WebCore/ChangeLog

   1 2007-03-15  Beth Dakin  <bdakin@apple.com>
   2
   3         Reviewed by Hyatt.
   4
   5         Fix for <rdar://problem/5065396> REGRESSION: leaks in
   6         RenderBlock::layoutInlineChildren seen on buildbot
   7
   8         This leak appeared after http://trac.webkit.org/projects/webkit/
   9         changeset/20188. This change shifted line boxes around in
  10         removeChild(). But since removeChild() calls
  11         setNeedsLayoutAndMinMaxRecalc(), all of the line boxes will be
  12         removed once we actually lay out anyway. So this patch fixes the
  13         leak by deleting the line boxes instead of shifting them around.
  14
  15         * editing/IndentOutdentCommand.cpp:
  16         (WebCore::IndentOutdentCommand::outdentParagraph): Call into
  17         updateLayout(). This fixes an assertion I got in editing/
  18         execCommand/4976800.html This is very similar to the line box fix I
  19         made recently (http://trac.webkit.org/projects/webkit/changeset/
  20         20177). We need to update layout before relying on VisiblePositions
  21         after removing a node.
  22         * rendering/RenderBlock.cpp:
  23         (WebCore::RenderBlock::deleteLinesForBlock): New helper function
  24         since this functionality is needed in three places now.
  25         (WebCore::RenderBlock::makeChildrenNonInline): Call into new
  26         deleteLinesForBlock().
  27         (WebCore::RenderBlock::removeChild): Same.
  28         * rendering/RenderBlock.h:
  29
  30 2007-03-15  Timothy Hatcher  <timothy@apple.com>
  31
  32         Reviewed by John.
  33
  34         * Factored out most of our common build settings into .xcconfig files. Anything that was common in
  35           each build configuration was factored out into the shared .xcconfig file.
  36         * Adds a Version.xcconfig file to define the current framework version, to be used in other places.
  37         * Use the new $(BUNDLE_VERSION) (defined in Version.xcconfig) in the preprocessed Info.plist.
  38         * Use the versions defined in Version.xcconfig to set $(DYLIB_CURRENT_VERSION).
  39
  40         * Configurations/Base.xcconfig: Added.
  41         * Configurations/DebugRelease.xcconfig: Added.
  42         * Configurations/Version.xcconfig: Added.
  43         * Configurations/WebCore.xcconfig: Added.
  44         * Info.plist:
  45         * WebCore.xcodeproj/project.pbxproj:
  46
  47 2007-03-15  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
  48
  49         Not reviewed - gdk build fixes.
  50
  51         * platform/gdk/EditorClientGdk.cpp:
  52         (WebCore::EditorClientGdk::handleKeypress):
  53         * platform/graphics/cairo/ImageSourceCairo.cpp:
  54         (WebCore::ImageSource::setData):
  55         * platform/graphics/gdk/ImageGdk.cpp:
  56         (WebCore::Image::loadPlatformResource):
  57
  58 2007-03-14  Justin Garcia  <justin.garcia@apple.com>
  59
  60         Reviewed by john
  61
  62         ~2x speed up of 5k rich text paste:
  63         http://shakespeare.mit.edu/hamlet/full.html
  64
  65         * editing/ReplaceSelectionCommand.cpp:
  66         (WebCore::ReplaceSelectionCommand::doApply): Remove the top
  67         level style span if its unnecessary before inserting
  68         into the document, its faster than doing it after.
  69         * editing/ReplaceSelectionCommand.h: Exposed
  70         ReplacementFragment::removeNodePreservingChildren so that
  71         the style span can be removed using non-undoable removes,
  72         like the rest of the removes done on the ReplacementFragment.
  73         * editing/markup.cpp:
  74         (WebCore::createMarkup): Make the style span the top level
  75         element, otherwise it's useless.  This also facilitates the
  76         optimization mentioned above.
  77         When including markup for a fully selected root, include markup
  78         for all the nodes beneath that fully selected root, to preserve
  79         the structure and appearance of the copied markup.  Did this
  80         by merging with the code for adding markup for descendants of
  81         special commonAncestorBlocks.
  82
  83 2007-03-15  Beth Dakin  <bdakin@apple.com>
  84
  85         Rubber-stamped by Adele.
  86
  87         Rolling out http://trac.webkit.org/projects/webkit/changeset/20148
  88         (which is a fix for http://bugs.webkit.org/show_bug.cgi?id=12595
  89         and rdar://4722863) because it causes a horrible memory-trasher.
  90
  91         * bindings/js/JSHTMLFormElementCustom.cpp:
  92         (WebCore::JSHTMLFormElement::canGetItemsForName):
  93         (WebCore::JSHTMLFormElement::nameGetter):
  94         * bindings/js/kjs_dom.cpp:
  95         (KJS::):
  96         (KJS::DOMNamedNodesCollection::DOMNamedNodesCollection):
  97         (KJS::DOMNamedNodesCollection::lengthGetter):
  98         (KJS::DOMNamedNodesCollection::indexGetter):
  99         (KJS::DOMNamedNodesCollection::getOwnPropertySlot):
 100         * bindings/js/kjs_dom.h:
 101         (KJS::DOMNamedNodesCollection::classInfo):
 102         * bindings/js/kjs_html.cpp:
 103         (KJS::JSHTMLCollection::getNamedItems):
 104         * dom/ChildNodeList.cpp:
 105         (WebCore::ChildNodeList::ChildNodeList):
 106         (WebCore::ChildNodeList::length):
 107         (WebCore::ChildNodeList::item):
 108         (WebCore::ChildNodeList::nodeMatches):
 109         * dom/ChildNodeList.h:
 110         * dom/NameNodeList.cpp:
 111         (WebCore::NameNodeList::NameNodeList):
 112         (WebCore::NameNodeList::item):
 113         (WebCore::NameNodeList::nodeMatches):
 114         * dom/NameNodeList.h:
 115         (WebCore::NameNodeList::rootNodeAttributeChanged):
 116         * dom/Node.cpp:
 117         (WebCore::TagNodeList::TagNodeList):
 118         (WebCore::TagNodeList::nodeMatches):
 119         (WebCore::Node::registerNodeList):
 120         (WebCore::Node::unregisterNodeList):
 121         * dom/Node.h:
 122         * dom/NodeList.cpp:
 123         (WebCore::NodeList::NodeList):
 124         (WebCore::NodeList::~NodeList):
 125         (WebCore::NodeList::recursiveLength):
 126         (WebCore::NodeList::itemForwardsFromCurrent):
 127         (WebCore::NodeList::itemBackwardsFromCurrent):
 128         (WebCore::NodeList::recursiveItem):
 129         (WebCore::NodeList::itemWithName):
 130         (WebCore::NodeList::rootNodeChildrenChanged):
 131         * dom/NodeList.h:
 132         (WebCore::NodeList::rootNodeAttributeChanged):
 133         * html/HTMLFormElement.cpp:
 134         (WebCore::HTMLFormElement::HTMLFormElement):
 135         (WebCore::HTMLFormElement::~HTMLFormElement):
 136         (WebCore::HTMLFormElement::formData):
 137         (WebCore::HTMLFormElement::parseMappedAttribute):
 138         (WebCore::HTMLFormElement::removeFormElement):
 139         * html/HTMLFormElement.h:
 140         * html/HTMLGenericFormElement.cpp:
 141         (WebCore::HTMLGenericFormElement::parseMappedAttribute):
 142         (WebCore::HTMLGenericFormElement::insertedIntoTree):
 143         * html/HTMLGenericFormElement.h:
 144         * html/HTMLInputElement.cpp:
 145         (WebCore::HTMLInputElement::parseMappedAttribute):
 146
 147 2007-03-15  Geoffrey Garen  <ggaren@apple.com>
 148
 149         Added an assert to help catch a bug. Hopefully someone will hit it!
 150
 151         * bindings/js/kjs_proxy.cpp:
 152         (WebCore::KJSProxy::~KJSProxy):
 153
 154 2007-03-14  Oliver Hunt  <oliver@apple.com>
 155
 156         Reviewed by Adele.
 157
 158         Fix for rdar://problem/5061737.
 159
 160         This was a regression from the original Objective-C -> C++ conversion
 161         for the drag logic.  We don't need to call Range::startNode as we just
 162         need the document that contains the range.
 163
 164         * page/DragController.cpp:
 165         (WebCore::documentFragmentFromDragData):
 166
 167 2007-03-14  Mitz Pettel  <mitz@webkit.org>
 168
 169         Reviewed by Hyatt.
 170
 171         - http://bugs.webkit.org/show_bug.cgi?id=13071
 172           REGRESSION: Plain text files no longer wrap lines longer than the width of the browser window
 173
 174         Test: fast/loader/text-document-wrapping.html
 175
 176         * loader/TextDocument.cpp:
 177         (WebCore::TextTokenizer::write): Specified word-wrap:break-word for the <pre> element.
 178
 179 2007-03-14  Mitz Pettel  <mitz@webkit.org>
 180
 181         Reviewed by Hyatt.
 182
 183         - fix http://bugs.webkit.org/show_bug.cgi?id=13072
 184           REGRESSION (r15617): white-space: pre-wrap breaks off the last character of a wide word
 185
 186         Test: fast/text/whitespace/pre-wrap-last-char.html
 187
 188         * rendering/bidi.cpp:
 189         (WebCore::RenderBlock::findNextLineBreak): Undid the change from r15617.
 190
 191 2007-03-14  David Hyatt  <hyatt@apple.com>
 192
 193         Add asserts to help catch double refs and double derefs of CachedResources.
 194
 195         Reviewed by mjs
 196
 197         * loader/CachedResource.cpp:
 198         (WebCore::CachedResource::ref):
 199         (WebCore::CachedResource::deref):
 200
 201 2007-03-14  Alice Liu  <alice.liu@apple.com>
 202
 203         Rubber-stamped by Hyatt.
 204
 205         Adding null check to prevent the crash that happens on 2nd run of iBench HTML load test
 206
 207         * loader/icon/IconDataCache.cpp:
 208         (WebCore::IconDataCache::writeToDatabase):
 209
 210 2007-03-14  Adele Peterson  <adele@apple.com>
 211
 212         Reviewed by Adam.
 213
 214         Fix for <rdar://problem/5062898> REGRESSION: autocomplete window in text fields doesn't come up
 215
 216         When we moved the initialization of a bunch of variables in the HTMLFormElement constructor, m_autocomplete accidently got initialized to false.
 217
 218         * html/HTMLFormElement.cpp: (WebCore::HTMLFormElement::HTMLFormElement):
 219
 220 2007-03-14  Steve Falkenburg  <sfalken@apple.com>
 221
 222         Reviewed by Adam.
 223
 224         Added missing null check in case this is called on a subframe
 225         that hasn't started loading.
 226
 227         * page/Frame.cpp:
 228         (WebCore::Frame::setMarkedTextMatchesAreHighlighted): Check for a null document.
 229
 230 2007-03-14  Antti Koivisto  <antti@apple.com>
 231
 232         Reviewed by Geoff.
 233
 234         Fix http://bugs.webkit.org/show_bug.cgi?id=13060
 235         REGRESSION: Repro ASSERT failure in Cache::adjustSize running layout tests
 236         <rdar://5060208>
 237
 238         Script evaluation may have dereffed the CachedScript object already, causing double deref and
 239         eventually m_liveResourcesSize underflow.
 240
 241         * html/HTMLScriptElement.cpp:
 242         (WebCore::HTMLScriptElement::notifyFinished):
 243
 244 2007-03-14  Adele Peterson  <adele@apple.com>
 245
 246         Reviewed by Darin.
 247
 248         Added InsertTab, InsertBacktab, InsertLineBreak, and InsertNewline to the editing command table.
 249         Added Event parameter to execCommand, and to all of the enabled and exec functions.  Right now,
 250         the event is only used by the newly added commands.  But in the future, many (and possibly all) of
 251         these editing commands will need to consider the event so they are applied to the correct selection.
 252
 253         * WebCore.exp:
 254         * editing/Editor.cpp:
 255         (WebCore::execCopy):
 256         (WebCore::execCut):
 257         (WebCore::execDelete):
 258         (WebCore::execBackwardDelete):
 259         (WebCore::execForwardDelete):
 260         (WebCore::execMoveBackward):
 261         (WebCore::execMoveBackwardAndModifySelection):
 262         (WebCore::execMoveUpByPageAndModifyCaret):
 263         (WebCore::execMoveDown):
 264         (WebCore::execMoveDownAndModifySelection):
 265         (WebCore::execMoveForward):
 266         (WebCore::execMoveForwardAndModifySelection):
 267         (WebCore::execMoveDownByPageAndModifyCaret):
 268         (WebCore::execMoveLeft):
 269         (WebCore::execMoveLeftAndModifySelection):
 270         (WebCore::execMoveRight):
 271         (WebCore::execMoveRightAndModifySelection):
 272         (WebCore::execMoveToBeginningOfDocument):
 273         (WebCore::execMoveToBeginningOfDocumentAndModifySelection):
 274         (WebCore::execMoveToBeginningOfSentence):
 275         (WebCore::execMoveToBeginningOfSentenceAndModifySelection):
 276         (WebCore::execMoveToBeginningOfLine):
 277         (WebCore::execMoveToBeginningOfLineAndModifySelection):
 278         (WebCore::execMoveToBeginningOfParagraph):
 279         (WebCore::execMoveToBeginningOfParagraphAndModifySelection):
 280         (WebCore::execMoveToEndOfDocument):
 281         (WebCore::execMoveToEndOfDocumentAndModifySelection):
 282         (WebCore::execMoveToEndOfSentence):
 283         (WebCore::execMoveToEndOfSentenceAndModifySelection):
 284         (WebCore::execMoveToEndOfLine):
 285         (WebCore::execMoveToEndOfLineAndModifySelection):
 286         (WebCore::execMoveToEndOfParagraph):
 287         (WebCore::execMoveToEndOfParagraphAndModifySelection):
 288         (WebCore::execMoveParagraphBackwardAndModifySelection):
 289         (WebCore::execMoveParagraphForwardAndModifySelection):
 290         (WebCore::execMoveUp):
 291         (WebCore::execMoveUpAndModifySelection):
 292         (WebCore::execMoveWordBackward):
 293         (WebCore::execMoveWordBackwardAndModifySelection):
 294         (WebCore::execMoveWordForward):
 295         (WebCore::execMoveWordForwardAndModifySelection):
 296         (WebCore::execMoveWordLeft):
 297         (WebCore::execMoveWordLeftAndModifySelection):
 298         (WebCore::execMoveWordRight):
 299         (WebCore::execMoveWordRightAndModifySelection):
 300         (WebCore::execPaste):
 301         (WebCore::execSelectAll):
 302         (WebCore::execToggleBold):
 303         (WebCore::execToggleItalic):
 304         (WebCore::execRedo):
 305         (WebCore::execUndo):
 306         (WebCore::execInsertTab):
 307         (WebCore::execInsertBacktab):
 308         (WebCore::execInsertNewline):
 309         (WebCore::execInsertLineBreak):
 310         (WebCore::enabled):
 311         (WebCore::canPaste):
 312         (WebCore::hasEditableSelection):
 313         (WebCore::hasEditableRangeSelection):
 314         (WebCore::hasRangeSelection):
 315         (WebCore::hasRichlyEditableSelection):
 316         (WebCore::canRedo):
 317         (WebCore::canUndo):
 318         (WebCore::CommandEntry::):
 319         (WebCore::Editor::toggleBold):
 320
 321         (WebCore::Editor::execCommand): Added optional event parameter.
 322         (WebCore::Editor::insertText): Added.  Calls handleTextInputEvent.
 323         (WebCore::Editor::insertTextWithoutSendingTextEvent): Renamed from insertText.
 324          Performs the actual insertion without dispatching any event.
 325         * editing/Editor.h:
 326         * page/EventHandler.cpp: (WebCore::EventHandler::defaultTextInputEventHandler):
 327           Call insertTextWithoutSendingTextEvent.
 328
 329 2007-03-14  Justin Garcia  <justin.garcia@apple.com>
 330
 331         Reviewed by darin
 332
 333         ~3x speedup pasting 5k lines of rich text:
 334         http://shakespeare.mit.edu/hamlet/full.html
 335         ~2x speedup pasting 10k lines of plain text
 336
 337         * css/CSSComputedStyleDeclaration.cpp:
 338         (WebCore::computedStyle): Added for convenience.
 339         * css/CSSComputedStyleDeclaration.h:
 340         * editing/ReplaceSelectionCommand.cpp:
 341         (WebCore::ReplaceSelectionCommand::removeRedundantStyles):
 342         The code that pushed down the top level style span had
 343         a bug in it that made it do unnecessary work.  Instead of
 344         fixing the bug I removed the code because it was used to
 345         help see more redundancies in second level style spans, but
 346         createMarkup now *only* creates a top level style span.
 347         Only remove redundant styles from style spans and only remove
 348         unstyled elements if they are style spans.  FF doesn't
 349         remove redundant styles from elements, or remove redundant
 350         font tags on copy/paste.  We could offer this functionality
 351         through a separate "cleanup" command.
 352         * editing/markup.cpp:
 353         (WebCore::createMarkup): Only add markup for ancestors of
 354         lastClosed if we're including markup for acommonAncestorBlock
 355         (we do this for commonAncestorBlocks like tables and lists),
 356         otherwise it's unnecessary/redundant.
 357
 358 2007-03-14  Anders Carlsson  <acarlsson@apple.com>
 359
 360         Reviewed by Ada.
 361
 362         Make sure to call ResourceLoader::didReceiveData to ensure that there's a shared buffer with the resource
 363         data, since SubresourceLoaderClients now make use of that buffer.
 364
 365         * loader/SubresourceLoader.cpp:
 366         (WebCore::SubresourceLoader::didReceiveData):
 367
 368 2007-03-14  Administrator  <acarlsson@apple.com>
 369
 370         Get the size from the shared buffer.
 371
 372         * platform/graphics/cg/ImageSourceCG.cpp:
 373         (WebCore::ImageSource::setData):
 374         * platform/graphics/cg/PDFDocumentImage.cpp:
 375         (WebCore::PDFDocumentImage::dataChanged):
 376
 377 2007-03-14  Mitz Pettel  <mitz@webkit.org>
 378
 379         Reviewed by Hyatt, thumbs up by Darin.
 380
 381         - fix http://bugs.webkit.org/show_bug.cgi?id=12782
 382           Reproducible crash in BidiContext::deref
 383
 384         Test: fast/dynamic/anonymous-block-orphaned-lines.html
 385
 386         * rendering/InlineBox.cpp:
 387         (WebCore::InlineBox::root): Added an assertion that we return a root inline box.
 388         * rendering/RenderBlock.cpp:
 389         (WebCore::RenderBlock::removeChild): Added code to adopt the line boxes of
 390         anonymous blocks being destroyed instead of leaving them orphaned, which
 391         is what caused this crash. The boxes will be deleted on the next layout, but
 392         this ensures consistency in the mean time.
 393
 394 2007-03-14  Lars Knoll <lars@trolltech.com>
 395
 396         Reviewed by Antti.
 397
 398         Fix the Qt build.
 399         Add a getter to SharedBuffer that returns a reference to the internal
 400         Vector to avoid an extra copy of the data.
 401
 402         * platform/SharedBuffer.h:
 403         (WebCore::SharedBuffer::buffer):
 404         * platform/graphics/qt/ImageQt.cpp:
 405         (WebCore::Image::loadPlatformResource):
 406         * platform/graphics/qt/ImageSourceQt.cpp:
 407         (WebCore::detectImageFormat):
 408         (WebCore::createDecoder):
 409         (WebCore::ImageSource::setData):
 410
 411 2007-03-14  Antti Koivisto  <antti@apple.com>
 412
 413         Reviewed by Mitz.
 414
 415         Fix <rdar://problem/5058774>
 416         REGRESSION: In Mail, caret appears oversized when typing in a To Do note
 417
 418         Horizontal and vertical were switched. Was regression from
 419         http://trac.webkit.org/projects/webkit/changeset/20103
 420
 421         * rendering/RootInlineBox.cpp:
 422         (WebCore::RootInlineBox::addHighlightOverflow):
 423
 424 2007-03-14  David Hyatt  <hyatt@apple.com>
 425
 426         Tweak the data() functions of stylesheets and scripts to be internally consistent.
 427
 428         * loader/CachedCSSStyleSheet.cpp:
 429         (WebCore::CachedCSSStyleSheet::data):
 430         * loader/CachedScript.cpp:
 431         (WebCore::CachedScript::data):
 432         * loader/CachedXSLStyleSheet.cpp:
 433         (WebCore::CachedXSLStyleSheet::data):
 434
 435 2007-03-14  David Hyatt  <hyatt@apple.com>
 436
 437         Fix Radar 5050688.
 438
 439         For large animated GIFs, destroy and recreate the source for every animation frame.  This keeps
 440         the memory consumption down while giant images are animating.
 441
 442         Reviewed by andersca
 443
 444         * platform/graphics/BitmapImage.cpp:
 445         (WebCore::BitmapImage::BitmapImage):
 446         (WebCore::BitmapImage::destroyDecodedData):
 447         (WebCore::BitmapImage::dataChanged):
 448         (WebCore::BitmapImage::advanceAnimation):
 449         * platform/graphics/BitmapImage.h:
 450
 451 2007-03-14  David Hyatt  <hyatt@apple.com>
 452
 453         Make sure to use CFDataCreateWithBytesNoCopy where we can.
 454
 455         Reviewed by andersca
 456
 457         * platform/graphics/cg/ImageSourceCG.cpp:
 458         (WebCore::ImageSource::setData):
 459         * platform/graphics/cg/PDFDocumentImage.cpp:
 460         (WebCore::PDFDocumentImage::dataChanged):
 461
 462 2007-03-13  David Hyatt  <hyatt@apple.com>
 463
 464         Fix for bugzilla bug 13050 and also radar p1 5050645.
 465
 466         This patch reworks resource loading to avoid having redundant buffers in the icon database and in cached
 467         images in the WebCore cache.  It also avoids overcopying in top-level image documents and in the icon
 468         database.
 469
 470         There is now only one SharedBuffer for a resource and everybody observes that buffer now instead of ever
 471         making their own.  Even ImageIO uses the SharedBuffer while decoding.
 472
 473         The page in 13050 dropped from 145mb down to 45mb of memory use with this change for a stunning savings
 474         of 100mb.
 475
 476         Reviewed by olliej, mjs
 477
 478         * WebCore.exp:
 479         * loader/CachedCSSStyleSheet.cpp:
 480         (WebCore::CachedCSSStyleSheet::data):
 481         * loader/CachedCSSStyleSheet.h:
 482         * loader/CachedImage.cpp:
 483         (WebCore::CachedImage::data):
 484         * loader/CachedImage.h:
 485         * loader/CachedResource.cpp:
 486         (WebCore::CachedResource::CachedResource):
 487         (WebCore::CachedResource::~CachedResource):
 488         * loader/CachedResource.h:
 489         (WebCore::CachedResource::data):
 490         * loader/CachedScript.cpp:
 491         (WebCore::CachedScript::data):
 492         * loader/CachedScript.h:
 493         * loader/CachedXSLStyleSheet.cpp:
 494         (WebCore::CachedXSLStyleSheet::data):
 495         * loader/CachedXSLStyleSheet.h:
 496         * loader/DocLoader.cpp:
 497         (WebCore::DocLoader::checkCacheObjectStatus):
 498         * loader/ImageDocument.cpp:
 499         (WebCore::ImageTokenizer::writeRawData):
 500         (WebCore::ImageTokenizer::finish):
 501         * loader/icon/IconDataCache.cpp:
 502         (WebCore::IconDataCache::setImageData):
 503         (WebCore::IconDataCache::writeToDatabase):
 504         * loader/icon/IconDataCache.h:
 505         * loader/icon/IconDatabase.cpp:
 506         (WebCore::IconDatabase::imageDataForIconURL):
 507         (WebCore::IconDatabase::iconForPageURL):
 508         (WebCore::IconDatabase::setIconDataForIconURL):
 509         (WebCore::IconDatabase::setHaveNoIconForIconURL):
 510         (WebCore::IconDatabase::imageDataForIconURLQuery):
 511         * loader/icon/IconDatabase.h:
 512         * loader/icon/IconLoader.cpp:
 513         (WebCore::IconLoader::startLoading):
 514         (WebCore::IconLoader::didReceiveResponse):
 515         (WebCore::IconLoader::didReceiveData):
 516         (WebCore::IconLoader::didFail):
 517         (WebCore::IconLoader::finishLoading):
 518         (WebCore::IconLoader::clearLoadingState):
 519         * loader/icon/IconLoader.h:
 520         * loader/icon/SQLStatement.cpp:
 521         (WebCore::SQLStatement::getColumnBlobAsVector):
 522         (WebCore::SQLStatement::isExpired):
 523         * loader/icon/SQLStatement.h:
 524         * loader/loader.cpp:
 525         (WebCore::Loader::didFinishLoading):
 526         (WebCore::Loader::didReceiveData):
 527         * page/mac/WebCoreFrameBridge.mm:
 528         (-[WebCoreFrameBridge getData:andResponse:forURL:]):
 529         (-[WebCoreFrameBridge getAllResourceDatas:andResponses:]):
 530         * platform/SharedBuffer.h:
 531         (WebCore::SharedBuffer::isEmpty):
 532         * platform/graphics/BitmapImage.cpp:
 533         (WebCore::BitmapImage::destroyDecodedData):
 534         (WebCore::BitmapImage::dataChanged):
 535         * platform/graphics/BitmapImage.h:
 536         * platform/graphics/Image.cpp:
 537         (WebCore::Image::setData):
 538         * platform/graphics/Image.h:
 539         (WebCore::Image::dataChanged):
 540         (WebCore::Image::data):
 541         * platform/graphics/ImageSource.h:
 542         * platform/graphics/cg/ImageSourceCG.cpp:
 543         (WebCore::ImageSource::setData):
 544         * platform/graphics/cg/PDFDocumentImage.cpp:
 545         (WebCore::PDFDocumentImage::dataChanged):
 546         * platform/graphics/cg/PDFDocumentImage.h:
 547         * platform/graphics/mac/ImageMac.mm:
 548         (WebCore::Image::loadPlatformResource):
 549         * platform/graphics/svg/SVGImage.cpp:
 550         (WebCore::SVGImage::setData):
 551         * platform/mac/PasteboardMac.mm:
 552         (WebCore::fileWrapperForImage):
 553
 554 2007-03-13  Justin Garcia  <justin.garcia@apple.com>
 555
 556         Reviewed by darin
 557
 558         <rdar://problem/5046875>
 559         Gmail Editor: Applying alignment to selected text in message also applies alignment to signature
 560
 561         * editing/ApplyStyleCommand.cpp:
 562         (WebCore::ApplyStyleCommand::doApply): Don't call applyBlockStyle unless
 563         there is a block style to apply.
 564         (WebCore::ApplyStyleCommand::applyBlockStyle): Don't do the remove step.
 565         It was unnecessary and removed properties from blocks that could contain
 566         content outside the range being operated on (added a testcase).
 567         (WebCore::ApplyStyleCommand::addBlockStyleIfNeeded): Used an early return
 568         instead of if-nesting.
 569         * editing/ApplyStyleCommand.h:
 570         * editing/CompositeEditCommand.cpp:
 571         (WebCore::CompositeEditCommand::moveParagraphContentsToNewBlockIfNecessary):
 572         Return the new block, if one was created.  Use moveParagraphs to move
 573         paragraphs into the new block, instead of moving nodes.  The old code moved
 574         too much (added a testcase).
 575         * editing/CompositeEditCommand.h:
 576
 577 2007-03-13  Oliver Hunt  <oliver@apple.com>
 578
 579         Reviewed by Brady.
 580
 581         To fix <rdar://problem/5044366> we now pass a NSString
 582         representation of the URL extracted with _web_originalDataAsString
 583         instead of relying on [NSURL absoluteString] in the bridge
 584
 585         * page/mac/WebCoreFrameBridge.h:
 586         * page/mac/WebCoreFrameBridge.mm:
 587         (-[WebCoreFrameBridge getData:andResponse:forURL:]):
 588
 589 2007-03-13  Brady Eidson  <beidson@apple.com>
 590
 591         Reviewed by Anders
 592
 593         <rdar://problem/5048818> - REGRESSION: Incompletely loaded resources being saved to the object cache
 594
 595         Due to a subtle change in loader behavior back in 10904, we would stop all loaders before calling
 596         didFail() on them in the Cache loader.  As a result, we basically cleared all of the Subresource Loaders
 597         out of the Cache loader before more properly failing them as errored out.  The result?  Partially loaded
 598         resources being cached.
 599
 600         Since Loader::didFail() both calls error() on the object *and* removes the loader, the solution is to call
 601         didFail() for all cancelled loaders instead of *only* removing them from the set of active loaders.
 602
 603         In addition, pages that didn't completely load were being saved to the back/forward cache.  To fix that,
 604         I added a null check on the DocumentLoader's error to see if the page ended in an error, or did indeed
 605         completely load.
 606
 607         Note that the layout test for this - if possible - will require other enhancements including possibly adding
 608         support for window.stop().  That task is documented in <rdar://problem/5061826>
 609
 610         * loader/FrameLoader.cpp:
 611         (WebCore::FrameLoader::provisionalLoadStarted): Fixed a few bugs relating to my original BFCache rewrite to
 612           more perfectly restore the original behavior - including only caching HTML documents via the
 613           m_client->canCachePage() call
 614         (WebCore::FrameLoader::canCachePage): Don't make the call to m_client->canCachePage() as that serves a different
 615           purpose
 616           - Check the mainDocumentError to see if the load ended in error as a further criteria in determining the
 617           cachability of a page
 618
 619         * loader/loader.cpp:
 620         (WebCore::Loader::cancelRequests): Call didFail(cancelledError()) instead of just removing the loaders from the
 621           loaders-in-progress set.  This adds the effect of properly cleaning up the cached object.
 622
 623 2007-03-13  Beth Dakin  <bdakin@apple.com>
 624
 625         Rendering part reviewed by Hyatt. Editing part consulted with and
 626         rubber stamped by Justin and Harrison.
 627
 628         Fix for <rdar://problem/5025925> A hang occurs in Safari when
 629         attempting to print page at http://www.pcadvisor.co.uk
 630
 631         * rendering/RenderBlock.cpp:
 632         (WebCore::RenderBlock::makeChildrenNonInline):
 633         RenderBlock::makeChildrenNonInline() takes a block's inline
 634         children and turns them into block children. If the children had
 635         line boxes, those boxes were being leaked. In the layout test I
 636         added with the change (and at pcadvisor.co.uk during printing)
 637         children were being made non-inline, and then they were being made
 638         inline again. This meant that some of the children ended up
 639         pointing to totally stale line boxes that are normally just leaked.
 640         This caused an infinite loop in RenderFlow::destroy(). This patch
 641         simply deletes everyone's line boxes in
 642         RenderBlock::makeChildrenNonInline()
 643
 644         * editing/InsertParagraphSeparatorCommand.cpp:
 645         (WebCore::InsertParagraphSeparatorCommand::doApply): The other part
 646         of this fix is that I added a call to updateLayout in
 647         InsertParagraphSeparatorCommand::doApply(). One layout test
 648         (editing/spelling/spelling.html) was changed by my patch to
 649         RenderBlock. doApply() inserts a node into the render tree. In at
 650         least one case in spelling.html, that caused some line boxes to be
 651         deleted. Back in doApply() this meant that the RenderTree was out-
 652         of-date, and we mistakenly thought we were at the end of the
 653         paragraph. This caused us to insert a RenderBR() at the end of the
 654         tree instead of an empty RenderText(). No one seems to know exactly
 655         why we insert either, or if the change is necessarily a problem. It
 656         is clear, though, that the RenderTree in doApply() is out-of-date
 657         after inserting the node and deleting some line boxes, so it seems
 658         prudent to call into updateLayout().
 659
 660 2007-03-13  Adam Roben  <aroben@apple.com>
 661
 662         Reviewed by Anders.
 663
 664         * platform/FontData.h: Added m_isSystemFont parameter to match NSFont.
 665
 666 2007-03-13  Beth Dakin  <bdakin@apple.com>
 667
 668         Reviewed by Maciej.
 669
 670         Export DocumentLoader::setFrame(). Part of fix for <rdar://
 671         problem/4277074> 8F32: Help Viewer crashed on clicking link -
 672         KHTMLView::viewportMouseReleaseEvent (12647)
 673
 674         * WebCore.exp:
 675
 676 2007-03-13  Darin Adler  <darin@apple.com>
 677
 678         Reviewed by Maciej.
 679
 680         - fix http://bugs.webkit.org/show_bug.cgi?id=12794
 681           <rdar://problem/5028154> REGRESSION: TripTik planner at aaa.com never
 682           finishes loading due to unclosed canvas tag (12794)
 683
 684         Change <canvas> elements so that their contents are parsed normally,
 685         but not rendered. This change fixes the bug, because normal parsing
 686         rules close the <canvas> element in that case. The special parser
 687         stuff was just getting in the way.
 688
 689         Also do some basic cleanup to the HTML parser. This was motivated by
 690         an earlier version of this patch that made even more changes to the
 691         parser, but the cleanup is still worth landing.
 692
 693         Test: fast/canvas/canvas-hides-fallback.html
 694         Test: fast/canvas/script-inside-canvas-fallback.html
 695         Test: fast/canvas/unclosed-canvas-1.html
 696         Test: fast/canvas/unclosed-canvas-2.html
 697         Test: fast/canvas/unclosed-canvas-3.html
 698         Test: fast/canvas/unclosed-canvas-4.html
 699
 700         * html/HTMLCanvasElement.h: Added a data member to keep track of whether the
 701         renderer is a RenderHTMLCanvas or not.
 702         * html/HTMLCanvasElement.cpp:
 703         (WebCore::HTMLCanvasElement::createRenderer): If JavaScript is enabled, create
 704         a RenderHTMLCanvas. If it's not, let the default code create the default type
 705         of renderer, which will result in fallback content being visible. The
 706         RenderHTMLCanvas class already hides all of its children. Set the m_rendererIsCanvas
 707         boolean accordingly. Since the actual storage for the canvas is allocated lazily
 708         when you actually get a drawing context, we don't need to do anything special
 709         to prevent it when JavaScript is disabled; the relevant functions won't be called.
 710         (WebCore::HTMLCanvasElement::reset): Protect the code that manipulates the
 711         RenderHTMLCanvas with a check of m_rendererIsCanvas. This is the only code inside
 712         the DOM element that relies on the renderer type.
 713
 714         * html/HTMLParser.h: Removed unneeded includes. Marked HTMLParser as
 715         Noncopyable. Changed the Document parameter to the constructor to instead
 716         be HTMLDocument. Renamed discard_until to m_skipModeTag for clarity.
 717         Removed unused noSpaces function and unneeded public doc() function.
 718         Moved data members all down to the end so you can see them together in order.
 719         Renamed map to m_currentMapElement and isindex to m_isindexElement.
 720         Removed unused end and headLoaded data members.  Renamed m_fragment to
 721         m_isParsingFragment to make it clearer that it's a boolean, not a fragment.
 722
 723         * html/HTMLParser.cpp:
 724         (WebCore::HTMLParser::HTMLParser): Changed to use member construction
 725         syntax instead of calling reset(). This is especially helpful in the
 726         fragment case, where calling reset() later on is illegal, so not using
 727         it in the constructor lets us assert.
 728         (WebCore::HTMLParser::~HTMLParser): Did an explicit deref instead of
 729         calling setCurrent for its side effect.
 730         (WebCore::HTMLParser::reset): Updated for member name changes and removal
 731         and to use document instead of doc().
 732         (WebCore::HTMLParser::setCurrent): Use document instead of doc().
 733         (WebCore::HTMLParser::setSkipMode): Added. No longer inline. Now sets the
 734         m_inCanvasBeforeFirstOpenTag data member to false.
 735         (WebCore::HTMLParser::parseToken): Tightened up the skip mode logic at the
 736         top of the function, and added a FIXME about the strange case there where
 737         we don't skip yet stay in skip mode. Updated for renaming and doc().
 738         (WebCore::HTMLParser::insertNode): Updated for renaming and doc().
 739         (WebCore::HTMLParser::handleError): Ditto.
 740         (WebCore::HTMLParser::framesetCreateErrorCheck): Ditto.
 741         (WebCore::HTMLParser::isindexCreateErrorCheck): Changed to use RefPtr.
 742         (WebCore::HTMLParser::noscriptCreateErrorCheck): Updated for renaming and doc().
 743         (WebCore::HTMLParser::mapCreateErrorCheck): Ditto.
 744         (WebCore::HTMLParser::getNode): Removed the special case for canvas here.
 745         Canvas fallback is now handled in the DOM, not the parser. Updated for
 746         renaming and doc().
 747         (WebCore::HTMLParser::allowNestedRedundantTag): Changed a #define into a C++
 748         constant.
 749         (WebCore::HTMLParser::processCloseTag): Updated for renaming and doc().
 750         (WebCore::HTMLParser::isInline): Ditto.
 751         (WebCore::HTMLParser::tagIsOnStack): Added. Used by new canvas logic.
 752         (WebCore::HTMLParser::popBlock): Updated for renaming and doc(). Also renamed
 753         the local variable Elem to elem.
 754         (WebCore::HTMLParser::createHead): Ditto.
 755         (WebCore::HTMLParser::handleIsindex): Changed to use RefPtr.
 756         (WebCore::HTMLParser::startBody): Updated for renaming and doc().
 757         (WebCore::HTMLParser::finished): Ditto.
 758
 759 2007-03-13  David Hyatt  <hyatt@apple.com>
 760
 761         Two more cleanup fixes to the cache.  Don't call destroyDecodedData in the BitmapImage destructor, since
 762         clearing the image source and calling setData on it again causes it to do an extra copy of the encoded
 763         data.  Since we're about to be destroyed this is just wasteful.
 764
 765         When the cache prunes, don't allow it to destroy the decoded data of an image that is still actively loading,
 766         since we've established that ImageIO can actually crash if you yank the rug out from under it like that.
 767
 768         Reviewed by ggaren
 769
 770         * loader/Cache.cpp:
 771         (WebCore::Cache::prune):
 772         * platform/graphics/BitmapImage.cpp:
 773         (WebCore::BitmapImage::~BitmapImage):
 774
 775 2007-03-13  Anders Carlsson  <acarlsson@apple.com>
 776
 777         Try fixing the Qt build.
 778
 779         * editing/Editor.h:
 780         (WebCore::Editor::setStartNewKillRingSequence):
 781
 782 2007-03-13  David Harrison  <harrison@apple.com>
 783
 784         Reviewed by Justin.
 785
 786         <rdar://problem/5031181> cntl-k at end of paragraph adds nothing to the kill ring
 787         <rdar://problem/5031189> REGRESSION: cntl-y yanks only the most recently killed content
 788
 789         For rdar://5031181, properly extend the selection before the killring handling, and
 790         make sure plainText of that selection returns a linefeed.
 791
 792         For rdar://5031189, restore Editor::deleteRange() code that continued current killring,
 793         even though the range deletion implicitly stopped it via changing the selection.
 794
 795         A byproduct of this change is the elimination of RUNDFINDER vs CONTENT TextIterator. The
 796         only difference between the two was whether to emit a newline when the range started
 797         with a blockflow element. No callers actually need that any more.
 798
 799         Tests added:
 800         * editing/pasteboard/emacs-ctrl-k-y-001-expected.checksum: Added.
 801         * editing/pasteboard/emacs-ctrl-k-y-001-expected.png: Added.
 802         * editing/pasteboard/emacs-ctrl-k-y-001-expected.txt: Added.
 803         * editing/pasteboard/emacs-ctrl-k-y-001.html: Added.
 804
 805         * editing/Editor.cpp:
 806         (WebCore::Editor::deleteRange):
 807         Clear the "start new kill ring sequence" setting, because it was set to true
 808         when the selection was updated by deleting the range.
 809
 810         (WebCore::Editor::deleteWithDirection):
 811         If extending the selection to the end of paragraph resulted in a caret selection,
 812         extend by character, to handle the case when the selection started as a caret at
 813         the end of paragraph.
 814
 815         * editing/TextIterator.cpp:
 816         (WebCore::TextIterator::TextIterator):
 817         Initialize new member variables for tracking handling of the beginning of the range.
 818
 819         (WebCore::TextIterator::advance):
 820         Call representNodeOffsetZero on the m_endContainer.
 821         Move visibility checks into handleTextNode and handleReplacedElement.
 822
 823         (WebCore::TextIterator::handleTextNode):
 824         (WebCore::TextIterator::handleTextBox):
 825         Call emitText.
 826
 827         (WebCore::TextIterator::handleReplacedElement):
 828         Moved visibility check into here.
 829
 830         (WebCore::shouldEmitNewlinesBeforeAndAfterNode):
 831
 832         (WebCore::TextIterator::shouldRepresentNodeOffsetZero):
 833         (WebCore::TextIterator::representNodeOffsetZero):
 834         New. Emits proper sequence when encountering offset 0 of a node, including the
 835         m_endContainer. Started with code from handleNonTextNode.
 836
 837         (WebCore::TextIterator::handleNonTextNode):
 838         Call representNodeOffsetZero.
 839
 840         (WebCore::TextIterator::exitNode):
 841         Similar to shouldRepresentNodeOffsetZero, do not emit the newline if the node
 842         was collapsed, and before any other emitted content.
 843
 844         (WebCore::TextIterator::emitCharacter):
 845
 846         (WebCore::TextIterator::emitText):
 847         New. Consolidates code used by handleText and handleTextBox.
 848
 849         (WebCore::CharacterIterator::CharacterIterator):
 850         Removed RUNFINDER.
 851
 852         (WebCore::WordAwareIterator::WordAwareIterator):
 853         Removed RUNFINDER.
 854
 855         (WebCore::WordAwareIterator::advance):
 856         Formatting.
 857
 858         (WebCore::TextIterator::rangeLength):
 859         Formatting.
 860
 861         * editing/TextIterator.h:
 862         Added member variables for tracking handling of the beginning of the range.
 863         Eliminated concept of RUNDFINDER vs CONTENT TextIterator.
 864
 865         * editing/visible_units.cpp:
 866         (WebCore::nextBoundary):
 867         Eliminated concept of RUNDFINDER vs CONTENT TextIterator.
 868
 869 2007-03-13  David Hyatt  <hyatt@apple.com>
 870
 871         Clean up the null image case in CachedImage::data to make sure the size totals will stay accurate.
 872         I'm not convinced this case can even be hit, but I'm cleaning it up just in case.
 873
 874         * loader/CachedImage.cpp:
 875         (WebCore::CachedImage::data):
 876
 877 2007-03-13  David Hyatt  <hyatt@apple.com>
 878
 879         Reorder the call to allReferencesRemoved, since otherwise the live object size will become
 880         incorrect.
 881
 882         * loader/CachedResource.cpp:
 883         (WebCore::CachedResource::deref):
 884
 885 2007-03-13  Antti Koivisto  <antti@apple.com>
 886
 887         Reviewed by Darin.
 888
 889         Fix http://bugs.webkit.org/show_bug.cgi?id=11083
 890         REGRESSION: Typing tab key fails to insert a tab character in Google Docs editable area
 891         <rdar://problem/4757650>
 892
 893         Allow inserting tabs in designMode. Backtab behaves like before (matches FF).
 894
 895         * page/EventHandler.cpp:
 896         (WebCore::EventHandler::defaultTabEventHandler):
 897
 898 2007-03-13  David Hyatt  <hyatt@apple.com>
 899
 900         Fix two counting errors in the decoded size of objects in the WebCore cache.  We need to explicitly
 901         destroy decoded data when clearing out a partially loaded image that had an error, since the destructor
 902         (although it does destroy the data) disconnects the observer so that the notification doesn't happen.
 903
 904         Some CachedImages aren't in the cache (like image documents).  When the decoded size of such an image
 905         changes we should not notify the cache, since - newsflash - we aren't actually in it.
 906
 907         Reviewed by ggaren
 908
 909         * loader/CachedImage.cpp:
 910         (WebCore::CachedImage::clear):
 911         (WebCore::CachedImage::decodedSizeChanged):
 912
 913 2007-03-13  Darin Adler  <darin@apple.com>
 914
 915         Reviewed by Tim Hatcher and John Sullivan.
 916
 917         - fix <rdar://problem/4915303> CrashTracer: 36 crashes in Safari at
 918           com.apple.AppKit: -[NSView getRectsBeingDrawn:count:] + 502
 919
 920         * page/mac/FrameMac.mm: (WebCore::Frame::imageFromRect):
 921         This was calling drawRect: directly, but NSView's getRectsBeingDrawn:count:
 922         method was never really safe to call unless it was AppKit that called your
 923         drawRect: method. Changed it to call drawSingleRect: instead. A little ugly,
 924         but seems to work and will almost certainly fix the bug.
 925
 926 2007-03-13  Nikolas Zimmermann  <zimmermann@kde.org>
 927
 928         Reviewed by Maciej.
 929
 930         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12974
 931
 932         Call forgetGenericContext in JSSVGPathSeg destructor, otherwhise
 933         we'll hit an ASSERT in a debug build, when running svg/custom/js-update-path-changes.svg
 934         a few dozen times in a single WebKit instance. The ASSERT is good and just warns that
 935         there was already a generic context pointer registered, and the new "to be registered"
 936         object already exists, but pointing to a different object. That's because garbage collection
 937         calls the JSSVGPathSeg destructor, but that didn't cleanup the generic context map.
 938
 939         Only JSSVGPathSeg is hit by this problem, as it's the only non SVGAnimated* type
 940         using the generic context system while using a custom JSSVGPathSegList implementation.
 941
 942         Also cleanup JSSVGPathSegListCustom code to call the static forgetGenericContext
 943         method instead of doing the same using custom code.
 944
 945         * bindings/js/JSSVGPathSegListCustom.cpp:
 946         (WebCore::removeFromPathSegContextMap):
 947         (WebCore::JSSVGPathSegList::clear):
 948         (WebCore::JSSVGPathSegList::removeItem):
 949         * bindings/scripts/CodeGeneratorJS.pm:
 950
 951 2007-03-13  Darin Adler  <darin@apple.com>
 952
 953         Reviewed by Adele.
 954
 955         - fix http://bugs.webkit.org/show_bug.cgi?id=12595
 956           <rdar://problem/4722863> REGRESSION: Can't add item to cart at lnt.com
 957           (JS type error) (12595)
 958
 959         Test: fast/forms/old-names.html
 960
 961         * bindings/js/JSHTMLFormElementCustom.cpp:
 962         (WebCore::JSHTMLFormElement::canGetItemsForName): If the form collection has
 963         nothing for a given name, try the form's oldNamedElement function.
 964         (WebCore::JSHTMLFormElement::nameGetter): Ditto.
 965
 966         * bindings/js/kjs_dom.h: Removed the DOMNamedNodesCollection. Instead we will use
 967         a class derived from NodeList.
 968         * bindings/js/kjs_dom.cpp: Ditto.
 969
 970         * bindings/js/kjs_html.cpp:
 971         (KJS::VectorNodeList::VectorNodeList): Added. Constructor for a new class derived
 972         from NodeList to be used for the named items result from a collection -- uses a
 973         vector of node pointers.
 974         (KJS::VectorNodeList::length): Added.
 975         (KJS::VectorNodeList::item): Added.
 976         (KJS::JSHTMLCollection::getNamedItems): Use VectorNodeList and the existing wrapper
 977         for NodeList rather than a custom JavaScript class, DOMNamedNodesCollection.
 978
 979         * dom/ChildNodeList.h:
 980         * dom/ChildNodeList.cpp:
 981         (WebCore::ChildNodeList::ChildNodeList): Updated to derive from TreeNodeList,
 982         since NodeList is now a simpler class.
 983         (WebCore::ChildNodeList::elementMatches): Updated for name and parameter change.
 984
 985         * dom/NameNodeList.h:
 986         * dom/NameNodeList.cpp:
 987         (WebCore::NameNodeList::NameNodeList): Updated to derive from TreeNodeList,
 988         since NodeList is now a simpler class.
 989         (WebCore::NameNodeList::rootNodeAttributeChanged): Updated for name and
 990         parameter change.
 991
 992         * dom/Node.h: Change register/unregister functions to take TreeNodeList.
 993         * dom/Node.cpp:
 994         (WebCore::TagNodeList::TagNodeList): Updated to derive from TreeNodeList,
 995         since NodeList is now a simpler abstract class.
 996         (WebCore::TagNodeList::elementMatches): Updated for name and parameter change.
 997         (WebCore::Node::registerNodeList): Changed type from NodeList to TreeNodeList.
 998         (WebCore::Node::unregisterNodeList): Ditto.
 999
1000         * dom/NodeList.h: Broke NodeList into a simpler base class and a derived class
1001         with the machinery for iterating a tree, called TreeNodeList.
1002         * dom/NodeList.cpp:
1003         (WebCore::NodeList::~NodeList): Added.
1004         (WebCore::NodeList::itemWithName): Factored out of the old itemWithName.
1005         (WebCore::TreeNodeList::TreeNodeList): Renamed from NodeList.
1006         (WebCore::TreeNodeList::~TreeNodeList): Ditto.
1007         (WebCore::TreeNodeList::recursiveLength): Ditto.
1008         (WebCore::TreeNodeList::itemForwardsFromCurrent): Ditto.
1009         (WebCore::TreeNodeList::itemBackwardsFromCurrent): Ditto.
1010         (WebCore::TreeNodeList::recursiveItem): Ditto.
1011         (WebCore::TreeNodeList::itemWithName): Factored half of this into this function,
1012         the other half in NodeList::itemWithName.
1013         (WebCore::TreeNodeList::rootNodeAttributeChanged): Added. No longer inline.
1014         (WebCore::TreeNodeList::rootNodeChildrenChanged): Renamed from NodeList.
1015
1016         * html/HTMLFormElement.h: Added formElementNameChanged and oldNamedElement
1017         fucntions, and a map called m_oldNames. Also removed m_boundary, which I
1018         thought I had already done.
1019         * html/HTMLFormElement.cpp:
1020         (WebCore::HTMLFormElement::HTMLFormElement): Initialize m_oldNames to 0.
1021         Switched the rest of the members to initialization syntax.
1022         (WebCore::HTMLFormElement::~HTMLFormElement): Delete m_oldNames.
1023         (WebCore::HTMLFormElement::formElementNameChanged): Added. Stores a reference
1024         to one element under each of its old names.
1025         (WebCore::HTMLFormElement::oldNamedElement): Added. Returns the old element
1026         that once had a given name.
1027
1028         * html/HTMLGenericFormElement.h:
1029         * html/HTMLGenericFormElement.cpp:
1030         (WebCore::HTMLGenericFormElement::parseMappedAttribute): When the name
1031         attribute changes, tell the form about the old name.
1032         (WebCore::HTMLGenericFormElement::insertedIntoTree): When telling a form
1033         about an element, also store away the old name so that we can use it
1034         when the name changes later.
1035
1036         * html/HTMLInputElement.cpp:
1037         (WebCore::HTMLInputElement::parseMappedAttribute): Added a call to the
1038         base class in the nameAttr case, so the code in HTMLGenericFormElement
1039         above will get called in the input element case.
1040
1041 2007-03-13  Antti Koivisto  <antti@apple.com>
1042
1043         Reviewed by Alexey.
1044
1045         Alexey spotted a DOS by using string of 64k unbreakable character in
1046         fix for http://bugs.webkit.org/show_bug.cgi?id=12833
1047
1048         * dom/Text.cpp:
1049         (WebCore::Text::createWithLengthLimit):
1050
1051 2007-03-13  Lars Knoll <lars@trolltech.com>
1052
1053         Fix the Qt build
1054
1055         * WebCore.pro:
1056
1057 2007-03-13  Rob Buis  <buis@kde.org>
1058
1059         Reviewed by Darin.
1060
1061         http://bugs.webkit.org/show_bug.cgi?id=12576
1062         WebKit does not support xlink:show attributes
1063
1064         Make an exception for non-empty targets, these should be opened
1065         in a new window, unless the value is _self.
1066
1067         * ksvg2/svg/SVGAElement.cpp:
1068         (WebCore::SVGAElement::defaultEventHandler):
1069
1070 2007-03-13  David Hyatt  <hyatt@apple.com>
1071
1072         - fix cache issues seen in http://bugs.webkit.org/show_bug.cgi?id=13050
1073           6 objects and ~200MB leaked after opening then closing tab
1074           <rdar://problem/5058714>
1075
1076         Double the encoded size of images for now.  We do this to account for a bug in ImageIO where they hold
1077         a separate copy of image data.  See <rdar://problem/5050645>.
1078
1079         Reviewed by aroben.
1080
1081         * loader/CachedImage.cpp:
1082         (WebCore::CachedImage::data):
1083
1084 2007-03-12  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
1085
1086         Not reviewed - gdk build fix.
1087
1088         * platform/gdk/TemporaryLinkStubs.cpp:
1089         (SearchPopupMenu::enabled): add missing stub.
1090
1091 2007-03-12  David Kilzer  <ddkilzer@kilzer.net>
1092
1093         Reviewed by Darin.
1094
1095         - fix http://bugs.webkit.org/show_bug.cgi?id=13045
1096           REGRESSION: Blackboard CourseWare Error with Nightlies after Mar 8
1097
1098         * html/HTMLFormElement.cpp:
1099         (WebCore::getUniqueBoundaryString): Speculative fix: remove dashes from
1100         the boundary prefix that appear after non-dash characters.
1101
1102 2007-03-12  Mitz Pettel  <mitz@webkit.org>
1103
1104         Reviewed by Anders.
1105
1106         - fix http://bugs.webkit.org/show_bug.cgi?id=13015
1107           REGRESSION (r17233-r17241): Repro crash when leaving a page whose unload handler submits a form
1108
1109         Test: fast/loader/onunload-form-submit-crash-2.html
1110
1111         * loader/DocumentLoader.cpp:
1112         (WebCore::DocumentLoader::finishedLoading): Added null check.
1113
1114 2007-03-12  Anders Carlsson  <acarlsson@apple.com>
1115
1116         Reviewed by Darin.
1117
1118         <rdar://problem/4900071>
1119         http://bugs.webkit.org/show_bug.cgi?id=6454
1120         ASSERTION: Navigating 'back' in frameset: !_private->previousItem (6454)
1121
1122         Change back the behavior of checkLoadComplete to traverse the entire frame tree instead of
1123         just the parent frames of the current frame. This is needed in order to reset the previous history item
1124         for all frames when doing a frame navigation. (This was changed in revision 11819)
1125
1126         * loader/FrameLoader.cpp:
1127         (WebCore::FrameLoader::recursiveCheckLoadComplete):
1128         (WebCore::FrameLoader::checkLoadComplete):
1129         * loader/FrameLoader.h:
1130
1131 2007-03-12  Justin Garcia  <justin.garcia@apple.com>
1132
1133         Reviewed by darin
1134
1135         <rdar://problem/5056619>
1136         REGRESSION: Gmail Editor: Dragging text into Reply (textarea) field results in a crash at WebCore::InsertNodeBeforeCommand::doApply()
1137
1138         * editing/SelectionController.cpp:
1139         (WebCore::removingNodeRemovesPosition): Added.
1140         (WebCore::SelectionController::nodeWillBeRemoved): Clear the
1141         selection if it's inside a shadow tree.
1142         * page/DragController.cpp:
1143         (WebCore::setSelectionToDragCaret): Return false to signal to
1144         clients that a drop shouldn't be performed if the second attempt
1145         to set a selection ends up in non-editable content.
1146
1147 2007-03-12  Darin Adler  <darin@apple.com>
1148
1149         Reviewed by Tim Hatcher.
1150
1151         - fixed JavaScript wrapper classes to be correct for a variety of cases
1152           that a new test uncovered: was broken for at least 5 classes
1153
1154         - fixed Objective-C wrapper classes to be correct for a variety of cases
1155           that a test case uncovered: was broken for ast least 50 classes
1156
1157         - added missing DOM API for creating OverflowEvent and WheelEvent instances
1158
1159         Test: fast/dom/wrapper-classes.html
1160
1161         * DerivedSources.make: Added missing bindings: HTMLCanvasElement for ObjC,
1162         CDATASection, Comment, and EntityReference for JavaScript.
1163         * WebCore.xcodeproj/project.pbxproj: Added those new generated files.
1164
1165         * bindings/js/JSHTMLElementWrapperFactory.cpp:
1166         (WebCore::createJSHTMLWrapper): Corrected the wrapper classes for <keygen>,
1167         which needs an HTMLSelectElement wrapper, and <xmp>, which needs an
1168         HTMLPreElement wrapper.
1169
1170         * bindings/objc/DOMInternal.h: Updated for new naming scheme.
1171         Also moved createDOMWrapper from the KJS namespace to the WebCore namespace.
1172         * bindings/objc/DOMUtility.mm:
1173         (KJS::createDOMWrapper): Broke the core function into a separate one, and
1174         left it in the KJS namespace because Objective-C++ rules make it impossible
1175         for it to work in the WebCore namespace. Used a macro-based implementation
1176         to cut down on repeated code, and added missing cases for Counter,
1177         HTMLOptionsCollection, Range, XPathExpression, XPathResult, Event, RGBColor,
1178         Rect, Window, DOMImplementation, NodeIterator, TreeWalker, and HTMLCollection.
1179         (WebCore::createDOMWrapper): The other half of the function.
1180
1181         * bindings/objc/DOM.mm:
1182         (WebCore::createElementClassMap): Corrected the wrapper classes for
1183         <canvas>, which needs a DOMHTMLCanvasElement wrapper, <del>, which needs
1184         a DOMHTMLModElement wrapper, <embed>, which needs a DOMHTMLEmbedElement
1185         wrapper, <ins>, which needs a DOMHTMLModElement wrapper, <th>, which needs
1186         a DOMHTMLTableCellElement wrapper, and <xmp>, which needs an
1187         DOMHTMLPreElement wrapper.
1188         (+[DOMNode _wrapNode:]): Updated for new naming scheme.
1189         (+[DOMNode _wrapEventTarget:]): Ditto.
1190         (+[DOMNodeFilter _wrapNodeFilter:]): Ditto.
1191         (ObjCNodeFilterCondition::acceptNode): Ditto.
1192         (-[DOMDocument createNodeIterator:whatToShow:filter:expandEntityReferences:]): Ditto.
1193         (-[DOMDocument createTreeWalker:whatToShow:filter:expandEntityReferences:]): Ditto.
1194         (WebCore::ObjCEventListener::handleEvent): Ditto.
1195
1196         * dom/Document.cpp: (WebCore::Document::createEvent):
1197         Added cases for OverflowEvent and WheelEvent.
1198
1199         * dom/OverflowEvent.h: Added empty constructor and initOverflowEvent.
1200         * dom/OverflowEvent.cpp:
1201         (WebCore::OverflowEvent::OverflowEvent): Added.
1202         (WebCore::OverflowEvent::initOverflowEvent): Added.
1203         * dom/OverflowEvent.idl: Added initOverflowEvent.
1204
1205         * bindings/objc/PublicDOMInterfaces.h: Added initOverflowEvent.
1206
1207         * bindings/objc/DOMCSS.mm:
1208         (+[DOMStyleSheet _wrapStyleSheet:]): Updated for new naming scheme.
1209         (+[DOMCSSRule _wrapCSSRule:]): Corrected wrapper for CSSUnknownRule.
1210         (+[DOMCSSValue _wrapCSSValue:]): Updated for new naming scheme.
1211
1212         * bindings/js/kjs_css.h:
1213         (KJS::DOMRGBColor::impl): Added. Used when making an ObjC wrapper.
1214         (KJS::DOMRect::impl): Ditto.
1215
1216         * bindings/js/kjs_dom.cpp: (KJS::toJS): Corrected the wrapper classes for
1217         CDATASection, Comment, and EntityReference.
1218
1219         * bindings/js/kjs_html.cpp: Corrected the class name for HTMLElement
1220         (was "DOMHTMLElement") and HTMLCollection (was "Collection").
1221
1222         * bindings/objc/DOMImplementationFront.h:
1223         * bindings/objc/DOMImplementationFront.cpp:
1224         (WebCore::implementationFront): Added new overload that returns a front given
1225         a JavaScript wrapper. Needed by the code that makes the Objective-C wrapper.
1226
1227         * bindings/objc/WebScriptObject.mm:
1228         (+[WebScriptObject _convertValueToObjcValue:originRootObject:rootObject:]):
1229         Updated since createDOMWrapper is now in the WebCore namespace.
1230
1231         * bindings/scripts/CodeGeneratorObjC.pm: Update the naming scheme for the
1232         wrapper creation functions to use _wrapElement: rather then _elementWith:
1233         style. Removed now-unneeded special cases for things that needed to stay
1234         upper-case, since we don't have to do the whole lcfirst thing.
1235
1236         * html/HTMLCanvasElement.idl: Added #if so we can successfully generate the
1237         Objective-C wrapper for this class, even though we can't yet handle the
1238         DOMObject return type.
1239
1240         * html/HTMLOptionsCollection.idl: Added GenerateNativeConverter so we get
1241         an appropriate toJS function.
1242
1243         * bindings/objc/DOMEvents.mm:
1244         (+[DOMEvent _wrapEvent:]):
1245         * bindings/objc/DOMHTML.mm:
1246         (-[DOMHTMLDocument createDocumentFragmentWithMarkupString:baseURL:]):
1247         (-[DOMHTMLDocument createDocumentFragmentWithText:]):
1248         * bindings/objc/DOMObject.mm:
1249         (-[DOMObject sheet]):
1250         * bindings/objc/DOMRGBColor.mm:
1251         (-[DOMRGBColor red]):
1252         (-[DOMRGBColor green]):
1253         (-[DOMRGBColor blue]):
1254         (-[DOMRGBColor alpha]):
1255         (+[DOMRGBColor _wrapRGBColor:]):
1256         * bindings/objc/DOMSVGPathSegInternal.mm:
1257         (+[DOMSVGPathSeg _wrapSVGPathSeg:]):
1258         * bindings/objc/DOMXPath.mm:
1259         (+[DOMNativeXPathNSResolver _wrapXPathNSResolver:]):
1260         * page/mac/WebCoreFrameBridge.mm:
1261         (-[WebCoreFrameBridge nodesFromList:]):
1262         (-[WebCoreFrameBridge elementWithName:inForm:]):
1263         (-[WebCoreFrameBridge formForElement:]):
1264         (-[WebCoreFrameBridge currentForm]):
1265         (-[WebCoreFrameBridge controlsInForm:]):
1266         (-[WebCoreFrameBridge rangeByAlteringCurrentSelection:SelectionController::direction:SelectionController::granularity:]):
1267         (-[WebCoreFrameBridge convertNSRangeToDOMRange:]):
1268         (-[WebCoreFrameBridge markDOMRange]):
1269         (-[WebCoreFrameBridge markedTextDOMRange]):
1270         (-[WebCoreFrameBridge smartDeleteRangeForProposedRange:]):
1271         (-[WebCoreFrameBridge documentFragmentWithMarkupString:baseURLString:]):
1272         (-[WebCoreFrameBridge documentFragmentWithText:inContext:]):
1273         (-[WebCoreFrameBridge documentFragmentWithNodesAsParagraphs:]):
1274         (-[WebCoreFrameBridge replaceSelectionWithNode:selectReplacement:smartReplace:matchStyle:]):
1275         (-[WebCoreFrameBridge replaceSelectionWithText:selectReplacement:smartReplace:]):
1276         (-[WebCoreFrameBridge increaseSelectionListLevel]):
1277         (-[WebCoreFrameBridge increaseSelectionListLevelOrdered]):
1278         (-[WebCoreFrameBridge increaseSelectionListLevelUnordered]):
1279         (-[WebCoreFrameBridge dragCaretDOMRange]):
1280         (-[WebCoreFrameBridge editableDOMRangeForPoint:]):
1281         (-[WebCoreFrameBridge characterRangeAtPoint:]):
1282         (-[WebCoreFrameBridge typingStyle]):
1283         (-[WebCoreFrameBridge rangeOfCharactersAroundCaret]):
1284         * platform/mac/ClipboardMac.mm:
1285         (WebCore::ClipboardMac::declareAndWriteDragImage):
1286         * platform/mac/PasteboardMac.mm:
1287         (WebCore::Pasteboard::writeSelection):
1288         Updated for new naming scheme.
1289
1290 2007-03-12  Timothy Hatcher  <timothy@apple.com>
1291
1292         Reviewed by Darin.
1293
1294         <rdar://problem/4990691> REGRESSION: Selecting text in Adium's Messages field causes horizontal shift
1295
1296         Corrected the recursive point conversion and scrolling done in scrollPointRecursively to use the
1297         document view of each NSClipView.
1298
1299         * platform/mac/ScrollViewMac.mm:
1300         (WebCore::ScrollView::scrollPointRecursively): Call convertPoint:fromView: on the document views,
1301         not the clip views as we encounter them. Then call constrainScrollPoint: on the converted point to
1302         constrain to the document view bounds. And finally call scrollPoint: on the document view, not the clip view.
1303
1304 2007-03-12  Antti Koivisto  <antti@apple.com>
1305
1306         Reviewed by Alexey.
1307
1308         Fix http://bugs.webkit.org/show_bug.cgi?id=12833
1309         REGRESSION: Selecting text in 6.6MB txt file is sluggish as of the Feb 19th nightly
1310         <rdar://problem/5028159>
1311
1312         Divide large text blocks (>64kB) over multiple text nodes. This limits linebox searches to
1313         a manageable subset.
1314
1315         * dom/Text.cpp:
1316         (WebCore::Text::createWithLengthLimit):
1317         * dom/Text.h:
1318         * html/HTMLParser.cpp:
1319         (WebCore::HTMLParser::parseToken):
1320         * loader/TextDocument.cpp:
1321         (WebCore::TextTokenizer::write):
1322
1323 2007-03-12  David Hyatt  <hyatt@apple.com>
1324
1325         Fix a regression in printing.  Printer fonts need to be part of the
1326         font cache key, since printer fonts have different glyph widths.
1327         (Integer antialiasing is used for screen fonts but not for printer fonts.)
1328
1329         Reviewed by aroben
1330
1331         * platform/FontCache.cpp:
1332         (WebCore::FontPlatformDataCacheKey::FontPlatformDataCacheKey):
1333         (WebCore::FontPlatformDataCacheKey::operator==):
1334         (WebCore::computeHash):
1335         (WebCore::FontCache::getCachedFontPlatformData):
1336
1337 2007-03-12  Lars Knoll <lars@trolltech.com>
1338
1339         add missing symbol to fix the Qt build again.
1340
1341         * platform/qt/SearchPopupMenuQt.cpp:
1342         (WebCore::SearchPopupMenu::enabled):
1343
1344 2007-03-12  Adele Peterson  <adele@apple.com>
1345
1346         Reviewed by Oliver.
1347
1348         Adding the ability to enable or disable a SearchPopupMenu.
1349
1350         * platform/SearchPopupMenu.h:
1351         * platform/mac/SearchPopupMenuMac.mm: (WebCore::SearchPopupMenu::enabled):
1352         * rendering/RenderTextControl.cpp: (WebCore::RenderTextControl::showPopup):
1353
1354 2007-03-12  Rob Buis  <buis@kde.org>
1355
1356         Reviewed by Nikolas Zimmermann.
1357
1358         http://bugs.webkit.org/show_bug.cgi?id=12500
1359         SVG fails to correctly handle all link style selectors
1360         http://bugs.webkit.org/show_bug.cgi?id=12567
1361         <text> elements ignore <a> children
1362
1363         Allow <a> inside svg text and handle xlink:show.
1364
1365         * ksvg2/svg/SVGAElement.cpp:
1366         (WebCore::SVGAElement::createRenderer):
1367         (WebCore::SVGAElement::defaultEventHandler):
1368         (WebCore::SVGAElement::childShouldCreateRenderer):
1369         * ksvg2/svg/SVGAElement.h:
1370         * ksvg2/svg/SVGElement.h:
1371         (WebCore::SVGElement::isTextContent):
1372         * ksvg2/svg/SVGTextContentElement.h:
1373         (WebCore::SVGTextContentElement::isTextContent):
1374         * ksvg2/svg/SVGTextElement.cpp:
1375         (WebCore::SVGTextElement::childShouldCreateRenderer):
1376         * rendering/SVGInlineFlowBox.cpp:
1377         (WebCore::translateBox):
1378         (WebCore::placePositionedBoxesHorizontally):
1379         (WebCore::placeBoxesVerticallyWithAbsBaseline):
1380
1381 2007-03-12  Adele Peterson  <adele@apple.com>
1382
1383         Reviewed by Oliver.
1384
1385         Add a missing parameter to the constructor.
1386
1387         * platform/PlatformKeyboardEvent.h:
1388         * platform/win/KeyEventWin.cpp: (WebCore::PlatformKeyboardEvent::PlatformKeyboardEvent):
1389
1390 2007-03-12  Lars Knoll <lars@trolltech.com>
1391
1392         Fix the Qt build. I still get crashes though :/
1393
1394         * platform/graphics/qt/ImageDecoderQt.cpp:
1395         (WebCore::ImageDecoderQt::clearFrame):
1396         * platform/graphics/qt/ImageDecoderQt.h:
1397         * platform/graphics/qt/ImageSourceQt.cpp:
1398         (WebCore::ImageSource::setData):
1399         (WebCore::ImageSource::frameIsCompleteAtIndex):
1400         (WebCore::ImageSource::clear):
1401         (WebCore::ImageSource::destroyFrameAtIndex):
1402         * platform/qt/ClipboardQt.cpp:
1403         (WebCore::ClipboardQt::ClipboardQt):
1404         * platform/qt/ClipboardQt.h:
1405         * platform/qt/DragDataQt.cpp:
1406         (WebCore::DragData::createClipboard):
1407
1408 2007-03-11  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
1409
1410         Not reviewed - gdk build fixes.
1411
1412         * platform/gdk/EditorClientGdk.cpp:
1413         (WebCore::EditorClientGdk::respondToChangedSelection): add stub
1414         * platform/gdk/EditorClientGdk.h:
1415         * platform/gdk/FrameGdk.cpp:
1416         * platform/gdk/TemporaryLinkStubs.cpp:
1417         (Editor::markMisspellings): add stub
1418
1419 2007-03-11  Alexey Proskuryakov  <ap@webkit.org>
1420
1421         Reviewed by Adele.
1422
1423         http://bugs.webkit.org/show_bug.cgi?id=12560
1424         W3C XPath test Text_Nodes.svg fails
1425
1426         * xml/XPathStep.cpp:
1427         (WebCore::XPath::Step::nodeTestMatches): Revert the fix, as the behavior doesn't appear
1428         all that desirable as it did at first glance.
1429
1430 2007-03-11  Oliver Hunt  <oliver@apple.com>
1431
1432         Reviewed by hyatt.
1433
1434         Fix for <rdar://problem/5055690> ASSERTION failure on drop into
1435         editable element with content changed on drop
1436
1437         After setting the selection for a drop into an editable region
1438         we make sure the we succeeded.  If we didn't we assume a focus handler
1439         or similar altered the element contents and try again, if the second
1440         attempt fails we bail out.
1441
1442         * page/DragController.cpp:
1443         (WebCore::setSelectionToDragCaret):
1444         (WebCore::DragController::concludeDrag):
1445
1446 2007-03-11  Oliver Hunt  <oliver@apple.com>
1447
1448         Reviewed by Adele.
1449
1450         Moving Frame{Mac}::respondToChangedSelection to Frame.cpp
1451         Added new EditorClient method to handle old bridge function
1452
1453         * bridge/EditorClient.h:
1454            Added respondToChangedSelection to replace old bridge function
1455         * editing/Editor.cpp:
1456         (WebCore::Editor::respondToChangedSelection):
1457            Add client call to replace old bridge call from Frame::respondToChangedSelection
1458         * editing/SelectionController.cpp:
1459         (WebCore::SelectionController::setSelection):
1460            No longer directly call Editor as Frame::respondToChangedSelection
1461            makes the call
1462         * page/Frame.cpp:
1463         (WebCore::Frame::respondToChangedSelection):
1464            Moved from FrameMac, replaced bridge call with call to Editor
1465         * page/mac/FrameMac.mm:
1466            Moved respondToChangedSelection to Frame.cpp
1467         * page/mac/WebCoreFrameBridge.h:
1468            Removed respondToChangedSelection from bridge
1469         * page/qt/FrameQt.cpp:
1470            Remove stub method for respondToChangedSelection
1471
1472 2007-03-11  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
1473
1474         Reviewed by Brady Eidson.
1475
1476         Linux/gdk fix.
1477
1478         * loader/gdk/FrameLoaderClientGdk.cpp:
1479         (WebCore::FrameLoaderClientGdk::shouldGoToHistoryItem): return true so
1480         that FrameLoader:goBackOrForwards() works.
1481
1482 2007-03-10  Geoffrey Garen  <ggaren@apple.com>
1483
1484         Reviewed by Darin Adler.
1485
1486         Fixed <rdar://problem/4587763> PAC file: lock inversion between QT and
1487         JSCore causes a hang @ www.panoramas.dk
1488
1489         See JavaScriptCore ChangeLog for details.
1490
1491         * bindings/objc/WebScriptObject.mm:
1492         (_didExecute): Added helpful ASSERT.
1493         (+[WebScriptObject throwException:]): Added missing JSLock.
1494
1495 2007-03-11  Antti Koivisto  <antti@apple.com>
1496
1497         Reviewed by Hyatt.
1498
1499         Optimize linebox memory consumption:
1500         - move all bitfields to baseclass compacting them
1501         - make InlineTextBox::m_truncation unsigned short and make it relative to m_start
1502         - remove extremely rarely used EllipsisBox pointer from RootInlineBox and instead
1503           use a global hashmap to store it if needed
1504         - use minimum required number of bits to store BidiStatus enum variables in RootInlineBox
1505         - move overflow variables in RootInlineBox to a separate struct that is instantiated
1506           only if any of the variables is set to a value that can't trivially be derived from
1507           box x, y, width and height
1508
1509         As a result line box objects shrink:
1510             InlineBox: 44 -> 44 bytes
1511             InlineTextBox: 68 -> 60 bytes
1512             InlineFlowBox: 68 -> 64 bytes
1513             RootInlineBox: 128 -> 88 bytes
1514
1515         The optimizations possiblity was noticed when debugging http://bugs.webkit.org/show_bug.cgi?id=12833
1516         Bug 12833: REGRESSION: Selecting text in 6.6MB txt file is sluggish as of the Feb 19th nightly
1517         <rdar://problem/5028159>
1518
1519         On that page the patch saves 11.5MB or some 21% of linebox memory consumption. It also
1520         actually improves selection performance somewhat by improving memory locality.
1521
1522         * rendering/InlineBox.h:
1523         (WebCore::InlineBox::InlineBox):
1524         * rendering/InlineFlowBox.h:
1525         (WebCore::InlineFlowBox::InlineFlowBox):
1526         * rendering/InlineTextBox.cpp:
1527         (WebCore::InlineTextBox::placeEllipsisBox):
1528         (WebCore::InlineTextBox::nodeAtPoint):
1529         (WebCore::InlineTextBox::paint):
1530         (WebCore::InlineTextBox::paintDecoration):
1531         (WebCore::InlineTextBox::paintSpellingOrGrammarMarker):
1532         (WebCore::InlineTextBox::paintMarkedTextUnderline):
1533         * rendering/InlineTextBox.h:
1534         (WebCore::InlineTextBox::InlineTextBox):
1535         * rendering/RootInlineBox.cpp:
1536         (WebCore::throw):
1537         (WebCore::RootInlineBox::Overflow::operator delete):
1538         (WebCore::RootInlineBox::Overflow::destroy):
1539         (WebCore::RootInlineBox::destroy):
1540         (WebCore::RootInlineBox::detachEllipsisBox):
1541         (WebCore::RootInlineBox::clearTruncation):
1542         (WebCore::RootInlineBox::placeEllipsis):
1543         (WebCore::RootInlineBox::paintEllipsisBox):
1544         (WebCore::RootInlineBox::addHighlightOverflow):
1545         (WebCore::RootInlineBox::nodeAtPoint):
1546         (WebCore::RootInlineBox::adjustPosition):
1547         (WebCore::RootInlineBox::selectionTop):
1548         (WebCore::RootInlineBox::setLineBreakInfo):
1549         (WebCore::RootInlineBox::ellipsisBox):
1550         (WebCore::RootInlineBox::setVerticalOverflowPositions):
1551         (WebCore::RootInlineBox::setHorizontalOverflowPositions):
1552         (WebCore::RootInlineBox::setVerticalSelectionPositions):
1553         * rendering/RootInlineBox.h:
1554         (WebCore::RootInlineBox::RootInlineBox):
1555         (WebCore::RootInlineBox::topOverflow):
1556         (WebCore::RootInlineBox::bottomOverflow):
1557         (WebCore::RootInlineBox::leftOverflow):
1558         (WebCore::RootInlineBox::rightOverflow):
1559         (WebCore::RootInlineBox::lineBreakBidiStatus):
1560         (WebCore::RootInlineBox::selectionBottom):
1561         (WebCore::RootInlineBox::Overflow::Overflow):
1562
1563 2007-03-11  Alexey Proskuryakov  <ap@webkit.org>
1564
1565         Reviewed by Darin.
1566
1567         A partial fix for http://bugs.webkit.org/show_bug.cgi?id=13021
1568         XPath can be very slow
1569
1570         * xml/XPathExpression.cpp:
1571         (WebCore::XPathExpression::evaluate): Cache evaluationContext in a local variable.
1572
1573         * xml/XPathExpressionNode.cpp:
1574         (WebCore::XPath::Expression::evaluationContext):
1575         * xml/XPathExpressionNode.h:
1576         (WebCore::XPath::Expression::addSubExpression):
1577         (WebCore::XPath::Expression::subExprCount):
1578         (WebCore::XPath::Expression::subExpr):
1579         * xml/XPathFunctions.cpp:
1580         * xml/XPathFunctions.h:
1581         (WebCore::XPath::Function::setName):
1582         (WebCore::XPath::Function::arg):
1583         (WebCore::XPath::Function::argCount):
1584         (WebCore::XPath::Function::name):
1585         Made one-liners critical for performance inline.
1586
1587         * xml/XPathGrammar.y: Fully parse NodeTests, so that strings are no longer passed for what is
1588         essentially an enum. Use LocationPath accessors to add steps, instead of directly manipulating
1589         internal data members.
1590
1591         * xml/XPathParser.cpp:
1592         (WebCore::XPath::Parser::parseStatement):
1593         (WebCore::XPath::Parser::registerNodeTest):
1594         (WebCore::XPath::Parser::deleteNodeTest):
1595         * xml/XPathParser.h:
1596         Added support methods for changes in XPathGrammar.y.
1597
1598         * xml/XPathPath.cpp:
1599         (WebCore::XPath::Filter::evaluate): Cache evaluationContext in a local variable. Use swap() to avoid
1600         performing vector assignments.
1601         (WebCore::XPath::LocationPath::evaluate): Use swap() to avoid performing vector assignments.
1602         (WebCore::XPath::LocationPath::optimizeStepPair): This new method is called during LocationPath construction,
1603         to simplify the path as it's being built. Currently, the only optimized case is "//*" - it is a basis for
1604         important operations that cannot be efficiently written in XPath 1.0, but can be optimized with a little bit
1605         of XPath 2.0.
1606         (WebCore::XPath::LocationPath::appendStep): A new accessor that modifies m_steps and calls optimizeStepPair().
1607         (WebCore::XPath::LocationPath::insertFirstStep): Ditto.
1608         * xml/XPathPath.h:
1609         (WebCore::XPath::LocationPath::setAbsolute): A new accessor.
1610
1611         * xml/XPathStep.h:
1612         (WebCore::XPath::Step::NodeTest::):
1613         (WebCore::XPath::Step::NodeTest::NodeTest):
1614         (WebCore::XPath::Step::NodeTest::kind):
1615         (WebCore::XPath::Step::NodeTest::data):
1616         Step::NodeTest is a new sub-class that represents a fully parsed NodeTest.
1617         (WebCore::XPath::Step::axis):
1618         (WebCore::XPath::Step::nodeTest):
1619         (WebCore::XPath::Step::nodeTestData):
1620         (WebCore::XPath::Step::namespaceURI):
1621         (WebCore::XPath::Step::predicates):
1622         (WebCore::XPath::Step::setAxis):
1623         (WebCore::XPath::Step::setNodeTest):
1624         (WebCore::XPath::Step::setNodeTestData):
1625         (WebCore::XPath::Step::setNamespaceURI):
1626         (WebCore::XPath::Step::setPredicates):
1627         New accessors that let optimizeStepPair() manipulate Step data.
1628
1629         * xml/XPathStep.cpp:
1630         (WebCore::XPath::Step::Step): Use the new NodeTest class.
1631         (WebCore::XPath::Step::evaluate): Cache evaluationContext in a local variable. Use swap() to avoid
1632         performing unneeded vector assignments.
1633         (WebCore::XPath::Step::nodesInAxis): Cosmetic changes.
1634         (WebCore::XPath::Step::nodeTestMatches): Use NodeTest instead of parsing the test from string each time.
1635         Added a partial implementation of XPath 2.0 element() node test.
1636
1637 2007-03-10  Alexey Proskuryakov  <ap@webkit.org>
1638
1639         Reviewed by Darin.
1640
1641         http://bugs.webkit.org/show_bug.cgi?id=12249
1642         FCKeditor: <hr>, <ul> and <ol> have id="undefined"
1643
1644         This fixes the attached reduction, but not the original issue.
1645
1646         Test: editing/execCommand/default-parameters.html
1647
1648         * dom/Document.h:
1649         * dom/Document.idl:
1650         Make second and third execCommand() parameters optional.
1651
1652 2007-03-10  Adele Peterson  <adele@apple.com>
1653
1654         Reviewed by Maciej.
1655
1656         Fix for http://bugs.webkit.org/show_bug.cgi?id=13028
1657         REGRESSION: textField:doCommandBySelector:inFrame: not being called properly
1658
1659         * html/HTMLInputElement.cpp: (WebCore::HTMLInputElement::defaultEventHandler):
1660           Before calling the base class defaultEventHandler, which will call handleKeypress, call doTextFieldCommandFromEvent.
1661
1662 2007-03-10  Mitz Pettel  <mitz@webkit.org>
1663
1664         Reviewed by Darin.
1665
1666         - fix http://bugs.webkit.org/show_bug.cgi?id=13013
1667           REGRESSION: Selection box does not scroll to where the focus jumps when pressing an alphanumeric key
1668
1669         Test: fast/forms/listbox-typeahead-scroll.html
1670
1671         * html/HTMLSelectElement.cpp:
1672         (WebCore::HTMLSelectElement::setSelectedIndex): Reordered to set the active selection's
1673         anchor and end before selecting the option, since the active selection is used to
1674         decide where to scroll when the selection is made.
1675         (WebCore::HTMLSelectElement::defaultEventHandler): Removed redundant check.
1676
1677 2007-03-10  Mitz Pettel  <mitz@webkit.org>
1678
1679         Reviewed by Darin.
1680
1681         - fix http://bugs.webkit.org/show_bug.cgi?id=12973
1682           REGRESSION: Reproducible assert while loading this test file if css is already in the cache
1683
1684         Test: fast/dom/css-cached-import-rule.html
1685
1686         Replaced some direct calls to document->stylesheetLoaded() with calls to
1687         the sheet's checkLoaded(). The latter calls back to the element's sheetLoaded() --
1688         which notifies the document of the load -- and then updates the sheet's
1689         loadCompleted() flag, ensuring that it stays in sync with whether the stylesheet
1690         is still considered pending by the document.
1691
1692         * dom/ProcessingInstruction.cpp:
1693         (WebCore::ProcessingInstruction::parseStyleSheet):
1694         * dom/StyleElement.cpp:
1695         (WebCore::StyleElement::childrenChanged):
1696         * html/HTMLLinkElement.cpp:
1697         (WebCore::HTMLLinkElement::process):
1698         (WebCore::HTMLLinkElement::setCSSStyleSheet):
1699         * ksvg2/svg/SVGStyleElement.cpp:
1700         (WebCore::SVGStyleElement::sheetLoaded):
1701         * ksvg2/svg/SVGStyleElement.h:
1702
1703 2007-03-10  David Kilzer  <ddkilzer@webkit.org>
1704
1705         Reviewed by Darin.
1706
1707         - fix http://bugs.webkit.org/show_bug.cgi?id=9609
1708           REGRESSION: Missing image icon needs to be moved back to WebKit
1709
1710         * WebCore.exp: Export WebCore::Image::loadPlatformResource(const char*) for use in
1711         [WebHTMLView _startDraggingImage:at:operation:event:sourceIsDHTML:DHTMLWroteData:].
1712
1713 2007-03-09  Darin Adler  <darin@apple.com>
1714
1715         Reviewed by Justin.
1716
1717         - fix http://bugs.webkit.org/show_bug.cgi?id=8928
1718           <rdar://problem/5045708> REPRODUCIBLE ASSERT: Cannot paste HTML into a
1719           contenteditable region in an XHTML document (8928)
1720
1721         Test: editing/pasteboard/paste-xml.xhtml
1722
1723         * editing/markup.cpp: (WebCore::createFragmentFromMarkup): Added a check for 0
1724         here, since createContextualFragment can return 0 for XML documents that fail
1725         to parse. In my testing, callers all seem equipped to handle 0.
1726
1727 2007-03-09  Mitz Pettel  <mitz@webkit.org>
1728
1729         Reviewed by Darin.
1730
1731         - fix http://bugs.webkit.org/show_bug.cgi?id=9929
1732           REGRESSION: crash on logging in on mijnpostbank.nl
1733
1734         Test: http/tests/misc/onload-remove-iframe-crash-2.html
1735
1736         The resulted from an iframe's load event handler removing the iframe
1737         from the document.
1738
1739         * dom/Document.cpp:
1740         (WebCore::Document::implicitClose): Bail out early if an event handler
1741         removed the frame.
1742         * loader/FrameLoader.cpp:
1743         (WebCore::FrameLoader::FrameLoader):
1744         (WebCore::FrameLoader::clear):
1745         (WebCore::FrameLoader::checkCompleted): Protect the frame from deletion
1746         by event handlers.
1747         (WebCore::FrameLoader::checkCompletedTimerFired):
1748         (WebCore::FrameLoader::scheduleCheckCompleted):
1749         (WebCore::FrameLoader::detachFromParent): Schedule a completion check
1750         on the parent (in case the child is what has been keeping it from completing).
1751         * loader/FrameLoader.h:
1752
1753 2007-03-08  David Kilzer  <ddkilzer@webkit.org>
1754
1755         Reviewed by Beth.
1756
1757         - fix http://bugs.webkit.org/show_bug.cgi?id=13019
1758           REGRESSION (r20074): Forms don't submit on a variety of websites
1759
1760         No tests added since LayoutTests/fast/forms/document-write.html was timing out
1761         and causing a layout test failure.
1762
1763         * html/HTMLFormElement.cpp:
1764         (WebCore::HTMLFormElement::submit): Removed stray code.
1765
1766 2007-03-08  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
1767
1768         Not reviewed - gdk build fix.
1769
1770         * platform/graphics/cairo/ImageSourceCairo.cpp:
1771         (WebCore::ImageSource::frameIsCompleteAtIndex): add empty stub
1772
1773 2007-03-08  Justin Garcia  <justin.garcia@apple.com>
1774
1775         Reviewed by harrison
1776
1777         <rdar://problem/4903193>
1778         On particular page, Copy is so slow it seems like a hang
1779
1780         ~2x speedup copying:
1781         http://shakespeare.mit.edu/henryviii/full.html
1782         Also produces less bloated markup.
1783
1784         * editing/markup.cpp:
1785         (WebCore::startMarkup): Don't wrap text nodes in style spans.
1786         For Elements, don't inline styles inherited from ancestors.
1787         (WebCore::createMarkup): No longer necessary to find
1788         the root's default style and pass it to startMarkup.
1789         Add a wrapper span around the markup with the styles
1790         that all nodes in the markup inherit (the inheritable
1791         styles from the common ancestor container's computed style).
1792         Added a FIXME about unecessary markup for inline ancestors
1793         up to the commonAncestorBlock.
1794
1795 2007-03-08  Darin Adler  <darin@apple.com>
1796
1797         Reviewed by Adele.
1798
1799         - <rdar://problem/4470381> multipart/form-data boundary security vulnerability
1800
1801         By making the form data boundary a string with some random data in it, we reduce
1802         the possibility that anyone could take advantage of it by creating a file that
1803         intentionally has the boundary string in it.
1804
1805         * html/HTMLFormElement.h: Removed boundary(), setBoundary(), and m_boundary.
1806         Marked a lot more stuff private.
1807         * html/HTMLFormElement.cpp:
1808         (WebCore::HTMLFormElement::HTMLFormElement): Removed code to initialize
1809         m_boundary.
1810         (WebCore::randomNumber): Added. Function that returns a random number, including
1811         seeding the random number generator the first time it's called. For now, usees the more
1812         random function random() on Mac OS X and the more-standard rand() on other platforms.
1813         (WebCore::HTMLFormElement::formData): Take a parameter with the form boundary string,
1814         and use that instead of m_boundary.
1815         (WebCore::getUniqueBoundaryString): Added. Makes a boundary string using random numbers
1816         and base 64 encoding.
1817         (WebCore::HTMLFormElement::submit): Call getUniqueBoundaryString and pass the boundary
1818         string into formData for multipart form posts.
1819
1820 2007-03-08  Maciej Stachowiak  <mjs@apple.com>
1821
1822         Reviewed by Adele.
1823
1824         <rdar://problem/4646563> REGRESSION: Unable to send text message from Verizon text message website: vtext.com (12588)
1825         http://bugs.webkit.org/show_bug.cgi?id=12588
1826
1827         Carefully revised which focus operations restore previous selection, which clear it, and which
1828         select the whole control contents.
1829
1830         Tests:
1831         fast/forms/focus-selection-input.html
1832         fast/forms/focus-selection-textarea.html
1833
1834         * dom/Element.cpp:
1835         (WebCore::Element::focus):
1836         * dom/Element.h:
1837         * html/HTMLInputElement.cpp:
1838         (WebCore::HTMLInputElement::focus):
1839         (WebCore::HTMLInputElement::accessKeyAction):
1840         * html/HTMLInputElement.h:
1841         * html/HTMLLabelElement.cpp:
1842         (WebCore::HTMLLabelElement::focus):
1843         (WebCore::HTMLLabelElement::accessKeyAction):
1844         * html/HTMLLabelElement.h:
1845         * html/HTMLLegendElement.cpp:
1846         (WebCore::HTMLLegendElement::focus):
1847         * html/HTMLLegendElement.h:
1848         * html/HTMLTextAreaElement.cpp:
1849         (WebCore::HTMLTextAreaElement::focus):
1850         * html/HTMLTextAreaElement.h:
1851         * page/FocusController.cpp:
1852         (WebCore::FocusController::advanceFocus):
1853
1854 2007-03-08  Justin Garcia  <justin.garcia@apple.com>
1855
1856         Reviewed by harrison
1857
1858         <http://bugs.webkit.org/show_bug.cgi?id=12244>
1859         FCKeditor: Find dialog doesn't work
1860
1861         * bindings/js/kjs_window.cpp:
1862         (KJS::Window::find): Added.  This function doesn't yet
1863         support whole word searches, searching in subframes, or
1864         opening the find dialog.
1865         (KJS::WindowFunc::callAsFunction):
1866         * bindings/js/kjs_window.h:
1867         (KJS::Window::):
1868
1869 2007-03-08  David Hyatt  <hyatt@apple.com>
1870
1871         Fix regression from throwing away frames of large animated images.  Alter
1872         animated images so that they refuse to advance the animation until the
1873         current displayed frame has been fully decoded.
1874
1875         Reviewed by ggaren
1876
1877         * platform/graphics/BitmapImage.cpp:
1878         (WebCore::BitmapImage::startAnimation):
1879         (WebCore::BitmapImage::advanceAnimation):
1880         * platform/graphics/ImageSource.h:
1881         * platform/graphics/cg/ImageSourceCG.cpp:
1882         (WebCore::ImageSource::frameIsCompleteAtIndex):
1883
1884 2007-03-08  David Hyatt  <hyatt@apple.com>
1885
1886         Fix 2% performance regression on the PLT.  Increase the large animated
1887         image cutoff from 1MB to 5MB.
1888
1889         In addition when pruning we will aggressively discard image sources.
1890
1891         Reviewed by ggaren
1892
1893         * platform/graphics/BitmapImage.cpp:
1894         (WebCore::BitmapImage::destroyDecodedData):
1895
1896 2007-03-08  Timothy Hatcher  <timothy@apple.com>
1897
1898         Reviewed by John.
1899
1900         <rdar://problem/4664697> highlighter SPI needs a node parameter to give more context
1901
1902         Pass the RenderObject's node to customHighlightLineRect and paintCustomHighlight.
1903
1904         * page/Frame.h:
1905         * page/mac/FrameMac.mm:
1906         (WebCore::Frame::customHighlightLineRect):
1907         (WebCore::Frame::paintCustomHighlight):
1908         * page/mac/WebCoreFrameBridge.h:
1909         * rendering/InlineTextBox.cpp:
1910         (WebCore::InlineTextBox::paintCustomHighlight):
1911         * rendering/RenderBox.cpp:
1912         (WebCore::RenderBox::paintCustomHighlight):
1913         * rendering/RootInlineBox.cpp:
1914         (WebCore::RootInlineBox::addHighlightOverflow):
1915         (WebCore::RootInlineBox::paintCustomHighlight):
1916
1917 2007-03-08  Justin Garcia  <justin.garcia@apple.com>
1918
1919         Reviewed by harrison
1920
1921         <http://bugs.webkit.org/show_bug.cgi?id=13000>
1922         Range.createContextualFragment is not supported
1923
1924         * dom/Range.cpp:
1925         (WebCore::Range::createContextualFragment): The
1926         "startContainer" may not be a container, if the
1927         range starts inside text.  In that case, look
1928         to the parent of the start node for an HTMLElement.
1929
1930 2007-03-08  Justin Garcia  <justin.garcia@apple.com>
1931
1932         Reviewed by harrison
1933
1934         <rdar://problem/5049671>
1935         Gmail Editor: With linked text, Remove Formatting doesn't always remove underline
1936
1937         * editing/Editor.cpp:
1938         (WebCore::Editor::removeFormattingAndStyle): Clear removed
1939         anchors after the deletion.
1940
1941 2007-03-08  David Kilzer  <ddkilzer@webkit.org>
1942
1943         Reviewed by NOBODY (build fix).
1944
1945         Added missing file for r20059:
1946         <rdar://problem/4708689> -- REGRESSION: Some symbols with 2-byte display as garbage in Hotmail.
1947
1948         File was taken from this svn repository which contained versions of ucnv.h and ucnv_err.h
1949         that were identical to ours:
1950         http://source.icu-project.org/repos/icu/icu/tags/release-3-2/source/common/unicode/ucnv_cb.h
1951
1952         * icu/unicode/ucnv_cb.h: Added.
1953
1954 2007-03-08  Shrikant Gangoda  <shrikant.gangoda@celunite.com>
1955
1956         Gdk build fix.
1957
1958         * platform/graphics/cairo/ImageSourceCairo.cpp:
1959         (WebCore::ImageSource::~ImageSource):
1960         (WebCore::ImageSource::clear):
1961
1962 2007-03-08  Oliver Hunt  <oliver@apple.com>
1963
1964         Reviewed by Adam.
1965
1966         To match old TEC behaviour when using ICU we need to use
1967         a few manual fallback encodings for the GBK/EUC-CN charsets
1968
1969         <rdar://problem/4708689> -- REGRESSION: Some symbols with 2-byte display as garbage in Hotmail.
1970
1971         * platform/TextCodecICU.cpp:
1972         (WebCore::TextCodecICU::TextCodecICU):
1973         (WebCore::TextCodecICU::createICUConverter):
1974         (WebCore::gbkEscapes):
1975         (WebCore::gbkCallbackEscape):
1976         (WebCore::gbkCallbackSubstitute):
1977         (WebCore::TextCodecICU::encode):
1978         * platform/TextCodecICU.h:
1979         (WebCore::TextCodecICU::needsGBKFallbacks):
1980         (WebCore::TextCodecICU::setNeedsGBKFallbacks):
1981
1982 2007-03-08  Alexey Proskuryakov  <ap@webkit.org>
1983
1984         Build fix.
1985
1986         * xml/XPathUtil.cpp:
1987         (WebCore::XPath::stringValue):
1988
1989 2007-03-08  David Hyatt  <hyatt@apple.com>
1990
1991         This patch dramatically reduces the memory consumed by animated images.  For large animated GIFs (defined for
1992         now as >1mb in terms of decoded frame buffer size), we will now aggressively flush previous frames of the
1993         animated GIF and just re-decode them on the fly if the animation loops.
1994
1995         Whenever a large animated GIF has its animation reset, we will also just throw out everything and start
1996         the animation over (in order to get rid of any cached detritus held in the ImageSource).
1997
1998         With this patch and the sample GIF used to test, WebKit's memory consumption went from 160MB down to 16MB.
1999
2000         Reviewed by mjs
2001
2002         * platform/graphics/BitmapImage.cpp:
2003         (WebCore::BitmapImage::destroyDecodedData):
2004         (WebCore::BitmapImage::resetAnimation):
2005         (WebCore::BitmapImage::advanceAnimation):
2006         * platform/graphics/ImageSource.h:
2007         * platform/graphics/cg/ImageSourceCG.cpp:
2008         (WebCore::ImageSource::~ImageSource):
2009         (WebCore::ImageSource::clear):
2010
2011 2007-03-08  Alexey Proskuryakov  <ap@webkit.org>
2012
2013         Reviewed by Darin.
2014
2015         http://bugs.webkit.org/show_bug.cgi?id=13006
2016         XPath string-value is broken for some node types
2017
2018         Test: fast/xpath/string-value.html
2019
2020         * xml/XPathUtil.cpp:
2021         (WebCore::XPath::stringValue): Fix it :-)
2022
2023 2007-03-07  Anders Carlsson  <acarlsson@apple.com>
2024
2025         Reviewed by Adam.
2026
2027         <rdar://problem/4981000>
2028         http://bugs.webkit.org/show_bug.cgi?id=12634
2029         REGRESSION: crash loading web archive (12634)
2030
2031         The reason this bug wasn't always reproducible is that it involved sending an event to a plugin while
2032         the page was loading. Before we send the event to the plugin we defer loads. The problem was that
2033         MainResourceLoader::setDefersLoad would not work with data loads.
2034
2035         * loader/DocumentLoader.cpp:
2036         (WebCore::DocumentLoader::setRequest):
2037         Only set m_committed to false if we also have a valid unreachable URL.
2038
2039         * loader/MainResourceLoader.cpp:
2040         (WebCore::MainResourceLoader::setDefersLoading):
2041         Make sure to stop and start data loads.
2042
2043 2007-03-07  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
2044
2045         Not reviewed.
2046
2047         Gdk build fixes.
2048
2049         * platform/gdk/EditorClientGdk.cpp:
2050         (WebCore::EditorClientGdk::handleKeypress): renamed
2051         (WebCore::EditorClientGdk::handleInputMethodKeypress): add a stub
2052         * platform/gdk/EditorClientGdk.h:
2053         * platform/graphics/cairo/ImageSourceCairo.cpp:
2054         (WebCore::ImageSource::destroyFrameAtIndex): add a stub
2055
2056 2007-03-07  Alexey Proskuryakov  <ap@webkit.org>
2057
2058         Reviewed by Darin.
2059
2060         http://bugs.webkit.org/show_bug.cgi?id=13004
2061         Repeatedly calling XPathExpression.evaluate() causes crashes or memory leaks
2062
2063         Removed XPath::Expression::optimize() and related methods, since they were buggy and almost useless.
2064         Merged doEvaluate() into evaluate(), since this was all evaluate() was doing after the above changes.
2065
2066         Test: fast/xpath/evaluate-twice.html
2067
2068         * xml/XPathExpression.cpp:
2069         (WebCore::XPathExpression::evaluate):
2070         * xml/XPathExpressionNode.cpp:
2071         (WebCore::XPath::Expression::Expression):
2072         (WebCore::XPath::Expression::~Expression):
2073         * xml/XPathExpressionNode.h:
2074         * xml/XPathFunctions.cpp:
2075         (WebCore::XPath::FunLast::evaluate):
2076         (WebCore::XPath::FunPosition::evaluate):
2077         (WebCore::XPath::FunId::evaluate):
2078         (WebCore::XPath::FunLocalName::evaluate):
2079         (WebCore::XPath::FunNamespaceURI::evaluate):
2080         (WebCore::XPath::FunName::evaluate):
2081         (WebCore::XPath::FunCount::evaluate):
2082         (WebCore::XPath::FunString::evaluate):
2083         (WebCore::XPath::FunConcat::evaluate):
2084         (WebCore::XPath::FunStartsWith::evaluate):
2085         (WebCore::XPath::FunContains::evaluate):
2086         (WebCore::XPath::FunSubstringBefore::evaluate):
2087         (WebCore::XPath::FunSubstringAfter::evaluate):
2088         (WebCore::XPath::FunSubstring::evaluate):
2089         (WebCore::XPath::FunStringLength::evaluate):
2090         (WebCore::XPath::FunNormalizeSpace::evaluate):
2091         (WebCore::XPath::FunTranslate::evaluate):
2092         (WebCore::XPath::FunBoolean::evaluate):
2093         (WebCore::XPath::FunNot::evaluate):
2094         (WebCore::XPath::FunTrue::evaluate):
2095         (WebCore::XPath::FunLang::evaluate):
2096         (WebCore::XPath::FunFalse::evaluate):
2097         (WebCore::XPath::FunNumber::evaluate):
2098         (WebCore::XPath::FunSum::evaluate):
2099         (WebCore::XPath::FunFloor::evaluate):
2100         (WebCore::XPath::FunCeiling::evaluate):
2101         (WebCore::XPath::FunRound::evaluate):
2102         * xml/XPathPath.cpp:
2103         (WebCore::XPath::Filter::evaluate):
2104         (WebCore::XPath::LocationPath::evaluate):
2105         (WebCore::XPath::Path::evaluate):
2106         * xml/XPathPath.h:
2107         * xml/XPathPredicate.cpp:
2108         (WebCore::XPath::Number::evaluate):
2109         (WebCore::XPath::StringExpression::evaluate):
2110         (WebCore::XPath::Negative::evaluate):
2111         (WebCore::XPath::NumericOp::evaluate):
2112         (WebCore::XPath::EqTestOp::evaluate):
2113         (WebCore::XPath::LogicalOp::evaluate):
2114         (WebCore::XPath::Union::evaluate):
2115         * xml/XPathPredicate.h:
2116         * xml/XPathStep.cpp:
2117         * xml/XPathStep.h:
2118         * xml/XPathVariableReference.cpp:
2119         (WebCore::XPath::VariableReference::evaluate):
2120         * xml/XPathVariableReference.h:
2121
2122 2007-03-07  Sam Weinig  <sam@webkit.org>
2123
2124         Reviewed by Tim H.
2125
2126         Remove unused #import from Objective-C bindings and cleanup the order of #imports.
2127
2128         * bindings/scripts/CodeGeneratorObjC.pm:
2129
2130 2007-03-07  Sam Weinig  <sam@webkit.org>
2131
2132         Reviewed by Tim H.
2133
2134         Make sure the baseURI attribute generates for private Objective-C bindings.
2135
2136         * dom/Node.idl:
2137
2138 2007-03-07  Anders Carlsson  <acarlsson@apple.com>
2139
2140         Reviewed by Maciej.
2141
2142         <rdar://problem/4874059>
2143         REGRESSION: Painter IX:register - Crash in WebCore:: ResourceLoader::willSendRequest()
2144
2145         If a load is done from inside of an error delegate method that is called because we cancel another load,
2146         the first load should be ignored since this is what shipping WebKit does.
2147
2148         (Actually, it does load the page in the data source but doesn't do anything with it since the data source
2149         won't have a web frame).
2150
2151         * loader/FrameLoader.cpp:
2152         (WebCore::FrameLoader::load):
2153         Just bail out if m_isStoppingLoad is true.
2154
2155 2007-03-07  David Hyatt  <hyatt@apple.com>
2156
2157         Use CGImageRelease instead of CFRelease.
2158
2159         Reviewed by aroben
2160
2161         * platform/graphics/cg/ImageCG.cpp:
2162         (WebCore::FrameData::clear):
2163         * platform/graphics/cg/ImageSourceCG.cpp:
2164         (WebCore::ImageSource::destroyFrameAtIndex):
2165
2166 2007-03-07  David Hyatt  <hyatt@apple.com>
2167
2168         Fix a regression where the cache size overflows because of a double
2169         subtraction per resource when they got removed from the cache.  Add an
2170         assert to adjustSize to detect this case in the future.
2171
2172         Fix ImageSourceCG so that when we flush decoded data from our cache that
2173         we also flush it from the ImageSource.
2174
2175         Reviewed by mjs
2176
2177         * loader/Cache.cpp:
2178         (WebCore::Cache::adjustSize):
2179         * platform/graphics/BitmapImage.cpp:
2180         (WebCore::BitmapImage::~BitmapImage):
2181         (WebCore::BitmapImage::destroyDecodedData):
2182         * platform/graphics/Image.h:
2183         * platform/graphics/ImageSource.h:
2184         * platform/graphics/cg/ImageSourceCG.cpp:
2185         (WebCore::ImageSource::setData):
2186         (WebCore::ImageSource::destroyFrameAtIndex):
2187
2188 2007-03-07  Mitz Pettel  <mitz@webkit.org>
2189
2190         Reviewed by Darin.
2191
2192         - fix http://bugs.webkit.org/show_bug.cgi?id=13002
2193           Incomplete repaint of inset outlines
2194
2195         Test: fast/repaint/outline-inset.html
2196
2197         * rendering/RenderObject.cpp:
2198         (WebCore::RenderObject::repaintAfterLayoutIfNeeded):
2199
2200 2007-03-07  Oliver Hunt  <oliver@apple.com>
2201
2202         Reviewed by Antti.
2203
2204         Reset mouse down/drag state variables on mouse button release
2205
2206         Fixes <rdar://problem/5044654>: Drag out of some QuickTime plug-ins converting into image drag
2207
2208         * page/EventHandler.cpp:
2209         (WebCore::EventHandler::handleMouseReleaseEvent):
2210
2211 2007-03-07  Adele Peterson  <adele@apple.com>
2212
2213         Reviewed by Darin.
2214
2215         WebCore part of fix for:
2216         http://bugs.webkit.org/show_bug.cgi?id=10871
2217         http://bugs.webkit.org/show_bug.cgi?id=12677
2218         <rdar://problem/4823129> REGRESSION: IME key events different in nightly
2219         <rdar://problem/4759563> REGRESSION: Return key is always sent when you confirm a clause in kotoeri
2220
2221         * page/EventHandler.cpp: (WebCore::EventHandler::defaultKeyboardEventHandler): Call handleInputMethodKeypress before actually dispatching the keypress event
2222           so that input methods have a chance to handle the event.
2223           If the input method handles the event (by marking or unmarking text), then we don't need to send the keypress event.
2224           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)
2225           when we dispatch the keypress event.
2226
2227         * dom/KeyboardEvent.h: Added Mac-specific KeypressCommand struct, so we can store command info during handleInputMethodKeypress, and use it during handleKeypress.
2228         (WebCore::KeyboardEvent::keypressCommand):
2229         (WebCore::KeyboardEvent::setKeypressCommand):
2230
2231         * bridge/EditorClient.h:
2232         * editing/Editor.cpp:
2233         (WebCore::Editor::handleKeypress): Changed handleKeyPress to handleKeypress.
2234         (WebCore::Editor::handleInputMethodKeypress): Added.
2235         * editing/Editor.h:
2236
2237         * platform/graphics/svg/SVGImageEmptyClients.h:
2238         (WebCore::SVGEmptyEditorClient::handleKeypress): Changed handleKeyPress to handleKeypress.
2239         (WebCore::SVGEmptyEditorClient::handleInputMethodKeypress): Added.
2240
2241 2007-03-07  Rob Buis  <buis@kde.org>
2242
2243         Reviewed by Darin.
2244
2245         http://bugs.webkit.org/show_bug.cgi?id=12579
2246         WebKit fails SVG xml:base test
2247
2248         Implement DOM3 properties baseURI and documentURI to fix
2249         the testcase in bug 12579.
2250
2251         * bindings/js/kjs_dom.cpp:
2252         (KJS::DOMNode::getValueProperty):
2253         * bindings/js/kjs_domnode.h:
2254         (KJS::DOMNode::):
2255         * dom/Document.cpp:
2256         (WebCore::Document::documentURI):
2257         (WebCore::Document::setDocumentURI):
2258         (WebCore::Document::baseURI):
2259         * dom/Document.h:
2260         * dom/Document.idl:
2261         * dom/DocumentType.cpp:
2262         (WebCore::DocumentType::baseURI):
2263         * dom/DocumentType.h:
2264         * dom/Element.cpp:
2265         (WebCore::Element::baseURI):
2266         * dom/Element.h:
2267         * dom/Node.cpp:
2268         (WebCore::Node::baseURI):
2269         * dom/Node.h:
2270         * ksvg2/misc/SVGImageLoader.cpp:
2271         (WebCore::SVGImageLoader::updateFromElement):
2272         * ksvg2/svg/SVGImageElement.cpp:
2273         (WebCore::SVGImageElement::parseMappedAttribute):
2274         (WebCore::SVGImageElement::attach):
2275
2276 2007-03-07  Anders Carlsson  <acarlsson@apple.com>
2277
2278         Reviewed by Brady.
2279
2280         Remove some methods in FrameLoader that just calls down to the active document loader. Since each
2281         resource loader now has a pointer to its document loader, we can just call directly to the
2282         document loader.
2283
2284         * WebCore.exp:
2285         * loader/FrameLoader.cpp:
2286         * loader/FrameLoader.h:
2287         * loader/MainResourceLoader.cpp:
2288         (WebCore::MainResourceLoader::willSendRequest):
2289         (WebCore::MainResourceLoader::didReceiveResponse):
2290         * loader/SubresourceLoader.cpp:
2291         (WebCore::SubresourceLoader::SubresourceLoader):
2292         (WebCore::SubresourceLoader::didFinishLoading):
2293         (WebCore::SubresourceLoader::didFail):
2294         (WebCore::SubresourceLoader::didCancel):
2295         * loader/mac/NetscapePlugInStreamLoaderMac.mm:
2296         (WebCore::NetscapePlugInStreamLoader::didFinishLoading):
2297         (WebCore::NetscapePlugInStreamLoader::didFail):
2298         (WebCore::NetscapePlugInStreamLoader::didCancel):
2299
2300 2007-03-06  Geoffrey Garen  <ggaren@apple.com>
2301
2302         Reviewed by Maciej Stachowiak.
2303
2304         Fixed <rdar://problem/4576242> | http://bugs.webkit.org/show_bug.cgi?id=12586
2305         PAC file: malloc deadlock sometimes causes a hang @ www.apple.com/pro/profiles/ (12586)
2306
2307         No test because this is very difficult to repro, and the new ASSERTs in
2308         JavaScriptCore catch the underlying cause while running normal layout tests.
2309
2310         This is a modified version of r14752 on the branch.
2311
2312         The fix is to use a bit inside each node, instead of a hash table, to track
2313         which node subtrees are in the process of being marked. This avoids a call
2314         to malloc inside mark().
2315
2316         * bindings/js/kjs_binding.cpp:
2317         (KJS::domObjects):
2318         (KJS::domNodesPerDocument):
2319         * bindings/js/kjs_dom.cpp:
2320         (KJS::DOMNode::mark):
2321         * dom/Node.cpp:
2322         (WebCore::Node::Node):
2323         * dom/Node.h:
2324
2325 2007-03-06  David Hyatt  <hyatt@apple.com>
2326
2327         This patch reworks the WebCore memory cache to significantly reduce the amount of memory consumed by
2328         images in the cache and to enhance the accuracy of the cache size as an absolute bound for the objects
2329         contained within it.  WebCore's memory use over time should significantly improve as a result of these
2330         changes.
2331
2332         Cached resources now have both an encoded size (the original data stream) and a decoded size (an estimate of
2333         the amount of memory consumed by an expanded version of that resource, e.g., the decoded frames of an image).
2334         Both sizes now count towards the total size of the object and towards the allowed memory cache total.
2335
2336         By including both totals the reported size of resources will now be larger, and the cache will therefore become
2337         much more aggressive about flushing.
2338
2339         Objects are stored in size-adjusted and popularity-aware LRU lists as before, but encoded size is now always
2340         used when determining the correct LRU list.
2341
2342         The flush algorithm for the memory cache has been rewritten to first destroy decoded data before evicting
2343         resources.  By being able to compact its resources without evicting them, the memory cache can now hold many more
2344         unique resources (encoded) in the same amount of space.  Depending on how much of a hit we want to take from
2345         re-decoding images, the memory cache could in theory have its size significantly reduced now while still holding
2346         more resources than it did at the larger size!
2347
2348         Reviewed by mjs
2349
2350         * WebCore.xcodeproj/project.pbxproj:
2351         * loader/Cache.cpp:
2352         (WebCore::Cache::requestResource):
2353         (WebCore::Cache::prune):
2354         (WebCore::Cache::remove):
2355         (WebCore::Cache::lruListFor):
2356         (WebCore::Cache::adjustSize):
2357         * loader/Cache.h:
2358         * loader/CachedCSSStyleSheet.cpp:
2359         (WebCore::CachedCSSStyleSheet::data):
2360         * loader/CachedImage.cpp:
2361         (WebCore::CachedImage::CachedImage):
2362         (WebCore::CachedImage::allReferencesRemoved):
2363         (WebCore::CachedImage::clear):
2364         (WebCore::CachedImage::data):
2365         (WebCore::CachedImage::destroyDecodedData):
2366         (WebCore::CachedImage::decodedSize):
2367         (WebCore::CachedImage::decodedSizeChanged):
2368         (WebCore::CachedImage::shouldPauseAnimation):
2369         * loader/CachedImage.h:
2370         * loader/CachedResource.cpp:
2371         (WebCore::CachedResource::CachedResource):
2372         (WebCore::CachedResource::deref):
2373         (WebCore::CachedResource::setEncodedSize):
2374         * loader/CachedResource.h:
2375         (WebCore::CachedResource::allReferencesRemoved):
2376         (WebCore::CachedResource::size):
2377         (WebCore::CachedResource::encodedSize):
2378         (WebCore::CachedResource::decodedSize):
2379         (WebCore::CachedResource::destroyDecodedData):
2380         * loader/CachedScript.cpp:
2381         (WebCore::CachedScript::data):
2382         * loader/CachedXSLStyleSheet.cpp:
2383         (WebCore::CachedXSLStyleSheet::data):
2384         * platform/graphics/BitmapImage.cpp:
2385         (WebCore::BitmapImage::BitmapImage):
2386         (WebCore::BitmapImage::~BitmapImage):
2387         (WebCore::BitmapImage::destroyDecodedData):
2388         (WebCore::BitmapImage::pruneDecodedDataIfNeeded):
2389         (WebCore::BitmapImage::cacheFrame):
2390         (WebCore::BitmapImage::setNativeData):
2391         (WebCore::BitmapImage::shouldAnimate):
2392         (WebCore::BitmapImage::advanceAnimation):
2393         * platform/graphics/BitmapImage.h:
2394         (WebCore::BitmapImage::decodedSize):
2395         * platform/graphics/Image.cpp:
2396         (WebCore::Image::Image):
2397         * platform/graphics/Image.h:
2398         (WebCore::Image::destroyDecodedData):
2399         (WebCore::Image::decodedSize):
2400         (WebCore::Image::imageObserver):
2401         * platform/graphics/ImageAnimationObserver.h: Removed.
2402         * platform/graphics/ImageObserver.h: Added.
2403         (WebCore::ImageObserver::~ImageObserver):
2404         * platform/graphics/svg/SVGImage.cpp:
2405         (WebCore::SVGImage::SVGImage):
2406         * platform/graphics/svg/SVGImage.h:
2407
2408 2007-03-06  Alexey Proskuryakov  <ap@webkit.org>
2409
2410         Reviewed by Sam Weinig.
2411
2412         http://bugs.webkit.org/show_bug.cgi?id=12987
2413         Fix and import 4XPath test_numeric_expr.html
2414
2415         * xml/XPathPredicate.cpp:
2416         (WebCore::XPath::Negative::doEvaluate): Convert the argument to number.
2417         (WebCore::XPath::NumericOp::doEvaluate): Convert the arguments to numbers. Use a correct operation for mod.
2418         * xml/XPathStep.cpp:
2419         (WebCore::XPath::Step::nodesInAxis): Do not append parent node if there is none.
2420         * xml/XPathValue.cpp:
2421         (WebCore::XPath::Value::toNumber): Do not convert to DeprecatedString just to trim whitespace and to convert to double.
2422         * platform/DeprecatedString.cpp:
2423         (WebCore::DeprecatedStringData::makeAscii): Added a FIXME about unreliable makeAscii() behavior.
2424
2425 2007-03-06  Maciej Stachowiak  <mjs@apple.com>
2426
2427         Reviewed by Adele.
2428
2429         <rdar://problem/4619663> REGRESSION (NativePopup): Popup menu doesn't draw at the correct vertical position (9816)
2430
2431         * platform/mac/PopupMenuMac.mm:
2432         (WebCore::PopupMenu::show): Make a temporary dummy view with the
2433         passed in rect, since AppKit will use the view bounds to determine
2434         what area to exclude when popping up a menu moved to the top of
2435         the screen.
2436
2437 2007-03-06  Geoffrey Garen  <ggaren@apple.com>
2438
2439         Reviewed by Maciej Stachowiak.
2440
2441         Fixed all known crashers exposed by run-webkit-tests --threaded [*]. See
2442         JavaScriptCore ChangeLog for more details.
2443
2444         * bindings/js/kjs_binding.cpp:
2445         (KJS::domNodesPerDocument): Added thread safety ASSERT.
2446         (KJS::ScriptInterpreter::mark): Removed obsolete logic for marking unsafe
2447         objects when collecting on a secondary thread. The Collector takes care
2448         of this now.
2449
2450         * bindings/js/kjs_binding.h:
2451         (KJS::DOMObject::DOMObject): Used new API for specifying that WebCore
2452         objects should be garbage collected on the main thread only.
2453
2454         * bindings/js/kjs_window.cpp:
2455         (KJS::ScheduledAction::execute): Moved JSLock to cover implementedsCall() call,
2456         which, for some subclasses, ends up allocating garbage collected objects.
2457         (This fix was speculative. I didn't actually see a crash from this.)
2458         (KJS::Window::timerFired): Added JSLock around ScheduleAction destruction,
2459         since it destroys a KJS::List.
2460
2461         * bindings/objc/WebScriptObject.mm:
2462         (-[WebScriptObject setException:]): Added JSLock. (This fix was speculative.
2463         I didn't actually see a crash from this.)
2464
2465         * bridge/mac/WebCoreScriptDebugger.mm:
2466         (-[WebCoreScriptCallFrame evaluateWebScript:]): Added JSLock. (This fix
2467         was speculative. I didn't actually see a crash from this.)
2468
2469         * dom/Document.cpp:
2470         (WebCore::Document::~Document): Added JSLock around modification to
2471         domNodesPerDocument(), which can be accessed concurrently during garbage
2472         collection.
2473         * dom/Node.cpp:
2474         (WebCore::Node::setDocument): ditto.
2475
2476         [*] fast/js/toString-stack-overflow.html is an exception. --threaded mode
2477         crashes this test because it causes the garbage collector to run frequently,
2478         and this test crashes if you happen to garbage collect while it's running.
2479         This is a known issue with stack overflow during the mark phase. It's
2480         not related to threading.
2481
2482 2007-03-06  Mark Rowe  <mrowe@apple.com>
2483
2484         Reviewed by Sam Weinig.
2485
2486         Fix http://bugs.webkit.org/show_bug.cgi?id=12942
2487         Bug 12942: ASSERTION FAILURE: qantas.com.au changing selected item in <select> via JS
2488
2489         Test: fast/dom/select-selectedIndex-bug-12942.html.
2490
2491         * html/HTMLSelectElement.cpp:
2492         (WebCore::HTMLSelectElement::recalcListItems): Reset m_lastOnChangeIndex when recalculating list items.
2493         * html/HTMLSelectElement.h:
2494
2495 2007-03-06  Brady Eidson  <beidson@apple.com>
2496
2497         Rubberstamped by Kevin Decker
2498
2499         20,000!
2500
2501         * ChangeLog: Point out revision 20,000
2502
2503 2007-03-06  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
2504
2505         Not reviewed.
2506
2507         Gdk build fix.
2508
2509         * loader/gdk/FrameLoaderClientGdk.cpp: update userAgent() signature.
2510         (WebCore::FrameLoaderClientGdk::userAgent):
2511         * loader/gdk/FrameLoaderClientGdk.h: ditto.
2512
2513 2007-03-06  Mitz Pettel  <mitz@webkit.org>
2514
2515         Reviewed by Adele.
2516
2517         - fix http://bugs.webkit.org/show_bug.cgi?id=12986
2518           REGRESSION(NativeListBox): Listboxes not updated when resized dynamically
2519
2520         Test: fast/forms/select-change-listbox-size.html
2521
2522         * html/HTMLSelectElement.cpp:
2523         (WebCore::HTMLSelectElement::parseMappedAttribute): Reattach on list box size change.
2524
2525 2007-03-06  Mitz Pettel  <mitz@webkit.org>
2526
2527         Reviewed by Dave Hyatt.
2528
2529         - fix http://bugs.webkit.org/show_bug.cgi?id=12885
2530           REGRESSION (r19696): Incomplete background repaint
2531
2532         Tests: fast/repaint/content-into-overflow.html
2533                fast/repaint/overflow-into-content.html
2534
2535         Changed repaintAfterLayoutIfNeeded() to take, in addition to the clipped overflow
2536         rect, the unclipped border box plus outline, and to repaint any areas that
2537         were added or removed from that box, in addition to any areas added or removed
2538         from the clipped overflow rect.
2539
2540         * platform/graphics/svg/SVGResourceMarker.cpp:
2541         (WebCore::SVGResourceMarker::draw):
2542         * rendering/RenderBlock.cpp:
2543         (WebCore::RenderBlock::layoutBlock):
2544         * rendering/RenderBox.cpp:
2545         (WebCore::RenderBox::absoluteClippedOverflowRect): Renamed getAbsoluteRepaintRect() to
2546         this.
2547         * rendering/RenderBox.h:
2548         * rendering/RenderFlexibleBox.cpp:
2549         (WebCore::RenderFlexibleBox::layoutBlock):
2550         * rendering/RenderFlow.cpp:
2551         (WebCore::RenderFlow::absoluteClippedOverflowRect):
2552         * rendering/RenderFlow.h:
2553         * rendering/RenderForeignObject.cpp:
2554         (WebCore::RenderForeignObject::layout):
2555         * rendering/RenderHTMLCanvas.cpp:
2556         (WebCore::RenderHTMLCanvas::layout):
2557         * rendering/RenderImage.cpp:
2558         (WebCore::RenderImage::layout):
2559         * rendering/RenderLayer.cpp:
2560         (WebCore::RenderLayer::RenderLayer):
2561         (WebCore::RenderLayer::checkForRepaintOnResize):
2562         (WebCore::RenderLayer::updateLayerPositions):
2563         * rendering/RenderLayer.h:
2564         * rendering/RenderObject.cpp:
2565         (WebCore::RenderObject::repaint):
2566         (WebCore::RenderObject::repaintAfterLayoutIfNeeded):
2567         (WebCore::RenderObject::getAbsoluteRepaintRectWithOutline):
2568         (WebCore::RenderObject::absoluteClippedOverflowRect):
2569         (WebCore::RenderObject::absoluteOutlineBox):
2570         * rendering/RenderObject.h:
2571         * rendering/RenderPath.cpp:
2572         (WebCore::RenderPath::layout):
2573         (WebCore::RenderPath::absoluteClippedOverflowRect):
2574         (WebCore::RenderPath::absoluteRects):
2575         (WebCore::RenderPath::drawMarkersIfNeeded):
2576         * rendering/RenderPath.h:
2577         * rendering/RenderSVGContainer.cpp:
2578         (WebCore::RenderSVGContainer::layout):
2579         (WebCore::RenderSVGContainer::absoluteClippedOverflowRect):
2580         (WebCore::RenderSVGContainer::absoluteRects):
2581         * rendering/RenderSVGContainer.h:
2582         * rendering/RenderSVGHiddenContainer.cpp:
2583         (WebCore::RenderSVGHiddenContainer::absoluteClippedOverflowRect):
2584         * rendering/RenderSVGHiddenContainer.h:
2585         * rendering/RenderSVGImage.cpp:
2586         (WebCore::RenderSVGImage::imageChanged):
2587         (WebCore::RenderSVGImage::absoluteClippedOverflowRect):
2588         (WebCore::RenderSVGImage::absoluteRects):
2589         * rendering/RenderSVGImage.h:
2590         * rendering/RenderSVGText.cpp:
2591         (WebCore::RenderSVGText::absoluteClippedOverflowRect):
2592         (WebCore::RenderSVGText::layout):
2593         (WebCore::RenderSVGText::absoluteRects):
2594         * rendering/RenderSVGText.h:
2595         * rendering/RenderTable.cpp:
2596         (WebCore::RenderTable::layout):
2597         * rendering/RenderTableCell.cpp:
2598         (WebCore::RenderTableCell::absoluteClippedOverflowRect):
2599         * rendering/RenderTableCell.h:
2600         * rendering/RenderTableCol.cpp:
2601         (WebCore::RenderTableCol::absoluteClippedOverflowRect):
2602         * rendering/RenderTableCol.h:
2603         * rendering/RenderTableRow.cpp:
2604         (WebCore::RenderTableRow::absoluteClippedOverflowRect):
2605         * rendering/RenderTableRow.h:
2606         * rendering/RenderText.cpp:
2607         (WebCore::RenderText::absoluteClippedOverflowRect):
2608         * rendering/RenderText.h:
2609
2610 2007-03-06  Ian Eng <ian.eng.webkit@gmail.com>
2611
2612         Reviewed by Maciej.
2613
2614         - fixed http://bugs.webkit.org/show_bug.cgi?id=12720
2615         Bug 12720: Re-defining window.location.toString function keeps re-loading forever
2616
2617         * bindings/js/kjs_window.cpp:
2618         Disallow replacing functions in LocationTable, and return early without updating URL.
2619
2620 2007-03-06  Kevin McCullough  <kmccullough@apple.com>
2621
2622         Reviewed by Darin.
2623
2624         <http://bugs.webkit.org/show_bug.cgi?id=12686>
2625         REGRESSION: Bloglines.com Feeds tab cannot expand folders in TOT
2626         - Now all class constructors implement implementsHasInstance.
2627
2628         * bindings/scripts/CodeGeneratorJS.pm:
2629
2630 2007-03-07  Nikolas Zimmermann  <zimmermann@kde.org>
2631
2632         Reviewed by Rob & Oliver.
2633
2634         Preparations for the new SVG text engine.
2635
2636         Handle baseline-shift / kerning css values correctly.
2637         Recognize missing svg presentation attribute "font-size-adjust".
2638         Add all missing svg<->css property mappings in mapToEntry().
2639
2640         Doesn't affect any layout test (as these properties are not used w/o my text patch).
2641         While I'm at it, unify the macro names (RS_ -> SVG_RS_) & cleanup style a bit.
2642
2643         * ksvg2/css/CSSPropertyNames.in:
2644         * ksvg2/css/SVGCSSParser.cpp:
2645         (WebCore::CSSParser::parseSVGValue):
2646         * ksvg2/css/SVGCSSStyleSelector.cpp:
2647         (WebCore::CSSStyleSelector::applySVGProperty):
2648         * ksvg2/css/SVGRenderStyle.cpp:
2649         (WebCore::SVGRenderStyle::SVGRenderStyle):
2650         (WebCore::SVGRenderStyle::operator==):
2651         (WebCore::SVGRenderStyle::inheritedNotEqual):
2652         (WebCore::SVGRenderStyle::inheritFrom):
2653         * ksvg2/css/SVGRenderStyle.h:
2654         (WebCore::SVGRenderStyle::NonInheritedFlags::):
2655         (WebCore::SVGRenderStyle::setBitDefaults):
2656         * ksvg2/css/SVGRenderStyleDefs.cpp:
2657         (StyleTextData::StyleTextData):
2658         (StyleTextData::operator==):
2659         (StyleMiscData::StyleMiscData):
2660         (StyleMiscData::operator==):
2661         * ksvg2/css/SVGRenderStyleDefs.h:
2662         (WebCore::):
2663         (WebCore::StyleTextData::operator!=):
2664         * ksvg2/svg/SVGStyledElement.cpp:
2665         (WebCore::SVGStyledElement::cssPropertyIdForSVGAttributeName):
2666
2667 2007-03-06  Justin Garcia  <justin.garcia@apple.com>
2668
2669         Reviewed by kevin
2670
2671         <http://bugs.webkit.org/show_bug.cgi?id=12245>
2672         FCKeditor: Remove Format sometimes doesn't work
2673         <rdar://problem/4786404>
2674         Underline style is not removed from selection after performing Remove Format
2675
2676         * editing/Editor.cpp:
2677         (WebCore::Editor::removeFormattingAndStyle): Re-wrote this.
2678
2679 2007-03-07  Nikolas Zimmermann  <zimmermann@kde.org>
2680
2681         Reviewed by Oliver.
2682
2683         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12979
2684         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12981
2685
2686         Guard against registering pending resources with empty id.
2687         Be careful with calling recalcStyle() in SVGUseElement, when change is "Detach" we
2688         have to use special code, as calling attach() on the shadow tree root element will crash,
2689         because it has no (direct) parent node, only a shadow parent node element.
2690
2691         * ksvg2/misc/SVGDocumentExtensions.cpp:
2692         (WebCore::SVGDocumentExtensions::addPendingResource):
2693         * ksvg2/svg/SVGUseElement.cpp:
2694         (WebCore::SVGUseElement::recalcStyle):
2695
2696 2007-03-06  Kevin Decker  <kdecker@apple.com>
2697
2698         Reviewed by Adele.
2699
2700         Fixed: <rdar://problem/5041660> REGRESSION: <keygen> element broken, prevents users from signing up for Thawte email certs
2701
2702         * bindings/objc/DOM.mm:
2703         (WebCore::createElementClassMap): Added <keygen> to the DOM bindings so it can be accessed from Objective-C.
2704         * css/html4.css: Apply the look of the <select> element to <keygen>.
2705         * html/HTMLElementFactory.cpp:
2706         (WebCore::keygenConstructor): Added.
2707         (WebCore::createFunctionMap): Added keygen.
2708         * html/HTMLSelectElement.cpp: Made const typeAheadTimeout variable static const.
2709         (WebCore::HTMLSelectElement::HTMLSelectElement):  Fix both HTMLSelectElement constructors to initialize the
2710          same number of member variables. The fact that some fields were not initialized could (and would) crash the
2711          keygen element when selecting different items. Also removed m_typedString(String()) from the constructor
2712          initialization because this is not needed.
2713
2714 2007-03-06  Kevin McCullough  <kmccullough@apple.com>
2715
2716         Reviewed by Darin.
2717
2718         - Rename a function to clarify its purpose.
2719
2720         * WebCore.exp:
2721         * loader/FrameLoader.cpp:
2722         (WebCore::FrameLoader::registerURLSchemeAsLocal):
2723         * loader/FrameLoader.h:
2724
2725 2007-03-06  Adam Roben  <aroben@apple.com>
2726
2727         Build fix.
2728
2729         * WebCore.exp: Updated symbols.
2730
2731 2007-03-06  Adam Roben  <aroben@apple.com>
2732
2733         Reviewed by Anders.
2734
2735         Added a parameter to all StringTruncator methods to specify whether
2736         rounding hacks should be on or off.
2737
2738         No layout test possible.
2739
2740         * platform/StringTruncator.cpp:
2741         (WebCore::stringWidth): Added disableRoundingHacks parameter.
2742         (WebCore::truncateString): Ditto.
2743         (WebCore::StringTruncator::centerTruncate): Ditto.
2744         (WebCore::StringTruncator::rightTruncate): Ditto.
2745         (WebCore::StringTruncator::width): Ditto.
2746         * platform/StringTruncator.h: Ditto.
2747         * platform/mac/FileChooserMac.mm:
2748         (WebCore::FileChooser::basenameForWidth): Pass in false to
2749         centerTruncate so that the truncation matches the way the text will be
2750         rendered.
2751
2752 2007-03-06  Anders Carlsson  <acarlsson@apple.com>
2753
2754         Reviewed by Maciej.
2755
2756         <rdar://problem/5035045>
2757         REGRESSION: WebKit browser doesn't display image at http://www.metoffice.gov.uk/weather/satellite/index.html
2758
2759         It turns out WinIE does allow you to access images by their id as special document properties. However, this is only
2760         allowed when the element also has a name attribute. The value of the name attribute is ignored and can even be empty!
2761
2762         * bindings/js/kjs_html.cpp:
2763         (KJS::JSHTMLDocument::namedItemGetter):
2764         Return jsUndefined() if the collection is empty.
2765
2766         * html/HTMLImageElement.cpp:
2767         (WebCore::HTMLImageElement::parseMappedAttribute):
2768         (WebCore::HTMLImageElement::insertedIntoDocument):
2769         (WebCore::HTMLImageElement::removedFromDocument):
2770         * html/HTMLImageElement.h:
2771         Add the id attribute value to the extra named item map.
2772
2773         * html/HTMLNameCollection.cpp:
2774         (WebCore::HTMLNameCollection::traverseNextItem):
2775         Check for images with name attributes that match, as well as elements with id attributes that match where
2776         the element also has a name attribute.
2777
2778 2007-03-06  Anders Carlsson  <acarlsson@apple.com>
2779
2780         Reviewed by Adam.
2781
2782         WebCore part of patch to make it possible to have different user agents for different URLs.
2783
2784         * bindings/js/kjs_navigator.cpp:
2785         (KJS::Navigator::getValueProperty):
2786         * bindings/js/kjs_proxy.cpp:
2787         (WebCore::KJSProxy::initScriptIfNeeded):
2788         * loader/DocumentLoader.cpp:
2789         (WebCore::DocumentLoader::setLoading):
2790         * loader/FrameLoader.cpp:
2791         (WebCore::FrameLoader::userAgent):
2792         (WebCore::FrameLoader::loadResourceSynchronously):
2793         (WebCore::FrameLoader::applyUserAgent):
2794         * loader/FrameLoader.h:
2795         * loader/FrameLoaderClient.h:
2796         * platform/graphics/svg/SVGImageEmptyClients.h:
2797         (WebCore::SVGEmptyFrameLoaderClient::userAgent):
2798
2799 2007-03-05  Kevin McCullough  <kmccullough@apple.com>
2800
2801         Reviewed by Mark and Dave H.
2802
2803         - rdar://problem/5038491
2804         An oversight of the security fix that prevented remote from loading local is that it
2805         prevents user style sheets when the site is remote.  This fixes that.
2806
2807         * loader/Cache.cpp: Propogate and check user style sheet flag.
2808         (WebCore::createResource):
2809         (WebCore::Cache::requestResource):
2810         * loader/Cache.h: Propogate user style sheet flag.
2811         * loader/CachedCSSStyleSheet.cpp: Propogate user style sheet flag.
2812         (WebCore::CachedCSSStyleSheet::CachedCSSStyleSheet):
2813         * loader/CachedCSSStyleSheet.h: Propogate user style sheet flag.
2814         * loader/DocLoader.cpp: Propogate user style sheet flag.
2815         (WebCore::DocLoader::requestResource):
2816         * loader/SubresourceLoader.cpp: Propogate and check user style sheet flag.
2817         (WebCore::SubresourceLoader::create):
2818         * loader/SubresourceLoader.h: Add check for user style sheet flag.
2819         * loader/loader.cpp: Propogate user style sheet flag.
2820         (WebCore::Loader::load):
2821         (WebCore::Loader::servePendingRequests):
2822         * loader/loader.h: Propogate user style sheet flag.
2823
2824 2007-03-06  Nikolas Zimmermann  <zimmermann@kde.org>
2825
2826         Reviewed by Darin.
2827
2828         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12936 (Master bug used to track all current use problems.)
2829
2830         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12926 (transform attribute not respected in nested <use> elements)
2831         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12267 (getElementById broken for <use>)
2832         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12916 (use instance in symbol definition does not work)
2833         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12917 (mouseout event does not occur after scaling use instance)
2834         Fixes: http://bugs.webkit.org/show_bug.cgi?id=12838 (SVG <use> DOM broken for script)
2835
2836         Added test: svg/custom/use-nested-transform.svg
2837         Added test: svg/custom/struct-use-09-b.svg (not yet official W3C-SVG-1.1 testcase)
2838         Fixes test: svg/custom/use-elementInstance-event-target.svg (rectangle now turns green on first click)
2839
2840         Fix all (except one) known <use> bug. The remaining bug (12630) is not crucical at all.
2841         Enable <use> again as default SVG feature, as discussed on webkit-dev.
2842
2843         dom/Element.cpp needed following tweak: set hasParentStyle to true, if there is no parentNode
2844         available - which happens for <use> nodes, as it's a shadow node. This fixes recalcStyle behaviour.
2845         Override recalcStyle() in SVGUseElement, and properly forward the call to the shadow tree root element.
2846         The shadow tree now receives proper style updates (without having to recreate the whole tree!).
2847
2848         Override attributeChanged() in SVGUseElement and only call notifyAttributeChange if one of x/y/width/height/xlink:href
2849         attribute changed - otherwhise we'll end up recreating the tree for every transform/style/(non-existing-attribute) change.
2850
2851         Do not override transform in nested use situations, but correctly append (right-sided) the translation of the <use> element.
2852         Some cosmetic fixes: don't add transform="translate(0 0)" attributes if both x/y values are null in the shadow tree.
2853
2854         Factor out logic for replacing symbol/svg tags in the shadow tree, and also invoke it during expandUseElementsInShadowTree -
2855         otherwhise <symbol><use xlink:href="#someOtherSymbol"></symbol> the <use> gets expanded to a <symbol>. Though no one expands
2856         the <symbol> element - and we're end up in hitting an assertion. Avoid that.
2857
2858         * bindings/js/JSSVGElementWrapperFactory.cpp:
2859         * bindings/js/kjs_dom.cpp:
2860         (KJS::toJS):
2861         * bindings/objc/DOM.mm:
2862         (WebCore::createElementClassMap):
2863         * dom/Element.cpp:
2864         (WebCore::Element::recalcStyle):
2865         * ksvg2/svg/SVGElement.cpp:
2866         (WebCore::shadowTreeParentElementForShadowTreeElement):
2867         (WebCore::SVGElement::dispatchEvent):
2868         * ksvg2/svg/SVGElementInstance.cpp:
2869         * ksvg2/svg/SVGElementInstance.h:
2870         * ksvg2/svg/SVGElementInstance.idl:
2871         * ksvg2/svg/SVGElementInstanceList.cpp:
2872         * ksvg2/svg/SVGElementInstanceList.h:
2873         * ksvg2/svg/SVGElementInstanceList.idl:
2874         * ksvg2/svg/SVGStyledElement.cpp:
2875         (WebCore::SVGStyledElement::notifyAttributeChange):
2876         (WebCore::SVGStyledElement::updateElementInstance):
2877         * ksvg2/svg/SVGStyledElement.h:
2878         * ksvg2/svg/SVGUseElement.cpp:
2879         (WebCore::SVGUseElement::attributeChanged):
2880         (WebCore::SVGUseElement::notifyAttributeChange):
2881         (WebCore::SVGUseElement::recalcStyle):
2882         (WebCore::SVGUseElement::buildPendingResource):
2883         (WebCore::SVGUseElement::buildShadowTreeForSymbolTag):
2884         (WebCore::SVGUseElement::alterShadowTreeForSVGTag):
2885         (WebCore::SVGUseElement::buildShadowTree):
2886         (WebCore::SVGUseElement::expandUseElementsInShadowTree):
2887         (WebCore::SVGUseElement::attachShadowTree):
2888         * ksvg2/svg/SVGUseElement.h:
2889         * ksvg2/svg/SVGUseElement.idl:
2890         * ksvg2/svg/svgtags.in:
2891
2892 2007-03-05  Brady Eidson  <beidson@apple.com>
2893
2894         Reviewed by John
2895
2896         Fixes <rdar://problem/4974258>
2897         Adds some key null checking
2898
2899         * bindings/js/kjs_html.cpp:
2900         (KJS::JSHTMLElement::implementsCall): Null check doc/frame
2901         * bindings/objc/DOMInternal.mm:
2902         (-[WebScriptObject _initializeScriptDOMNodeImp]): Null check doc/frame
2903
2904 2007-03-06  Nikolas Zimmermann  <zimmermann@kde.org>
2905
2906         Reviewed by Darin.
2907
2908         Path::normalAngleAtLength() / Path::pointAtLength() don't work correctly.
2909         pointAtLength() was not implemented, basically and normalAngleAtLength()
2910         had a bug in the tangent slope calculation.
2911
2912         The normalAngleAtLength() stuff can only be tested with my local textPath support.
2913         New LayoutTest: svg/custom/path-textPath-simulation.svg
2914
2915         * platform/graphics/Path.cpp:
2916         (WebCore::pathLengthApplierFunction):
2917         * platform/graphics/PathTraversalState.cpp:
2918         (WebCore::PathTraversalState::quadraticBezierTo):
2919         (WebCore::PathTraversalState::cubicBezierTo):
2920         * platform/graphics/PathTraversalState.h:
2921         (WebCore::PathTraversalState::):
2922
2923 2007-03-05  Alexey Proskuryakov  <ap@webkit.org>
2924
2925         Reviewed by Darin.
2926
2927         http://bugs.webkit.org/show_bug.cgi?id=12970
2928         Fix and import 4XPath test_core_functions.html test
2929
2930         * xml/XPathExpression.cpp:
2931         (WebCore::XPathExpression::evaluate): Fully initialize the evaluation context.
2932
2933         * xml/XPathFunctions.cpp:
2934         (WebCore::XPath::FunSubstring::doEvaluate): Fixed handling of edge cases.
2935         (WebCore::XPath::FunRound::round): Reimplemented to match the spec; exposed FunRound::round() to be used in
2936         other functions.
2937
2938 2007-03-05  Alexey Proskuryakov  <ap@webkit.org>
2939
2940         Reviewed by Darin.
2941
2942         http://bugs.webkit.org/show_bug.cgi?id=12954
2943         XPath relative operations are implemented incorrectly
2944
2945         * xml/XPathPredicate.cpp:
2946         (WebCore::XPath::NumericOp::doEvaluate):
2947         (WebCore::XPath::EqTestOp::compare):
2948         (WebCore::XPath::EqTestOp::doEvaluate):
2949         Reimplemented relative equality operations to match the spec.
2950
2951         * xml/XPathPredicate.h:
2952         (WebCore::XPath::NumericOp::):
2953         (WebCore::XPath::EqTestOp::):
2954         Moved relative operations to EqTestOp.
2955
2956         * xml/XPathGrammar.y:
2957         * xml/XPathParser.cpp:
2958         (WebCore::XPath::Parser::nextTokenInternal):
2959         (WebCore::XPath::Parser::lex):
2960         Adapted for the above changes.
2961
2962 2007-03-05  Mark Rowe  <mrowe@apple.com>
2963
2964         Reviewed by Lars.
2965
2966         Fix http://bugs.webkit.org/show_bug.cgi?id=12947
2967         Bug 12947: REGRESSION: ASSERTION FAILED: maxWidth >= 0 in StringTruncator.cpp:109 in WebCore::truncateString()
2968
2969         Handle nil window correctly in toUserSpace and toDeviceSpace.  On Intel Macs a message to nil that returns a
2970         float will return 0.0.  We use this as the divisor in calculating a scale factor, which results in NaN being
2971         introduced into our rect.
2972
2973         * platform/mac/ScreenMac.mm:
2974         (WebCore::toUserSpace):
2975         (WebCore::toDeviceSpace):
2976
2977 2007-03-05  Rob Buis  <buis@kde.org>
2978
2979         Reviewed by Darin.
2980
2981         http://bugs.webkit.org/show_bug.cgi?id=12868
2982         parts of the CSS classes in this simple SVG example are not applied
2983
2984         Make sure the xml stylesheets are parsed in strict mode.
2985
2986         * dom/ProcessingInstruction.cpp:
2987         (WebCore::ProcessingInstruction::parseStyleSheet):
2988
2989 2007-03-04  Adele Peterson  <adele@apple.com>
2990
2991         Reviewed by Darin.
2992
2993         Change to dispatch the keypress event during the defaultEventHandler for keydown events.  This matches IE behavior.
2994         This is preparation for fixing event dispatch with input methods (http://bugs.webkit.org/show_bug.cgi?id=10871)
2995
2996         Test: fast/events/keydown-keypress-preventDefault.html
2997
2998         * dom/EventTargetNode.cpp: (WebCore::EventTargetNode::defaultEventHandler): Call the defaultKeyboardEventHandler for keydown events.
2999         * page/EventHandler.cpp:
3000         (WebCore::eventTargetNodeForDocument): Return 0 instead of false since the return type in EventTargetNode.
3001         (WebCore::EventHandler::keyEvent): Removed dispatch of keypress event, since this is now done in the default event handler.
3002         (WebCore::EventHandler::defaultKeyboardEventHandler): For keydown events, create and dispatch a keypress event.
3003
3004 2007-03-04  Alexey Proskuryakov  <ap@webkit.org>
3005
3006         Reviewed by Nikolas Zimmermann (yay!).
3007
3008         http://bugs.webkit.org/show_bug.cgi?id=12962
3009         4XPath tests crash on lang() function
3010
3011         Covered by 4XPath tests, to be landed later.
3012
3013         * platform/StringImpl.cpp:
3014         (WebCore::StringImpl::reverseFind): Do not crash with empty strings.
3015         * xml/XPathFunctions.cpp:
3016         (WebCore::XPath::FunLang::doEvaluate): Do not crash when an element has no
3017         attributes. Use a proper namespace for xml:lang (not sure where "xms" came from).
3018         Rewrote the algorithm for suffix removing to match the spec.
3019
3020 2007-03-02  Anders Carlsson  <acarlsson@apple.com>
3021
3022         Reviewed by Darin.
3023
3024         <rdar://problem/5028165>
3025         http://bugs.webkit.org/show_bug.cgi?id=12915
3026         REGRESSION: XMLHttpRequest.abort() does not stop loading (12915)
3027
3028         * loader/DocumentLoader.cpp:
3029         (WebCore::DocumentLoader::stopLoading):
3030         Save the value of m_loading since calling FrameLoader::stopLoading could set it to false.
3031
3032         * loader/SubresourceLoader.cpp:
3033         (WebCore::SubresourceLoader::didCancel):
3034         * loader/SubresourceLoader.h:
3035         Get rid of didCancel now, it's not needed anymore.
3036
3037         * xml/xmlhttprequest.cpp:
3038         (WebCore::XMLHttpRequest::abort):
3039         Call cancel() instead of stopLoading(). Also, set m_aborted to true so the XMLHttpRequest object
3040         won't be dereferenced in didFail when aborting.
3041
3042 2007-03-04  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
3043
3044         Reviewed by Nikolas Zimmermann.
3045
3046         Move ScrollView stubs to ScrollViewGdk.cpp
3047
3048         * platform/gdk/FrameGdk.h: remove comment that no longer makes sense
3049         * platform/gdk/ScrollViewGdk.cpp:
3050         (WebCore::ScrollView::addChild):
3051         (WebCore::ScrollView::removeChild):
3052         (WebCore::ScrollView::scrollPointRecursively):
3053         (WebCore::ScrollView::inWindow):
3054         (WebCore::ScrollView::wheelEvent):
3055         (WebCore::ScrollView::updateScrollbars):
3056         (WebCore::ScrollView::updateScrollInfo):
3057         (WebCore::ScrollView::windowToContents):
3058         (WebCore::ScrollView::contentsToWindow):
3059         (WebCore::ScrollView::scrollbarUnderMouse):
3060         * platform/gdk/TemporaryLinkStubs.cpp:
3061
3062 2007-03-02  Kevin McCullough  <kmccullough@apple.com>
3063
3064         Reviewed by Geoff.
3065
3066         - rdar://problem/4922454
3067         - This fixes a security issue by making remote referrers not able to access local
3068         resources, unless they register their schemes to be treated as local. The result is
3069         that those schemes can access local resources and cannot be accessed by remote
3070         referrers.
3071         Because this behavior is new a link-on-or-after check is made to determine if the
3072         app should use the older, less safe, behavior.
3073
3074         * WebCore.exp: added exported functions
3075         * bindings/objc/DOM.mm: consolodated function to base class
3076         (-[DOMElement image]):
3077         (-[DOMElement _imageTIFFRepresentation]):
3078         * dom/Document.cpp: Cache the document's ability to load local resources.
3079         (WebCore::Document::Document):
3080         (WebCore::Document::setURL):
3081         (WebCore::Document::shouldBeAllowedToLoadLocalResources):
3082         (WebCore::Document::stylesheetLoaded):
3083         * dom/Document.h: Cache the docuent's ability to load local resources.
3084         (WebCore::Document::getPendingSheet):
3085         (WebCore::Document::isAllowedToLoadLocalResources):
3086         * html/HTMLImageLoader.cpp: Moved functionality into base class.
3087         (WebCore::HTMLImageLoader::updateFromElement):
3088         (WebCore::HTMLImageLoader::dispatchLoadEvent):
3089         * html/HTMLLinkElement.cpp: Handles null returns correctly now.
3090         * html/HTMLTokenizer.cpp: Moved functionality into base class.
3091         (WebCore::HTMLTokenizer::notifyFinished):
3092         * ksvg2/misc/SVGImageLoader.cpp: Moved functionality into base class.
3093         (WebCore::SVGImageLoader::dispatchLoadEvent):
3094         * loader/Cache.cpp: Checks if the cached resource can be loaded.
3095         (WebCore::Cache::requestResource):
3096         * loader/CachedCSSStyleSheet.cpp: Moved functionality into base class.
3097         (WebCore::CachedCSSStyleSheet::ref):
3098         (WebCore::CachedCSSStyleSheet::error):
3099         * loader/CachedImage.cpp: Moved functionality into base class.
3100         (WebCore::CachedImage::CachedImage):
3101         * loader/CachedImage.h: Moved functionality into base class.
3102         (WebCore::CachedImage::canRender):
3103         * loader/CachedResource.cpp: Cache if the CachedResource should be treated as local
3104         (WebCore::CachedResource::CachedResource):
3105         * loader/CachedResource.h: Moved functionality into base class.
3106         (WebCore::CachedResource::errorOccurred):
3107         (WebCore::CachedResource::shouldTreatAsLocal):
3108         * loader/CachedScript.cpp: Moved functionality into base class.
3109         (WebCore::CachedScript::CachedScript):
3110         * loader/CachedScript.h: Moved functionality into base class.
3111         (WebCore::CachedScript::schedule):
3112         * loader/CachedXBLDocument.cpp: Moved functionality into base class.
3113         (WebCore::CachedXBLDocument::error):
3114         * loader/CachedXSLStyleSheet.cpp: Moved functionality into base class.
3115         (WebCore::CachedXSLStyleSheet::error):
3116         * loader/FrameLoader.cpp: See comments for each function below.
3117         (WebCore::FrameLoader::loadSubframe): Use new canLoad.
3118         (WebCore::FrameLoader::restrictAccessToLocal): return value of linked-on-or-after check.
3119         (WebCore::FrameLoader::setRestrictAccessToLocal): set value for linked-on-or-after check.
3120         (WebCore::localSchemes): Return set of schemes that are to be treated as local.
3121         (WebCore::FrameLoader::loadPlugin): Use new canLoad.
3122         (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.
3123         (WebCore::FrameLoader::shouldHideReferrer): Extracted out the logic to determine if the referrer should be hidden so it is only calculated when needed.
3124         (WebCore::FrameLoader::loadResourceSynchronously): No longer calls canLoad to get hideReferrer info.
3125         (WebCore::FrameLoader::registerSchemeAsLocal): Functionality to register a scheme to be treated as local.
3126         (WebCore::FrameLoader::treatURLAsLocal): Given a URL this function determines if it should be treated as local.
3127         * loader/FrameLoader.h: Declared functions for this security fix.  See above.
3128         * loader/MainResourceLoader.cpp: Optized order of bools to regain performance.
3129         (WebCore::MainResourceLoader::continueAfterContentPolicy):
3130         * loader/SubresourceLoader.cpp: Now restricts remote from loading local resources.
3131         (WebCore::SubresourceLoader::create):
3132         * page/EventHandler.cpp: Moved functionality into base class.
3133         (WebCore::selectCursor):
3134         * platform/KURL.cpp: KURLs need to check all the registered schemes now.
3135         (WebCore::KURL::isLocalFile):
3136         * rendering/HitTestResult.cpp: Moved functionality into base class.
3137         (WebCore::HitTestResult::image):
3138         * rendering/RenderImage.cpp: Moved functionality into base class.
3139         (WebCore::RenderImage::setCachedImage):
3140         (WebCore::RenderImage::imageChanged):
3141         (WebCore::RenderImage::paint):
3142         (WebCore::RenderImage::layout):
3143         (WebCore::RenderImage::calcAspectRatioWidth):
3144         (WebCore::RenderImage::calcAspectRatioHeight):
3145         * rendering/RenderImage.h: Moved functionality into base class.
3146         (WebCore::RenderImage::errorOccurred):
3147         * rendering/RenderListItem.cpp: Moved functionality into base class.
3148         (WebCore::RenderListItem::setStyle):
3149         * rendering/RenderListMarker.cpp: Moved functionality into base class.
3150         (WebCore::RenderListMarker::isImage):
3151         * xml/xmlhttprequest.cpp: Check doc's cached value instead of determining independently.
3152         (WebCore::XMLHttpRequest::urlMatchesDocumentDomain):
3153
3154 2007-03-02  Justin Garcia  <justin.garcia@apple.com>
3155
3156         Reviewed by kevin
3157
3158         <rdar://problem/5028447>
3159         REGRESSION: Gmail Editor: Copied message text pastes at the wrong font size
3160
3161         * editing/markup.cpp:
3162         (WebCore::createMarkup): The style of the div that holds
3163         a fully selected body's styles didn't include styles inherited
3164         from the body's ancestors.
3165
3166 2007-03-02  Justin Garcia  <justin.garcia@apple.com>
3167
3168         Reviewed by harrison
3169
3170         <rdar://problem/4545040>
3171         innerHTML does not HTML-escape text nodes inside PRE elements
3172         <rdar://problem/5027857>
3173         Pasting into Mail from Safari's view-source window renders the HTML
3174
3175         * editing/HTMLInterchange.cpp:
3176         (WebCore::convertHTMLTextToInterchangeFormat): Send this function
3177         the node that the text comes from as a parameter.  It shouldn't convert
3178         '\n's to spaces/nbsps if the text is coming from text where newlines are
3179         preserved.
3180         * editing/HTMLInterchange.h:
3181         * editing/markup.cpp:
3182         (WebCore::startMarkup): Escape text inside the children of PREs.
3183
3184 2007-03-02  Sam Weinig  <sam@webkit.org>
3185
3186         Reviewed by Anders.
3187
3188         Try to fix the Qt build.
3189
3190         * platform/qt/TemporaryLinkStubs.cpp: Add stubs.
3191         (WebCore::searchMenuNoRecentSearchesText):
3192         (WebCore::searchMenuRecentSearchesText):
3193         (WebCore::searchMenuClearRecentSearchesText):
3194         (WebCore::AXWebAreaText):
3195         (WebCore::AXLinkText):
3196         (WebCore::AXListMarkerText):
3197         (WebCore::AXImageMapText):
3198         (WebCore::AXHeadingText):
3199
3200 2007-03-02  David Harrison  <harrison@apple.com>
3201
3202         Suggested by Darin.
3203
3204         A more efficient solution to rdar://4961431.
3205
3206         * bridge/mac/WebCoreAXObject.mm:
3207         (-[WebCoreAXObject accessibilityIsIgnored]):
3208
3209 2007-03-01  Justin Garcia  <justin.garcia@apple.com>
3210
3211         Reviewed by john
3212
3213         <rdar://problem/5032095>
3214         Gmail Editor: Copied text pastes on a new line instead of current line
3215
3216         Start merge failed to occur because positionAtStartOfInsertedContent
3217         had a bug.
3218
3219         * editing/ReplaceSelectionCommand.cpp:
3220         (WebCore::ReplaceSelectionCommand::positionAtStartOfInsertedContent):
3221         Was failing when inserting <span><div>foo/div></span>. Return the
3222         inserted content's first VisiblePosition.
3223
3224 2007-03-02  Dave Hyatt  <hyatt@apple.com>
3225
3226         Fix crasher in glyph map code (buffer overrun).
3227
3228         Reviewed by darin
3229
3230         * platform/win/GlyphPageTreeNodeWin.cpp:
3231         (WebCore::GlyphPage::fill):
3232
3233 2007-03-01  Antti Koivisto  <antti@apple.com>
3234
3235         Reviewed by Maciej.
3236
3237         Fix for http://bugs.webkit.org/show_bug.cgi?id=12895
3238         REGRESSION: imagemap: pointer cursor is shown everywhere
3239         <rdar://problem/5028163>
3240
3241         Image with imagemap should never itself be URLElement in hit test results.
3242
3243         No layout test, cursor state can't be captured.
3244
3245         * rendering/RenderLayer.cpp:
3246         (WebCore::RenderLayer::hitTest):
3247
3248 2007-03-01  Antti Koivisto  <antti@apple.com>
3249
3250         Reviewed by Maciej.
3251
3252         Fix http://bugs.webkit.org/show_bug.cgi?id=12690
3253         REGRESSION: can not log in to bank of america with TOT webkit
3254         <rdar://problem/4990044>
3255
3256         and http://bugs.webkit.org/show_bug.cgi?id=12604
3257         REGRESSION: After closing the "Would you like to save password" sheet, the form fails
3258         to submit automatically at http://www.mac.com/WebObjects/HomePage.woa
3259         <rdar://problem/4871752>
3260
3261         and http://bugs.webkit.org/show_bug.cgi?id=12020
3262         REGRESSION: Flickr uploading broken
3263         <rdar://problem/4928662>
3264
3265         Turn protection against multiple forms submission back on. This approach is buggy
3266         but it is way better than not having it at all. Not protecting against this
3267         breaks number of major sites.
3268
3269         * bridge/mac/WebCoreAXObject.mm:
3270         (-[WebCoreAXObject accessibilityPerformAction:]):
3271         * loader/FrameLoader.cpp:
3272         (WebCore::FrameLoader::resetMultipleFormSubmissionProtection):
3273         (WebCore::FrameLoader::submitForm):
3274         (WebCore::FrameLoader::receivedMainResourceError):
3275         * loader/FrameLoader.h:
3276         * page/EventHandler.cpp:
3277         (WebCore::EventHandler::keyEvent):
3278         * page/Frame.cpp:
3279         (WebCore::Frame::setView):
3280         * page/mac/EventHandlerMac.mm:
3281         (WebCore::EventHandler::mouseDown):
3282
3283 2007-03-01  Kevin McCullough  <kmccullough@apple.com>
3284
3285         Reviewed by Adam.
3286
3287         - Added the test case: external-script-URL-location.html
3288         - Fix an issue where the url of a document is null after an open if the document
3289         has no parent.
3290
3291         * dom/Document.cpp:
3292         (WebCore::Document::open):
3293
3294 2007-03-01  Anders Carlsson  <acarlsson@apple.com>
3295
3296         Reviewed by Darin.
3297
3298         <rdar://problem/4960250>
3299         http://bugs.webkit.org/show_bug.cgi?id=11627
3300         REGRESSION: Reproducible crash at IMDb in WebCore::FrameLoader::stopLoadingSubframes
3301
3302         In rare cases, we could end up calling checkLoadComplete twice for the same frame. This would cause the
3303         didFailProvisionalLoad delegate method to be called twice for the same frame, and also cause the provisional document loader
3304         to be reset to null when other code wasn't expecting it.
3305
3306         This regressed in revision 10904 with the fix for <rdar://problem/4184719>. The fix is to only call stopLoading on the frame
3307         if either the document loader is loading, or the document is still being parsed. I've verified that the bug is still fixed and
3308         that no leaks occur.
3309
3310         * loader/DocumentLoader.cpp:
3311         (WebCore::DocumentLoader::stopLoading):
3312
3313 2007-03-01  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
3314
3315         Reviewed by Darin.
3316
3317         Fix ARM crash due to accessing non-4-byte-aligned memory
3318         as 32-bit values.
3319
3320         * platform/AtomicString.cpp:
3321         (WebCore::UCharBufferTranslator::equal):
3322
3323 2007-03-01  David Harrison  <harrison@apple.com>
3324
3325         Reviewed by Darin.
3326
3327         <rdar://problem/5033905> Have the DOM secondary thread check raise an exception by default
3328
3329         * platform/mac/ThreadCheck.mm:
3330         (WebCore::_WebCoreThreadViolationCheck):
3331         Initialize threadViolationIsException to true.
3332
3333 2007-03-01  Brady Eidson  <beidson@apple.com>
3334
3335         Reviewed by Darin
3336
3337         <rdar://problem/5030628> - Crash opening a new window with the
3338         "New windows open to the same page" pref set
3339
3340         * history/HistoryItem.cpp:
3341         (WebCore::HistoryItem::HistoryItem): Set the m_subItems vector capacity correctly
3342
3343 2007-03-01  Alexey Proskuryakov  <ap@webkit.org>
3344
3345         Reviewed by Maciej.
3346
3347         http://bugs.webkit.org/show_bug.cgi?id=12801
3348         Assertion failure in createMarkup() (root) when doing Select All, Copy in an SVG document
3349
3350         Test: editing/pasteboard/createMarkup-assert.xml
3351
3352         * editing/markup.cpp:
3353         (WebCore::createMarkup): Removed the assertion, because it's wrong in a non-HTML world.
3354
3355 2007-02-28  Adam Roben  <aroben@apple.com>
3356
3357         Reviewed by Oliver and Hyatt.
3358
3359         Fix <rdar://problem/5024233> Crash while using Find on empty document
3360
3361         No layout test possible, as this involves a null document.
3362
3363         * dom/Range.cpp:
3364         (WebCore::rangeOfContents): Added an ASSERT.
3365         * page/Frame.cpp:
3366         (WebCore::Frame::findString): Added a null-check for document().
3367         (WebCore::Frame::markAllMatchesForText): Ditto.
3368
3369 2007-02-28  Krzysztof Kowalczyk  <kkowalczyk@gmail.com>
3370
3371         Not reviewed - simple gdk build fix.
3372
3373         Add stubs for newly introduced localized strings.
3374
3375         * platform/gdk/TemporaryLinkStubs.cpp:
3376         (WebCore::searchMenuNoRecentSearchesText):
3377         (WebCore::searchMenuRecentSearchesText):
3378         (WebCore::searchMenuClearRecentSearchesText):
3379
3380 2007-02-28  Brady Eidson  <beidson@apple.com>
3381
3382         Reviewed by Beth
3383
3384         Move the Thread Safety Check functions into their own header for export to WebKit
3385
3386         * WebCore.exp: Export the function
3387         * WebCore.xcodeproj/project.pbxproj:
3388         * bindings/scripts/CodeGeneratorObjC.pm: Include the new header
3389         * platform/Logging.h: Move stuff to ThreadCheck.h
3390         * platform/ThreadCheck.h: Added.
3391         * platform/mac/LoggingMac.mm: Move stuff to ThreadCheck.mm
3392         * platform/mac/ThreadCheck.mm: Added.
3393         (WebCore::_WebCoreThreadViolationCheck):
3394         (WebCoreReportThreadViolation):
3395
3396 2007-02-28  Beth Dakin  <bdakin@apple.com>
3397
3398         Reviewed by Brady.
3399
3400         Fix for http://bugs.webkit.org/show_bug.cgi?id=12923 REGRESSION:
3401         Assertion failure copying standalone image
3402
3403         * platform/mac/PasteboardMac.mm:
3404         (WebCore::Pasteboard::writeImage): Use the CachedImage as the
3405         resource. This makes more sense anyway. There is no need to null-
3406         check the renderer or the CachedImage since we return early if
3407         there is no Image* and the HitTestResult::image() function checks
3408         for these things.
3409
3410 2007-02-28  Adele Peterson  <adele@apple.com>
3411
3412         Reviewed by Beth.
3413
3414         Fix for <rdar://problem/4887423> REGRESSION: search results popup menu strings are not localized
3415         and <rdar://problem/3517227> accessibility-related strings in WebCore are not localized
3416
3417         Use localized strings from WebKit instead of hard coded strings.
3418
3419         * bridge/mac/WebCoreAXObject.mm: (-[WebCoreAXObject roleDescription]):
3420         * page/mac/WebCoreViewFactory.h:
3421         * platform/LocalizedStrings.h:
3422         * platform/mac/LocalizedStringsMac.mm:
3423         (WebCore::searchMenuNoRecentSearchesText):
3424         (WebCore::searchMenuRecentSearchesText):
3425         (WebCore::searchMenuClearRecentSearchesText):
3426         (WebCore::AXWebAreaText):
3427         (WebCore::AXLinkText):
3428         (WebCore::AXListMarkerText):
3429         (WebCore::AXImageMapText):
3430         (WebCore::AXHeadingText):
3431         * rendering/RenderTextControl.cpp: (WebCore::RenderTextControl::itemText):
3432
3433 2007-02-27  Antti Koivisto  <antti@apple.com>
3434
3435         Reviewed by Maciej.
3436
3437         - fix http://bugs.webkit.org/show_bug.cgi?id=12911
3438         GoogleDocs: Ordered lists don't update immediately when start attribute changed
3439
3440         Update list marker value when start attribute changes.
3441
3442         * html/HTMLOListElement.cpp:
3443         (WebCore::HTMLOListElement::parseMappedAttribute):
3444
3445 2007-02-28  Nikolas Zimmermann  <zimmermann@kde.org>
3446
3447         Reviewed by Maciej.
3448
3449         Corrected the use of the new ENABLE() macros in some ksvg2/svg files.
3450         Add new build-webkit flags "--(no-)xpath" / "--(no-)xslt", to be able
3451         to switch off build features easily. Also add "--(no-)svg-experimental-features"
3452         flag, to be able to test filters/animations/use/foreignObject easily.
3453
3454         * DerivedSources.make:
3455         * ksvg2/scripts/make_names.pl:
3456         * ksvg2/svg/SVGFEDiffuseLightingElement.cpp:
3457         * ksvg2/svg/SVGStyledElement.cpp:
3458         * ksvg2/svg/SVGUseElement.cpp:
3459
3460 2007-02-28  Rob Buis  <buis@kde.org>
3461
3462         Reviewed by Maciej.
3463
3464         http://bugs.webkit.org/show_bug.cgi?id=12609
3465         Any SVG element will create renderers even when children of HTML elements
3466
3467         Allow creation of svg renderers only when parent is SVG, except for
3468         the <svg> element.
3469
3470         * ksvg2/svg/SVGAElement.h:
3471         * ksvg2/svg/SVGAnimationElement.h:
3472         * ksvg2/svg/SVGCircleElement.h:
3473         * ksvg2/svg/SVGClipPathElement.h:
3474         (WebCore::SVGClipPathElement::rendererIsNeeded):
3475         * ksvg2/svg/SVGDefsElement.cpp:
3476         * ksvg2/svg/SVGDefsElement.h:
3477         * ksvg2/svg/SVGDescElement.h:
3478         (WebCore::SVGDescElement::rendererIsNeeded):
3479         * ksvg2/svg/SVGElement.cpp:
3480         * ksvg2/svg/SVGEllipseElement.h:
3481         * ksvg2/svg/SVGFilterElement.h:
3482         (WebCore::SVGFilterElement::rendererIsNeeded):
3483         * ksvg2/svg/SVGFilterPrimitiveStandardAttributes.h:
3484         (WebCore::SVGFilterPrimitiveStandardAttributes::rendererIsNeeded):
3485         * ksvg2/svg/SVGForeignObjectElement.h:
3486         * ksvg2/svg/SVGGElement.h:
3487         * ksvg2/svg/SVGGradientElement.h:
3488         * ksvg2/svg/SVGImageElement.h:
3489         * ksvg2/svg/SVGLineElement.h:
3490         * ksvg2/svg/SVGMarkerElement.h:
3491         * ksvg2/svg/SVGMaskElement.h:
3492         * ksvg2/svg/SVGPathElement.h:
3493         * ksvg2/svg/SVGPatternElement.h:
3494         * ksvg2/svg/SVGPolyElement.h:
3495         * ksvg2/svg/SVGRectElement.h:
3496         * ksvg2/svg/SVGStopElement.h:
3497         * ksvg2/svg/SVGStyledElement.cpp:
3498         (WebCore::SVGStyledElement::rendererIsNeeded):
3499         * ksvg2/svg/SVGStyledElement.h:
3500         * ksvg2/svg/SVGSwitchElement.h:
3501         * ksvg2/svg/SVGSymbolElement.h:
3502         (WebCore::SVGSymbolElement::rendererIsNeeded):
3503         * ksvg2/svg/SVGTRefElement.h:
3504         * ksvg2/svg/SVGTSpanElement.h:
3505         * ksvg2/svg/SVGTextElement.h:
3506         * ksvg2/svg/SVGTitleElement.h:
3507         (WebCore::SVGTitleElement::rendererIsNeeded):
3508         * ksvg2/svg/SVGUseElement.h:
3509         * ksvg2/svg/SVGViewElement.h:
3510         (WebCore::SVGViewElement::rendererIsNeeded):
3511
3512 2007-02-28  Rob Buis  <buis@kde.org>
3513
3514         Reviewed by Maciej.
3515
3516         http://bugs.webkit.org/show_bug.cgi?id=12913
3517         Markers do not render in webkit when it misses markerWidth or markerHeight attribute
3518
3519         Set defaults for markerWidth/markerHeight so markers that do not specify them render.
3520
3521         * ksvg2/svg/SVGMarkerElement.cpp:
3522         (WebCore::SVGMarkerElement::SVGMarkerElement):
3523
3524 2007-02-27  Anders Carlsson  <acarlsson@apple.com>