests/applet-param-no-name.html:
 195         * manual-tests/bidi-parens.html:
 196         * manual-tests/bugzilla-3855.html:
 197         * manual-tests/bugzilla-4840.html:
 198         * manual-tests/bugzilla-6821.html:
 199         * manual-tests/containing-block-position-chage.html:
 200         * manual-tests/contenteditable-link.html:
 201         * manual-tests/css3-cursor-fallback-quirks.html:
 202         * manual-tests/css3-cursor-fallback-strict.html:
 203         * manual-tests/custom-cursors.html:
 204         * manual-tests/dictionary-scrolled-iframe.html:
 205         * manual-tests/dom-manipulation-on-resize.html:
 206         * manual-tests/drag-image-to-address-bar.html:
 207         * manual-tests/empty-link-target.html:
 208         * manual-tests/empty-title-popup.html:
 209         * manual-tests/first-line-style-crash.html:
 210         * manual-tests/invalid-mouse-event.html:
 211         * manual-tests/left-overflow-repaint.html:
 212         * manual-tests/linkjump-3.html:
 213         * manual-tests/log-keypress-events.html:
 214         * manual-tests/named-window-blank-target.html:
 215         * manual-tests/plain-text-paste.html:
 216         * manual-tests/plugin-controller-datasource.html:
 217         * manual-tests/pre-tab-selection-rect.html:
 218         * manual-tests/redirection-target.html:
 219         * manual-tests/redraw-page-cache-visited-links.html:
 220         * manual-tests/reset-initiatedDrag.html:
 221         * manual-tests/resources/named-window-blank-target-step2.html:
 222         * manual-tests/resources/named-window-blank-target-step3.html:
 223         * manual-tests/resources/named-window-blank-target-step4.html:
 224         * manual-tests/resources/redraw-page-cache-visited-links-2.html:
 225         * manual-tests/scrollbar-hittest.html:
 226         * manual-tests/scrollbar-hittest2.html:
 227         * manual-tests/subview-click-assertion.html:
 228         * manual-tests/tabbing-input-google.html:
 229         * manual-tests/text-field-autoscroll.html:
 230         * manual-tests/textarea-after-stylesheet-link.html:
 231         * manual-tests/textarea-focus.html:
 232         * manual-tests/whitespace-pre-affinity.html:
 233
 234 2007-03-18  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
 235
 236         Not reviewed - gdk build fix.
 237
 238         * platform/gdk/ScrollViewGdk.cpp:
 239         (WebCore::ScrollView::update): add a stub
 240
 241 2007-03-18  Brady Eidson  <beidson@apple.com>
 242
 243         Reviewed by Hyatt
 244
 245         http://bugs.webkit.org/show_bug.cgi?id=13111
 246         When stopped a load before it completes, partial images become the
 247         broken image icon
 248
 249         * loader/loader.cpp:
 250         (WebCore::Loader::didFail): Split off into a method that knows the
 251           difference between failed and cancelled.  If a load is cancelled,
 252           don't call error() on the object - only remove it from the cache
 253         (WebCore::Loader::cancelRequests): Call didFail(loader, true)
 254         * loader/loader.h: Added didFail(SubresourceLoader*, bool cancelled)
 255
 256 2007-03-18  Mitz Pettel  <mitz@webkit.org>
 257
 258         Reviewed by Adele.
 259
 260         - fix http://bugs.webkit.org/show_bug.cgi?id=13101
 261           REGRESSION (NativeTextField): Specifying word-wrap:break-word makes text field wrap
 262
 263         Test: fast/forms/input-text-word-wrap.html
 264
 265         * rendering/RenderTextControl.cpp:
 266         (WebCore::RenderTextControl::createInnerTextStyle): Force the inner block to
 267         'word-wrap:normal'.
 268
 269 2007-03-18  Dan Waylonis  <waylonis@mac.com>
 270
 271         Reviewed by Tim Hatcher.
 272
 273         Fix http://bugs.webkit.org/show_bug.cgi?id=13005
 274         Bug 13005: WebScriptObject +throwException needs NULL check.
 275
 276         Add checking for NULL interpreter before throwing exception.
 277
 278         * bindings/objc/WebScriptObject.mm:
 279         (+[WebScriptObject throwException:]):
 280
 281 2007-03-18  Geoffrey Garen  <ggaren@apple.com>
 282
 283         Reviewed by Oliver Hunt.
 284
 285         Fixed http://bugs.webkit.org/show_bug.cgi?id=13106
 286         REGRESSION: Incomplete document.all implementation breaks abtelectronics.com
 287         (Style Change Through JavaScript Blanks Content)
 288
 289         When indexing into collections, treat the empty string as an invalid argument,
 290         instead of the number 0.
 291
 292         * bindings/js/kjs_html.cpp:
 293         (KJS::JSHTMLCollection::getOwnPropertySlot):
 294         (KJS::JSHTMLCollection::callAsFunction):
 295         (KJS::JSHTMLCollectionPrototypeFunction::callAsFunction):
 296
 297 2007-03-18  David Hyatt  <hyatt@apple.com>
 298
 299         Use fillRect instead of drawLine, since drawLine has some odd behavior (it's only used for borders).
 300
 301         Reviewed by aroben
 302
 303         * platform/graphics/IntRect.h:
 304         * rendering/RenderFrameSet.cpp:
 305         (WebCore::RenderFrameSet::paintColumnBorder):
 306         (WebCore::RenderFrameSet::paintRowBorder):
 307
 308 2007-03-18  David Hyatt  <hyatt@apple.com>
 309
 310         Change the frameborder color to be light gray for framesets.
 311
 312         Reviewed by aroben
 313
 314         * rendering/RenderFrameSet.cpp:
 315         (WebCore::borderFillColor):
 316         (WebCore::RenderFrameSet::paintColumnBorder):
 317         (WebCore::RenderFrameSet::paintRowBorder):
 318
 319 2007-03-18  David Hyatt  <hyatt@apple.com>
 320
 321         Make iframes work properly with frameborder="0".
 322
 323         Reviewed by aroben
 324
 325         * html/HTMLIFrameElement.cpp:
 326         (WebCore::HTMLIFrameElement::mapToEntry):
 327         (WebCore::HTMLIFrameElement::parseMappedAttribute):
 328
 329 2007-03-18  David Hyatt  <hyatt@apple.com>
 330
 331         Add a 2px border by default to <iframe>s for compatibility with other browsers.
 332
 333         Reviewed by aroben
 334
 335         * css/html4.css:
 336
 337 2007-03-18  David Hyatt  <hyatt@apple.com>
 338
 339         Move frame borders out of WebKit and into WebCore.
 340
 341         Reviewed by aroben, olliej
 342
 343         * bridge/mac/FrameViewMac.mm:
 344         * css/html4.css:
 345         * html/HTMLFrameSetElement.cpp:
 346         (WebCore::HTMLFrameSetElement::HTMLFrameSetElement):
 347         (WebCore::HTMLFrameSetElement::mapToEntry):
 348         (WebCore::HTMLFrameSetElement::parseMappedAttribute):
 349         (WebCore::HTMLFrameSetElement::attach):
 350         * html/HTMLFrameSetElement.h:
 351         (WebCore::HTMLFrameSetElement::hasBorderColor):
 352         * page/FrameView.cpp:
 353         (WebCore::FrameViewPrivate::FrameViewPrivate):
 354         * page/FrameView.h:
 355         * page/mac/WebCoreFrameBridge.h:
 356         * platform/graphics/IntRect.h:
 357         (WebCore::IntRect::topLeft):
 358         (WebCore::IntRect::topRight):
 359         (WebCore::IntRect::bottomLeft):
 360         (WebCore::IntRect::bottomRight):
 361         * rendering/RenderFrame.cpp:
 362         (WebCore::RenderFrame::viewCleared):
 363         * rendering/RenderFrameSet.cpp:
 364         (WebCore::borderStartEdgeColor):
 365         (WebCore::borderEndEdgeColor):
 366         (WebCore::RenderFrameSet::paintColumnBorder):
 367         (WebCore::RenderFrameSet::paintRowBorder):
 368         (WebCore::RenderFrameSet::paint):
 369         * rendering/RenderFrameSet.h:
 370         * rendering/RenderPartObject.cpp:
 371         (WebCore::RenderPartObject::viewCleared):
 372         * rendering/RenderView.cpp:
 373         (WebCore::RenderView::repaintViewRectangle):
 374
 375 2007-03-17  David Hyatt  <hyatt@apple.com>
 376
 377         Make onload do an explicit paint when a top-level document is ready (to ensure that a first paint always
 378         happens before any timers set by the onload can fire).
 379
 380         Reviewed by aroben
 381
 382         * dom/Document.cpp:
 383         (WebCore::Document::implicitClose):
 384         * platform/ScrollView.h:
 385         * platform/mac/ScrollViewMac.mm:
 386         (WebCore::ScrollView::update):
 387
 388 2007-03-17  Dave Hyatt  <hyatt@apple.com>
 389
 390         Prevent starvation of user input and painting when processing timers.
 391         If we are already in the processing of a custom timer message, don't allow that
 392         processing to do another PostMessage.  Force SetTimer to be used instead.
 393
 394         Reviewed by ggaren
 395
 396        * platform/win/SharedTimerWin.cpp:
 397         (WebCore::TimerWindowWndProc):
 398         (WebCore::setSharedTimerFireTime):
 399
 400 2007-03-17  Geoffrey Garen  <ggaren@apple.com>
 401
 402         Reviewed by Beth Dakin.
 403
 404         Fixed <rdar://problem/5070967> REGRESSION (r20211): Repro crash when
 405         closing View Source window
 406
 407         I award myself 0 points, and may God have mercy on my soul.
 408
 409         * bindings/js/kjs_proxy.cpp:
 410         (WebCore::KJSProxy::~KJSProxy):
 411
 412 2007-03-17  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
 413
 414         Not reviewed - gdk build fix.
 415
 416         * platform/gdk/TemporaryLinkStubs.cpp:
 417         (Pasteboard::writeURL):
 418
 419 2007-03-17  David Hyatt  <hyatt@apple.com>
 420
 421         Fix tiny PLT regression.  Don't examine the border/background data for RenderImage when images change
 422         unless we actually have a border/background.
 423
 424         Reviewed by Brady
 425
 426         * rendering/RenderImage.cpp:
 427         (WebCore::RenderImage::imageChanged):
 428
 429 2007-03-17  Nikolas Zimmermann  <zimmermann@kde.org>
 430
 431         Reviewed by Sam Weinig.
 432
 433         Fixes: http://bugs.webkit.org/show_bug.cgi?id=13046 (CSS styles on hover and focus are broken for <use>)
 434         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12838 (SVG <use> CSS styles broken)
 435         Added test: svg/custom/use-css-events.svg
 436
 437         In general this fixes the peepo.co.uk website.
 438
 439         Forgot to pass the actual StyleChange to the recalcStyle() function,
 440         when calling it on the shadow tree root element.
 441
 442         * ksvg2/svg/SVGUseElement.cpp:
 443         (WebCore::SVGUseElement::recalcStyle):
 444
 445 2007-03-17  Timothy Hatcher  <timothy@apple.com>
 446
 447         Reviewed by Mark Rowe.
 448
 449         Made Version.xcconfig smarter when building for different configurations.
 450         Now uses the 522+ OpenSource version for Debug and Release, while using the
 451         full 522.4 version for Production builds. The system prefix is also computed
 452         based on the current system, so 4522.4 on Tiger and 5522.4 on Leopard.
 453
 454         * Configurations/Version.xcconfig:
 455         * Configurations/WebCore.xcconfig:
 456
 457 2007-03-17  Antti Koivisto  <antti@apple.com>
 458
 459         Reviewed by Adele.
 460
 461         Fix http://bugs.webkit.org/show_bug.cgi?id=12595
 462         REGRESSION: Can't add item to cart at lnt.com (JS type error)
 463         <rdar://problem/4722863>
 464
 465         Emulate Firefox behavior where form elements accessed by a name
 466         can be accessed with that name later even if the name changes or
 467         even if element is removed from the document.
 468
 469         This is loosely based on Darin's earlier patch for the same problem but
 470         is much less expansive. It takes somewhat different approach to more closely
 471         mimic Firefox behavior. Includes expanded test case.
 472
 473         * bindings/js/JSHTMLFormElementCustom.cpp:
 474         (WebCore::JSHTMLFormElement::canGetItemsForName):
 475             Use new the HTMLFormElement::getNamedElements() method
 476         (WebCore::JSHTMLFormElement::nameGetter):
 477             Use new the HTMLFormElement::getNamedElements() method
 478         * html/HTMLFormElement.cpp:
 479         (WebCore::HTMLFormElement::HTMLFormElement):
 480         (WebCore::HTMLFormElement::~HTMLFormElement):
 481         (WebCore::HTMLFormElement::elementForAlias):
 482         (WebCore::HTMLFormElement::addElementAlias):
 483             Maintain a map of known element aliases
 484         (WebCore::HTMLFormElement::getNamedElements):
 485             Get a list of elements matching the name, based both their
 486             current names and known aliases (earlier names).
 487             Keep the alias list in sync.
 488         * html/HTMLFormElement.h:
 489
 490 2007-03-17  Adele Peterson  <adele@apple.com>
 491
 492         Reviewed by Hyatt.
 493
 494         Fix for <rdar://problem/4990050> REGRESSION: onchange gets fired when clicking on a programmatically selected element in a listbox
 495         http://bugs.webkit.org/show_bug.cgi?id=12725
 496
 497         Test: updated fast/forms/listbox-onchange.html
 498
 499         Added HTMLSelectElement::saveLastSelection that is called before changing a selection that could result
 500         in onChange being called.  m_lastOnChangeIndex and m_lastOnChangeSelection no longer have to be up-to date all the time,
 501         they just have to be up-to-date before we execute an action that may trigger onChange.
 502
 503         * html/HTMLOptionElement.cpp: (WebCore::HTMLOptionElement::setSelectedState): Added.
 504           The HTMLSelectElement will only set an option's selected state with this method.  This ensures
 505           that notifyOptionSelected won't get called when the call originates from the select element.
 506         * html/HTMLOptionElement.h:
 507
 508         * html/HTMLSelectElement.cpp:
 509         (WebCore::HTMLSelectElement::deselectItems): Calls setSelectedState.
 510         (WebCore::HTMLSelectElement::setSelectedIndex): ditto. Don't update the last selection variables here.  Scroll to the new selection.
 511          The scrolling call used to only be in notifyOptionSelected.  There's no reason we shouldn't scroll when the selection is set through
 512          the HTMLSelectElement.
 513         (WebCore::HTMLSelectElement::setValue): Call setSelectedIndex to update all options selected state.
 514         (WebCore::HTMLSelectElement::restoreState): Call setSelectedState.
 515         (WebCore::HTMLSelectElement::selectAll): Call saveLastSelection before making the selection, and calling onChange.
 516         (WebCore::HTMLSelectElement::recalcListItems): Call setSelectedState.  Don't need to save selection here anymore,
 517          since it will get saved before we call onChange.
 518         (WebCore::HTMLSelectElement::reset): ditto.
 519         (WebCore::HTMLSelectElement::dispatchFocusEvent): Added. Call saveLastSelection for menu lists, since onChange can be fired
 520          on blur.
 521         (WebCore::HTMLSelectElement::dispatchBlurEvent): Call menuListOnChange.
 522         (WebCore::HTMLSelectElement::menuListDefaultEventHandler): Call saveLastSelection before showing the popup window.
 523         (WebCore::HTMLSelectElement::listBoxDefaultEventHandler):  Call saveLastSelection during mousedown (to prepare for an
 524          onchange during mouseup, or after autoscroll).
 525         (WebCore::HTMLSelectElement::updateListBoxSelection): Call setSelectedState.
 526
 527         (WebCore::HTMLSelectElement::menuListOnChange): Added. Compares the m_lastOnChangeIndex to the selectedIndex().
 528         (WebCore::HTMLSelectElement::listBoxOnChange): Move the m_lastOnChangeSelection creation to saveLastSelection.
 529         (WebCore::HTMLSelectElement::saveLastSelection): Added. Sets m_lastOnChangeIndex or m_lastOnChangeSelection.
 530         * html/HTMLSelectElement.h:
 531         * platform/PopupMenu.h:
 532
 533 2007-03-16  Oliver Hunt  <oliver@apple.com>
 534
 535         Reviewed by Hyatt.
 536
 537         The old canSaveAsWebArchive call was necessary as stand alone
 538         images used to be rendered by ImageDocument.
 539
 540         Fixes rdar://problem/5061252
 541
 542         * dom/Clipboard.cpp:
 543         * dom/Clipboard.h:
 544         (WebCore::Clipboard::setDragHasStarted):
 545         * page/DragClient.h:
 546         (WebCore::DragClient::declareAndWriteDragImage):
 547         * platform/mac/ClipboardMac.mm:
 548         (WebCore::ClipboardMac::declareAndWriteDragImage):
 549
 550 2007-03-16  Anders Carlsson  <acarlsson@apple.com>
 551
 552         Reviewed by Maciej.
 553
 554         <rdar://problem/4869095>
 555         default content type changed for XMLHttpRequest POSTs changed (breaks Flickrator 0.1 widget)
 556
 557         * xml/xmlhttprequest.cpp:
 558         (WebCore::XMLHttpRequest::send):
 559
 560 2007-03-16  Brady Eidson  <beidson@apple.com>
 561
 562         Rubberstamped by Tim Hatcher
 563
 564         Update the hash table header for Window object properties
 565
 566         * bindings/js/kjs_window.cpp:
 567
 568 2007-03-16  Brady Eidson  <beidson@apple.com>
 569
 570         Reviewed by Anders
 571
 572         <rdar://problem/5061826> and
 573         http://bugs.webkit.org/show_bug.cgi?id=12863
 574         Implement window.stop()
 575
 576         * bindings/js/kjs_window.cpp:
 577         (KJS::WindowFunc::callAsFunction): Add case Window::Stop
 578         * bindings/js/kjs_window.h:
 579         (KJS::Window::): Add "Stop"
 580
 581 2007-03-16  Geoffrey Garen  <ggaren@apple.com>
 582
 583         Reviewed by Brady Eidson.
 584
 585         Fixed dir creation to account for already existing dirs and missing leading
 586         dirs in path.
 587
 588         * loader/icon/IconDatabase.cpp:
 589         (WebCore::makeAllDirectories):
 590
 591 2007-03-16  David Hyatt  <hyatt@apple.com>
 592
 593         Fix for 13084, assertion failure in the Cache.  Convert the client list
 594         to a HashCountedSet so that multiple refs and derefs are allowed.
 595
 596         Fix RenderImage so that if it has the same image used as a background/border
 597         and as the foreground that it will repaint properly (can be tested using
 598         border-image and a foreground image).
 599
 600         Optimize list marker so that it doesn't waste time in the base class method,
 601         since list markers don't support background or border images.
 602
 603         Reviewed by andersca
 604
 605         * ChangeLog:
 606         * loader/CachedResource.cpp:
 607         (WebCore::CachedResource::ref):
 608         * loader/CachedResource.h:
 609         * loader/CachedResourceClientWalker.cpp:
 610         (WebCore::CachedResourceClientWalker::CachedResourceClientWalker):
 611         * loader/CachedResourceClientWalker.h:
 612         * rendering/RenderImage.cpp:
 613         (WebCore::RenderImage::imageChanged):
 614         * rendering/RenderListMarker.cpp:
 615         (WebCore::RenderListMarker::imageChanged):
 616
 617 2007-03-16  Geoffrey Garen  <ggaren@apple.com>
 618
 619         Reviewed by Anders Carlsson.
 620
 621         Fixed <rdar://problem/5065399> REGRESSION: leaks in Frame::bindingRootObject
 622         seen on buildbot
 623
 624         The problem was that we were initializing the same WebScriptObject twice.
 625         This caused it to leak its original set of ivars.
 626
 627         I think some refactoring could prevent this situation from arising in the
 628         first place, but I'm just adding a check at the call site for now, to do
 629         the simplest thing.
 630
 631         * bindings/objc/WebScriptObject.mm:
 632         (-[WebScriptObject _setImp:originRootObject:rootObject:]): Added ASSERTs
 633         against multiple calls.
 634
 635         * bindings/objc/WebScriptObjectPrivate.h: Renamed _initializeWithObjectImp
 636         to setImp because "init" vs "initialize" was a too subtle indication that
 637         one was a Cocoa initializer and one was not.
 638
 639 2007-03-16  Lars Knoll <lars@trolltech.com>
 640
 641         don't use #import in .cpp files.
 642
 643         * editing/qt/EditorQt.cpp:
 644
 645 2007-03-16  Lars Knoll <lars@trolltech.com>
 646
 647         Fix the Qt build once again.
 648
 649         * platform/qt/PasteboardQt.cpp:
 650         (WebCore::Pasteboard::Pasteboard):
 651         (WebCore::Pasteboard::writeSelection):
 652         (WebCore::Pasteboard::plainText):
 653         (WebCore::Pasteboard::documentFragment):
 654         (WebCore::Pasteboard::writeURL):
 655         (WebCore::Pasteboard::writeImage):
 656         (WebCore::Pasteboard::clear):
 657
 658 2007-03-15  Shrikant Gangoda  <shrikant.gangoda@celunite.com>
 659
 660         Gdk build fix.
 661
 662         * loader/gdk/FrameLoaderClientGdk.cpp:
 663         (WebCore::FrameLoaderClientGdk::blockedError):
 664         * loader/gdk/FrameLoaderClientGdk.h:
 665
 666 2007-03-15  Beth Dakin  <bdakin@apple.com>
 667
 668         Reviewed by Geoff.
 669
 670         Fix for http://bugs.webkit.org/show_bug.cgi?id=13088 REGRESSION
 671         (r19761-19779): Copy image no longer includes image address
 672         (rdar://5067927)
 673
 674         writeURL() does not need an isImage parameter. Now that it accepts
 675         the types as a parameter, we can just declare the appropriate image
 676         types within writeImage and send them to writeURL(). Also,
 677         declaring the types twice is what broke this.
 678
 679         * platform/Pasteboard.h:
 680         * platform/mac/PasteboardMac.mm:
 681         (WebCore::Pasteboard::writeURL):
 682         (WebCore::Pasteboard::writeImage):
 683
 684 2007-03-15  Adele Peterson  <adele@apple.com>
 685
 686         Reviewed by Kevin Decker.
 687
 688         Fix for <rdar://problem/4926179> Text in menulist control should never update if menu is open
 689
 690         * rendering/RenderMenuList.cpp: (WebCore::RenderMenuList::updateFromElement):
 691           Only setTextFromOption if the menu isn't visible.
 692
 693 2007-03-15  Brady Eidson  <beidson@apple.com>
 694
 695         Reviewed by Oliver
 696
 697         A few platform specific tweaks
 698
 699         * platform/cf/RetainPtr.h:
 700         (WebCore::RetainPtr::releaseRef): Fixed releaseRef to work with CF
 701
 702         * platform/network/ResourceHandle.h: Added a "releaseRef" style call
 703
 704 2007-03-15  Maciej Stachowiak  <mjs@apple.com>
 705
 706         Reviewed by Geoff and Steve.
 707
 708         * config.h: Remove unneeded hack.
 709
 710 2007-03-15  Justin Garcia  <justin.garcia@apple.com>
 711
 712         Reviewed by john
 713
 714         <rdar://problem/5062376>
 715         REGRESSION: In Mail and Gmail, can't change alignment to text after it has been applied
 716
 717         Bring back the remove step in applyBlockStyle.  It's
 718         necessary because addBlockStyleIfNeeded assumes that
 719         the properties it adds aren't already on the block that
 720         it adds them to.
 721
 722         * editing/ApplyStyleCommand.cpp:
 723         (WebCore::ApplyStyleCommand::applyBlockStyle):
 724         Bring back the remove step (added a testcase).
 725         Don't do the add step if m_removeOnly is true (no testcase
 726         because there aren't any clients using removeOnly functionality
 727         to remove styles yet, only styled elemets).
 728         Moved the code for creating new blocks up one level
 729         to this function so that we can pass blocks to removeCSSStyle.
 730         When converting VisiblePositions to indices and vice versa,
 731         use the highest node in the shadow tree if we're in one as
 732         the scope (working on a testcase).
 733         (WebCore::ApplyStyleCommand::addBlockStyle): Moved code to
 734         applyBlockStyle.
 735         * editing/ApplyStyleCommand.h:
 736
 737 2007-03-15  Brady Eidson  <beidson@apple.com>
 738
 739         Reviewed by Maciej
 740
 741         <rdar://problem/4429701>
 742         Implements a port blocking black list that matches Firefox's
 743
 744         * loader/FrameLoader.cpp:
 745         (WebCore::FrameLoader::blockedError): Call through to the client for blockedError
 746         * loader/FrameLoader.h:
 747
 748         * loader/FrameLoaderClient.h: Get the "port blocked" error for the current platform
 749
 750         * loader/ResourceLoader.cpp:
 751         (WebCore::ResourceLoader::wasBlocked): ResourceHandleClient method to pass on the didFail(error)
 752         (WebCore::ResourceLoader::blockedError): Following the pattern of "CancelledError()", get the
 753           error to fail with for the didFail() call
 754         * loader/ResourceLoader.h:
 755
 756         * platform/graphics/svg/SVGImageEmptyClients.h:
 757         (WebCore::SVGEmptyFrameLoaderClient::blockedError): Added stub
 758
 759         * platform/network/ResourceHandle.cpp:
 760         (WebCore::ResourceHandle::create): If the port is blocked, create the handle but schedule it for
 761           deferred failure on a timer
 762         (WebCore::ResourceHandle::scheduleBlockedFailure): Do the timer scheduling
 763         (WebCore::ResourceHandle::fireBlockedFailure): Fire the timer here
 764         (WebCore::ResourceHandle::portAllowed): Implements checking of Mozilla's
 765         * platform/network/ResourceHandle.h:
 766         * platform/network/ResourceHandleClient.h:
 767         (WebCore::ResourceHandleClient::wasBlocked): Virtual for clients to get the "blocked" message
 768
 769 2007-03-15  Beth Dakin  <bdakin@apple.com>
 770
 771         Reviewed by Hyatt.
 772
 773         Fix for <rdar://problem/5065396> REGRESSION: leaks in
 774         RenderBlock::layoutInlineChildren seen on buildbot
 775
 776         This leak appeared after http://trac.webkit.org/projects/webkit/
 777         changeset/20188. This change shifted line boxes around in
 778         removeChild(). But since removeChild() calls
 779         setNeedsLayoutAndMinMaxRecalc(), all of the line boxes will be
 780         removed once we actually lay out anyway. So this patch fixes the
 781         leak by deleting the line boxes instead of shifting them around.
 782
 783         * editing/IndentOutdentCommand.cpp:
 784         (WebCore::IndentOutdentCommand::outdentParagraph): Call into
 785         updateLayout(). This fixes an assertion I got in editing/
 786         execCommand/4976800.html This is very similar to the line box fix I
 787         made recently (http://trac.webkit.org/projects/webkit/changeset/
 788         20177). We need to update layout before relying on VisiblePositions
 789         after removing a node.
 790         * rendering/RenderBlock.cpp:
 791         (WebCore::RenderBlock::deleteLinesForBlock): New helper function
 792         since this functionality is needed in three places now.
 793         (WebCore::RenderBlock::makeChildrenNonInline): Call into new
 794         deleteLinesForBlock().
 795         (WebCore::RenderBlock::removeChild): Same.
 796         * rendering/RenderBlock.h:
 797
 798 2007-03-15  Timothy Hatcher  <timothy@apple.com>
 799
 800         Reviewed by John.
 801
 802         * Factored out most of our common build settings into .xcconfig files. Anything that was common in
 803           each build configuration was factored out into the shared .xcconfig file.
 804         * Adds a Version.xcconfig file to define the current framework version, to be used in other places.
 805         * Use the new $(BUNDLE_VERSION) (defined in Version.xcconfig) in the preprocessed Info.plist.
 806         * Use the versions defined in Version.xcconfig to set $(DYLIB_CURRENT_VERSION).
 807
 808         * Configurations/Base.xcconfig: Added.
 809         * Configurations/DebugRelease.xcconfig: Added.
 810         * Configurations/Version.xcconfig: Added.
 811         * Configurations/WebCore.xcconfig: Added.
 812         * Info.plist:
 813         * WebCore.xcodeproj/project.pbxproj:
 814
 815 2007-03-15  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
 816
 817         Not reviewed - gdk build fixes.
 818
 819         * platform/gdk/EditorClientGdk.cpp:
 820         (WebCore::EditorClientGdk::handleKeypress):
 821         * platform/graphics/cairo/ImageSourceCairo.cpp:
 822         (WebCore::ImageSource::setData):
 823         * platform/graphics/gdk/ImageGdk.cpp:
 824         (WebCore::Image::loadPlatformResource):
 825
 826 2007-03-14  Justin Garcia  <justin.garcia@apple.com>
 827
 828         Reviewed by john
 829
 830         ~2x speed up of 5k rich text paste:
 831         http://shakespeare.mit.edu/hamlet/full.html
 832
 833         * editing/ReplaceSelectionCommand.cpp:
 834         (WebCore::ReplaceSelectionCommand::doApply): Remove the top
 835         level style span if its unnecessary before inserting
 836         into the document, its faster than doing it after.
 837         * editing/ReplaceSelectionCommand.h: Exposed
 838         ReplacementFragment::removeNodePreservingChildren so that
 839         the style span can be removed using non-undoable removes,
 840         like the rest of the removes done on the ReplacementFragment.
 841         * editing/markup.cpp:
 842         (WebCore::createMarkup): Make the style span the top level
 843         element, otherwise it's useless.  This also facilitates the
 844         optimization mentioned above.
 845         When including markup for a fully selected root, include markup
 846         for all the nodes beneath that fully selected root, to preserve
 847         the structure and appearance of the copied markup.  Did this
 848         by merging with the code for adding markup for descendants of
 849         special commonAncestorBlocks.
 850
 851 2007-03-15  Beth Dakin  <bdakin@apple.com>
 852
 853         Rubber-stamped by Adele.
 854
 855         Rolling out http://trac.webkit.org/projects/webkit/changeset/20148
 856         (which is a fix for http://bugs.webkit.org/show_bug.cgi?id=12595
 857         and rdar://4722863) because it causes a horrible memory-trasher.
 858
 859         * bindings/js/JSHTMLFormElementCustom.cpp:
 860         (WebCore::JSHTMLFormElement::canGetItemsForName):
 861         (WebCore::JSHTMLFormElement::nameGetter):
 862         * bindings/js/kjs_dom.cpp:
 863         (KJS::):
 864         (KJS::DOMNamedNodesCollection::DOMNamedNodesCollection):
 865         (KJS::DOMNamedNodesCollection::lengthGetter):
 866         (KJS::DOMNamedNodesCollection::indexGetter):
 867         (KJS::DOMNamedNodesCollection::getOwnPropertySlot):
 868         * bindings/js/kjs_dom.h:
 869         (KJS::DOMNamedNodesCollection::classInfo):
 870         * bindings/js/kjs_html.cpp:
 871         (KJS::JSHTMLCollection::getNamedItems):
 872         * dom/ChildNodeList.cpp:
 873         (WebCore::ChildNodeList::ChildNodeList):
 874         (WebCore::ChildNodeList::length):
 875         (WebCore::ChildNodeList::item):
 876         (WebCore::ChildNodeList::nodeMatches):
 877         * dom/ChildNodeList.h:
 878         * dom/NameNodeList.cpp:
 879         (WebCore::NameNodeList::NameNodeList):
 880         (WebCore::NameNodeList::item):
 881         (WebCore::NameNodeList::nodeMatches):
 882         * dom/NameNodeList.h:
 883         (WebCore::NameNodeList::rootNodeAttributeChanged):
 884         * dom/Node.cpp:
 885         (WebCore::TagNodeList::TagNodeList):
 886         (WebCore::TagNodeList::nodeMatches):
 887         (WebCore::Node::registerNodeList):
 888         (WebCore::Node::unregisterNodeList):
 889         * dom/Node.h:
 890         * dom/NodeList.cpp:
 891         (WebCore::NodeList::NodeList):
 892         (WebCore::NodeList::~NodeList):
 893         (WebCore::NodeList::recursiveLength):
 894         (WebCore::NodeList::itemForwardsFromCurrent):
 895         (WebCore::NodeList::itemBackwardsFromCurrent):
 896         (WebCore::NodeList::recursiveItem):
 897         (WebCore::NodeList::itemWithName):
 898         (WebCore::NodeList::rootNodeChildrenChanged):
 899         * dom/NodeList.h:
 900         (WebCore::NodeList::rootNodeAttributeChanged):
 901         * html/HTMLFormElement.cpp:
 902         (WebCore::HTMLFormElement::HTMLFormElement):
 903         (WebCore::HTMLFormElement::~HTMLFormElement):
 904         (WebCore::HTMLFormElement::formData):
 905         (WebCore::HTMLFormElement::parseMappedAttribute):
 906         (WebCore::HTMLFormElement::removeFormElement):
 907         * html/HTMLFormElement.h:
 908         * html/HTMLGenericFormElement.cpp:
 909         (WebCore::HTMLGenericFormElement::parseMappedAttribute):
 910         (WebCore::HTMLGenericFormElement::insertedIntoTree):
 911         * html/HTMLGenericFormElement.h:
 912         * html/HTMLInputElement.cpp:
 913         (WebCore::HTMLInputElement::parseMappedAttribute):
 914
 915 2007-03-15  Geoffrey Garen  <ggaren@apple.com>
 916
 917         Added an assert to help catch a bug. Hopefully someone will hit it!
 918
 919         * bindings/js/kjs_proxy.cpp:
 920         (WebCore::KJSProxy::~KJSProxy):
 921
 922 2007-03-14  Oliver Hunt  <oliver@apple.com>
 923
 924         Reviewed by Adele.
 925
 926         Fix for rdar://problem/5061737.
 927
 928         This was a regression from the original Objective-C -> C++ conversion
 929         for the drag logic.  We don't need to call Range::startNode as we just
 930         need the document that contains the range.
 931
 932         * page/DragController.cpp:
 933         (WebCore::documentFragmentFromDragData):
 934
 935 2007-03-14  Mitz Pettel  <mitz@webkit.org>
 936
 937         Reviewed by Hyatt.
 938
 939         - http://bugs.webkit.org/show_bug.cgi?id=13071
 940           REGRESSION: Plain text files no longer wrap lines longer than the width of the browser window
 941
 942         Test: fast/loader/text-document-wrapping.html
 943
 944         * loader/TextDocument.cpp:
 945         (WebCore::TextTokenizer::write): Specified word-wrap:break-word for the <pre> element.
 946
 947 2007-03-14  Mitz Pettel  <mitz@webkit.org>
 948
 949         Reviewed by Hyatt.
 950
 951         - fix http://bugs.webkit.org/show_bug.cgi?id=13072
 952           REGRESSION (r15617): white-space: pre-wrap breaks off the last character of a wide word
 953
 954         Test: fast/text/whitespace/pre-wrap-last-char.html
 955
 956         * rendering/bidi.cpp:
 957         (WebCore::RenderBlock::findNextLineBreak): Undid the change from r15617.
 958
 959 2007-03-14  David Hyatt  <hyatt@apple.com>
 960
 961         Add asserts to help catch double refs and double derefs of CachedResources.
 962
 963         Reviewed by mjs
 964
 965         * loader/CachedResource.cpp:
 966         (WebCore::CachedResource::ref):
 967         (WebCore::CachedResource::deref):
 968
 969 2007-03-14  Alice Liu  <alice.liu@apple.com>
 970
 971         Rubber-stamped by Hyatt.
 972
 973         Adding null check to prevent the crash that happens on 2nd run of iBench HTML load test
 974
 975         * loader/icon/IconDataCache.cpp:
 976         (WebCore::IconDataCache::writeToDatabase):
 977
 978 === Safari-5522.4 ===
 979
 980 2007-03-14  Adele Peterson  <adele@apple.com>
 981
 982         Reviewed by Adam.
 983
 984         Fix for <rdar://problem/5062898> REGRESSION: autocomplete window in text fields doesn't come up
 985
 986         When we moved the initialization of a bunch of variables in the HTMLFormElement constructor, m_autocomplete accidently got initialized to false.
 987
 988         * html/HTMLFormElement.cpp: (WebCore::HTMLFormElement::HTMLFormElement):
 989
 990 2007-03-14  Steve Falkenburg  <sfalken@apple.com>
 991
 992         Reviewed by Adam.
 993
 994         Added missing null check in case this is called on a subframe
 995         that hasn't started loading.
 996
 997         * page/Frame.cpp:
 998         (WebCore::Frame::setMarkedTextMatchesAreHighlighted): Check for a null document.
 999
1000 2007-03-14  Antti Koivisto  <antti@apple.com>
1001
1002         Reviewed by Geoff.
1003
1004         Fix http://bugs.webkit.org/show_bug.cgi?id=13060
1005         REGRESSION: Repro ASSERT failure in Cache::adjustSize running layout tests
1006         <rdar://5060208>
1007
1008         Script evaluation may have dereffed the CachedScript object already, causing double deref and
1009         eventually m_liveResourcesSize underflow.
1010
1011         * html/HTMLScriptElement.cpp:
1012         (WebCore::HTMLScriptElement::notifyFinished):
1013
1014 2007-03-14  Adele Peterson  <adele@apple.com>
1015
1016         Reviewed by Darin.
1017
1018         Added InsertTab, InsertBacktab, InsertLineBreak, and InsertNewline to the editing command table.
1019         Added Event parameter to execCommand, and to all of the enabled and exec functions.  Right now,
1020         the event is only used by the newly added commands.  But in the future, many (and possibly all) of
1021         these editing commands will need to consider the event so they are applied to the correct selection.
1022
1023         * WebCore.exp:
1024         * editing/Editor.cpp:
1025         (WebCore::execCopy):
1026         (WebCore::execCut):
1027         (WebCore::execDelete):
1028         (WebCore::execBackwardDelete):
1029         (WebCore::execForwardDelete):
1030         (WebCore::execMoveBackward):
1031         (WebCore::execMoveBackwardAndModifySelection):
1032         (WebCore::execMoveUpByPageAndModifyCaret):
1033         (WebCore::execMoveDown):
1034         (WebCore::execMoveDownAndModifySelection):
1035         (WebCore::execMoveForward):
1036         (WebCore::execMoveForwardAndModifySelection):
1037         (WebCore::execMoveDownByPageAndModifyCaret):
1038         (WebCore::execMoveLeft):
1039         (WebCore::execMoveLeftAndModifySelection):
1040         (WebCore::execMoveRight):
1041         (WebCore::execMoveRightAndModifySelection):
1042         (WebCore::execMoveToBeginningOfDocument):
1043         (WebCore::execMoveToBeginningOfDocumentAndModifySelection):
1044         (WebCore::execMoveToBeginningOfSentence):
1045         (WebCore::execMoveToBeginningOfSentenceAndModifySelection):
1046         (WebCore::execMoveToBeginningOfLine):
1047         (WebCore::execMoveToBeginningOfLineAndModifySelection):
1048         (WebCore::execMoveToBeginningOfParagraph):
1049         (WebCore::execMoveToBeginningOfParagraphAndModifySelection):
1050         (WebCore::execMoveToEndOfDocument):
1051         (WebCore::execMoveToEndOfDocumentAndModifySelection):
1052         (WebCore::execMoveToEndOfSentence):
1053         (WebCore::execMoveToEndOfSentenceAndModifySelection):
1054         (WebCore::execMoveToEndOfLine):
1055         (WebCore::execMoveToEndOfLineAndModifySelection):
1056         (WebCore::execMoveToEndOfParagraph):
1057         (WebCore::execMoveToEndOfParagraphAndModifySelection):
1058         (WebCore::execMoveParagraphBackwardAndModifySelection):
1059         (WebCore::execMoveParagraphForwardAndModifySelection):
1060         (WebCore::execMoveUp):
1061         (WebCore::execMoveUpAndModifySelection):
1062         (WebCore::execMoveWordBackward):
1063         (WebCore::execMoveWordBackwardAndModifySelection):
1064         (WebCore::execMoveWordForward):
1065         (WebCore::execMoveWordForwardAndModifySelection):
1066         (WebCore::execMoveWordLeft):
1067         (WebCore::execMoveWordLeftAndModifySelection):
1068         (WebCore::execMoveWordRight):
1069         (WebCore::execMoveWordRightAndModifySelection):
1070         (WebCore::execPaste):
1071         (WebCore::execSelectAll):
1072         (WebCore::execToggleBold):
1073         (WebCore::execToggleItalic):
1074         (WebCore::execRedo):
1075         (WebCore::execUndo):
1076         (WebCore::execInsertTab):
1077         (WebCore::execInsertBacktab):
1078         (WebCore::execInsertNewline):
1079         (WebCore::execInsertLineBreak):
1080         (WebCore::enabled):
1081         (WebCore::canPaste):
1082         (WebCore::hasEditableSelection):
1083         (WebCore::hasEditableRangeSelection):
1084         (WebCore::hasRangeSelection):
1085         (WebCore::hasRichlyEditableSelection):
1086         (WebCore::canRedo):
1087         (WebCore::canUndo):
1088         (WebCore::CommandEntry::):
1089         (WebCore::Editor::toggleBold):
1090
1091         (WebCore::Editor::execCommand): Added optional event parameter.
1092         (WebCore::Editor::insertText): Added.  Calls handleTextInputEvent.
1093         (WebCore::Editor::insertTextWithoutSendingTextEvent): Renamed from insertText.
1094          Performs the actual insertion without dispatching any event.
1095         * editing/Editor.h:
1096         * page/EventHandler.cpp: (WebCore::EventHandler::defaultTextInputEventHandler):
1097           Call insertTextWithoutSendingTextEvent.
1098
1099 2007-03-14  Justin Garcia  <justin.garcia@apple.com>
1100
1101         Reviewed by darin
1102
1103         ~3x speedup pasting 5k lines of rich text:
1104         http://shakespeare.mit.edu/hamlet/full.html
1105         ~2x speedup pasting 10k lines of plain text
1106
1107         * css/CSSComputedStyleDeclaration.cpp:
1108         (WebCore::computedStyle): Added for convenience.
1109         * css/CSSComputedStyleDeclaration.h:
1110         * editing/ReplaceSelectionCommand.cpp:
1111         (WebCore::ReplaceSelectionCommand::removeRedundantStyles):
1112         The code that pushed down the top level style span had
1113         a bug in it that made it do unnecessary work.  Instead of
1114         fixing the bug I removed the code because it was used to
1115         help see more redundancies in second level style spans, but
1116         createMarkup now *only* creates a top level style span.
1117         Only remove redundant styles from style spans and only remove
1118         unstyled elements if they are style spans.  FF doesn't
1119         remove redundant styles from elements, or remove redundant
1120         font tags on copy/paste.  We could offer this functionality
1121         through a separate "cleanup" command.
1122         * editing/markup.cpp:
1123         (WebCore::createMarkup): Only add markup for ancestors of
1124         lastClosed if we're including markup for acommonAncestorBlock
1125         (we do this for commonAncestorBlocks like tables and lists),
1126         otherwise it's unnecessary/redundant.
1127
1128 2007-03-14  Anders Carlsson  <acarlsson@apple.com>
1129
1130         Reviewed by Ada.
1131
1132         Make sure to call ResourceLoader::didReceiveData to ensure that there's a shared buffer with the resource
1133         data, since SubresourceLoaderClients now make use of that buffer.
1134
1135         * loader/SubresourceLoader.cpp:
1136         (WebCore::SubresourceLoader::didReceiveData):
1137
1138 2007-03-14  Administrator  <acarlsson@apple.com>
1139
1140         Get the size from the shared buffer.
1141
1142         * platform/graphics/cg/ImageSourceCG.cpp:
1143         (WebCore::ImageSource::setData):
1144         * platform/graphics/cg/PDFDocumentImage.cpp:
1145         (WebCore::PDFDocumentImage::dataChanged):
1146
1147 2007-03-14  Mitz Pettel  <mitz@webkit.org>
1148
1149         Reviewed by Hyatt, thumbs up by Darin.
1150
1151         - fix http://bugs.webkit.org/show_bug.cgi?id=12782
1152           Reproducible crash in BidiContext::deref
1153
1154         Test: fast/dynamic/anonymous-block-orphaned-lines.html
1155
1156         * rendering/InlineBox.cpp:
1157         (WebCore::InlineBox::root): Added an assertion that we return a root inline box.
1158         * rendering/RenderBlock.cpp:
1159         (WebCore::RenderBlock::removeChild): Added code to adopt the line boxes of
1160         anonymous blocks being destroyed instead of leaving them orphaned, which
1161         is what caused this crash. The boxes will be deleted on the next layout, but
1162         this ensures consistency in the mean time.
1163
1164 2007-03-14  Lars Knoll <lars@trolltech.com>
1165
1166         Reviewed by Antti.
1167
1168         Fix the Qt build.
1169         Add a getter to SharedBuffer that returns a reference to the internal
1170         Vector to avoid an extra copy of the data.
1171
1172         * platform/SharedBuffer.h:
1173         (WebCore::SharedBuffer::buffer):
1174         * platform/graphics/qt/ImageQt.cpp:
1175         (WebCore::Image::loadPlatformResource):
1176         * platform/graphics/qt/ImageSourceQt.cpp:
1177         (WebCore::detectImageFormat):
1178         (WebCore::createDecoder):
1179         (WebCore::ImageSource::setData):
1180
1181 2007-03-14  Antti Koivisto  <antti@apple.com>
1182
1183         Reviewed by Mitz.
1184
1185         Fix <rdar://problem/5058774>
1186         REGRESSION: In Mail, caret appears oversized when typing in a To Do note
1187
1188         Horizontal and vertical were switched. Was regression from
1189         http://trac.webkit.org/projects/webkit/changeset/20103
1190
1191         * rendering/RootInlineBox.cpp:
1192         (WebCore::RootInlineBox::addHighlightOverflow):
1193
1194 2007-03-14  David Hyatt  <hyatt@apple.com>
1195
1196         Tweak the data() functions of stylesheets and scripts to be internally consistent.
1197
1198         * loader/CachedCSSStyleSheet.cpp:
1199         (WebCore::CachedCSSStyleSheet::data):
1200         * loader/CachedScript.cpp:
1201         (WebCore::CachedScript::data):
1202         * loader/CachedXSLStyleSheet.cpp:
1203         (WebCore::CachedXSLStyleSheet::data):
1204
1205 2007-03-14  David Hyatt  <hyatt@apple.com>
1206
1207         Fix Radar 5050688.
1208
1209         For large animated GIFs, destroy and recreate the source for every animation frame.  This keeps
1210         the memory consumption down while giant images are animating.
1211
1212         Reviewed by andersca
1213
1214         * platform/graphics/BitmapImage.cpp:
1215         (WebCore::BitmapImage::BitmapImage):
1216         (WebCore::BitmapImage::destroyDecodedData):
1217         (WebCore::BitmapImage::dataChanged):
1218         (WebCore::BitmapImage::advanceAnimation):
1219         * platform/graphics/BitmapImage.h:
1220
1221 2007-03-14  David Hyatt  <hyatt@apple.com>
1222
1223         Make sure to use CFDataCreateWithBytesNoCopy where we can.
1224
1225         Reviewed by andersca
1226
1227         * platform/graphics/cg/ImageSourceCG.cpp:
1228         (WebCore::ImageSource::setData):
1229         * platform/graphics/cg/PDFDocumentImage.cpp:
1230         (WebCore::PDFDocumentImage::dataChanged):
1231
1232 2007-03-13  David Hyatt  <hyatt@apple.com>
1233
1234         Fix for bugzilla bug 13050 and also radar p1 5050645.
1235
1236         This patch reworks resource loading to avoid having redundant buffers in the icon database and in cached
1237         images in the WebCore cache.  It also avoids overcopying in top-level image documents and in the icon
1238         database.
1239
1240         There is now only one SharedBuffer for a resource and everybody observes that buffer now instead of ever
1241         making their own.  Even ImageIO uses the SharedBuffer while decoding.
1242
1243         The page in 13050 dropped from 145mb down to 45mb of memory use with this change for a stunning savings
1244         of 100mb.
1245
1246         Reviewed by olliej, mjs
1247
1248         * WebCore.exp:
1249         * loader/CachedCSSStyleSheet.cpp:
1250         (WebCore::CachedCSSStyleSheet::data):
1251         * loader/CachedCSSStyleSheet.h:
1252         * loader/CachedImage.cpp:
1253         (WebCore::CachedImage::data):
1254         * loader/CachedImage.h:
1255         * loader/CachedResource.cpp:
1256         (WebCore::CachedResource::CachedResource):
1257         (WebCore::CachedResource::~CachedResource):
1258         * loader/CachedResource.h:
1259         (WebCore::CachedResource::data):
1260         * loader/CachedScript.cpp:
1261         (WebCore::CachedScript::data):
1262         * loader/CachedScript.h:
1263         * loader/CachedXSLStyleSheet.cpp:
1264         (WebCore::CachedXSLStyleSheet::data):
1265         * loader/CachedXSLStyleSheet.h:
1266         * loader/DocLoader.cpp:
1267         (WebCore::DocLoader::checkCacheObjectStatus):
1268         * loader/ImageDocument.cpp:
1269         (WebCore::ImageTokenizer::writeRawData):
1270         (WebCore::ImageTokenizer::finish):
1271         * loader/icon/IconDataCache.cpp:
1272         (WebCore::IconDataCache::setImageData):
1273         (WebCore::IconDataCache::writeToDatabase):
1274         * loader/icon/IconDataCache.h:
1275         * loader/icon/IconDatabase.cpp:
1276         (WebCore::IconDatabase::imageDataForIconURL):
1277         (WebCore::IconDatabase::iconForPageURL):
1278         (WebCore::IconDatabase::setIconDataForIconURL):
1279         (WebCore::IconDatabase::setHaveNoIconForIconURL):
1280         (WebCore::IconDatabase::imageDataForIconURLQuery):
1281         * loader/icon/IconDatabase.h:
1282         * loader/icon/IconLoader.cpp:
1283         (WebCore::IconLoader::startLoading):
1284         (WebCore::IconLoader::didReceiveResponse):
1285         (WebCore::IconLoader::didReceiveData):
1286         (WebCore::IconLoader::didFail):
1287         (WebCore::IconLoader::finishLoading):
1288         (WebCore::IconLoader::clearLoadingState):
1289         * loader/icon/IconLoader.h:
1290         * loader/icon/SQLStatement.cpp:
1291         (WebCore::SQLStatement::getColumnBlobAsVector):
1292         (WebCore::SQLStatement::isExpired):
1293         * loader/icon/SQLStatement.h:
1294         * loader/loader.cpp:
1295         (WebCore::Loader::didFinishLoading):
1296         (WebCore::Loader::didReceiveData):
1297         * page/mac/WebCoreFrameBridge.mm:
1298         (-[WebCoreFrameBridge getData:andResponse:forURL:]):
1299         (-[WebCoreFrameBridge getAllResourceDatas:andResponses:]):
1300         * platform/SharedBuffer.h:
1301         (WebCore::SharedBuffer::isEmpty):
1302         * platform/graphics/BitmapImage.cpp:
1303         (WebCore::BitmapImage::destroyDecodedData):
1304         (WebCore::BitmapImage::dataChanged):
1305         * platform/graphics/BitmapImage.h:
1306         * platform/graphics/Image.cpp:
1307         (WebCore::Image::setData):
1308         * platform/graphics/Image.h:
1309         (WebCore::Image::dataChanged):
1310         (WebCore::Image::data):
1311         * platform/graphics/ImageSource.h:
1312         * platform/graphics/cg/ImageSourceCG.cpp:
1313         (WebCore::ImageSource::setData):
1314         * platform/graphics/cg/PDFDocumentImage.cpp:
1315         (WebCore::PDFDocumentImage::dataChanged):
1316         * platform/graphics/cg/PDFDocumentImage.h:
1317         * platform/graphics/mac/ImageMac.mm:
1318         (WebCore::Image::loadPlatformResource):
1319         * platform/graphics/svg/SVGImage.cpp:
1320         (WebCore::SVGImage::setData):
1321         * platform/mac/PasteboardMac.mm:
1322         (WebCore::fileWrapperForImage):
1323
1324 2007-03-13  Justin Garcia  <justin.garcia@apple.com>
1325
1326         Reviewed by darin
1327
1328         <rdar://problem/5046875>
1329         Gmail Editor: Applying alignment to selected text in message also applies alignment to signature
1330
1331         * editing/ApplyStyleCommand.cpp:
1332         (WebCore::ApplyStyleCommand::doApply): Don't call applyBlockStyle unless
1333         there is a block style to apply.
1334         (WebCore::ApplyStyleCommand::applyBlockStyle): Don't do the remove step.
1335         It was unnecessary and removed properties from blocks that could contain
1336         content outside the range being operated on (added a testcase).
1337         (WebCore::ApplyStyleCommand::addBlockStyleIfNeeded): Used an early return
1338         instead of if-nesting.
1339         * editing/ApplyStyleCommand.h:
1340         * editing/CompositeEditCommand.cpp:
1341         (WebCore::CompositeEditCommand::moveParagraphContentsToNewBlockIfNecessary):
1342         Return the new block, if one was created.  Use moveParagraphs to move
1343         paragraphs into the new block, instead of moving nodes.  The old code moved
1344         too much (added a testcase).
1345         * editing/CompositeEditCommand.h:
1346
1347 2007-03-13  Oliver Hunt  <oliver@apple.com>
1348
1349         Reviewed by Brady.
1350
1351         To fix <rdar://problem/5044366> we now pass a NSString
1352         representation of the URL extracted with _web_originalDataAsString
1353         instead of relying on [NSURL absoluteString] in the bridge
1354
1355         * page/mac/WebCoreFrameBridge.h:
1356         * page/mac/WebCoreFrameBridge.mm:
1357         (-[WebCoreFrameBridge getData:andResponse:forURL:]):
1358
1359 2007-03-13  Brady Eidson  <beidson@apple.com>
1360
1361         Reviewed by Anders
1362
1363         <rdar://problem/5048818> - REGRESSION: Incompletely loaded resources being saved to the object cache
1364
1365         Due to a subtle change in loader behavior back in 10904, we would stop all loaders before calling
1366         didFail() on them in the Cache loader.  As a result, we basically cleared all of the Subresource Loaders
1367         out of the Cache loader before more properly failing them as errored out.  The result?  Partially loaded
1368         resources being cached.
1369
1370         Since Loader::didFail() both calls error() on the object *and* removes the loader, the solution is to call
1371         didFail() for all cancelled loaders instead of *only* removing them from the set of active loaders.
1372
1373         In addition, pages that didn't completely load were being saved to the back/forward cache.  To fix that,
1374         I added a null check on the DocumentLoader's error to see if the page ended in an error, or did indeed
1375         completely load.
1376
1377         Note that the layout test for this - if possible - will require other enhancements including possibly adding
1378         support for window.stop().  That task is documented in <rdar://problem/5061826>
1379
1380         * loader/FrameLoader.cpp:
1381         (WebCore::FrameLoader::provisionalLoadStarted): Fixed a few bugs relating to my original BFCache rewrite to
1382           more perfectly restore the original behavior - including only caching HTML documents via the
1383           m_client->canCachePage() call
1384         (WebCore::FrameLoader::canCachePage): Don't make the call to m_client->canCachePage() as that serves a different
1385           purpose
1386           - Check the mainDocumentError to see if the load ended in error as a further criteria in determining the
1387           cachability of a page
1388
1389         * loader/loader.cpp:
1390         (WebCore::Loader::cancelRequests): Call didFail(cancelledError()) instead of just removing the loaders from the
1391           loaders-in-progress set.  This adds the effect of properly cleaning up the cached object.
1392
1393 2007-03-13  Beth Dakin  <bdakin@apple.com>
1394
1395         Rendering part reviewed by Hyatt. Editing part consulted with and
1396         rubber stamped by Justin and Harrison.
1397
1398         Fix for <rdar://problem/5025925> A hang occurs in Safari when
1399         attempting to print page at http://www.pcadvisor.co.uk
1400
1401         * rendering/RenderBlock.cpp:
1402         (WebCore::RenderBlock::makeChildrenNonInline):
1403         RenderBlock::makeChildrenNonInline() takes a block's inline
1404         children and turns them into block children. If the children had
1405         line boxes, those boxes were being leaked. In the layout test I
1406         added with the change (and at pcadvisor.co.uk during printing)
1407         children were being made non-inline, and then they were being made
1408         inline again. This meant that some of the children ended up
1409         pointing to totally stale line boxes that are normally just leaked.
1410         This caused an infinite loop in RenderFlow::destroy(). This patch
1411         simply deletes everyone's line boxes in
1412         RenderBlock::makeChildrenNonInline()
1413
1414         * editing/InsertParagraphSeparatorCommand.cpp:
1415         (WebCore::InsertParagraphSeparatorCommand::doApply): The other part
1416         of this fix is that I added a call to updateLayout in
1417         InsertParagraphSeparatorCommand::doApply(). One layout test
1418         (editing/spelling/spelling.html) was changed by my patch to
1419         RenderBlock. doApply() inserts a node into the render tree. In at
1420         least one case in spelling.html, that caused some line boxes to be
1421         deleted. Back in doApply() this meant that the RenderTree was out-
1422         of-date, and we mistakenly thought we were at the end of the
1423         paragraph. This caused us to insert a RenderBR() at the end of the
1424         tree instead of an empty RenderText(). No one seems to know exactly
1425         why we insert either, or if the change is necessarily a problem. It
1426         is clear, though, that the RenderTree in doApply() is out-of-date
1427         after inserting the node and deleting some line boxes, so it seems
1428         prudent to call into updateLayout().
1429
1430 2007-03-13  Adam Roben  <aroben@apple.com>
1431
1432         Reviewed by Anders.
1433
1434         * platform/FontData.h: Added m_isSystemFont parameter to match NSFont.
1435
1436 2007-03-13  Beth Dakin  <bdakin@apple.com>
1437
1438         Reviewed by Maciej.
1439
1440         Export DocumentLoader::setFrame(). Part of fix for <rdar://
1441         problem/4277074> 8F32: Help Viewer crashed on clicking link -
1442         KHTMLView::viewportMouseReleaseEvent (12647)
1443
1444         * WebCore.exp:
1445
1446 2007-03-13  Darin Adler  <darin@apple.com>
1447
1448         Reviewed by Maciej.
1449
1450         - fix http://bugs.webkit.org/show_bug.cgi?id=12794
1451           <rdar://problem/5028154> REGRESSION: TripTik planner at aaa.com never
1452           finishes loading due to unclosed canvas tag (12794)
1453
1454         Change <canvas> elements so that their contents are parsed normally,
1455         but not rendered. This change fixes the bug, because normal parsing
1456         rules close the <canvas> element in that case. The special parser
1457         stuff was just getting in the way.
1458
1459         Also do some basic cleanup to the HTML parser. This was motivated by
1460         an earlier version of this patch that made even more changes to the
1461         parser, but the cleanup is still worth landing.
1462
1463         Test: fast/canvas/canvas-hides-fallback.html
1464         Test: fast/canvas/script-inside-canvas-fallback.html
1465         Test: fast/canvas/unclosed-canvas-1.html
1466         Test: fast/canvas/unclosed-canvas-2.html
1467         Test: fast/canvas/unclosed-canvas-3.html
1468         Test: fast/canvas/unclosed-canvas-4.html
1469
1470         * html/HTMLCanvasElement.h: Added a data member to keep track of whether the
1471         renderer is a RenderHTMLCanvas or not.
1472         * html/HTMLCanvasElement.cpp:
1473         (WebCore::HTMLCanvasElement::createRenderer): If JavaScript is enabled, create
1474         a RenderHTMLCanvas. If it's not, let the default code create the default type
1475         of renderer, which will result in fallback content being visible. The
1476         RenderHTMLCanvas class already hides all of its children. Set the m_rendererIsCanvas
1477         boolean accordingly. Since the actual storage for the canvas is allocated lazily
1478         when you actually get a drawing context, we don't need to do anything special
1479         to prevent it when JavaScript is disabled; the relevant functions won't be called.
1480         (WebCore::HTMLCanvasElement::reset): Protect the code that manipulates the
1481         RenderHTMLCanvas with a check of m_rendererIsCanvas. This is the only code inside
1482         the DOM element that relies on the renderer type.
1483
1484         * html/HTMLParser.h: Removed unneeded includes. Marked HTMLParser as
1485         Noncopyable. Changed the Document parameter to the constructor to instead
1486         be HTMLDocument. Renamed discard_until to m_skipModeTag for clarity.
1487         Removed unused noSpaces function and unneeded public doc() function.
1488         Moved data members all down to the end so you can see them together in order.
1489         Renamed map to m_currentMapElement and isindex to m_isindexElement.
1490         Removed unused end and headLoaded data members.  Renamed m_fragment to
1491         m_isParsingFragment to make it clearer that it's a boolean, not a fragment.
1492
1493         * html/HTMLParser.cpp:
1494         (WebCore::HTMLParser::HTMLParser): Changed to use member construction
1495         syntax instead of calling reset(). This is especially helpful in the
1496         fragment case, where calling reset() later on is illegal, so not using
1497         it in the constructor lets us assert.
1498         (WebCore::HTMLParser::~HTMLParser): Did an explicit deref instead of
1499         calling setCurrent for its side effect.
1500         (WebCore::HTMLParser::reset): Updated for member name changes and removal
1501         and to use document instead of doc().
1502         (WebCore::HTMLParser::setCurrent): Use document instead of doc().
1503         (WebCore::HTMLParser::setSkipMode): Added. No longer inline. Now sets the
1504         m_inCanvasBeforeFirstOpenTag data member to false.
1505         (WebCore::HTMLParser::parseToken): Tightened up the skip mode logic at the
1506         top of the function, and added a FIXME about the strange case there where
1507         we don't skip yet stay in skip mode. Updated for renaming and doc().
1508         (WebCore::HTMLParser::insertNode): Updated for renaming and doc().
1509         (WebCore::HTMLParser::handleError): Ditto.
1510         (WebCore::HTMLParser::framesetCreateErrorCheck): Ditto.
1511         (WebCore::HTMLParser::isindexCreateErrorCheck): Changed to use RefPtr.
1512         (WebCore::HTMLParser::noscriptCreateErrorCheck): Updated for renaming and doc().
1513         (WebCore::HTMLParser::mapCreateErrorCheck): Ditto.
1514         (WebCore::HTMLParser::getNode): Removed the special case for canvas here.
1515         Canvas fallback is now handled in the DOM, not the parser. Updated for
1516         renaming and doc().
1517         (WebCore::HTMLParser::allowNestedRedundantTag): Changed a #define into a C++
1518         constant.
1519         (WebCore::HTMLParser::processCloseTag): Updated for renaming and doc().
1520         (WebCore::HTMLParser::isInline): Ditto.
1521         (WebCore::HTMLParser::tagIsOnStack): Added. Used by new canvas logic.
1522         (WebCore::HTMLParser::popBlock): Updated for renaming and doc(). Also renamed
1523         the local variable Elem to elem.
1524         (WebCore::HTMLParser::createHead): Ditto.
1525         (WebCore::HTMLParser::handleIsindex): Changed to use RefPtr.
1526         (WebCore::HTMLParser::startBody): Updated for renaming and doc().
1527         (WebCore::HTMLParser::finished): Ditto.
1528
1529 2007-03-13  David Hyatt  <hyatt@apple.com>
1530
1531         Two more cleanup fixes to the cache.  Don't call destroyDecodedData in the BitmapImage destructor, since
1532         clearing the image source and calling setData on it again causes it to do an extra copy of the encoded
1533         data.  Since we're about to be destroyed this is just wasteful.
1534
1535         When the cache prunes, don't allow it to destroy the decoded data of an image that is still actively loading,
1536         since we've established that ImageIO can actually crash if you yank the rug out from under it like that.
1537
1538         Reviewed by ggaren
1539
1540         * loader/Cache.cpp:
1541         (WebCore::Cache::prune):
1542         * platform/graphics/BitmapImage.cpp:
1543         (WebCore::BitmapImage::~BitmapImage):
1544
1545 2007-03-13  Anders Carlsson  <acarlsson@apple.com>
1546
1547         Try fixing the Qt build.
1548
1549         * editing/Editor.h:
1550         (WebCore::Editor::setStartNewKillRingSequence):
1551
1552 2007-03-13  David Harrison  <harrison@apple.com>
1553
1554         Reviewed by Justin.
1555
1556         <rdar://problem/5031181> cntl-k at end of paragraph adds nothing to the kill ring
1557         <rdar://problem/5031189> REGRESSION: cntl-y yanks only the most recently killed content
1558
1559         For rdar://5031181, properly extend the selection before the killring handling, and
1560         make sure plainText of that selection returns a linefeed.
1561
1562         For rdar://5031189, restore Editor::deleteRange() code that continued current killring,
1563         even though the range deletion implicitly stopped it via changing the selection.
1564
1565         A byproduct of this change is the elimination of RUNDFINDER vs CONTENT TextIterator. The
1566         only difference between the two was whether to emit a newline when the range started
1567         with a blockflow element. No callers actually need that any more.
1568
1569         Tests added:
1570         * editing/pasteboard/emacs-ctrl-k-y-001-expected.checksum: Added.
1571         * editing/pasteboard/emacs-ctrl-k-y-001-expected.png: Added.
1572         * editing/pasteboard/emacs-ctrl-k-y-001-expected.txt: Added.
1573         * editing/pasteboard/emacs-ctrl-k-y-001.html: Added.
1574
1575         * editing/Editor.cpp:
1576         (WebCore::Editor::deleteRange):
1577         Clear the "start new kill ring sequence" setting, because it was set to true
1578         when the selection was updated by deleting the range.
1579
1580         (WebCore::Editor::deleteWithDirection):
1581         If extending the selection to the end of paragraph resulted in a caret selection,
1582         extend by character, to handle the case when the selection started as a caret at
1583         the end of paragraph.
1584
1585         * editing/TextIterator.cpp:
1586         (WebCore::TextIterator::TextIterator):
1587         Initialize new member variables for tracking handling of the beginning of the range.
1588
1589         (WebCore::TextIterator::advance):
1590         Call representNodeOffsetZero on the m_endContainer.
1591         Move visibility checks into handleTextNode and handleReplacedElement.
1592
1593         (WebCore::TextIterator::handleTextNode):
1594         (WebCore::TextIterator::handleTextBox):
1595         Call emitText.
1596
1597         (WebCore::TextIterator::handleReplacedElement):
1598         Moved visibility check into here.
1599
1600         (WebCore::shouldEmitNewlinesBeforeAndAfterNode):
1601
1602         (WebCore::TextIterator::shouldRepresentNodeOffsetZero):
1603         (WebCore::TextIterator::representNodeOffsetZero):
1604         New. Emits proper sequence when encountering offset 0 of a node, including the
1605         m_endContainer. Started with code from handleNonTextNode.
1606
1607         (WebCore::TextIterator::handleNonTextNode):
1608         Call representNodeOffsetZero.
1609
1610         (WebCore::TextIterator::exitNode):
1611         Similar to shouldRepresentNodeOffsetZero, do not emit the newline if the node
1612         was collapsed, and before any other emitted content.
1613
1614         (WebCore::TextIterator::emitCharacter):
1615
1616         (WebCore::TextIterator::emitText):
1617         New. Consolidates code used by handleText and handleTextBox.
1618
1619         (WebCore::CharacterIterator::CharacterIterator):
1620         Removed RUNFINDER.
1621
1622         (WebCore::WordAwareIterator::WordAwareIterator):
1623         Removed RUNFINDER.
1624
1625         (WebCore::WordAwareIterator::advance):
1626         Formatting.
1627
1628         (WebCore::TextIterator::rangeLength):
1629         Formatting.
1630
1631         * editing/TextIterator.h:
1632         Added member variables for tracking handling of the beginning of the range.
1633         Eliminated concept of RUNDFINDER vs CONTENT TextIterator.
1634
1635         * editing/visible_units.cpp:
1636         (WebCore::nextBoundary):
1637         Eliminated concept of RUNDFINDER vs CONTENT TextIterator.
1638
1639 2007-03-13  David Hyatt  <hyatt@apple.com>
1640
1641         Clean up the null image case in CachedImage::data to make sure the size totals will stay accurate.
1642         I'm not convinced this case can even be hit, but I'm cleaning it up just in case.
1643
1644         * loader/CachedImage.cpp:
1645         (WebCore::CachedImage::data):
1646
1647 2007-03-13  David Hyatt  <hyatt@apple.com>
1648
1649         Reorder the call to allReferencesRemoved, since otherwise the live object size will become
1650         incorrect.
1651
1652         * loader/CachedResource.cpp:
1653         (WebCore::CachedResource::deref):
1654
1655 2007-03-13  Antti Koivisto  <antti@apple.com>
1656
1657         Reviewed by Darin.
1658
1659         Fix http://bugs.webkit.org/show_bug.cgi?id=11083
1660         REGRESSION: Typing tab key fails to insert a tab character in Google Docs editable area
1661         <rdar://problem/4757650>
1662
1663         Allow inserting tabs in designMode. Backtab behaves like before (matches FF).
1664
1665         * page/EventHandler.cpp:
1666         (WebCore::EventHandler::defaultTabEventHandler):
1667
1668 2007-03-13  David Hyatt  <hyatt@apple.com>
1669
1670         Fix two counting errors in the decoded size of objects in the WebCore cache.  We need to explicitly
1671         destroy decoded data when clearing out a partially loaded image that had an error, since the destructor
1672         (although it does destroy the data) disconnects the observer so that the notification doesn't happen.
1673
1674         Some CachedImages aren't in the cache (like image documents).  When the decoded size of such an image
1675         changes we should not notify the cache, since - newsflash - we aren't actually in it.
1676
1677         Reviewed by ggaren
1678
1679         * loader/CachedImage.cpp:
1680         (WebCore::CachedImage::clear):
1681         (WebCore::CachedImage::decodedSizeChanged):
1682
1683 2007-03-13  Darin Adler  <darin@apple.com>
1684
1685         Reviewed by Tim Hatcher and John Sullivan.
1686
1687         - fix <rdar://problem/4915303> CrashTracer: 36 crashes in Safari at
1688           com.apple.AppKit: -[NSView getRectsBeingDrawn:count:] + 502
1689
1690         * page/mac/FrameMac.mm: (WebCore::Frame::imageFromRect):
1691         This was calling drawRect: directly, but NSView's getRectsBeingDrawn:count:
1692         method was never really safe to call unless it was AppKit that called your
1693         drawRect: method. Changed it to call drawSingleRect: instead. A little ugly,
1694         but seems to work and will almost certainly fix the bug.
1695
1696 2007-03-13  Nikolas Zimmermann  <zimmermann@kde.org>
1697
1698         Reviewed by Maciej.
1699
1700         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12974
1701
1702         Call forgetGenericContext in JSSVGPathSeg destructor, otherwhise
1703         we'll hit an ASSERT in a debug build, when running svg/custom/js-update-path-changes.svg
1704         a few dozen times in a single WebKit instance. The ASSERT is good and just warns that
1705         there was already a generic context pointer registered, and the new "to be registered"
1706         object already exists, but pointing to a different object. That's because garbage collection
1707         calls the JSSVGPathSeg destructor, but that didn't cleanup the generic context map.
1708
1709         Only JSSVGPathSeg is hit by this problem, as it's the only non SVGAnimated* type
1710         using the generic context system while using a custom JSSVGPathSegList implementation.
1711
1712         Also cleanup JSSVGPathSegListCustom code to call the static forgetGenericContext
1713         method instead of doing the same using custom code.
1714
1715         * bindings/js/JSSVGPathSegListCustom.cpp:
1716         (WebCore::removeFromPathSegContextMap):
1717         (WebCore::JSSVGPathSegList::clear):
1718         (WebCore::JSSVGPathSegList::removeItem):
1719         * bindings/scripts/CodeGeneratorJS.pm:
1720
1721 2007-03-13  Darin Adler  <darin@apple.com>
1722
1723         Reviewed by Adele.
1724
1725         - fix http://bugs.webkit.org/show_bug.cgi?id=12595
1726           <rdar://problem/4722863> REGRESSION: Can't add item to cart at lnt.com
1727           (JS type error) (12595)
1728
1729         Test: fast/forms/old-names.html
1730
1731         * bindings/js/JSHTMLFormElementCustom.cpp:
1732         (WebCore::JSHTMLFormElement::canGetItemsForName): If the form collection has
1733         nothing for a given name, try the form's oldNamedElement function.
1734         (WebCore::JSHTMLFormElement::nameGetter): Ditto.
1735
1736         * bindings/js/kjs_dom.h: Removed the DOMNamedNodesCollection. Instead we will use
1737         a class derived from NodeList.
1738         * bindings/js/kjs_dom.cpp: Ditto.
1739
1740         * bindings/js/kjs_html.cpp:
1741         (KJS::VectorNodeList::VectorNodeList): Added. Constructor for a new class derived
1742         from NodeList to be used for the named items result from a collection -- uses a
1743         vector of node pointers.
1744         (KJS::VectorNodeList::length): Added.
1745         (KJS::VectorNodeList::item): Added.
1746         (KJS::JSHTMLCollection::getNamedItems): Use VectorNodeList and the existing wrapper
1747         for NodeList rather than a custom JavaScript class, DOMNamedNodesCollection.
1748
1749         * dom/ChildNodeList.h:
1750         * dom/ChildNodeList.cpp:
1751         (WebCore::ChildNodeList::ChildNodeList): Updated to derive from TreeNodeList,
1752         since NodeList is now a simpler class.
1753         (WebCore::ChildNodeList::elementMatches): Updated for name and parameter change.
1754
1755         * dom/NameNodeList.h:
1756         * dom/NameNodeList.cpp:
1757         (WebCore::NameNodeList::NameNodeList): Updated to derive from TreeNodeList,
1758         since NodeList is now a simpler class.
1759         (WebCore::NameNodeList::rootNodeAttributeChanged): Updated for name and
1760         parameter change.
1761
1762         * dom/Node.h: Change register/unregister functions to take TreeNodeList.
1763         * dom/Node.cpp:
1764         (WebCore::TagNodeList::TagNodeList): Updated to derive from TreeNodeList,
1765         since NodeList is now a simpler abstract class.
1766         (WebCore::TagNodeList::elementMatches): Updated for name and parameter change.
1767         (WebCore::Node::registerNodeList): Changed type from NodeList to TreeNodeList.
1768         (WebCore::Node::unregisterNodeList): Ditto.
1769
1770         * dom/NodeList.h: Broke NodeList into a simpler base class and a derived class
1771         with the machinery for iterating a tree, called TreeNodeList.
1772         * dom/NodeList.cpp:
1773         (WebCore::NodeList::~NodeList): Added.
1774         (WebCore::NodeList::itemWithName): Factored out of the old itemWithName.
1775         (WebCore::TreeNodeList::TreeNodeList): Renamed from NodeList.
1776         (WebCore::TreeNodeList::~TreeNodeList): Ditto.
1777         (WebCore::TreeNodeList::recursiveLength): Ditto.
1778         (WebCore::TreeNodeList::itemForwardsFromCurrent): Ditto.
1779         (WebCore::TreeNodeList::itemBackwardsFromCurrent): Ditto.
1780         (WebCore::TreeNodeList::recursiveItem): Ditto.
1781         (WebCore::TreeNodeList::itemWithName): Factored half of this into this function,
1782         the other half in NodeList::itemWithName.
1783         (WebCore::TreeNodeList::rootNodeAttributeChanged): Added. No longer inline.
1784         (WebCore::TreeNodeList::rootNodeChildrenChanged): Renamed from NodeList.
1785
1786         * html/HTMLFormElement.h: Added formElementNameChanged and oldNamedElement
1787         fucntions, and a map called m_oldNames. Also removed m_boundary, which I
1788         thought I had already done.
1789         * html/HTMLFormElement.cpp:
1790         (WebCore::HTMLFormElement::HTMLFormElement): Initialize m_oldNames to 0.
1791         Switched the rest of the members to initialization syntax.
1792         (WebCore::HTMLFormElement::~HTMLFormElement): Delete m_oldNames.
1793         (WebCore::HTMLFormElement::formElementNameChanged): Added. Stores a reference
1794         to one element under each of its old names.
1795         (WebCore::HTMLFormElement::oldNamedElement): Added. Returns the old element
1796         that once had a given name.
1797
1798         * html/HTMLGenericFormElement.h:
1799         * html/HTMLGenericFormElement.cpp:
1800         (WebCore::HTMLGenericFormElement::parseMappedAttribute): When the name
1801         attribute changes, tell the form about the old name.
1802         (WebCore::HTMLGenericFormElement::insertedIntoTree): When telling a form
1803         about an element, also store away the old name so that we can use it
1804         when the name changes later.
1805
1806         * html/HTMLInputElement.cpp:
1807         (WebCore::HTMLInputElement::parseMappedAttribute): Added a call to the
1808         base class in the nameAttr case, so the code in HTMLGenericFormElement
1809         above will get called in the input element case.
1810
1811 2007-03-13  Antti Koivisto  <antti@apple.com>
1812
1813         Reviewed by Alexey.
1814
1815         Alexey spotted a DOS by using string of 64k unbreakable character in
1816         fix for http://bugs.webkit.org/show_bug.cgi?id=12833
1817
1818         * dom/Text.cpp:
1819         (WebCore::Text::createWithLengthLimit):
1820
1821 2007-03-13  Lars Knoll <lars@trolltech.com>
1822
1823         Fix the Qt build
1824
1825         * WebCore.pro:
1826
1827 2007-03-13  Rob Buis  <buis@kde.org>
1828
1829         Reviewed by Darin.
1830
1831         http://bugs.webkit.org/show_bug.cgi?id=12576
1832         WebKit does not support xlink:show attributes
1833
1834         Make an exception for non-empty targets, these should be opened
1835         in a new window, unless the value is _self.
1836
1837         * ksvg2/svg/SVGAElement.cpp:
1838         (WebCore::SVGAElement::defaultEventHandler):
1839
1840 2007-03-13  David Hyatt  <hyatt@apple.com>
1841
1842         - fix cache issues seen in http://bugs.webkit.org/show_bug.cgi?id=13050
1843           6 objects and ~200MB leaked after opening then closing tab
1844           <rdar://problem/5058714>
1845
1846         Double the encoded size of images for now.  We do this to account for a bug in ImageIO where they hold
1847         a separate copy of image data.  See <rdar://problem/5050645>.
1848
1849         Reviewed by aroben.
1850
1851         * loader/CachedImage.cpp:
1852         (WebCore::CachedImage::data):
1853
1854 2007-03-12  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
1855
1856         Not reviewed - gdk build fix.
1857
1858         * platform/gdk/TemporaryLinkStubs.cpp:
1859         (SearchPopupMenu::enabled): add missing stub.
1860
1861 2007-03-12  David Kilzer  <ddkilzer@kilzer.net>
1862
1863         Reviewed by Darin.
1864
1865         - fix http://bugs.webkit.org/show_bug.cgi?id=13045
1866           REGRESSION: Blackboard CourseWare Error with Nightlies after Mar 8
1867
1868         * html/HTMLFormElement.cpp:
1869         (WebCore::getUniqueBoundaryString): Speculative fix: remove dashes from
1870         the boundary prefix that appear after non-dash characters.
1871
1872 2007-03-12  Mitz Pettel  <mitz@webkit.org>
1873
1874         Reviewed by Anders.
1875
1876         - fix http://bugs.webkit.org/show_bug.cgi?id=13015
1877           REGRESSION (r17233-r17241): Repro crash when leaving a page whose unload handler submits a form
1878
1879         Test: fast/loader/onunload-form-submit-crash-2.html
1880
1881         * loader/DocumentLoader.cpp:
1882         (WebCore::DocumentLoader::finishedLoading): Added null check.
1883
1884 2007-03-12  Anders Carlsson  <acarlsson@apple.com>
1885
1886         Reviewed by Darin.
1887
1888         <rdar://problem/4900071>
1889         http://bugs.webkit.org/show_bug.cgi?id=6454
1890         ASSERTION: Navigating 'back' in frameset: !_private->previousItem (6454)
1891
1892         Change back the behavior of checkLoadComplete to traverse the entire frame tree instead of
1893         just the parent frames of the current frame. This is needed in order to reset the previous history item
1894         for all frames when doing a frame navigation. (This was changed in revision 11819)
1895
1896         * loader/FrameLoader.cpp:
1897         (WebCore::FrameLoader::recursiveCheckLoadComplete):
1898         (WebCore::FrameLoader::checkLoadComplete):
1899         * loader/FrameLoader.h:
1900
1901 2007-03-12  Justin Garcia  <justin.garcia@apple.com>
1902
1903         Reviewed by darin
1904
1905         <rdar://problem/5056619>
1906         REGRESSION: Gmail Editor: Dragging text into Reply (textarea) field results in a crash at WebCore::InsertNodeBeforeCommand::doApply()
1907
1908         * editing/SelectionController.cpp:
1909         (WebCore::removingNodeRemovesPosition): Added.
1910         (WebCore::SelectionController::nodeWillBeRemoved): Clear the
1911         selection if it's inside a shadow tree.
1912         * page/DragController.cpp:
1913         (WebCore::setSelectionToDragCaret): Return false to signal to
1914         clients that a drop shouldn't be performed if the second attempt
1915         to set a selection ends up in non-editable content.
1916
1917 2007-03-12  Darin Adler  <darin@apple.com>
1918
1919         Reviewed by Tim Hatcher.
1920
1921         - fixed JavaScript wrapper classes to be correct for a variety of cases
1922           that a new test uncovered: was broken for at least 5 classes
1923
1924         - fixed Objective-C wrapper classes to be correct for a variety of cases
1925           that a test case uncovered: was broken for ast least 50 classes
1926
1927         - added missing DOM API for creating OverflowEvent and WheelEvent instances
1928
1929         Test: fast/dom/wrapper-classes.html
1930
1931         * DerivedSources.make: Added missing bindings: HTMLCanvasElement for ObjC,
1932         CDATASection, Comment, and EntityReference for JavaScript.
1933         * WebCore.xcodeproj/project.pbxproj: Added those new generated files.
1934
1935         * bindings/js/JSHTMLElementWrapperFactory.cpp:
1936         (WebCore::createJSHTMLWrapper): Corrected the wrapper classes for <keygen>,
1937         which needs an HTMLSelectElement wrapper, and <xmp>, which needs an
1938         HTMLPreElement wrapper.
1939
1940         * bindings/objc/DOMInternal.h: Updated for new naming scheme.
1941         Also moved createDOMWrapper from the KJS namespace to the WebCore namespace.
1942         * bindings/objc/DOMUtility.mm:
1943         (KJS::createDOMWrapper): Broke the core function into a separate one, and
1944         left it in the KJS namespace because Objective-C++ rules make it impossible
1945         for it to work in the WebCore namespace. Used a macro-based implementation
1946         to cut down on repeated code, and added missing cases for Counter,
1947         HTMLOptionsCollection, Range, XPathExpression, XPathResult, Event, RGBColor,
1948         Rect, Window, DOMImplementation, NodeIterator, TreeWalker, and HTMLCollection.
1949         (WebCore::createDOMWrapper): The other half of the function.
1950
1951         * bindings/objc/DOM.mm:
1952         (WebCore::createElementClassMap): Corrected the wrapper classes for
1953         <canvas>, which needs a DOMHTMLCanvasElement wrapper, <del>, which needs
1954         a DOMHTMLModElement wrapper, <embed>, which needs a DOMHTMLEmbedElement
1955         wrapper, <ins>, which needs a DOMHTMLModElement wrapper, <th>, which needs
1956         a DOMHTMLTableCellElement wrapper, and <xmp>, which needs an
1957         DOMHTMLPreElement wrapper.
1958         (+[DOMNode _wrapNode:]): Updated for new naming scheme.
1959         (+[DOMNode _wrapEventTarget:]): Ditto.
1960         (+[DOMNodeFilter _wrapNodeFilter:]): Ditto.
1961         (ObjCNodeFilterCondition::acceptNode): Ditto.
1962         (-[DOMDocument createNodeIterator:whatToShow:filter:expandEntityReferences:]): Ditto.
1963         (-[DOMDocument createTreeWalker:whatToShow:filter:expandEntityReferences:]): Ditto.
1964         (WebCore::ObjCEventListener::handleEvent): Ditto.
1965
1966         * dom/Document.cpp: (WebCore::Document::createEvent):
1967         Added cases for OverflowEvent and WheelEvent.
1968
1969         * dom/OverflowEvent.h: Added empty constructor and initOverflowEvent.
1970         * dom/OverflowEvent.cpp:
1971         (WebCore::OverflowEvent::OverflowEvent): Added.
1972         (WebCore::OverflowEvent::initOverflowEvent): Added.
1973         * dom/OverflowEvent.idl: Added initOverflowEvent.
1974
1975         * bindings/objc/PublicDOMInterfaces.h: Added initOverflowEvent.
1976
1977         * bindings/objc/DOMCSS.mm:
1978         (+[DOMStyleSheet _wrapStyleSheet:]): Updated for new naming scheme.
1979         (+[DOMCSSRule _wrapCSSRule:]): Corrected wrapper for CSSUnknownRule.
1980         (+[DOMCSSValue _wrapCSSValue:]): Updated for new naming scheme.
1981
1982         * bindings/js/kjs_css.h:
1983         (KJS::DOMRGBColor::impl): Added. Used when making an ObjC wrapper.
1984         (KJS::DOMRect::impl): Ditto.
1985
1986         * bindings/js/kjs_dom.cpp: (KJS::toJS): Corrected the wrapper classes for
1987         CDATASection, Comment, and EntityReference.
1988
1989         * bindings/js/kjs_html.cpp: Corrected the class name for HTMLElement
1990         (was "DOMHTMLElement") and HTMLCollection (was "Collection").
1991
1992         * bindings/objc/DOMImplementationFront.h:
1993         * bindings/objc/DOMImplementationFront.cpp:
1994         (WebCore::implementationFront): Added new overload that returns a front given
1995         a JavaScript wrapper. Needed by the code that makes the Objective-C wrapper.
1996
1997         * bindings/objc/WebScriptObject.mm:
1998         (+[WebScriptObject _convertValueToObjcValue:originRootObject:rootObject:]):
1999         Updated since createDOMWrapper is now in the WebCore namespace.
2000
2001         * bindings/scripts/CodeGeneratorObjC.pm: Update the naming scheme for the
2002         wrapper creation functions to use _wrapElement: rather then _elementWith:
2003         style. Removed now-unneeded special cases for things that needed to stay
2004         upper-case, since we don't have to do the whole lcfirst thing.
2005
2006         * html/HTMLCanvasElement.idl: Added #if so we can successfully generate the
2007         Objective-C wrapper for this class, even though we can't yet handle the
2008         DOMObject return type.
2009
2010         * html/HTMLOptionsCollection.idl: Added GenerateNativeConverter so we get
2011         an appropriate toJS function.
2012
2013         * bindings/objc/DOMEvents.mm:
2014         (+[DOMEvent _wrapEvent:]):
2015         * bindings/objc/DOMHTML.mm:
2016         (-[DOMHTMLDocument createDocumentFragmentWithMarkupString:baseURL:]):
2017         (-[DOMHTMLDocument createDocumentFragmentWithText:]):
2018         * bindings/objc/DOMObject.mm:
2019         (-[DOMObject sheet]):
2020         * bindings/objc/DOMRGBColor.mm:
2021         (-[DOMRGBColor red]):
2022         (-[DOMRGBColor green]):
2023         (-[DOMRGBColor blue]):
2024         (-[DOMRGBColor alpha]):
2025         (+[DOMRGBColor _wrapRGBColor:]):
2026         * bindings/objc/DOMSVGPathSegInternal.mm:
2027         (+[DOMSVGPathSeg _wrapSVGPathSeg:]):
2028         * bindings/objc/DOMXPath.mm:
2029         (+[DOMNativeXPathNSResolver _wrapXPathNSResolver:]):
2030         * page/mac/WebCoreFrameBridge.mm:
2031         (-[WebCoreFrameBridge nodesFromList:]):
2032         (-[WebCoreFrameBridge elementWithName:inForm:]):
2033         (-[WebCoreFrameBridge formForElement:]):
2034         (-[WebCoreFrameBridge currentForm]):
2035         (-[WebCoreFrameBridge controlsInForm:]):
2036         (-[WebCoreFrameBridge rangeByAlteringCurrentSelection:SelectionController::direction:SelectionController::granularity:]):
2037         (-[WebCoreFrameBridge convertNSRangeToDOMRange:]):
2038         (-[WebCoreFrameBridge markDOMRange]):
2039         (-[WebCoreFrameBridge markedTextDOMRange]):
2040         (-[WebCoreFrameBridge smartDeleteRangeForProposedRange:]):
2041         (-[WebCoreFrameBridge documentFragmentWithMarkupString:baseURLString:]):
2042         (-[WebCoreFrameBridge documentFragmentWithText:inContext:]):
2043         (-[WebCoreFrameBridge documentFragmentWithNodesAsParagraphs:]):
2044         (-[WebCoreFrameBridge replaceSelectionWithNode:selectReplacement:smartReplace:matchStyle:]):
2045         (-[WebCoreFrameBridge replaceSelectionWithText:selectReplacement:smartReplace:]):
2046         (-[WebCoreFrameBridge increaseSelectionListLevel]):
2047         (-[WebCoreFrameBridge increaseSelectionListLevelOrdered]):
2048         (-[WebCoreFrameBridge increaseSelectionListLevelUnordered]):
2049         (-[WebCoreFrameBridge dragCaretDOMRange]):
2050         (-[WebCoreFrameBridge editableDOMRangeForPoint:]):
2051         (-[WebCoreFrameBridge characterRangeAtPoint:]):
2052         (-[WebCoreFrameBridge typingStyle]):
2053         (-[WebCoreFrameBridge rangeOfCharactersAroundCaret]):
2054         * platform/mac/ClipboardMac.mm:
2055         (WebCore::ClipboardMac::declareAndWriteDragImage):
2056         * platform/mac/PasteboardMac.mm:
2057         (WebCore::Pasteboard::writeSelection):
2058         Updated for new naming scheme.
2059
2060 2007-03-12  Timothy Hatcher  <timothy@apple.com>
2061
2062         Reviewed by Darin.
2063
2064         <rdar://problem/4990691> REGRESSION: Selecting text in Adium's Messages field causes horizontal shift
2065
2066         Corrected the recursive point conversion and scrolling done in scrollPointRecursively to use the
2067         document view of each NSClipView.
2068
2069         * platform/mac/ScrollViewMac.mm:
2070         (WebCore::ScrollView::scrollPointRecursively): Call convertPoint:fromView: on the document views,
2071         not the clip views as we encounter them. Then call constrainScrollPoint: on the converted point to
2072         constrain to the document view bounds. And finally call scrollPoint: on the document view, not the clip view.
2073
2074 2007-03-12  Antti Koivisto  <antti@apple.com>
2075
2076         Reviewed by Alexey.
2077
2078         Fix http://bugs.webkit.org/show_bug.cgi?id=12833
2079         REGRESSION: Selecting text in 6.6MB txt file is sluggish as of the Feb 19th nightly
2080         <rdar://problem/5028159>
2081
2082         Divide large text blocks (>64kB) over multiple text nodes. This limits linebox searches to
2083         a manageable subset.
2084
2085         * dom/Text.cpp:
2086         (WebCore::Text::createWithLengthLimit):
2087         * dom/Text.h:
2088         * html/HTMLParser.cpp:
2089         (WebCore::HTMLParser::parseToken):
2090         * loader/TextDocument.cpp:
2091         (WebCore::TextTokenizer::write):
2092
2093 2007-03-12  David Hyatt  <hyatt@apple.com>
2094
2095         Fix a regression in printing.  Printer fonts need to be part of the
2096         font cache key, since printer fonts have different glyph widths.
2097         (Integer antialiasing is used for screen fonts but not for printer fonts.)
2098
2099         Reviewed by aroben
2100
2101         * platform/FontCache.cpp:
2102         (WebCore::FontPlatformDataCacheKey::FontPlatformDataCacheKey):
2103         (WebCore::FontPlatformDataCacheKey::operator==):
2104         (WebCore::computeHash):
2105         (WebCore::FontCache::getCachedFontPlatformData):
2106
2107 2007-03-12  Lars Knoll <lars@trolltech.com>
2108
2109         add missing symbol to fix the Qt build again.
2110
2111         * platform/qt/SearchPopupMenuQt.cpp:
2112         (WebCore::SearchPopupMenu::enabled):
2113
2114 2007-03-12  Adele Peterson  <adele@apple.com>
2115
2116         Reviewed by Oliver.
2117
2118         Adding the ability to enable or disable a SearchPopupMenu.
2119
2120         * platform/SearchPopupMenu.h:
2121         * platform/mac/SearchPopupMenuMac.mm: (WebCore::SearchPopupMenu::enabled):
2122         * rendering/RenderTextControl.cpp: (WebCore::RenderTextControl::showPopup):
2123
2124 2007-03-12  Rob Buis  <buis@kde.org>
2125
2126         Reviewed by Nikolas Zimmermann.
2127
2128         http://bugs.webkit.org/show_bug.cgi?id=12500
2129         SVG fails to correctly handle all link style selectors
2130         http://bugs.webkit.org/show_bug.cgi?id=12567
2131         <text> elements ignore <a> children
2132
2133         Allow <a> inside svg text and handle xlink:show.
2134
2135         * ksvg2/svg/SVGAElement.cpp:
2136         (WebCore::SVGAElement::createRenderer):
2137         (WebCore::SVGAElement::defaultEventHandler):
2138         (WebCore::SVGAElement::childShouldCreateRenderer):
2139         * ksvg2/svg/SVGAElement.h:
2140         * ksvg2/svg/SVGElement.h:
2141         (WebCore::SVGElement::isTextContent):
2142         * ksvg2/svg/SVGTextContentElement.h:
2143         (WebCore::SVGTextContentElement::isTextContent):
2144         * ksvg2/svg/SVGTextElement.cpp:
2145         (WebCore::SVGTextElement::childShouldCreateRenderer):
2146         * rendering/SVGInlineFlowBox.cpp:
2147         (WebCore::translateBox):
2148         (WebCore::placePositionedBoxesHorizontally):
2149         (WebCore::placeBoxesVerticallyWithAbsBaseline):
2150
2151 2007-03-12  Adele Peterson  <adele@apple.com>
2152
2153         Reviewed by Oliver.
2154
2155         Add a missing parameter to the constructor.
2156
2157         * platform/PlatformKeyboardEvent.h:
2158         * platform/win/KeyEventWin.cpp: (WebCore::PlatformKeyboardEvent::PlatformKeyboardEvent):
2159
2160 2007-03-12  Lars Knoll <lars@trolltech.com>
2161
2162         Fix the Qt build. I still get crashes though :/
2163
2164         * platform/graphics/qt/ImageDecoderQt.cpp:
2165         (WebCore::ImageDecoderQt::clearFrame):
2166         * platform/graphics/qt/ImageDecoderQt.h:
2167         * platform/graphics/qt/ImageSourceQt.cpp:
2168         (WebCore::ImageSource::setData):
2169         (WebCore::ImageSource::frameIsCompleteAtIndex):
2170         (WebCore::ImageSource::clear):
2171         (WebCore::ImageSource::destroyFrameAtIndex):
2172         * platform/qt/ClipboardQt.cpp:
2173         (WebCore::ClipboardQt::ClipboardQt):
2174         * platform/qt/ClipboardQt.h:
2175         * platform/qt/DragDataQt.cpp:
2176         (WebCore::DragData::createClipboard):
2177
2178 2007-03-11  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
2179
2180         Not reviewed - gdk build fixes.
2181
2182         * platform/gdk/EditorClientGdk.cpp:
2183         (WebCore::EditorClientGdk::respondToChangedSelection): add stub
2184         * platform/gdk/EditorClientGdk.h:
2185         * platform/gdk/FrameGdk.cpp:
2186         * platform/gdk/TemporaryLinkStubs.cpp:
2187         (Editor::markMisspellings): add stub
2188
2189 2007-03-11  Alexey Proskuryakov  <ap@webkit.org>
2190
2191         Reviewed by Adele.
2192
2193         http://bugs.webkit.org/show_bug.cgi?id=12560
2194         W3C XPath test Text_Nodes.svg fails
2195
2196         * xml/XPathStep.cpp:
2197         (WebCore::XPath::Step::nodeTestMatches): Revert the fix, as the behavior doesn't appear
2198         all that desirable as it did at first glance.
2199
2200 2007-03-11  Oliver Hunt  <oliver@apple.com>
2201
2202         Reviewed by hyatt.
2203
2204         Fix for <rdar://problem/5055690> ASSERTION failure on drop into
2205         editable element with content changed on drop
2206
2207         After setting the selection for a drop into an editable region
2208         we make sure the we succeeded.  If we didn't we assume a focus handler
2209         or similar altered the element contents and try again, if the second
2210         attempt fails we bail out.
2211
2212         * page/DragController.cpp:
2213         (WebCore::setSelectionToDragCaret):
2214         (WebCore::DragController::concludeDrag):
2215
2216 2007-03-11  Oliver Hunt  <oliver@apple.com>
2217
2218         Reviewed by Adele.
2219
2220         Moving Frame{Mac}::respondToChangedSelection to Frame.cpp
2221         Added new EditorClient method to handle old bridge function
2222
2223         * bridge/EditorClient.h:
2224            Added respondToChangedSelection to replace old bridge function
2225         * editing/Editor.cpp:
2226         (WebCore::Editor::respondToChangedSelection):
2227            Add client call to replace old bridge call from Frame::respondToChangedSelection
2228         * editing/SelectionController.cpp:
2229         (WebCore::SelectionController::setSelection):
2230            No longer directly call Editor as Frame::respondToChangedSelection
2231            makes the call
2232         * page/Frame.cpp:
2233         (WebCore::Frame::respondToChangedSelection):
2234            Moved from FrameMac, replaced bridge call with call to Editor
2235         * page/mac/FrameMac.mm:
2236            Moved respondToChangedSelection to Frame.cpp
2237         * page/mac/WebCoreFrameBridge.h:
2238            Removed respondToChangedSelection from bridge
2239         * page/qt/FrameQt.cpp:
2240            Remove stub method for respondToChangedSelection
2241
2242 2007-03-11  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
2243
2244         Reviewed by Brady Eidson.
2245
2246         Linux/gdk fix.
2247
2248         * loader/gdk/FrameLoaderClientGdk.cpp:
2249         (WebCore::FrameLoaderClientGdk::shouldGoToHistoryItem): return true so
2250         that FrameLoader:goBackOrForwards() works.
2251
2252 2007-03-10  Geoffrey Garen  <ggaren@apple.com>
2253
2254         Reviewed by Darin Adler.
2255
2256         Fixed <rdar://problem/4587763> PAC file: lock inversion between QT and
2257         JSCore causes a hang @ www.panoramas.dk
2258
2259         See JavaScriptCore ChangeLog for details.
2260
2261         * bindings/objc/WebScriptObject.mm:
2262         (_didExecute): Added helpful ASSERT.
2263         (+[WebScriptObject throwException:]): Added missing JSLock.
2264
2265 2007-03-11  Antti Koivisto  <antti@apple.com>
2266
2267         Reviewed by Hyatt.
2268
2269         Optimize linebox memory consumption:
2270         - move all bitfields to baseclass compacting them
2271         - make InlineTextBox::m_truncation unsigned short and make it relative to m_start
2272         - remove extremely rarely used EllipsisBox pointer from RootInlineBox and instead
2273           use a global hashmap to store it if needed
2274         - use minimum required number of bits to store BidiStatus enum variables in RootInlineBox
2275         - move overflow variables in RootInlineBox to a separate struct that is instantiated
2276           only if any of the variables is set to a value that can't trivially be derived from
2277           box x, y, width and height
2278
2279         As a result line box objects shrink:
2280             InlineBox: 44 -> 44 bytes
2281             InlineTextBox: 68 -> 60 bytes
2282             InlineFlowBox: 68 -> 64 bytes
2283             RootInlineBox: 128 -> 88 bytes
2284
2285         The optimizations possiblity was noticed when debugging http://bugs.webkit.org/show_bug.cgi?id=12833
2286         Bug 12833: REGRESSION: Selecting text in 6.6MB txt file is sluggish as of the Feb 19th nightly
2287         <rdar://problem/5028159>
2288
2289         On that page the patch saves 11.5MB or some 21% of linebox memory consumption. It also
2290         actually improves selection performance somewhat by improving memory locality.
2291
2292         * rendering/InlineBox.h:
2293         (WebCore::InlineBox::InlineBox):
2294         * rendering/InlineFlowBox.h:
2295         (WebCore::InlineFlowBox::InlineFlowBox):
2296         * rendering/InlineTextBox.cpp:
2297         (WebCore::InlineTextBox::placeEllipsisBox):
2298         (WebCore::InlineTextBox::nodeAtPoint):
2299         (WebCore::InlineTextBox::paint):
2300         (WebCore::InlineTextBox::paintDecoration):
2301         (WebCore::InlineTextBox::paintSpellingOrGrammarMarker):
2302         (WebCore::InlineTextBox::paintMarkedTextUnderline):
2303         * rendering/InlineTextBox.h:
2304         (WebCore::InlineTextBox::InlineTextBox):
2305         * rendering/RootInlineBox.cpp:
2306         (WebCore::throw):
2307         (WebCore::RootInlineBox::Overflow::operator delete):
2308         (WebCore::RootInlineBox::Overflow::destroy):
2309         (WebCore::RootInlineBox::destroy):
2310         (WebCore::RootInlineBox::detachEllipsisBox):
2311         (WebCore::RootInlineBox::clearTruncation):
2312         (WebCore::RootInlineBox::placeEllipsis):
2313         (WebCore::RootInlineBox::paintEllipsisBox):
2314         (WebCore::RootInlineBox::addHighlightOverflow):
2315         (WebCore::RootInlineBox::nodeAtPoint):
2316         (WebCore::RootInlineBox::adjustPosition):
2317         (WebCore::RootInlineBox::selectionTop):
2318         (WebCore::RootInlineBox::setLineBreakInfo):
2319         (WebCore::RootInlineBox::ellipsisBox):
2320         (WebCore::RootInlineBox::setVerticalOverflowPositions):
2321         (WebCore::RootInlineBox::setHorizontalOverflowPositions):
2322         (WebCore::RootInlineBox::setVerticalSelectionPositions):
2323         * rendering/RootInlineBox.h:
2324         (WebCore::RootInlineBox::RootInlineBox):
2325         (WebCore::RootInlineBox::topOverflow):
2326         (WebCore::RootInlineBox::bottomOverflow):
2327         (WebCore::RootInlineBox::leftOverflow):
2328         (WebCore::RootInlineBox::rightOverflow):
2329         (WebCore::RootInlineBox::lineBreakBidiStatus):
2330         (WebCore::RootInlineBox::selectionBottom):
2331         (WebCore::RootInlineBox::Overflow::Overflow):
2332
2333 2007-03-11  Alexey Proskuryakov  <ap@webkit.org>
2334
2335         Reviewed by Darin.
2336
2337         A partial fix for http://bugs.webkit.org/show_bug.cgi?id=13021
2338         XPath can be very slow
2339
2340         * xml/XPathExpression.cpp:
2341         (WebCore::XPathExpression::evaluate): Cache evaluationContext in a local variable.
2342
2343         * xml/XPathExpressionNode.cpp:
2344         (WebCore::XPath::Expression::evaluationContext):
2345         * xml/XPathExpressionNode.h:
2346         (WebCore::XPath::Expression::addSubExpression):
2347         (WebCore::XPath::Expression::subExprCount):
2348         (WebCore::XPath::Expression::subExpr):
2349         * xml/XPathFunctions.cpp:
2350         * xml/XPathFunctions.h:
2351         (WebCore::XPath::Function::setName):
2352         (WebCore::XPath::Function::arg):
2353         (WebCore::XPath::Function::argCount):
2354         (WebCore::XPath::Function::name):
2355         Made one-liners critical for performance inline.
2356
2357         * xml/XPathGrammar.y: Fully parse NodeTests, so that strings are no longer passed for what is
2358         essentially an enum. Use LocationPath accessors to add steps, instead of directly manipulating
2359         internal data members.
2360
2361         * xml/XPathParser.cpp:
2362         (WebCore::XPath::Parser::parseStatement):
2363         (WebCore::XPath::Parser::registerNodeTest):
2364         (WebCore::XPath::Parser::deleteNodeTest):
2365         * xml/XPathParser.h:
2366         Added support methods for changes in XPathGrammar.y.
2367
2368         * xml/XPathPath.cpp:
2369         (WebCore::XPath::Filter::evaluate): Cache evaluationContext in a local variable. Use swap() to avoid
2370         performing vector assignments.
2371         (WebCore::XPath::LocationPath::evaluate): Use swap() to avoid performing vector assignments.
2372         (WebCore::XPath::LocationPath::optimizeStepPair): This new method is called during LocationPath construction,
2373         to simplify the path as it's being built. Currently, the only optimized case is "//*" - it is a basis for
2374         important operations that cannot be efficiently written in XPath 1.0, but can be optimized with a little bit
2375         of XPath 2.0.
2376         (WebCore::XPath::LocationPath::appendStep): A new accessor that modifies m_steps and calls optimizeStepPair().
2377         (WebCore::XPath::LocationPath::insertFirstStep): Ditto.
2378         * xml/XPathPath.h:
2379         (WebCore::XPath::LocationPath::setAbsolute): A new accessor.
2380
2381         * xml/XPathStep.h:
2382         (WebCore::XPath::Step::NodeTest::):
2383         (WebCore::XPath::Step::NodeTest::NodeTest):
2384         (WebCore::XPath::Step::NodeTest::kind):
2385         (WebCore::XPath::Step::NodeTest::data):
2386         Step::NodeTest is a new sub-class that represents a fully parsed NodeTest.
2387         (WebCore::XPath::Step::axis):
2388         (WebCore::XPath::Step::nodeTest):
2389         (WebCore::XPath::Step::nodeTestData):
2390         (WebCore::XPath::Step::namespaceURI):
2391         (WebCore::XPath::Step::predicates):
2392         (WebCore::XPath::Step::setAxis):
2393         (WebCore::XPath::Step::setNodeTest):
2394         (WebCore::XPath::Step::setNodeTestData):
2395         (WebCore::XPath::Step::setNamespaceURI):
2396         (WebCore::XPath::Step::setPredicates):
2397         New accessors that let optimizeStepPair() manipulate Step data.
2398
2399         * xml/XPathStep.cpp:
2400         (WebCore::XPath::Step::Step): Use the new NodeTest class.
2401         (WebCore::XPath::Step::evaluate): Cache evaluationContext in a local variable. Use swap() to avoid
2402         performing unneeded vector assignments.
2403         (WebCore::XPath::Step::nodesInAxis): Cosmetic changes.
2404         (WebCore::XPath::Step::nodeTestMatches): Use NodeTest instead of parsing the test from string each time.
2405         Added a partial implementation of XPath 2.0 element() node test.
2406
2407 2007-03-10  Alexey Proskuryakov  <ap@webkit.org>
2408
2409         Reviewed by Darin.
2410
2411         http://bugs.webkit.org/show_bug.cgi?id=12249
2412         FCKeditor: <hr>, <ul> and <ol> have id="undefined"
2413
2414         This fixes the attached reduction, but not the original issue.
2415
2416         Test: editing/execCommand/default-parameters.html
2417
2418         * dom/Document.h:
2419         * dom/Document.idl:
2420         Make second and third execCommand() parameters optional.
2421
2422 2007-03-10  Adele Peterson  <adele@apple.com>
2423
2424         Reviewed by Maciej.
2425
2426         Fix for http://bugs.webkit.org/show_bug.cgi?id=13028
2427         REGRESSION: textField:doCommandBySelector:inFrame: not being called properly
2428
2429         * html/HTMLInputElement.cpp: (WebCore::HTMLInputElement::defaultEventHandler):
2430           Before calling the base class defaultEventHandler, which will call handleKeypress, call doTextFieldCommandFromEvent.
2431
2432 2007-03-10  Mitz Pettel  <mitz@webkit.org>
2433
2434         Reviewed by Darin.
2435
2436         - fix http://bugs.webkit.org/show_bug.cgi?id=13013
2437           REGRESSION: Selection box does not scroll to where the focus jumps when pressing an alphanumeric key
2438
2439         Test: fast/forms/listbox-typeahead-scroll.html
2440
2441         * html/HTMLSelectElement.cpp:
2442         (WebCore::HTMLSelectElement::setSelectedIndex): Reordered to set the active selection's
2443         anchor and end before selecting the option, since the active selection is used to
2444         decide where to scroll when the selection is made.
2445         (WebCore::HTMLSelectElement::defaultEventHandler): Removed redundant check.
2446
2447 2007-03-10  Mitz Pettel  <mitz@webkit.org>
2448
2449         Reviewed by Darin.
2450
2451         - fix http://bugs.webkit.org/show_bug.cgi?id=12973
2452           REGRESSION: Reproducible assert while loading this test file if css is already in the cache
2453
2454         Test: fast/dom/css-cached-import-rule.html
2455
2456         Replaced some direct calls to document->stylesheetLoaded() with calls to
2457         the sheet's checkLoaded(). The latter calls back to the element's sheetLoaded() --
2458         which notifies the document of the load -- and then updates the sheet's
2459         loadCompleted() flag, ensuring that it stays in sync with whether the stylesheet
2460         is still considered pending by the document.
2461
2462         * dom/ProcessingInstruction.cpp:
2463         (WebCore::ProcessingInstruction::parseStyleSheet):
2464         * dom/StyleElement.cpp:
2465         (WebCore::StyleElement::childrenChanged):
2466         * html/HTMLLinkElement.cpp:
2467         (WebCore::HTMLLinkElement::process):
2468         (WebCore::HTMLLinkElement::setCSSStyleSheet):
2469         * ksvg2/svg/SVGStyleElement.cpp:
2470         (WebCore::SVGStyleElement::sheetLoaded):
2471         * ksvg2/svg/SVGStyleElement.h:
2472
2473 2007-03-10  David Kilzer  <ddkilzer@webkit.org>
2474
2475         Reviewed by Darin.
2476
2477         - fix http://bugs.webkit.org/show_bug.cgi?id=9609
2478           REGRESSION: Missing image icon needs to be moved back to WebKit
2479
2480         * WebCore.exp: Export WebCore::Image::loadPlatformResource(const char*) for use in
2481         [WebHTMLView _startDraggingImage:at:operation:event:sourceIsDHTML:DHTMLWroteData:].
2482
2483 2007-03-09  Darin Adler  <darin@apple.com>
2484
2485         Reviewed by Justin.
2486
2487         - fix http://bugs.webkit.org/show_bug.cgi?id=8928
2488           <rdar://problem/5045708> REPRODUCIBLE ASSERT: Cannot paste HTML into a
2489           contenteditable region in an XHTML document (8928)
2490
2491         Test: editing/pasteboard/paste-xml.xhtml
2492
2493         * editing/markup.cpp: (WebCore::createFragmentFromMarkup): Added a check for 0
2494         here, since createContextualFragment can return 0 for XML documents that fail
2495         to parse. In my testing, callers all seem equipped to handle 0.
2496
2497 2007-03-09  Mitz Pettel  <mitz@webkit.org>
2498
2499         Reviewed by Darin.
2500
2501         - fix http://bugs.webkit.org/show_bug.cgi?id=9929
2502           REGRESSION: crash on logging in on mijnpostbank.nl
2503
2504         Test: http/tests/misc/onload-remove-iframe-crash-2.html
2505
2506         The resulted from an iframe's load event handler removing the iframe
2507         from the document.
2508
2509         * dom/Document.cpp:
2510         (WebCore::Document::implicitClose): Bail out early if an event handler
2511         removed the frame.
2512         * loader/FrameLoader.cpp:
2513         (WebCore::FrameLoader::FrameLoader):
2514         (WebCore::FrameLoader::clear):
2515         (WebCore::FrameLoader::checkCompleted): Protect the frame from deletion
2516         by event handlers.
2517         (WebCore::FrameLoader::checkCompletedTimerFired):
2518         (WebCore::FrameLoader::scheduleCheckCompleted):
2519         (WebCore::FrameLoader::detachFromParent): Schedule a completion check
2520         on the parent (in case the child is what has been keeping it from completing).
2521         * loader/FrameLoader.h:
2522
2523 2007-03-08  David Kilzer  <ddkilzer@webkit.org>
2524
2525         Reviewed by Beth.
2526
2527         - fix http://bugs.webkit.org/show_bug.cgi?id=13019
2528           REGRESSION (r20074): Forms don't submit on a variety of websites
2529
2530         No tests added since LayoutTests/fast/forms/document-write.html was timing out
2531         and causing a layout test failure.
2532
2533         * html/HTMLFormElement.cpp:
2534         (WebCore::HTMLFormElement::submit): Removed stray code.
2535
2536 2007-03-08  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
2537
2538         Not reviewed - gdk build fix.
2539
2540         * platform/graphics/cairo/ImageSourceCairo.cpp:
2541         (WebCore::ImageSource::frameIsCompleteAtIndex): add empty stub
2542
2543 2007-03-08  Justin Garcia  <justin.garcia@apple.com>
2544
2545         Reviewed by harrison
2546
2547         <rdar://problem/4903193>
2548         On particular page, Copy is so slow it seems like a hang
2549
2550         ~2x speedup copying:
2551         http://shakespeare.mit.edu/henryviii/full.html
2552         Also produces less bloated markup.
2553
2554         * editing/markup.cpp:
2555         (WebCore::startMarkup): Don't wrap text nodes in style spans.
2556         For Elements, don't inline styles inherited from ancestors.
2557         (WebCore::createMarkup): No longer necessary to find
2558         the root's default style and pass it to startMarkup.
2559         Add a wrapper span around the markup with the styles
2560         that all nodes in the markup inherit (the inheritable
2561         styles from the common ancestor container's computed style).
2562         Added a FIXME about unecessary markup for inline ancestors
2563         up to the commonAncestorBlock.
2564
2565 2007-03-08  Darin Adler  <darin@apple.com>
2566
2567         Reviewed by Adele.
2568
2569         - <rdar://problem/4470381> multipart/form-data boundary security vulnerability
2570
2571         By making the form data boundary a string with some random data in it, we reduce
2572         the possibility that anyone could take advantage of it by creating a file that
2573         intentionally has the boundary string in it.
2574
2575         * html/HTMLFormElement.h: Removed boundary(), setBoundary(), and m_boundary.
2576         Marked a lot more stuff private.
2577         * html/HTMLFormElement.cpp:
2578         (WebCore::HTMLFormElement::HTMLFormElement): Removed code to initialize
2579         m_boundary.
2580         (WebCore::randomNumber): Added. Function that returns a random number, including
2581         seeding the random number generator the first time it's called. For now, usees the more
2582         random function random() on Mac OS X and the more-standard rand() on other platforms.
2583         (WebCore::HTMLFormElement::formData): Take a parameter with the form boundary string,
2584         and use that instead of m_boundary.
2585         (WebCore::getUniqueBoundaryString): Added. Makes a boundary string using random numbers
2586         and base 64 encoding.
2587         (WebCore::HTMLFormElement::submit): Call getUniqueBoundaryString and pass the boundary
2588         string into formData for multipart form posts.
2589
2590 2007-03-08  Maciej Stachowiak  <mjs@apple.com>
2591
2592         Reviewed by Adele.
2593
2594         <rdar://problem/4646563> REGRESSION: Unable to send text message from Verizon text message website: vtext.com (12588)
2595         http://bugs.webkit.org/show_bug.cgi?id=12588
2596
2597         Carefully revised which focus operations restore previous selection, which clear it, and which
2598         select the whole control contents.
2599
2600         Tests:
2601         fast/forms/focus-selection-input.html
2602         fast/forms/focus-selection-textarea.html
2603
2604         * dom/Element.cpp:
2605         (WebCore::Element::focus):
2606         * dom/Element.h:
2607         * html/HTMLInputElement.cpp:
2608         (WebCore::HTMLInputElement::focus):
2609         (WebCore::HTMLInputElement::accessKeyAction):
2610         * html/HTMLInputElement.h:
2611         * html/HTMLLabelElement.cpp:
2612         (WebCore::HTMLLabelElement::focus):
2613         (WebCore::HTMLLabelElement::accessKeyAction):
2614         * html/HTMLLabelElement.h:
2615         * html/HTMLLegendElement.cpp:
2616         (WebCore::HTMLLegendElement::focus):
2617         * html/HTMLLegendElement.h:
2618         * html/HTMLTextAreaElement.cpp:
2619         (WebCore::HTMLTextAreaElement::focus):
2620         * html/HTMLTextAreaElement.h:
2621         * page/FocusController.cpp:
2622         (WebCore::FocusController::advanceFocus):
2623
2624 2007-03-08  Justin Garcia  <justin.garcia@apple.com>
2625
2626         Reviewed by harrison
2627
2628         <http://bugs.webkit.org/show_bug.cgi?id=12244>
2629         FCKeditor: Find dialog doesn't work
2630
2631         * bindings/js/kjs_window.cpp:
2632         (KJS::Window::find): Added.  This function doesn't yet
2633         support whole word searches, searching in subframes, or
2634         opening the find dialog.
2635         (KJS::WindowFunc::callAsFunction):
2636         * bindings/js/kjs_window.h:
2637         (KJS::Window::):
2638
2639 2007-03-08  David Hyatt  <hyatt@apple.com>
2640
2641         Fix regression from throwing away frames of large animated images.  Alter
2642         animated images so that they refuse to advance the animation until the
2643         current displayed frame has been fully decoded.
2644
2645         Reviewed by ggaren
2646
2647         * platform/graphics/BitmapImage.cpp:
2648         (WebCore::BitmapImage::startAnimation):
2649         (WebCore::BitmapImage::advanceAnimation):
2650         * platform/graphics/ImageSource.h:
2651         * platform/graphics/cg/ImageSourceCG.cpp:
2652         (WebCore::ImageSource::frameIsCompleteAtIndex):
2653
2654 2007-03-08  David Hyatt  <hyatt@apple.com>
2655
2656         Fix 2% performance regression on the PLT.  Increase the large animated
2657         image cutoff from 1MB to 5MB.
2658
2659         In addition when pruning we will aggressively discard image sources.
2660
2661         Reviewed by ggaren
2662
2663         * platform/graphics/BitmapImage.cpp:
2664         (WebCore::BitmapImage::destroyDecodedData):
2665
2666 2007-03-08  Timothy Hatcher  <timothy@apple.com>
2667
2668         Reviewed by John.
2669
2670         <rdar://problem/4664697> highlighter SPI needs a node parameter to give more context
2671
2672         Pass the RenderObject's node to customHighlightLineRect and paintCustomHighlight.
2673
2674         * page/Frame.h:
2675         * page/mac/FrameMac.mm:
2676         (WebCore::Frame::customHighlightLineRect):
2677         (WebCore::Frame::paintCustomHighlight):
2678         * page/mac/WebCoreFrameBridge.h:
2679         * rendering/InlineTextBox.cpp:
2680         (WebCore::InlineTextBox::paintCustomHighlight):
2681         * rendering/RenderBox.cpp:
2682         (WebCore::RenderBox::paintCustomHighlight):
2683         * rendering/RootInlineBox.cpp:
2684         (WebCore::RootInlineBox::addHighlightOverflow):
2685         (WebCore::RootInlineBox::paintCustomHighlight):
2686
2687 2007-03-08  Justin Garcia  <justin.garcia@apple.com>
2688
2689         Reviewed by harrison
2690
2691         <http://bugs.webkit.org/show_bug.cgi?id=13000>
2692         Range.createContextualFragment is not supported
2693
2694         * dom/Range.cpp:
2695         (WebCore::Range::createContextualFragment): The
2696         "startContainer" may not be a container, if the
2697         range starts inside text.  In that case, look
2698         to the parent of the start node for an HTMLElement.
2699
2700 2007-03-08  Justin Garcia  <justin.garcia@apple.com>
2701
2702         Reviewed by harrison
2703
2704         <rdar://problem/5049671>
2705         Gmail Editor: With linked text, Remove Formatting doesn't always remove underline
2706
2707         * editing/Editor.cpp:
2708         (WebCore::Editor::removeFormattingAndStyle): Clear removed
2709         anchors after the deletion.
2710
2711 2007-03-08  David Kilzer  <ddkilzer@webkit.org>
2712
2713         Reviewed by NOBODY (build fix).
2714
2715         Added missing file for r20059:
2716         <rdar://problem/4708689> -- REGRESSION: Some symbols with 2-byte display as garbage in Hotmail.
2717
2718         File was taken from this svn repository which contained versions of ucnv.h and ucnv_err.h
2719         that were identical to ours:
2720         http://source.icu-project.org/repos/icu/icu/tags/release-3-2/source/common/unicode/ucnv_cb.h
2721
2722         * icu/unicode/ucnv_cb.h: Added.
2723
2724 2007-03-08  Shrikant Gangoda  <shrikant.gangoda@celunite.com>
2725
2726         Gdk build fix.
2727
2728         * platform/graphics/cairo/ImageSourceCairo.cpp:
2729         (WebCore::ImageSource::~ImageSource):
2730         (WebCore::ImageSource::clear):
2731
2732 2007-03-08  Oliver Hunt  <oliver@apple.com>
2733
2734         Reviewed by Adam.
2735
2736         To match old TEC behaviour when using ICU we need to use
2737         a few manual fallback encodings for the GBK/EUC-CN charsets
2738
2739         <rdar://problem/4708689> -- REGRESSION: Some symbols with 2-byte display as garbage in Hotmail.
2740
2741         * platform/TextCodecICU.cpp:
2742         (WebCore::TextCodecICU::TextCodecICU):
2743         (WebCore::TextCodecICU::createICUConverter):
2744         (WebCore::gbkEscapes):
2745         (WebCore::gbkCallbackEscape):
2746         (WebCore::gbkCallbackSubstitute):
2747         (WebCore::TextCodecICU::encode):
2748         * platform/TextCodecICU.h:
2749         (WebCore::TextCodecICU::needsGBKFallbacks):
2750         (WebCore::TextCodecICU::setNeedsGBKFallbacks):
2751
2752 2007-03-08  Alexey Proskuryakov  <ap@webkit.org>
2753
2754         Build fix.
2755
2756         * xml/XPathUtil.cpp:
2757         (WebCore::XPath::stringValue):
2758
2759 2007-03-08  David Hyatt  <hyatt@apple.com>
2760
2761         This patch dramatically reduces the memory consumed by animated images.  For large animated GIFs (defined for
2762         now as >1mb in terms of decoded frame buffer size), we will now aggressively flush previous frames of the
2763         animated GIF and just re-decode them on the fly if the animation loops.
2764
2765         Whenever a large animated GIF has its animation reset, we will also just throw out everything and start
2766         the animation over (in order to get rid of any cached detritus held in the ImageSource).
2767
2768         With this patch and the sample GIF used to test, WebKit's memory consumption went from 160MB down to 16MB.
2769
2770         Reviewed by mjs
2771
2772         * platform/graphics/BitmapImage.cpp:
2773         (WebCore::BitmapImage::destroyDecodedData):
2774         (WebCore::BitmapImage::resetAnimation):
2775         (WebCore::BitmapImage::advanceAnimation):
2776         * platform/graphics/ImageSource.h:
2777         * platform/graphics/cg/ImageSourceCG.cpp:
2778         (WebCore::ImageSource::~ImageSource):
2779         (WebCore::ImageSource::clear):
2780
2781 2007-03-08  Alexey Proskuryakov  <ap@webkit.org>
2782
2783         Reviewed by Darin.
2784
2785         http://bugs.webkit.org/show_bug.cgi?id=13006
2786         XPath string-value is broken for some node types
2787
2788         Test: fast/xpath/string-value.html
2789
2790         * xml/XPathUtil.cpp:
2791         (WebCore::XPath::stringValue): Fix it :-)
2792
2793 2007-03-07  Anders Carlsson  <acarlsson@apple.com>
2794
2795         Reviewed by Adam.
2796
2797         <rdar://problem/4981000>
2798         http://bugs.webkit.org/show_bug.cgi?id=12634
2799         REGRESSION: crash loading web archive (12634)
2800
2801         The reason this bug wasn't always reproducible is that it involved sending an event to a plugin while
2802         the page was loading. Before we send the event to the plugin we defer loads. The problem was that
2803         MainResourceLoader::setDefersLoad would not work with data loads.
2804
2805         * loader/DocumentLoader.cpp:
2806         (WebCore::DocumentLoader::setRequest):
2807         Only set m_committed to false if we also have a valid unreachable URL.
2808
2809         * loader/MainResourceLoader.cpp:
2810         (WebCore::MainResourceLoader::setDefersLoading):
2811         Make sure to stop and start data loads.
2812
2813 2007-03-07  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
2814
2815         Not reviewed.
2816
2817         Gdk build fixes.
2818
2819         * platform/gdk/EditorClientGdk.cpp:
2820         (WebCore::EditorClientGdk::handleKeypress): renamed
2821         (WebCore::EditorClientGdk::handleInputMethodKeypress): add a stub
2822         * platform/gdk/EditorClientGdk.h:
2823         * platform/graphics/cairo/ImageSourceCairo.cpp:
2824         (WebCore::ImageSource::destroyFrameAtIndex): add a stub
2825
2826 2007-03-07  Alexey Proskuryakov  <ap@webkit.org>
2827
2828         Reviewed by Darin.
2829
2830         http://bugs.webkit.org/show_bug.cgi?id=13004
2831         Repeatedly calling XPathExpression.evaluate() causes crashes or memory leaks
2832
2833         Removed XPath::Expression::optimize() and related methods, since they were buggy and almost useless.
2834         Merged doEvaluate() into evaluate(), since this was all evaluate() was doing after the above changes.
2835
2836         Test: fast/xpath/evaluate-twice.html
2837
2838         * xml/XPathExpression.cpp:
2839         (WebCore::XPathExpression::evaluate):
2840         * xml/XPathExpressionNode.cpp:
2841         (WebCore::XPath::Expression::Expression):
2842         (WebCore::XPath::Expression::~Expression):
2843         * xml/XPathExpressionNode.h:
2844         * xml/XPathFunctions.cpp:
2845         (WebCore::XPath::FunLast::evaluate):
2846         (WebCore::XPath::FunPosition::evaluate):
2847         (WebCore::XPath::FunId::evaluate):
2848         (WebCore::XPath::FunLocalName::evaluate):
2849         (WebCore::XPath::FunNamespaceURI::evaluate):
2850         (WebCore::XPath::FunName::evaluate):
2851         (WebCore::XPath::FunCount::evaluate):
2852         (WebCore::XPath::FunString::evaluate):
2853         (WebCore::XPath::FunConcat::evaluate):
2854         (WebCore::XPath::FunStartsWith::evaluate):
2855         (WebCore::XPath::FunContains::evaluate):
2856         (WebCore::XPath::FunSubstringBefore::evaluate):
2857         (WebCore::XPath::FunSubstringAfter::evaluate):
2858         (WebCore::XPath::FunSubstring::evaluate):
2859         (WebCore::XPath::FunStringLength::evaluate):
2860         (WebCore::XPath::FunNormalizeSpace::evaluate):
2861         (WebCore::XPath::FunTranslate::evaluate):
2862         (WebCore::XPath::FunBoolean::evaluate):
2863         (WebCore::XPath::FunNot::evaluate):
2864         (WebCore::XPath::FunTrue::evaluate):
2865         (WebCore::XPath::FunLang::evaluate):
2866         (WebCore::XPath::FunFalse::evaluate):
2867         (WebCore::XPath::FunNumber::evaluate):
2868         (WebCore::XPath::FunSum::evaluate):
2869         (WebCore::XPath::FunFloor::evaluate):
2870         (WebCore::XPath::FunCeiling::evaluate):
2871         (WebCore::XPath::FunRound::evaluate):
2872         * xml/XPathPath.cpp:
2873         (WebCore::XPath::Filter::evaluate):
2874         (WebCore::XPath::LocationPath::evaluate):
2875         (WebCore::XPath::Path::evaluate):
2876         * xml/XPathPath.h:
2877         * xml/XPathPredicate.cpp:
2878         (WebCore::XPath::Number::evaluate):
2879         (WebCore::XPath::StringExpression::evaluate):
2880         (WebCore::XPath::Negative::evaluate):
2881         (WebCore::XPath::NumericOp::evaluate):
2882         (WebCore::XPath::EqTestOp::evaluate):
2883         (WebCore::XPath::LogicalOp::evaluate):
2884         (WebCore::XPath::Union::evaluate):
2885         * xml/XPathPredicate.h:
2886         * xml/XPathStep.cpp:
2887         * xml/XPathStep.h:
2888         * xml/XPathVariableReference.cpp:
2889         (WebCore::XPath::VariableReference::evaluate):
2890         * xml/XPathVariableReference.h:
2891
2892 2007-03-07  Sam Weinig  <sam@webkit.org>
2893
2894         Reviewed by Tim H.
2895
2896         Remove unused #import from Objective-C bindings and cleanup the order of #imports.
2897
2898         * bindings/scripts/CodeGeneratorObjC.pm:
2899
2900 2007-03-07  Sam Weinig  <sam@webkit.org>
2901
2902         Reviewed by Tim H.
2903
2904         Make sure the baseURI attribute generates for private Objective-C bindings.
2905
2906         * dom/Node.idl:
2907
2908 2007-03-07  Anders Carlsson  <acarlsson@apple.com>
2909
2910         Reviewed by Maciej.
2911
2912         <rdar://problem/4874059>
2913         REGRESSION: Painter IX:register - Crash in WebCore:: ResourceLoader::willSendRequest()
2914
2915         If a load is done from inside of an error delegate method that is called because we cancel another load,
2916         the first load should be ignored since this is what shipping WebKit does.
2917
2918         (Actually, it does load the page in the data source but doesn't do anything with it since the data source
2919         won't have a web frame).
2920
2921         * loader/FrameLoader.cpp:
2922         (WebCore::FrameLoader::load):
2923         Just bail out if m_isStoppingLoad is true.
2924
2925 2007-03-07  David Hyatt  <hyatt@apple.com>
2926
2927         Use CGImageRelease instead of CFRelease.
2928
2929         Reviewed by aroben
2930
2931         * platform/graphics/cg/ImageCG.cpp:
2932         (WebCore::FrameData::clear):
2933         * platform/graphics/cg/ImageSourceCG.cpp:
2934         (WebCore::ImageSource::destroyFrameAtIndex):
2935
2936 2007-03-07  David Hyatt  <hyatt@apple.com>
2937
2938         Fix a regression where the cache size overflows because of a double
2939         subtraction per resource when they got removed from the cache.  Add an
2940         assert to adjustSize to detect this case in the future.
2941
2942         Fix ImageSourceCG so that when we flush decoded data from our cache that
2943         we also flush it from the ImageSource.
2944
2945         Reviewed by mjs
2946
2947         * loader/Cache.cpp:
2948         (WebCore::Cache::adjustSize):
2949         * platform/graphics/BitmapImage.cpp:
2950         (WebCore::BitmapImage::~BitmapImage):
2951         (WebCore::BitmapImage::destroyDecodedData):
2952         * platform/graphics/Image.h:
2953         * platform/graphics/ImageSource.h:
2954         * platform/graphics/cg/ImageSourceCG.cpp:
2955         (WebCore::ImageSource::setData):
2956         (WebCore::ImageSource::destroyFrameAtIndex):
2957
2958 2007-03-07  Mitz Pettel  <mitz@webkit.org>
2959
2960         Reviewed by Darin.
2961
2962         - fix http://bugs.webkit.org/show_bug.cgi?id=13002
2963           Incomplete repaint of inset outlines
2964
2965         Test: fast/repaint/outline-inset.html
2966
2967         * rendering/RenderObject.cpp:
2968         (WebCore::RenderObject::repaintAfterLayoutIfNeeded):
2969
2970 2007-03-07  Oliver Hunt  <oliver@apple.com>
2971
2972         Reviewed by Antti.
2973
2974         Reset mouse down/drag state variables on mouse button release
2975
2976         Fixes <rdar://problem/5044654>: Drag out of some QuickTime plug-ins converting into image drag
2977
2978         * page/EventHandler.cpp:
2979         (WebCore::EventHandler::handleMouseReleaseEvent):
2980
2981 2007-03-07  Adele Peterson  <adele@apple.com>
2982
2983         Reviewed by Darin.
2984
2985         WebCore part of fix for:
2986         http://bugs.webkit.org/show_bug.cgi?id=10871
2987         http://bugs.webkit.org/show_bug.cgi?id=12677
2988         <rdar://problem/4823129> REGRESSION: IME key events different in nightly
2989         <rdar://problem/4759563> REGRESSION: Return key is always sent when you confirm a clause in kotoeri
2990
2991         * page/EventHandler.cpp: (WebCore::EventHandler::defaultKeyboardEventHandler): Call handleInputMethodKeypress before actually dispatching the keypress event
2992           so that input methods have a chance to handle the event.
2993           If the input method handles the event (by marking or unmarking text), then we don't need to send the keypress event.
2994           If an input method doesn't handle the event, then we'll save the data we need to perform the correct action (like what text to insert or what selector to use)
2995           when we dispatch the keypress event.
2996
2997         * dom/KeyboardEvent.h: Added Mac-specific KeypressCommand struct, so we can store command info during handleInputMethodKeypress, and use it during handleKeypress.
2998         (WebCore::KeyboardEvent::keypressCommand):
2999         (WebCore::KeyboardEvent::setKeypressCommand):
3000
3001         * bridge/EditorClient.h:
3002         * editing/Editor.cpp:
3003         (WebCore::Editor::handleKeypress): Changed handleKeyPress to handleKeypress.
3004         (WebCore::Editor::handleInputMethodKeypress): Added.
3005         * editing/Editor.h:
3006
3007         * platform/graphics/svg/SVGImageEmptyClients.h:
3008         (WebCore::SVGEmptyEditorClient::handleKeypress): Changed handleKeyPress to handleKeypress.
3009         (WebCore::SVGEmptyEditorClient::handleInputMethodKeypress): Added.
3010
3011 2007-03-07  Rob Buis  <buis@kde.org>
3012
3013         Reviewed by Darin.
3014
3015         http://bugs.webkit.org/show_bug.cgi?id=12579
3016         WebKit fails SVG xml:base test
3017
3018         Implement DOM3 properties baseURI and documentURI to fix
3019         the testcase in bug 12579.
3020
3021         * bindings/js/kjs_dom.cpp:
3022         (KJS::DOMNode::getValueProperty):
3023         * bindings/js/kjs_domnode.h:
3024         (KJS::DOMNode::):
3025         * dom/Document.cpp:
3026         (WebCore::Document::documentURI):
3027         (WebCore::Document::setDocumentURI):
3028         (WebCore::Document::baseURI):
3029         * dom/Document.h:
3030         * dom/Document.idl:
3031         * dom/DocumentType.cpp:
3032         (WebCore::DocumentType::baseURI):
3033         * dom/DocumentType.h:
3034         * dom/Element.cpp:
3035         (WebCore::Element::baseURI):
3036         * dom/Element.h:
3037         * dom/Node.cpp:
3038         (WebCore::Node::baseURI):
3039         * dom/Node.h:
3040         * ksvg2/misc/SVGImageLoader.cpp:
3041         (WebCore::SVGImageLoader::updateFromElement):
3042         * ksvg2/svg/SVGImageElement.cpp:
3043         (WebCore::SVGImageElement::parseMappedAttribute):
3044         (WebCore::SVGImageElement::attach):
3045
3046 2007-03-07  Anders Carlsson  <acarlsson@apple.com>
3047
3048         Reviewed by Brady.
3049
3050         Remove some methods in FrameLoader that just calls down to the active document loader. Since each
3051         resource loader now has a pointer to its document loader, we can just call directly to the
3052         document loader.
3053
3054         * WebCore.exp:
3055         * loader/FrameLoader.cpp:
3056         * loader/FrameLoader.h:
3057         * loader/MainResourceLoader.cpp:
3058         (WebCore::MainResourceLoader::willSendRequest):
3059         (WebCore::MainResourceLoader::didReceiveResponse):
3060         * loader/SubresourceLoader.cpp:
3061         (WebCore::SubresourceLoader::SubresourceLoader):
3062         (WebCore::SubresourceLoader::didFinishLoading):
3063         (WebCore::SubresourceLoader::didFail):
3064         (WebCore::SubresourceLoader::didCancel):
3065         * loader/mac/NetscapePlugInStreamLoaderMac.mm:
3066         (WebCore::NetscapePlugInStreamLoader::didFinishLoading):
3067         (WebCore::NetscapePlugInStreamLoader::didFail):
3068         (WebCore::NetscapePlugInStreamLoader::didCancel):
3069
3070 2007-03-06  Geoffrey Garen  <ggaren@apple.com>
3071
3072         Reviewed by Maciej Stachowiak.
3073
3074         Fixed <rdar://problem/4576242> | http://bugs.webkit.org/show_bug.cgi?id=12586
3075         PAC file: malloc deadlock sometimes causes a hang @ www.apple.com/pro/profiles/ (12586)
3076
3077         No test because this is very difficult to repro, and the new ASSERTs in
3078         JavaScriptCore catch the underlying cause while running normal layout tests.
3079
3080         This is a modified version of r14752 on the branch.
3081
3082         The fix is to use a bit inside each node, instead of a hash table, to track
3083         which node subtrees are in the process of being marked. This avoids a call
3084         to malloc inside mark().
3085
3086         * bindings/js/kjs_binding.cpp:
3087         (KJS::domObjects):
3088         (KJS::domNodesPerDocument):
3089         * bindings/js/kjs_dom.cpp:
3090         (KJS::DOMNode::mark):
3091         * dom/Node.cpp:
3092         (WebCore::Node::Node):
3093         * dom/Node.h:
3094
3095 2007-03-06  David Hyatt  <hyatt@apple.com>
3096
3097         This patch reworks the WebCore memory cache to significantly reduce the amount of memory consumed by
3098         images in the cache and to enhance the accuracy of the cache size as an absolute bound for the objects
3099         contained within it.  WebCore's memory use over time should significantly improve as a result of these
3100         changes.
3101
3102         Cached resources now have both an encoded size (the original data stream) and a decoded size (an estimate of
3103         the amount of memory consumed by an expanded version of that resource, e.g., the decoded frames of an image).
3104         Both sizes now count towards the total size of the object and towards the allowed memory cache total.
3105
3106         By including both totals the reported size of resources will now be larger, and the cache will therefore become
3107         much more aggressive about flushing.
3108
3109         Objects are stored in size-adjusted and popularity-aware LRU lists as before, but encoded size is now always
3110         used when determining the correct LRU list.
3111
3112         The flush algorithm for the memory cache has been rewritten to first destroy decoded data before evicting
3113         resources.  By being able to compact its resources without evicting them, the memory cache can now hold many more
3114         unique resources (encoded) in the same amount of space.  Depending on how much of a hit we want to take from
3115         re-decoding images, the memory cache could in theory have its size significantly reduced now while still holding
3116         more resources than it did at the larger size!
3117
3118         Reviewed by mjs
3119
3120         * WebCore.xcodeproj/project.pbxproj:
3121         * loader/Cache.cpp:
3122         (WebCore::Cache::requestResource):
3123         (WebCore::Cache::prune):
3124         (WebCore::Cache::remove):
3125         (WebCore::Cache::lruListFor):
3126         (WebCore::Cache::adjustSize):
3127         * loader/Cache.h:
3128         * loader/CachedCSSStyleSheet.cpp:
3129         (WebCore::CachedCSSStyleSheet::data):
3130         * loader/CachedImage.cpp:
3131         (WebCore::CachedImage::CachedImage):
3132         (WebCore::CachedImage::allReferencesRemoved):
3133         (WebCore::CachedImage::clear):
3134         (WebCore::CachedImage::data):
3135         (WebCore::CachedImage::destroyDecodedData):
3136         (WebCore::CachedImage::decodedSize):
3137         (WebCore::CachedImage::decodedSizeChanged):
3138         (WebCore::CachedImage::shouldPauseAnimation):
3139         * loader/CachedImage.h:
3140         * loader/CachedResource.cpp:
3141         (WebCore::CachedResource::CachedResource):
3142         (WebCore::CachedResource::deref):
3143         (WebCore::CachedResource::setEncodedSize):
3144         * loader/CachedResource.h:
3145         (WebCore::CachedResource::allReferencesRemoved):
3146         (WebCore::CachedResource::size):
3147         (WebCore::CachedResource::encodedSize):
3148         (WebCore::CachedResource::decodedSize):
3149         (WebCore::CachedResource::destroyDecodedData):
3150         * loader/CachedScript.cpp:
3151         (WebCore::CachedScript::data):
3152         * loader/CachedXSLStyleSheet.cpp:
3153         (WebCore::CachedXSLStyleSheet::data):
3154         * platform/graphics/BitmapImage.cpp:
3155         (WebCore::BitmapImage::BitmapImage):
3156         (WebCore::BitmapImage::~BitmapImage):
3157         (WebCore::BitmapImage::destroyDecodedData):
3158         (WebCore::BitmapImage::pruneDecodedDataIfNeeded):
3159         (WebCore::BitmapImage::cacheFrame):
3160         (WebCore::BitmapImage::setNativeData):
3161         (WebCore::BitmapImage::shouldAnimate):
3162         (WebCore::BitmapImage::advanceAnimation):
3163         * platform/graphics/BitmapImage.h:
3164         (WebCore::BitmapImage::decodedSize):
3165         * platform/graphics/Image.cpp:
3166         (WebCore::Image::Image):
3167         * platform/graphics/Image.h:
3168         (WebCore::Image::destroyDecodedData):
3169         (WebCore::Image::decodedSize):
3170         (WebCore::Image::imageObserver):
3171         * platform/graphics/ImageAnimationObserver.h: Removed.
3172         * platform/graphics/ImageObserver.h: Added.
3173         (WebCore::ImageObserver::~ImageObserver):
3174         * platform/graphics/svg/SVGImage.cpp:
3175         (WebCore::SVGImage::SVGImage):
3176         * platform/graphics/svg/SVGImage.h:
3177
3178 2007-03-06  Alexey Proskuryakov  <ap@webkit.org>
3179
3180         Reviewed by Sam Weinig.
3181
3182         http://bugs.webkit.org/show_bug.cgi?id=12987
3183         Fix and import 4XPath test_numeric_expr.html
3184
3185         * xml/XPathPredicate.cpp:
3186         (WebCore::XPath::Negative::doEvaluate): Convert the argument to number.
3187         (WebCore::XPath::NumericOp::doEvaluate): Convert the arguments to numbers. Use a correct operation for mod.
3188         * xml/XPathStep.cpp:
3189         (WebCore::XPath::Step::nodesInAxis): Do not append parent node if there is none.
3190         * xml/XPathValue.cpp:
3191         (WebCore::XPath::Value::toNumber): Do not convert to DeprecatedString just to trim whitespace and to convert to double.
3192         * platform/DeprecatedString.cpp:
3193         (WebCore::DeprecatedStringData::makeAscii): Added a FIXME about unreliable makeAscii() behavior.
3194
3195 2007-03-06  Maciej Stachowiak  <mjs@apple.com>
3196
3197         Reviewed by Adele.
3198
3199         <rdar://problem/4619663> REGRESSION (NativePopup): Popup menu doesn't draw at the correct vertical position (9816)
3200
3201         * platform/mac/PopupMenuMac.mm:
3202         (WebCore::PopupMenu::show): Make a temporary dummy view with the
3203         passed in rect, since AppKit will use the view bounds to determine
3204         what area to exclude when popping up a menu moved to the top of
3205         the screen.
3206
3207 2007-03-06  Geoffrey Garen  <ggaren@apple.com>
3208
3209         Reviewed by Maciej Stachowiak.
3210
3211         Fixed all known crashers exposed by run-webkit-tests --threaded [*]. See
3212         JavaScriptCore ChangeLog for more details.
3213
3214         * bindings/js/kjs_binding.cpp:
3215         (KJS::domNodesPerDocument): Added thread safety ASSERT.
3216         (KJS::ScriptInterpreter::mark): Removed obsolete logic for marking unsafe
3217         objects when collecting on a secondary thread. The Collector takes care
3218         of this now.
3219
3220         * bindings/js/kjs_binding.h:
3221         (KJS::DOMObject::DOMObject): Used new API for specifying that WebCore
3222         objects should be garbage collected on the main thread only.
3223
3224         * bindings/js/kjs_window.cpp:
3225         (KJS::ScheduledAction::execute): Moved JSLock to cover implementedsCall() call,
3226         which, for some subclasses, ends up allocating garbage collected objects.
3227         (This fix was speculative. I didn't actually see a crash from this.)
3228         (KJS::Window::timerFired): Added JSLock around ScheduleAction destruction,
3229         since it destroys a KJS::List.
3230
3231         * bindings/objc/WebScriptObject.mm:
3232         (-[WebScriptObject setException:]): Added JSLock. (This fix was speculative.
3233         I didn't actually see a crash from this.)
3234
3235         * bridge/mac/WebCoreScriptDebugger.mm:
3236         (-[WebCoreScriptCallFrame evaluateWebScript:]): Added JSLock. (This fix
3237         was speculative. I didn't actually see a crash from this.)
3238
3239         * dom/Document.cpp:
3240         (WebCore::Document::~Document): Added JSLock around modification to
3241         domNodesPerDocument(), which can be accessed concurrently during garbage
3242         collection.
3243         * dom/Node.cpp:
3244         (WebCore::Node::setDocument): ditto.
3245
3246         [*] fast/js/toString-stack-overflow.html is an exception. --threaded mode
3247         crashes this test because it causes the garbage collector to run frequently,
3248         and this test crashes if you happen to garbage collect while it's running.
3249         This is a known issue with stack overflow during the mark phase. It's
3250         not related to threading.
3251
3252 2007-03-06  Mark Rowe  <mrowe@apple.com>
3253
3254         Reviewed by Sam Weinig.
3255
3256         Fix http://bugs.webkit.org/show_bug.cgi?id=12942
3257         Bug 12942: ASSERTION FAILURE: qantas.com.au changing selected item in <select> via JS
3258
3259         Test: fast/dom/select-selectedIndex-bug-12942.html.
3260
3261         * html/HTMLSelectElement.cpp:
3262         (WebCore::HTMLSelectElement::recalcListItems): Reset m_lastOnChangeIndex when recalculating list items.
3263         * html/HTMLSelectElement.h:
3264
3265 2007-03-06  Brady Eidson  <beidson@apple.com>
3266
3267         Rubberstamped by Kevin Decker
3268
3269         20,000!
3270
3271         * ChangeLog: Point out revision 20,000
3272
3273 2007-03-06  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
3274
3275         Not reviewed.
3276
3277         Gdk build fix.
3278
3279         * loader/gdk/FrameLoaderClientGdk.cpp: update userAgent() signature.
3280         (WebCore::FrameLoaderClientGdk::userAgent):
3281         * loader/gdk/FrameLoaderClientGdk.h: ditto.
3282
3283 2007-03-06  Mitz Pettel  <mitz@webkit.org>
3284
3285         Reviewed by Adele.
3286
3287         - fix http://bugs.webkit.org/show_bug.cgi?id=12986
3288           REGRESSION(NativeListBox): Listboxes not updated when resized dynamically
3289
3290         Test: fast/forms/select-change-listbox-size.html
3291
3292         * html/HTMLSelectElement.cpp:
3293         (WebCore::HTMLSelectElement::parseMappedAttribute): Reattach on list box size change.
3294
3295 2007-03-06  Mitz Pettel  <mitz@webkit.org>
3296
3297         Reviewed by Dave Hyatt.
3298
3299         - fix http://bugs.webkit.org/show_bug.cgi?id=12885
3300           REGRESSION (r19696): Incomplete background repaint
3301
3302         Tests: fast/repaint/content-into-overflow.html
3303                fast/repaint/overflow-into-content.html
3304
3305         Changed repaintAfterLayoutIfNeeded() to take, in addition to the clipped overflow
3306         rect, the unclipped border box plus outline, and to repaint any areas that
3307         were added or removed from that box, in addition to any areas added or removed
3308         from the clipped overflow rect.
3309
3310         * platform/graphics/svg/SVGResourceMarker.cpp:
3311         (WebCore::SVGResourceMarker::draw):
3312         * rendering/RenderBlock.cpp:
3313         (WebCore::RenderBlock::layoutBlock):
3314         * rendering/RenderBox.cpp:
3315         (WebCore::RenderBox::absoluteClippedOverflowRect): Renamed getAbsoluteRepaintRect() to
3316         this.
3317         * rendering/RenderBox.h:
3318         * rendering/RenderFlexibleBox.cpp:
3319         (WebCore::RenderFlexibleBox::layoutBlock):
3320         * rendering/RenderFlow.cpp:
3321         (WebCore::RenderFlow::absoluteClippedOverflowRect):
3322         * rendering/RenderFlow.h:
3323         * rendering/RenderForeignObject.cpp:
3324         (WebCore::RenderForeignObject::layout):
3325         * rendering/RenderHTMLCanvas.cpp:
3326         (WebCore::RenderHTMLCanvas::layout):
3327         * rendering/RenderImage.cpp:
3328         (WebCore::RenderImage::layout):
3329         * rendering/RenderLayer.cpp:
3330         (WebCore::RenderLayer::RenderLayer):
3331         (WebCore::RenderLayer::checkForRepaintOnResize):
3332         (WebCore::RenderLayer::updateLayerPositions):
3333         * rendering/RenderLayer.h:
3334         * rendering/RenderObject.cpp:
3335         (WebCore::RenderObject::repaint):
3336         (WebCore::RenderObject::repaintAfterLayoutIfNeeded):
3337         (WebCore::RenderObject::getAbsoluteRepaintRectWithOutline):
3338         (WebCore::RenderObject::absoluteClippedOverflowRect):
3339         (WebCore::RenderObject::absoluteOutlineBox):
3340         * rendering/RenderObject.h:
3341         * rendering/RenderPath.cpp:
3342         (WebCore::RenderPath::layout):
3343         (WebCore::RenderPath::absoluteClippedOverflowRect):
3344         (WebCore::RenderPath::absoluteRects):
3345         (WebCore::RenderPath::drawMarkersIfNeeded):
3346         * rendering/RenderPath.h:
3347         * rendering/RenderSVGContainer.cpp:
3348         (WebCore::RenderSVGContainer::layout):
3349         (WebCore::RenderSVGContainer::absoluteClippedOverflowRect):
3350         (WebCore::RenderSVGContainer::absoluteRects):
3351         * rendering/RenderSVGContainer.h:
3352         * rendering/RenderSVGHiddenContainer.cpp:
3353         (WebCore::RenderSVGHiddenContainer::absoluteClippedOverflowRect):
3354         * rendering/RenderSVGHiddenContainer.h:
3355         * rendering/RenderSVGImage.cpp:
3356         (WebCore::RenderSVGImage::imageChanged):
3357         (WebCore::RenderSVGImage::absoluteClippedOverflowRect):
3358         (WebCore::RenderSVGImage::absoluteRects):
3359         * rendering/RenderSVGImage.h:
3360         * rendering/RenderSVGText.cpp:
3361         (WebCore::RenderSVGText::absoluteClippedOverflowRect):
3362         (WebCore::RenderSVGText::layout):
3363         (WebCore::RenderSVGText::absoluteRects):
3364         * rendering/RenderSVGText.h:
3365         * rendering/RenderTable.cpp:
3366         (WebCore::RenderTable::layout):
3367         * rendering/RenderTableCell.cpp:
3368         (WebCore::RenderTableCell::absoluteClippedOverflowRect):
3369         * rendering/RenderTableCell.h:
3370         * rendering/RenderTableCol.cpp:
3371         (WebCore::RenderTableCol::absoluteClippedOverflowRect):
3372         * rendering/RenderTableCol.h:
3373         * rendering/RenderTableRow.cpp:
3374         (WebCore::RenderTableRow::absoluteClippedOverflowRect):
3375         * rendering/RenderTableRow.h:
3376         * rendering/RenderText.cpp:
3377         (WebCore::RenderText::absoluteClippedOverflowRect):
3378         * rendering/RenderText.h:
3379
3380 2007-03-06  Ian Eng <ian.eng.webkit@gmail.com>
3381
3382         Reviewed by Maciej.
3383
3384         - fixed http://bugs.webkit.org/show_bug.cgi?id=12720
3385         Bug 12720: Re-defining window.location.toString function keeps re-loading forever
3386
3387         * bindings/js/kjs_window.cpp:
3388         Disallow replacing functions in LocationTable, and return early without updating URL.
3389
3390 2007-03-06  Kevin McCullough  <kmccullough@apple.com>
3391
3392         Reviewed by Darin.
3393
3394         <http://bugs.webkit.org/show_bug.cgi?id=12686>
3395         REGRESSION: Bloglines.com Feeds tab cannot expand folders in TOT
3396         - Now all class constructors implement implementsHasInstance.
3397
3398         * bindings/scripts/CodeGeneratorJS.pm:
3399
3400 2007-03-07  Nikolas Zimmermann  <zimmermann@kde.org>
3401
3402         Reviewed by Rob & Oliver.
3403
3404         Preparations for the new SVG text engine.
3405
3406         Handle baseline-shift / kerning css values correctly.
3407         Recognize missing svg presentation attribute "font-size-adjust".
3408         Add all missing svg<->css property mappings in mapToEntry().
3409
3410         Doesn't affect any layout test (as these properties are not used w/o my text patch).
3411         While I'm at it, unify the macro names (RS_ -> SVG_RS_) & cleanup style a bit.
3412
3413         * ksvg2/css/CSSPropertyNames.in:
3414         * ksvg2/css/SVGCSSParser.cpp:
3415         (WebCore::CSSParser::parseSVGValue):
3416         * ksvg2/css/SVGCSSStyleSelector.cpp:
3417         (WebCore::CSSStyleSelector::applySVGProperty):
3418         * ksvg2/css/SVGRenderStyle.cpp:
3419         (WebCore::SVGRenderStyle::SVGRenderStyle):
3420         (WebCore::SVGRenderStyle::operator==):
3421         (WebCore::SVGRenderStyle::inheritedNotEqual):
3422         (WebCore::SVGRenderStyle::inheritFrom):
3423         * ksvg2/css/SVGRenderStyle.h:
3424         (WebCore::SVGRenderStyle::NonInheritedFlags::):
3425         (WebCore::SVGRenderStyle::setBitDefaults):
3426         * ksvg2/css/SVGRenderStyleDefs.cpp:
3427         (StyleTextData::StyleTextData):
3428         (StyleTextData::operator==):
3429         (StyleMiscData::StyleMiscData):
3430         (StyleMiscData::operator==):
3431         * ksvg2/css/SVGRenderStyleDefs.h:
3432         (WebCore::):
3433         (WebCore::StyleTextData::operator!=):
3434         * ksvg2/svg/SVGStyledElement.cpp:
3435         (WebCore::SVGStyledElement::cssPropertyIdForSVGAttributeName):
3436
3437 2007-03-06  Justin Garcia  <justin.garcia@apple.com>
3438
3439         Reviewed by kevin
3440
3441         <http://bugs.webkit.org/show_bug.cgi?id=12245>
3442         FCKeditor: Remove Format sometimes doesn't work
3443         <rdar://problem/4786404>
3444         Underline style is not removed from selection after performing Remove Format
3445
3446         * editing/Editor.cpp:
3447         (WebCore::Editor::removeFormattingAndStyle): Re-wrote this.
3448
3449 2007-03-07  Nikolas Zimmermann  <zimmermann@kde.org>
3450
3451         Reviewed by Oliver.
3452
3453         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12979
3454         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12981
3455
3456         Guard against registering pending resources with empty id.
3457         Be careful with calling recalcStyle() in SVGUseElement, when change is "Detach" we
3458         have to use special code, as calling attach() on the shadow tree root element will crash,
3459         because it has no (direct) parent node, only a shadow parent node element.
3460
3461         * ksvg2/misc/SVGDocumentExtensions.cpp:
3462         (WebCore::SVGDocumentExtensions::addPendingResource):
3463         * ksvg2/svg/SVGUseElement.cpp:
3464         (WebCore::SVGUseElement::recalcStyle):
3465
3466 2007-03-06  Kevin Decker  <kdecker@apple.com>
3467
3468         Reviewed by Adele.
3469
3470         Fixed: <rdar://problem/5041660> REGRESSION: <keygen> element broken, prevents users from signing up for Thawte email certs
3471
3472         * bindings/objc/DOM.mm:
3473         (WebCore::createElementClassMap): Added <keygen> to the DOM bindings so it can be accessed from Objective-C.
3474         * css/html4.css: Apply the look of the <select> element to <keygen>.
3475         * html/HTMLElementFactory.cpp:
3476         (WebCore::keygenConstructor): Added.
3477         (WebCore::createFunctionMap): Added keygen.
3478         * html/HTMLSelectElement.cpp: Made const typeAheadTimeout variable static const.
3479         (WebCore::HTMLSelectElement::HTMLSelectElement):  Fix both HTMLSelectElement constructors to initialize the
3480          same number of member variables. The fact that some fields were not initialized could (and would) crash the
3481          keygen element when selecting different items. Also removed m_typedString(String()) from the constructor
3482          initialization because this is not needed.
3483
3484 2007-03-06  Kevin McCullough  <kmccullough@apple.com>
3485
3486         Reviewed by Darin.
3487
3488         - Rename a function to clarify its purpose.
3489
3490         * WebCore.exp:
3491         * loader/FrameLoader.cpp:
3492         (WebCore::FrameLoader::registerURLSchemeAsLocal):
3493         * loader/FrameLoader.h:
3494
3495 2007-03-06  Adam Roben  <aroben@apple.com>
3496
3497         Build fix.
3498
3499         * WebCore.exp: Updated symbols.
3500
3501 2007-03-06  Adam Roben  <aroben@apple.com>
3502
3503         Reviewed by Anders.
3504
3505         Added a parameter to all StringTruncator methods to specify whether
3506         rounding hacks should be on or off.
3507
3508         No layout test possible.
3509
3510         * platform/StringTruncator.cpp:
3511         (WebCore::stringWidth): Added disableRoundingHacks parameter.
3512         (WebCore::truncateString): Ditto.
3513         (WebCore::StringTruncator::centerTruncate): Ditto.
3514         (WebCore::StringTruncator::rightTruncate): Ditto.
3515         (WebCore::StringTruncator::width): Ditto.
3516         * platform/StringTruncator.h: Ditto.
3517         * platform/mac/FileChooserMac.mm:
3518         (WebCore::FileChooser::basenameForWidth): Pass in false to
3519         centerTruncate so that the truncation matches the way the text will be
3520         rendered.
3521
3522 2007-03-06  Anders Carlsson  <acarlsson@apple.com>
3523
3524         Reviewed by Maciej.
3525
3526         <rdar://problem/5035045>
3527         REGRESSION: WebKit browser doesn't display image at http://www.metoffice.gov.uk/weather/satellite/index.html
3528
3529         It turns out WinIE does allow you to access images by their id as special document properties. However, this is only
3530         allowed when the element also has a name attribute. The value of the name attribute is ignored and can even be empty!
3531
3532         * bindings/js/kjs_html.cpp:
3533         (KJS::JSHTMLDocument::namedItemGetter):
3534         Return jsUndefined() if the collection is empty.
3535
3536         * html/HTMLImageElement.cpp:
3537         (WebCore::HTMLImageElement::parseMappedAttribute):
3538         (WebCore::HTMLImageElement::insertedIntoDocument):
3539         (WebCore::HTMLImageElement::removedFromDocument):
3540         * html/HTMLImageElement.h:
3541         Add the id attribute value to the extra named item map.
3542
3543         * html/HTMLNameCollection.cpp:
3544         (WebCore::HTMLNameCollection::traverseNextItem):
3545         Check for images with name attributes that match, as well as elements with id attributes that match where
3546         the element also has a name attribute.
3547
3548 2007-03-06  Anders Carlsson  <acarlsson@apple.com>
3549
3550         Reviewed by Adam.
3551
3552         WebCore part of patch to make it possible to have different user agents for different URLs.
3553
3554         * bindings/js/kjs_navigator.cpp:
3555         (KJS::Navigator::getValueProperty):
3556         * bindings/js/kjs_proxy.cpp:
3557         (WebCore::KJSProxy::initScriptIfNeeded):
3558         * loader/DocumentLoader.cpp:
3559         (WebCore::DocumentLoader::setLoading):
3560         * loader/FrameLoader.cpp:
3561         (WebCore::FrameLoader::userAgent):
3562         (WebCore::FrameLoader::loadResourceSynchronously):
3563         (WebCore::FrameLoader::applyUserAgent):
3564         * loader/FrameLoader.h:
3565         * loader/FrameLoaderClient.h:
3566         * platform/graphics/svg/SVGImageEmptyClients.h:
3567         (WebCore::SVGEmptyFrameLoaderClient::userAgent):
3568
3569 2007-03-05  Kevin McCullough  <kmccullough@apple.com>
3570
3571         Reviewed by Mark and Dave H.
3572
3573         - rdar://problem/5038491
3574         An oversight of the security fix that prevented remote from loading local is that it
3575         prevents user style sheets when the site is remote.  This fixes that.
3576
3577         * loader/Cache.cpp: Propogate and check user style sheet flag.
3578         (WebCore::createResource):
3579         (WebCore::Cache::requestResource):
3580         * loader/Cache.h: Propogate user style sheet flag.
3581         * loader/CachedCSSStyleSheet.cpp: Propogate user style sheet flag.
3582         (WebCore::CachedCSSStyleSheet::CachedCSSStyleSheet):
3583         * loader/CachedCSSStyleSheet.h: Propogate user style sheet flag.
3584         * loader/DocLoader.cpp: Propogate user style sheet flag.
3585         (WebCore::DocLoader::requestResource):
3586         * loader/SubresourceLoader.cpp: Propogate and check user style sheet flag.
3587         (WebCore::SubresourceLoader::create):
3588         * loader/SubresourceLoader.h: Add check for user style sheet flag.
3589         * loader/loader.cpp: Propogate user style sheet flag.
3590         (WebCore::Loader::load):
3591         (WebCore::Loader::servePendingRequests):
3592         * loader/loader.h: Propogate user style sheet flag.
3593
3594 2007-03-06  Nikolas Zimmermann  <zimmermann@kde.org>
3595
3596         Reviewed by Darin.
3597
3598         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12936 (Master bug used to track all current use problems.)
3599
3600         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12926 (transform attribute not respected in nested <use> elements)
3601         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12267 (getElementById broken for <use>)
3602         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12916 (use instance in symbol definition does not work)
3603         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12917 (mouseout event does not occur after scaling use instance)
3604         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12838 (SVG <use> DOM broken for script)
3605
3606         Added test: svg/custom/use-nested-transform.svg
3607         Added test: svg/custom/struct-use-09-b.svg (not yet official W3C-SVG-1.1 testcase)
3608         Fixes test: svg/custom/use-elementInstance-event-target.svg (rectangle now turns green on first click)
3609
3610         Fix all (except one) known <use> bug. The remaining bug (12630) is not crucical at all.
3611         Enable <use> again as default SVG feature, as discussed on webkit-dev.
3612
3613         dom/Element.cpp needed following tweak: set hasParentStyle to true, if there is no parentNode
3614         available - which happens for <use> nodes, as it's a shadow node. This fixes recalcStyle behaviour.
3615         Override recalcStyle() in SVGUseElement, and properly forward the call to the shadow tree root element.
3616         The shadow tree now receives proper style updates (without having to recreate the whole tree!).
3617
3618         Override attributeChanged() in SVGUseElement and only call notifyAttributeChange if one of x/y/width/height/xlink:href
3619         attribute changed - otherwhise we'll end up recreating the tree for every transform/style/(non-existing-attribute) change.
3620
3621         Do not override transform in nested use situations, but correctly append (right-sided) the translation of the <use> element.
3622         Some cosmetic fixes: don't add transform="translate(0 0)" attributes if both x/y values are null in the shadow tree.
3623
3624         Factor out logic for replacing symbol/svg tags in the shadow tree, and also invoke it during expandUseElementsInShadowTree -
3625         otherwhise <symbol><use xlink:href="#someOtherSymbol"></symbol> the <use> gets expanded to a <symbol>. Though no one expands
3626         the <symbol> element - and we're end up in hitting an assertion. Avoid that.
3627
3628         * bindings/js/JSSVGElementWrapperFactory.cpp:
3629         * bindings/js/kjs_dom.cpp:
3630         (KJS::toJS):
3631         * bindings/objc/DOM.mm:
3632         (WebCore::createElementClassMap):
3633         * dom/Element.cpp:
3634         (WebCore::Element::recalcStyle):
3635         * ksvg2/svg/SVGElement.cpp:
3636         (WebCore::shadowTreeParentElementForShadowTreeElement):
3637         (WebCore::SVGElement::dispatchEvent):
3638         * ksvg2/svg/SVGElementInstance.cpp:
3639         * ksvg2/svg/SVGElementInstance.h:
3640         * ksvg2/svg/SVGElementInstance.idl:
3641         * ksvg2/svg/SVGElementInstanceList.cpp:
3642         * ksvg2/svg/SVGElementInstanceList.h:
3643         * ksvg2/svg/SVGElementInstanceList.idl:
3644         * ksvg2/svg/SVGStyledElement.cpp:
3645         (WebCore::SVGStyledElement::notifyAttributeChange):
3646         (WebCore::SVGStyledElement::updateElementInstance):
3647         * ksvg2/svg/SVGStyledElement.h:
3648         * ksvg2/svg/SVGUseElement.cpp:
3649         (WebCore::SVGUseElement::attributeChanged):
3650         (WebCore::SVGUseElement::notifyAttributeChange):
3651         (WebCore::SVGUseElement::recalcStyle):
3652         (WebCore::SVGUseElement::buildPendingResource):
3653         (WebCore::SVGUseElement::buildShadowTreeForSymbolTag):
3654         (WebCore::SVGUseElement::alterShadowTreeForSVGTag):
3655         (WebCore::SVGUseElement::buildShadowTree):
3656         (WebCore::SVGUseElement::expandUseElementsInShadowTree):
3657         (WebCore::SVGUseElement::attachShadowTree):
3658         * ksvg2/svg/SVGUseElement.h:
3659         * ksvg2/svg/SVGUseElement.idl:
3660         * ksvg2/svg/svgtags.in:
3661
3662 2007-03-05  Brady Eidson  <beidson@apple.com>
3663
3664         Reviewed by John
3665
3666         Fixes <rdar://problem/4974258>
3667         Adds some key null checking
3668
3669         * bindings/js/kjs_html.cpp:
3670         (KJS::JSHTMLElement::implementsCall): Null check doc/frame
3671         * bindings/objc/DOMInternal.mm:
3672         (-[WebScriptObject _initializeScriptDOMNodeImp]): Null check doc/frame
3673
3674 2007-03-06  Nikolas Zimmermann  <zimmermann@kde.org>
3675
3676         Reviewed by Darin.
3677
3678         Path::normalAngleAtLength() / Path::pointAtLength() don't work correctly.
3679         pointAtLength() was not implemented, basically and normalAngleAtLength()
3680         had a bug in the tangent slope calculation.
3681
3682         The normalAngleAtLength() stuff can only be tested with my local textPath support.
3683         New LayoutTest: svg/custom/path-textPath-simulation.svg
3684
3685         * platform/graphics/Path.cpp:
3686         (WebCore::pathLengthApplierFunction):
3687         * platform/graphics/PathTraversalState.cpp:
3688         (WebCore::PathTraversalState::quadraticBezierTo):
3689         (WebCore::PathTraversalState::cubicBezierTo):
3690         * platform/graphics/PathTraversalState.h:
3691         (WebCore::PathTraversalState::):
3692
3693 2007-03-05  Alexey Proskuryakov  <ap@webkit.org>
3694
3695         Reviewed by Darin.
3696
3697         http://bugs.webkit.org/show_bug.cgi?id=12970
3698         Fix and import 4XPath test_core_functions.html test
3699
3700         * xml/XPathExpression.cpp:
3701         (WebCore::XPathExpression::evaluate): Fully initialize the evaluation context.
3702
3703         * xml/XPathFunctions.cpp:
3704         (WebCore::XPath::FunSubstring::doEvaluate): Fixed handling of edge cases.
3705         (WebCore::XPath::FunRound::round): Reimplemented to match the spec; exposed FunRound::round() to be used in
3706         other functions.
3707
3708 2007-03-05  Alexey Proskuryakov  <ap@webkit.org>
3709
3710         Reviewed by Darin.
3711
3712         http://bugs.webkit.org/show_bug.cgi?id=12954
3713         XPath relative operations are implemented incorrectly
3714
3715         * xml/XPathPredicate.cpp:
3716         (WebCore::XPath::NumericOp::doEvaluate):
3717         (WebCore::XPath::EqTestOp::compare):
3718         (WebCore::XPath::EqTestOp::doEvaluate):
3719         Reimplemented relative equality operations to match the spec.
3720
3721         * xml/XPathPredicate.h:
3722         (WebCore::XPath::NumericOp::):
3723         (WebCore::XPath::EqTestOp::):
3724         Moved relative operations to EqTestOp.
3725
3726         * xml/XPathGrammar.y:
3727         * xml/XPathParser.cpp:
3728         (WebCore::XPath::Parser::nextTokenInternal):
3729         (WebCore::XPath::Parser::lex):
3730         Adapted for the above changes.
3731
3732 2007-03-05  Mark Rowe  <mrowe@apple.com>
3733
3734         Reviewed by Lars.
3735
3736         Fix http://bugs.webkit.org/show_bug.cgi?id=12947
3737         Bug 12947: REGRESSION: ASSERTION FAILED: maxWidth >= 0 in StringTruncator.cpp:109 in WebCore::truncateString()
3738
3739         Handle nil window correctly in toUserSpace and toDeviceSpace.  On Intel Macs a message to nil that returns a
3740         float will return 0.0.  We use this as the divisor in calculating a scale factor, which results in NaN being
3741         introduced into our rect.
3742
3743         * platform/mac/ScreenMac.mm:
3744         (WebCore::toUserSpace):
3745         (WebCore::toDeviceSpace):
3746
3747 2007-03-05  Rob Buis  <buis@kde.org>
3748
3749         Reviewed by Darin.
3750
3751         http://bugs.webkit.org/show_bug.cgi?id=12868
3752         parts of the CSS classes in this simple SVG example are not applied
3753
3754         Make sure the xml stylesheets are parsed in strict mode.
3755
3756         * dom/ProcessingInstruction.cpp:
3757         (WebCore::ProcessingInstruction::parseStyleSheet):
3758
3759 2007-03-04  Adele Peterson  <adele@apple.com>
3760
3761         Reviewed by Darin.
3762
3763         Change to dispatch the keypress event during the defaultEventHandler for keydown events.  This matches IE behavior.
3764         This is preparation for fixing event dispatch with input methods (http://bugs.webkit.org/show_bug.cgi?id=10871)
3765
3766         Test: fast/events/keydown-keypress-preventDefault.html
3767
3768         * dom/EventTargetNode.cpp: (WebCore::EventTargetNode::defaultEventHandler): Call the defaultKeyboardEventHandler for keydown events.
3769         * page/EventHandler.cpp:
3770         (WebCore::eventTargetNodeForDocument): Return 0 instead of false since the return type in EventTargetNode.
3771         (WebCore::EventHandler::keyEvent): Removed dispatch of keypress event, since this is now done in the default event handler.
3772         (WebCore::EventHandler::defaultKeyboardEventHandler): For keydown events, create and dispatch a keypress event.
3773
3774 2007-03-04  Alexey Proskuryakov  <ap@webkit.org>
3775
3776         Reviewed by Nikolas Zimmermann (yay!).
3777
3778         http://bugs.webkit.org/show_bug.cgi?id=12962
3779         4XPath tests crash on lang() function
3780
3781         Covered by 4XPath tests, to be landed later.
3782
3783         * platform/StringImpl.cpp:
3784         (WebCore::StringImpl::reverseFind): Do not crash with empty strings.
3785         * xml/XPathFunctions.cpp:
3786         (WebCore::XPath::FunLang::doEvaluate): Do not crash when an element has no
3787         attributes. Use a proper namespace for xml:lang (not sure where "xms" came from).
3788         Rewrote the algorithm for suffix removing to match the spec.
3789
3790 2007-03-02  Anders Carlsson  <acarlsson@apple.com>
3791
3792         Reviewed by Darin.
3793
3794         <rdar://problem/5028165>
3795         http://bugs.webkit.org/show_bug.cgi?id=12915
3796         REGRESSION: XMLHttpRequest.abort() does not stop loading (12915)
3797
3798         * loader/DocumentLoader.cpp:
3799         (WebCore::DocumentLoader::stopLoading):
3800         Save the value of m_loading since calling FrameLoader::stopLoading could set it to false.
3801
3802         * loader/SubresourceLoader.cpp:
3803         (WebCore::SubresourceLoader::didCancel):
3804         * loader/SubresourceLoader.h:
3805         Get rid of didCancel now, it's not needed anymore.
3806
3807         * xml/xmlhttprequest.cpp:
3808         (WebCore::XMLHttpRequest::abort):
3809         Call cancel() instead of stopLoading(). Also, set m_aborted to true so the XMLHttpRequest object
3810         won't be dereferenced in didFail when aborting.
3811
3812 2007-03-04  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
3813
3814         Reviewed by Nikolas Zimmermann.
3815
3816         Move ScrollView stubs to ScrollViewGdk.cpp
3817
3818         * platform/gdk/FrameGdk.h: remove comment that no longer makes sense
3819         * platform/gdk/ScrollViewGdk.cpp:
3820         (WebCore::ScrollView::addChild):
3821         (WebCore::ScrollView::removeChild):
3822         (WebCore::ScrollView::scrollPointRecursively):
3823         (WebCore::ScrollView::inWindow):
3824         (WebCore::ScrollView::wheelEvent):
3825         (WebCore::ScrollView::updateScrollbars):
3826         (WebCore::ScrollView::updateScrollInfo):
3827         (WebCore::ScrollView::windowToContents):
3828         (WebCore::ScrollView::contentsToWindow):
3829         (WebCore::ScrollView::scrollbarUnderMouse):
3830         * platform/gdk/TemporaryLinkStubs.cpp:
3831
3832 2007-03-02  Kevin McCullough  <kmccullough@apple.com>
3833
3834         Reviewed by Geoff.
3835
3836         - rdar://problem/4922454
3837         - This fixes a security issue by making remote referrers not able to access local
3838         resources, unless they register their schemes to be treated as local. The result is
3839         that those schemes can access local resources and cannot be accessed by remote
3840         referrers.
3841         Because this behavior is new a link-on-or-after check is made to determine if the
3842         app should use the older, less safe, behavior.
3843
3844         * WebCore.exp: added exported functions
3845         * bindings/objc/DOM.mm: consolodated function to base class
3846         (-[DOMElement image]):
3847         (-[DOMElement _imageTIFFRepresentation]):
3848         * dom/Document.cpp: Cache the document's ability to load local resources.
3849         (WebCore::Document::Document):
3850         (WebCore::Document::setURL):
3851         (WebCore::Document::shouldBeAllowedToLoadLocalResources):
3852         (WebCore::Document::stylesheetLoaded):
3853         * dom/Document.h: Cache the docuent's ability to load local resources.
3854         (WebCore::Document::getPendingSheet):
3855         (WebCore::Document::isAllowedToLoadLocalResources):
3856         * html/HTMLImageLoader.cpp: Moved functionality into base class.
3857         (WebCore::HTMLImageLoader::updateFromElement):
3858         (WebCore::HTMLImageLoader::dispatchLoadEvent):
3859         * html/HTMLLinkElement.cpp: Handles null returns correctly now.
3860         * html/HTMLTokenizer.cpp: Moved functionality into base class.
3861         (WebCore::HTMLTokenizer::notifyFinished):
3862         * ksvg2/misc/SVGImageLoader.cpp: Moved functionality into base class.
3863         (WebCore::SVGImageLoader::dispatchLoadEvent):
3864         * loader/Cache.cpp: Checks if the cached resource can be loaded.
3865         (WebCore::Cache::requestResource):
3866         * loader/CachedCSSStyleSheet.cpp: Moved functionality into base class.
3867         (WebCore::CachedCSSStyleSheet::ref):
3868         (WebCore::CachedCSSStyleSheet::error):
3869         * loader/CachedImage.cpp: Moved functionality into base class.
3870         (WebCore::CachedImage::CachedImage):
3871         * loader/CachedImage.h: Moved functionality into base class.
3872         (WebCore::CachedImage::canRender):
3873         * loader/CachedResource.cpp: Cache if the CachedResource should be treated as local
3874         (WebCore::CachedResource::CachedResource):
3875         * loader/CachedResource.h: Moved functionality into base class.
3876         (WebCore::CachedResource::errorOccurred):
3877         (WebCore::CachedResource::shouldTreatAsLocal):
3878         * loader/CachedScript.cpp: Moved functionality into base class.
3879         (WebCore::CachedScript::CachedScript):
3880         * loader/CachedScript.h: Moved functionality into base class.
3881         (WebCore::CachedScript::schedule):
3882         * loader/CachedXBLDocument.cpp: Moved functionality into base class.
3883         (WebCore::CachedXBLDocument::error):
3884         * loader/CachedXSLStyleSheet.cpp: Moved functionality into base class.
3885         (WebCore::CachedXSLStyleSheet::error):
3886         * loader/FrameLoader.cpp: See comments for each function below.
3887         (WebCore::FrameLoader::loadSubframe): Use new canLoad.
3888         (WebCore::FrameLoader::restrictAccessToLocal): return value of linked-on-or-after check.
3889         (WebCore::FrameLoader::setRestrictAccessToLocal): set value for linked-on-or-after check.
3890         (WebCore::localSchemes): Return set of schemes that are to be treated as local.
3891         (WebCore::FrameLoader::loadPlugin): Use new canLoad.
3892         (WebCore::FrameLoader::canLoad): Now multiple functions that each do the same work but some can take advantage of the cached values, if they were computed previously.
3893         (WebCore::FrameLoader::shouldHideReferrer): Extracted out the logic to determine if the referrer should be hidden so it is only calculated when needed.
3894         (WebCore::FrameLoader::loadResourceSynchronously): No longer calls canLoad to get hideReferrer info.
3895         (WebCore::FrameLoader::registerSchemeAsLocal): Functionality to register a scheme to be treated as local.
3896         (WebCore::FrameLoader::treatURLAsLocal): Given a URL this function determines if it should be treated as local.
3897         * loader/FrameLoader.h: Declared functions for this security fix.  See above.
3898         * loader/MainResourceLoader.cpp: Optized order of bools to regain performance.
3899         (WebCore::MainResourceLoader::continueAfterContentPolicy):
3900         * loader/SubresourceLoader.cpp: Now restricts remote from loading local resources.
3901         (WebCore::SubresourceLoader::create):
3902         * page/EventHandler.cpp: Moved functionality into base class.
3903         (WebCore::selectCursor):
3904         * platform/KURL.cpp: KURLs need to check all the registered schemes now.
3905         (WebCore::KURL::isLocalFile):
3906         * rendering/HitTestResult.cpp: Moved functionality into base class.
3907         (WebCore::HitTestResult::image):
3908         * rendering/RenderImage.cpp: Moved functionality into base class.
3909         (WebCore::RenderImage::setCachedImage):
3910         (WebCore::RenderImage::imageChanged):
3911         (WebCore::RenderImage::paint):
3912         (WebCore::RenderImage::layout):
3913         (WebCore::RenderImage::calcAspectRatioWidth):
3914         (WebCore::RenderImage::calcAspectRatioHeight):
3915         * rendering/RenderImage.h: Moved functionality into base class.
3916         (WebCore::RenderImage::errorOccurred):
3917         * rendering/RenderListItem.cpp: Moved functionality into base class.
3918         (WebCore::RenderListItem::setStyle):
3919         * rendering/RenderListMarker.cpp: Moved functionality into base class.
3920         (WebCore::RenderListMarker::isImage):
3921         * xml/xmlhttprequest.cpp: Check doc's cached value instead of determining independently.
3922         (WebCore::XMLHttpRequest::urlMatchesDocumentDomain):
3923
3924 2007-03-02  Justin Garcia  <justin.garcia@apple.com>
3925
3926         Reviewed by kevin
3927
3928         <rdar://problem/5028447>
3929         REGRESSION: Gmail Editor: Copied message text pastes at the wrong font size
3930
3931         * editing/markup.cpp:
3932         (WebCore::createMarkup): The style of the div that holds
3933         a fully selected body's styles didn't include styles inherited
3934         from the body's ancestors.
3935
3936 2007-03-02  Justin Garcia  <justin.garcia@apple.com>
3937
3938         Reviewed by harrison
3939
3940         <rdar://problem/4545040>
3941         innerHTML does not HTML-escape text nodes inside PRE elements
3942         <rdar://problem/5027857>
3943         Pasting into Mail from Safari's view-source window renders the HTML
3944
3945         * editing/HTMLInterchange.cpp:
3946         (WebCore::convertHTMLTextToInterchangeFormat): Send this function
3947         the node that the text comes from as a parameter.  It shouldn't convert
3948         '\n's to spaces/nbsps if the text is coming from text where newlines are
3949         preserved.
3950         * editing/HTMLInterchange.h:
3951         * editing/markup.cpp:
3952         (WebCore::startMarkup): Escape text inside the children of PREs.
3953
3954 2007-03-02  Sam Weinig  <sam@webkit.org>
3955
3956         Reviewed by Anders.
3957
3958         Try to fix the Qt build.
3959
3960         * platform/qt/TemporaryLinkStubs.cpp: Add stubs.
3961         (WebCore::searchMenuNoRecentSearchesText):
3962         (WebCore::searchMenuRecentSearchesText):
3963         (WebCore::searchMenuClearRecentSearchesText):
3964         (WebCore::AXWebAreaText):
3965         (WebCore::AXLinkText):
3966         (WebCore::AXListMarkerText):
3967         (WebCore::AXImageMapText):
3968         (WebCore::AXHeadingText):
3969
3970 2007-03-02  David Harrison  <harrison@apple.com>
3971
3972         Suggested by Darin.
3973
3974         A more efficient solution to rdar://4961431.
3975
3976         * bridge/mac/WebCoreAXObject.mm:
3977         (-[WebCoreAXObject accessibilityIsIgnored]):
3978
3979 2007-03-01  Justin Garcia  <justin.garcia@apple.com>
3980
3981         Reviewed by john
3982
3983         <rdar://problem/5032095>
3984         Gmail Editor: Copied text pastes on a new line instead of current line
3985
3986         Start merge failed to occur because positionAtStartOfInsertedContent
3987         had a bug.
3988
3989         * editing/ReplaceSelectionCommand.cpp:
3990         (WebCore::ReplaceSelectionCommand::positionAtStartOfInsertedContent):
3991         Was failing when inserting <span><div>foo/div></span>. Return the
3992         inserted content's first VisiblePosition.
3993
3994 2007-03-02  Dave Hyatt  <hyatt@apple.com>
3995
3996         Fix crasher in glyph map code (buffer overrun).
3997
3998         Reviewed by darin
3999
4000         * platform/win/GlyphPageTreeNodeWin.cpp:
4001         (WebCore::GlyphPage::fill):
4002
4003 2007-03-01  Antti Koivisto  <antti@apple.com>
4004
4005         Reviewed by Maciej.
4006
4007         Fix for http://bugs.webkit.org/show_bug.cgi?id=12895
4008         REGRESSION: imagemap: pointer cursor is shown everywhere
4009         <rdar://problem/5028163>
4010
4011         Image with imagemap should never itself be URLElement in hit test results.
4012
4013         No layout test, cursor state can't be captured.
4014
4015         * rendering/RenderLayer.cpp:
4016         (WebCore::RenderLayer::hitTest):
4017
4018 2007-03-01  Antti Koivisto  <antti@apple.com>
4019
4020         Reviewed by Maciej.
4021
4022         Fix http://bugs.webkit.org/show_bug.cgi?id=12690
4023         REGRESSION: can not log in to bank of america with TOT webkit
4024         <rdar://problem/4990044>
4025
4026         and http://bugs.webkit.org/show_bug.cgi?id=12604
4027         REGRESSION: After closing the "Would you like to save password" sheet, the form fails
4028         to submit automatically at http://www.mac.com/WebObjects/HomePage.woa
4029         <rdar://problem/4871752>
4030
4031         and http://bugs.webkit.org/show_bug.cgi?id=12020
4032         REGRESSION: Flickr uploading broken
4033         <rdar://problem/4928662>
4034
4035         Turn protection against multiple forms submission back on. This approach is buggy
4036         but it is way better than not having it at all. Not protecting against this
4037         breaks number of major sites.
4038
4039         * bridge/mac/WebCoreAXObject.mm:
4040         (-[WebCoreAXObject accessibilityPerformAction:]):
4041         * loader/FrameLoader.cpp:
4042         (WebCore::FrameLoader::resetMultipleFormSubmissionProtection):
4043         (WebCore::FrameLoader::submitForm):
4044         (WebCore::FrameLoader::receivedMainResourceError):
4045         * loader/FrameLoader.h:
4046         * page/EventHandler.cpp:
4047         (WebCore::EventHandler::keyEvent):
4048         * page/Frame.cpp:
4049         (WebCore::Frame::setView):
4050         * page/mac/EventHandlerMac.mm:
4051         (WebCore::EventHandler::mouseDown):
4052
4053 2007-03-01  Kevin McCullough  <kmccullough@apple.com>
4054
4055         Reviewed by Adam.
4056
4057         - Added the test case: external-script-URL-location.html
4058         - Fix an issue where the url of a document is null after an open if the document
4059         has no parent.
4060
4061         * dom/Document.cpp:
4062         (WebCore::Document::open):
4063
4064 2007-03-01  Anders Carlsson  <acarlsson@apple.com>
4065
4066         Reviewed by Darin.
4067
4068         <rdar://problem/4960250>
4069         http://bugs.webkit.org/show_bug.cgi?id=11627
4070         REGRESSION: Reproducible crash at IMDb in WebCore::FrameLoader::stopLoadingSubframes
4071
4072         In rare cases, we could end up calling checkLoadComplete twice for the same frame. This would cause the
4073         didFailProvisionalLoad delegate method to be called twice for the same frame, and also cause the provisional document loader
4074         to be reset to null when other code wasn't expecting it.
4075
4076         This regressed in revision 10904 with the fix for <rdar://problem/4184719>. The fix is to only call stopLoading on the frame
4077         if either the document loader is loading, or the document is still being parsed. I've verified that the bug is still fixed and
4078         that no leaks occur.
4079
4080         * loader/DocumentLoader.cpp:
4081         (WebCore::DocumentLoader::stopLoading):
4082
4083 2007-03-01  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
4084
4085         Reviewed by Darin.
4086
4087         Fix ARM crash due to accessing non-4-byte-aligned memory
4088         as 32-bit values.
4089
4090         * platform/AtomicString.cpp:
4091         (WebCore::UCharBufferTranslator::equal):
4092
4093 2007-03-01  David Harrison  <harrison@apple.com>
4094
4095         Reviewed by Darin.
4096
4097         <rdar://problem/5033905> Have the DOM secondary thread check raise an exception by default
4098
4099         * platform/mac/ThreadCheck.mm:
4100         (WebCore::_WebCoreThreadViolationCheck):
4101         Initialize threadViolationIsException to true.
4102
4103 2007-03-01  Brady Eidson  <beidson@apple.com>
4104
4105         Reviewed by Darin
4106
4107         <rdar://problem/5030628> - Crash opening a new window with the
4108         "New windows open to the same page" pref set
4109
4110         * history/HistoryItem.cpp:
4111         (WebCore::HistoryItem::HistoryItem): Set the m_subItems vector capacity correctly
4112
4113 2007-03-01  Alexey Proskuryakov  <ap@webkit.org>
4114
4115         Reviewed by Maciej.
4116
4117         http://bugs.webkit.org/show_bug.cgi?id=12801
4118         Assertion failure in createMarkup() (root) when doing Select All, Copy in an SVG document
4119
4120         Test: editing/pasteboard/createMarkup-assert.xml
4121
4122         * editing/markup.cpp:
4123         (WebCore::createMarkup): Removed the assertion, because it's wrong in a non-HTML world.
4124
4125 2007-02-28  Adam Roben  <aroben@apple.com>
4126
4127         Reviewed by Oliver and Hyatt.
4128
4129         Fix <rdar://problem/5024233> Crash while using Find on empty document
4130
4131         No layout test possible, as this involves a null document.
4132
4133         * dom/Range.cpp:
4134         (WebCore::rangeOfContents): Added an ASSERT.
4135         * page/Frame.cpp:
4136         (WebCore::Frame::findString): Added a null-check for document().
4137         (WebCore::Frame::markAllMatchesForText): Ditto.
4138
4139 2007-02-28  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
4140
4141         Not reviewed - simple gdk build fix.
4142
4143         Add stubs for newly introduced localized strings.
4144
4145         * platform/gdk/TemporaryLinkStubs.cpp:
4146         (WebCore::searchMenuNoRecentSearchesText):
4147         (WebCore::searchMenuRecentSearchesText):
4148         (WebCore::searchMenuClearRecentSearchesText):
4149
4150 2007-02-28  Brady Eidson  <beidson@apple.com>
4151
4152         Reviewed by Beth
4153
4154         Move the Thread Safety Check functions into their own header for export to WebKit
4155
4156         * WebCore.exp: Export the function
4157         * WebCore.xcodeproj/project.pbxproj: