20:54 Changeset [24792] by justing


Reviewed by Darin.

Crash due to infinite recursion in moveParagraphs deleting line from a Mail message

  • editing/deleting/5369009-expected.checksum: Added.
  • editing/deleting/5369009-expected.png: Added.
  • editing/deleting/5369009-expected.txt: Added.
  • editing/deleting/5369009.html: Added.


Reviewed by Darin.

<rdar://problem/5369009> Crash due to infinite recursion in moveParagraphs on delete

After the delete, we moved content into the previous block, and a style
rule turned a style span in the moved content into a block, throwing
moveParagraphs into infinite recursion, as it continually tried and failed
to get the style span into the same paragraph as the content just before it.

Added a method to ReplaceSelectionCommand to keep inserted style spans
from turning into blocks because of style rules. Will add code to prevent
other kinds of content from changing appearance because of style rules,
post-Leopard (5371536). Also added a bool to ReplaceSelectionCommand's
constructor to guard against infinite recursion.

  • editing/CompositeEditCommand.cpp: (WebCore::CompositeEditCommand::moveParagraphs):
  • editing/JSEditor.cpp:
  • editing/ReplaceSelectionCommand.cpp: (WebCore::ReplaceSelectionCommand::ReplaceSelectionCommand): (WebCore::ReplaceSelectionCommand::negateStyleRulesThatEffectAppearance): (WebCore::ReplaceSelectionCommand::doApply):
  • editing/ReplaceSelectionCommand.h:
20:44 Changeset [24791] by treat

Fix build.

20:18 Changeset [24790] by sullivan

Reviewed by Darin

  • fixed <rdar://problem/5216176> Need WebKit SPI to allow clients using embedded WebViews to avoid clipping ends of some printed pages

This provides Mail, and other clients that print views that embed WebViews, a way to ensure that
the HTML is laid out for printing before pagination occurs.

  • WebView/WebHTMLViewPrivate.h: new SPI method -_layoutForPrinting

  • WebView/WebHTMLView.mm: (-[WebHTMLView _web_setPrintingModeRecursiveAndAdjustViewSize]): new method, just like existing _web_setPrintingModeRecursive except passes YES for adjustViewSize (-[WebHTMLView _layoutForPrinting]): new SPI method, sets printing mode temporarily to adjust the view size for printing (-[NSView _web_setPrintingModeRecursiveAndAdjustViewSize]): new helper method to do the recursion
17:48 Changeset [24789] by andersca

Reviewed by Brady.

BitmapImage::getTIFFRepresentation() bug results in favicons not working at many sites

Don't bail out if a single frame is invalid.

  • platform/graphics/mac/ImageMac.mm: (WebCore::BitmapImage::getTIFFRepresentation):
17:29 Changeset [24788] by darin

Fix a bug number on an old check-in.

17:09 Changeset [24787] by adele

Reviewed by Darin.

Fix for <rdar://problem/5367919> A crash occurs at WebCore::Frame::isFrameSet() when attempting to print a iframe before it loads at http://www.monster.com/

  • WebView/WebHTMLView.mm: (-[WebHTMLView knowsPageRange:]): Nil check for frame.
16:56 Changeset [24786] by adele

Reviewed by Darin.

Fix for <rdar://problem/5274937> Safari 3 breaks Missing Sync's WebKit code by crashing

  • loader/MainResourceLoader.cpp: (WebCore::MainResourceLoader::continueAfterContentPolicy): Nil check for the FrameLoader.
16:24 Changeset [24785] by andersca


Reviewed by Darin.

REGRESSION: After switching from Bookmark view, the Find Banner won't appear while displaying a PDF file

Add a hasHTMLView to FrameLoaderClient. This is only useful for clients who can show different views for different
MIME types and URL schemes.

  • loader/FrameLoader.cpp: (WebCore::FrameLoader::transitionToCommitted): Call delegate methods here if the current view is not a HTML view.

  • loader/FrameLoaderClient.h: (WebCore::FrameLoaderClient::hasHTMLView):


Reviewed by Darin.

REGRESSION: After switching from Bookmark view, the Find Banner won't appear while displaying a PDF file

Implement hasHTMLView.

  • WebCoreSupport/WebFrameLoaderClient.h:
  • WebCoreSupport/WebFrameLoaderClient.mm: (WebFrameLoaderClient::hasHTMLView):
16:05 Changeset [24784] by darin


16:05 Changeset [24783] by darin

Reviewed by Tim Matcher.

  • DumpRenderTree/DumpRenderTree.m: (dump): Fix dumping for documents that include null characters. This turned out not to be needed for the test case that motivated me to do it, but it's nice to have this for the future.
15:29 Changeset [24782] by justing

Fixed a typo in my last checkin.

11:29 Changeset [24781] by weinig


Reviewed by Maciej.

Update tests for <rdar://problem/5363896>
REGRESSION: Setting document.domain does not override port-based cross-frame security checks -- breaks SAP NetWeaver

  • Also changes frame access test to use shared resource iframes.
  • http/tests/security/cross-frame-access-port-expected.txt:
  • http/tests/security/cross-frame-access-port-explicit-domain-expected.txt:
  • http/tests/security/cross-frame-access-port-explicit-domain.html:
  • http/tests/security/cross-frame-access-port.html:
  • http/tests/security/cross-frame-access-protocol-expected.txt:
  • http/tests/security/cross-frame-access-protocol-explicit-domain-expected.txt:
  • http/tests/security/cross-frame-access-protocol-explicit-domain.html:
  • http/tests/security/cross-frame-access-protocol.html:
  • http/tests/security/resources/cross-frame-iframe-for-port-explicit-domain-test.html: Removed.
  • http/tests/security/resources/cross-frame-iframe-for-port-test.html: Removed.
  • http/tests/security/resources/cross-frame-iframe-for-protocol-explicit-domain-test.html: Removed.
  • http/tests/security/resources/cross-frame-iframe-for-protocol-test.html: Removed.
  • http/tests/security/resources/cross-frame-iframe-with-explicit-domain-set.html: Added.
  • http/tests/security/resources/cross-frame-iframe.html: Added.


Reviewed by Maciej.

Fix for <rdar://problem/5363896>
REGRESSION: Setting document.domain does not override port-based cross-frame security checks -- breaks SAP NetWeaver

Updates the results for:


  • bindings/js/kjs_window.cpp: (KJS::createWindow): Use the new setDomainInternal method. (KJS::Window::isSafeScript): Don't set check the port or protocol if both documents have explicitly set document.domain in the DOM and the those domains are equal.
  • dom/Document.cpp: (WebCore::Document::Document): (WebCore::Document::setDomain): Remove force parameter and change all calls that relied on it use the new setDomainInternal method which does the same thing. (WebCore::Document::setDomainInternal):
  • dom/Document.h: Add new m_domainWasSetInDOM variable and accessor. (WebCore::Document::domainWasSetInDOM):
  • loader/FrameLoader.cpp: (WebCore::FrameLoader::checkCallImplicitClose): Use the new setDomainInternal method.
11:25 Changeset [24780] by justing


Reviewed by Tristan.

<rdar://problem/5098931> Attachments are lost when they are moved into a ToDo after a delete

Added an editor client method for asking whether or not it is
OK to merge content after a delete.

  • bridge/EditorClient.h:
  • editing/DeleteSelectionCommand.cpp: (WebCore::DeleteSelectionCommand::mergeParagraphs): Ask if the merge is allowed.
  • platform/gdk/EditorClientGdk.cpp: (WebCore::EditorClientGdk::shouldMoveRangeAfterDelete): Added a method stub.
  • platform/gdk/EditorClientGdk.h:
  • platform/graphics/svg/SVGImageEmptyClients.h: (WebCore::SVGEmptyEditorClient::shouldMoveRangeAfterDelete): Ditto.


Reviewed by Tristan.

<rdar://problem/5098931> Attachments are lost when they are moved into a ToDo after a delete

Mail needs to be asked if it is OK to do the content movement that happens
after a deleting in a situation like this one:
<div contenteditable="plaintext-only">foo</div><div>bar</div>
so that they can prevent the move or so that they can save content that
will be stripped by the move. This could have been done with shouldInsertNode
and a new WebViewInsertAction for "moves", but WebKit clients like Mail and
DashCode think that a shouldInsert* means that the user pasted something and
perform actions only appropriate for pastes. This change is less risky because
it won't require those clients to change their code.

  • DefaultDelegates/WebDefaultEditingDelegate.m: (-[WebDefaultEditingDelegate webView:shouldMoveRangeAfterDelete:replacingRange:]):
  • WebCoreSupport/WebEditorClient.h:
  • WebCoreSupport/WebEditorClient.mm: (WebEditorClient::shouldMoveRangeAfterDelete):
  • WebView/WebEditingDelegatePrivate.h:


Reviewed by Tristan.

<rdar://problem/5098931> Attachments are lost when they are moved into a ToDo after a delete

  • WebEditorClient.cpp: (WebEditorClient::shouldMoveRangeAfterDelete): Method stub.
  • WebEditorClient.h:
11:15 Changeset [24779] by andersca


Reviewed by Anders.

REGRESSION: sync XMLHttpRequest.send() raises an exception for failed authentication

  • http/tests/xmlhttprequest/failed-auth-expected.txt: Added.
  • http/tests/xmlhttprequest/failed-auth.html: Added. This test has other cases from bug 13075 commented out, as they still fail.
  • http/tests/xmlhttprequest/xmlhttprequest-missing-file-exception-expected.txt: Added.
  • http/tests/xmlhttprequest/xmlhttprequest-missing-file-exception.html: Added. This test already passed, but I think it's useful to cover this case, too.


Reviewed by Anders.

REGRESSION: sync XMLHttpRequest.send() raises an exception for failed authentication

Test: http/tests/xmlhttprequest/failed-auth.html

  • platform/network/mac/ResourceHandleMac.mm: (WebCore::ResourceHandle::loadResourceSynchronously): Make the fake response better match the real one in case of failed authentication. Unfortunately, NSURLConnection doesn't give us the real response.
  • xml/XMLHttpRequest.cpp: (WebCore::XMLHttpRequest::send): If the response has an HTTP code in it, then the error wasn't a network one, and an exception shouldn't be raised.
10:48 Changeset [24778] by tristan

Updating layout test results with new expected results.

10:28 Changeset [24777] by andersca

Reviewed by Darin.

<rdar://problem/4942372> REGRESSION: Anchor links in Mail don't work anymore

Make it possible for KURL to handle addig fragments to non-hierarchical URLs
such as mailto: and cid:.

  • platform/KURL.cpp: (WebCore::KURL::init): If the base URL is not hierarchical but the relative URL is a fragment, then allow parsing it.

If the URL is not hierarchical, set the fragment start and end positions correctly.

09:10 Changeset [24776] by hausmann

On Windows make sure the dll also ends up in $$OUTPUT_DIR/bin, so that QtLauncher can find it.

09:10 Changeset [24775] by hausmann

Link QtLauncher into $$OUTPUT_DIR/bin

08:30 Changeset [24774] by hausmann

Removed the BUILDING_QT ifdef in JSStringRef.h and changed UChar for the Qt build to use wchar_t on Windows.

08:30 Changeset [24773] by hausmann

Don't link against libxml/libxslt unless we really need it.

06:57 Changeset [24772] by zimmermann

Reviewed by Oliver & Rob.

Enable SVG textPath selection. Actually build chunks for them instead of
ignoring them - was easier than I thought. Respect per-character transformation
when calculating the chunk boundaries for the selection rectangles. Also fixes
selection rect drawing of rotated characters.

Next thing todo is to unify the glyph size calculations, as per-glyph transformations
are not respected when calculating flow box sizes in SVGRootInlineBox.

06:51 BuildingQtOnWindows edited by hausmann@webkit.org
06:46 BuildingQtOnWindows edited by hausmann@webkit.org
06:39 BuildingQtOnWindows edited by hausmann@webkit.org
06:34 BuildingQtOnWindows edited by hausmann@webkit.org
06:31 BuildingQtOnWindows edited by hausmann@webkit.org


23:33 Changeset [24771] by tristan

Reviewed by Oliver Hunt.

<rdar://problem/5369282> REGRESSION: DOMHTMLInput throwing exception due to missing _isEdited method

  • bindings/objc/DOMHTML.mm: (-[DOMHTMLInputElement _isEdited]): (-[DOMHTMLTextAreaElement _isEdited]):
  • bindings/objc/DOMPrivate.h: Renamed _isUserEdited to _isEdited.
21:46 Changeset [24770] by alp

2007-07-30 Diego Escalante Urrelo <diegoe@gnome.org>

Reviewed by Adam.

Implement can_go_backward and can_go_forward in webkitgtkpage.cpp

  • gtk/Api/webkitgtkpage.cpp: Implement webkit_gtk_page_can_go_backward() and webkit_gtk_page_can_go_forward() functions.
21:07 Changeset [24769] by alp

2007-07-30 Diego Escalante Urrelo <diegoe@gnome.org>

Reviewed by Alp Toker.

Bad n_params on load_finished's g_signal_new in webkitgtkpage.cpp

This is the proper fix for #14810.

  • gtk/Api/webkitgtkframe.cpp:
  • gtk/Api/webkitgtkpage.cpp: Fix the marshaller type of the "load_finished" signal and correct the class_offset parameters.
20:12 Changeset [24768] by alp

2007-07-29 Diego Escalante Urrelo <diegoe@gnome.org>

Reviewed by Alp Toker.

Bad n_params on load_finished's g_signal_new in webkitgtkpage.cpp

  • gtk/Api/webkitgtkpage.cpp: Use a correct n_params value and the corresponding types for them in the "load_finished" signal.
19:40 Changeset [24767] by treat

Paint and scrolling performance fixes and improvements.

16:25 Changeset [24766] by adele


Reviewed by John.

WebCore part of fix for <rdar://problem/5102522> REGRESSION: Can't tab to webview that doesn't have editable content (affects Safari, preview pane in Mail)

  • WebCore.exp:
  • page/FocusController.cpp: (WebCore::FocusController::setInitialFocus): Added. (WebCore::FocusController::advanceFocus): Don't give the focus back to the chrome if this is the initial focus.
  • page/FocusController.h:


Reviewed by John.

WebKit part of fix for <rdar://problem/5102522> REGRESSION: Can't tab to webview that doesn't have editable content

  • WebView/WebHTMLView.mm: (-[WebHTMLView becomeFirstResponder]): Call new setInitialFocus method instead of advanceFocus.
14:51 Changeset [24765] by zecke

2007-07-29 Brian Mastenbrook <brian@mastenbrook.net>

Build fix for WebKit/Gtk to compile on Fedora 7 as described in

  • WebKit.pri: cairo-ft is used by the WebKit/Gtk port
08:18 Changeset [24764] by treat

Respect the margins when creating frames.
Set the scroll mode to always off like we did before the rendered
frames patch and the other ports do now.

07:42 Changeset [24763] by treat

WebCore::Widget::setParent should be virtual.

07:22 Changeset [24762] by treat

We actually pass the mac versions expected results here.


19:34 Changeset [24761] by weinig

Reviewed by Mark Rowe.

  • Rename test for clarity and consistency.
  • Move resource from ssl/resources to security/resources as it is only being used by the security tests.
  • http/tests/security/cross-frame-access-port-domain-expected.txt: Removed.
  • http/tests/security/cross-frame-access-port-domain.html: Removed.
  • http/tests/security/cross-frame-access-port-explicit-domain-expected.txt: Copied from http/tests/security/cross-frame-access-port-domain-expected.txt.
  • http/tests/security/cross-frame-access-port-explicit-domain.html: Copied from http/tests/security/cross-frame-access-port-domain.html.
  • http/tests/security/cross-frame-access-protocol-expected.txt:
  • http/tests/security/cross-frame-access-protocol.html:
  • http/tests/security/resources/cross-frame-iframe-for-port-domain-test.html: Removed.
  • http/tests/security/resources/cross-frame-iframe-for-port-explicit-domain-test.html: Copied from http/tests/security/resources/cross-frame-iframe-for-port-domain-test.html.
  • http/tests/security/resources/cross-frame-iframe-for-protocol-test.html: Copied from http/tests/ssl/resources/cross-frame-access-protocol-iframe.html.
  • http/tests/ssl/resources: Removed.
  • http/tests/ssl/resources/cross-frame-access-protocol-iframe.html: Removed.
19:15 Changeset [24760] by weinig

Reviewed by Oliver Hunt.

  • Factor access testing code used in protocol and port tests to cross-frame-access.js.
  • Add cross-frame-iframe-for-protocol-explicit-domain-test.html which is parallel test to cross-frame-iframe-for-port-domain-test.html but for different protocols.
  • http/tests/security/cross-frame-access-port-domain-expected.txt:
  • http/tests/security/cross-frame-access-port-domain.html:
  • http/tests/security/cross-frame-access-port.html:
  • http/tests/security/cross-frame-access-protocol-explicit-domain-expected.txt: Added.
  • http/tests/security/cross-frame-access-protocol-explicit-domain.html: Added.
  • http/tests/security/cross-frame-access-protocol.html:
  • http/tests/security/resources/cross-frame-access.js:
  • http/tests/security/resources/cross-frame-iframe-for-port-domain-test.html:
  • http/tests/security/resources/cross-frame-iframe-for-port-test.html:
  • http/tests/security/resources/cross-frame-iframe-for-protocol-explicit-domain-test.html: Added.
19:02 Changeset [24759] by alp

Fix up reviewer name typo in ChangeLog

18:30 Changeset [24758] by alp

2007-07-29 Alp Toker <alp.toker@collabora.co.uk>

Reviewed by Brian Dash.

http://bugs.webkit.org/show_bug.cgi?id=14711 (revisited)
RenderThemeGdk's buttons are state-agnostic (pressed, hovered)

Generalize shadow state into a function.
use a GtkHBox to work around a theme bug (thanks to Nigel Tao).

  • platform/gdk/RenderThemeGdk.cpp: (WebCore::RenderThemeGdk::determineShadow): (WebCore::RenderThemeGdk::paintCheckbox): (WebCore::RenderThemeGdk::paintRadio): (WebCore::RenderThemeGdk::paintButton): (WebCore::RenderThemeGdk::gtkWindowContainer):
  • platform/gdk/RenderThemeGdk.h:
18:18 Changeset [24757] by bdash

2007-07-28 Mark Rowe <mrowe@apple.com>

Reviewed by Darin and John.

<rdar://problem/5229705> Thai characters render differently between leopard and tiger

Add Leopard-specific test results for tests containing Thai characters.

  • mac/leopard/Skipped:
  • mac/leopard/fast/text: Added.
  • mac/leopard/fast/text/atsui-multiple-renderers-expected.txt: Added.
  • mac/leopard/fast/text/international: Added.
  • mac/leopard/fast/text/international/thai-line-breaks-expected.txt: Added.
18:15 Changeset [24756] by bdash

2007-07-29 Mark Rowe <mrowe@apple.com>

Reviewed by Oliver.

<rdar://problem/5228270> SVG test failures due to different metrics for scaled text blocks

Add Leopard-specific results for three SVG tests.

  • mac/leopard/Skipped:
  • mac/leopard/svg: Added.
  • mac/leopard/svg/hixie: Added.
  • mac/leopard/svg/hixie/text: Added.
  • mac/leopard/svg/hixie/text/003-expected.txt: Added.
  • mac/leopard/svg/hixie/text/003a-expected.txt: Added.
  • mac/leopard/svg/hixie/viewbox: Added.
  • mac/leopard/svg/hixie/viewbox/preserveAspectRatio: Added.
  • mac/leopard/svg/hixie/viewbox/preserveAspectRatio/002-expected.txt: Added.
10:37 Changeset [24755] by bdash


10:37 Changeset [24754] by bdash

Merge r24658.

Reviewed by Oliver.

In Mail, a crash occurs at WTF::fastMalloc() when attempting to create a new HTML message from http://nytimes.com/

If the cache is disabled and there is already a resource with the given URL, return it instead of creating a new one.

  • loader/DocLoader.cpp: (WebCore::DocLoader::requestResource):
10:37 Changeset [24753] by bdash

Merge r24514.

Reviewed by Anders.

  • improve the fix for <rdar://problem/5298816> Crash redirecting message, in -[SharedBufferData initWithSharedBuffer:]; fixes a crash seen on the buildbot
  • loader/DocLoader.cpp: (WebCore::DocLoader::~DocLoader): Call setDocLoader(0) on all the resources so they don't try to use a stale DocLoader pointer later.
10:36 Changeset [24752] by bdash

Merge r24500.

Reviewed by Darin.

If the style sheet requested is a user style sheet, don't store it in the DocLoader's resource map;
user style sheets can outlive their doc loaders.

  • loader/DocLoader.cpp: (WebCore::DocLoader::requestCSSStyleSheet):
10:36 Changeset [24751] by bdash

Merge r24492.

Reviewed by Darin.

<rdar://problem/5298816> Crash redirecting message, in -[SharedBufferData initWithSharedBuffer:]

Have cached resources keep track of what DocLoader they belong to. This is so that they can remove themselves
from the DocLoader's resource map when they're destroyed when the cache is disabled.

  • html/HTMLImageLoader.cpp: (WebCore::HTMLImageLoader::updateFromElement):
  • loader/Cache.cpp: (WebCore::Cache::requestResource):
  • loader/CachedResource.cpp: (WebCore::CachedResource::CachedResource): (WebCore::CachedResource::~CachedResource):
  • loader/CachedResource.h: (WebCore::CachedResource::setDocLoader):
10:20 Changeset [24750] by bdash

New tag.

07:22 BuildingGdk edited by sylvain.pasche@gmail.com
link to BuildingOnLinux (diff)
07:20 BuildingOnLinux edited by sylvain.pasche@gmail.com
Ubuntu package requirements (diff)


17:54 Changeset [24749] by adele

Reviewed by Darin.

Temporary fix for Windows to prevent crashing on all frame pages. Darin
is working on a more permanent fix.

  • loader/FrameLoader.cpp:
16:57 Changeset [24748] by treat

Fix crash when using QtWebKit canvas support.

16:37 Changeset [24747] by ddkilzer


Reviewed by Geoff and Darin.

Use a subroutine for validating the --skipped switch.

  • Scripts/run-webkit-tests:
14:58 ApolloWebKit created by ddkilzer@webkit.org
Initial instructions on building Apollo WebKit.
14:56 WikiStart edited by ddkilzer@webkit.org
Added link to Apollo WebKit page. (diff)
14:49 Changeset [24746] by weinig

Reviewed by Geoff Garen.

  • Test that cross frame access to a window on a different port is denied.
  • Adds test for out behavior of the above if document.domain is explicitly set to the same domain for both frames. We currently fail this test and a radar has been filed.
  • Clean up cross-frame-access-protocol.html to work in FireFox as well.
  • http/conf/apache2-httpd.conf: Listen on port 8080 now as well
  • http/conf/cygwin-httpd.conf: Listen on port 8080 now as well
  • http/conf/httpd.conf: Listen on port 8080 now as well
  • http/tests/security/cross-frame-access-port-domain-expected.txt: Added.
  • http/tests/security/cross-frame-access-port-domain.html: Added.
  • http/tests/security/cross-frame-access-port-expected.txt: Added.
  • http/tests/security/cross-frame-access-port.html: Added.
  • http/tests/security/cross-frame-access-protocol.html:
  • http/tests/security/resources/cross-frame-iframe-for-port-domain-test.html: Added.
  • http/tests/security/resources/cross-frame-iframe-for-port-test.html: Added.
13:40 Changeset [24745] by darin
  • fix build
  • WebKitPrefix.h: Removed the USING_WEBCORE_XXX definitions.
  • WebView/WebHTMLView.mm: (-[WebHTMLView delete:]): Fixed this to use WebCore again; I accidentally revived a dead code path that didn't work in the last patch! (-[WebHTMLView deleteToMark:]): Ditto. (-[WebHTMLView copy:]): Removed the unused side of the ifdef. (-[WebHTMLView cut:]): Ditto. (-[WebHTMLView paste:]): Ditto.
13:27 Changeset [24744] by darin

Reviewed by Sam.

  • Scripts/run-webkit-tests: Remove exception for leaks bug that has been fixed on Leopard.
13:00 Changeset [24743] by weinig

Reviewed by Adam Roben.

Don't display the expected and resulting values for screenTop and screenY
as they depend on the physical screen size.

  • http/tests/security/cross-frame-access-put-expected.txt:
  • http/tests/security/resources/cross-frame-iframe-for-put-test.html:
12:13 Changeset [24742] by darin

Reviewed by Tim Hatcher and Oliver Hunt.

  • fix <rdar://problem/5355815> webView:doCommandBySelector: isn't getting called for copy:

Added code so that webView:doCommandBySelector: is called for every command.

  • WebView/WebHTMLView.mm: Made 44 of the commands use the WEBCORE_COMMAND macro instead of being handwritten. For all the others, added invocation of COMMAND_PROLOGUE macro at the start of the command. (-[WebHTMLView callDelegateDoCommandBySelectorIfNeeded:]): Added. (-[WebHTMLView callWebCoreCommand:]): Added. (-[WebHTMLView delete:]): Removed unused code path -- easy to bring back some day, but we don't need it compiled in. (-[WebHTMLView deleteBackwardByDecomposingPreviousCharacter:]): Changed so this doesn't call deleteBackward: any more so we don't call the delegate two times. (-[WebHTMLView deleteToMark:]): Changed so this doesn't call delete: any more so we don't call the delegate two times. (-[WebHTMLView selectToMark:]): Changed so this doesn't call setMark: any more so we don't call the delegate two times. (-[WebHTMLView doCommandBySelector:]): Added code to set the private variable selectorForDoCommandBySelector. This allows callDelegateDoCommandBySelectorIfNeeded to detect that we've already called the delegate and avoids calling it twice.
  • WebView/WebHTMLViewInternal.h: Added the selectorForDoCommandBySelector field.
12:09 Changeset [24741] by darin
  • rendering/RenderLayer.cpp: Rolled out accidentally-checked-in code.
12:07 Changeset [24740] by darin


Reviewed by John Sullivan.

  • fix <rdar://problem/5261393> In Mail the plain text alternatives to messages with attachments show weird little OBJ thingies
  • fast/text/zero-width-characters-expected.txt: Added.
  • fast/text/zero-width-characters.html: Added.


Reviewed by John Sullivan.

  • fix <rdar://problem/5261393> In Mail the plain text alternatives to messages with attachments show weird little OBJ thingies

The "weird little thingies" are U+FFFC characters. It's good to render them as zero-width, the way we do
with control characters -- helps Mail and doesn't hurt web browsing.

Test: fast/text/zero-width-characters.html

  • platform/CharacterNames.h: Add names for leftToRightMark, rightToLeftMark, and objectReplacementCharacter.
  • platform/GlyphPageTreeNode.cpp: (WebCore::GlyphPageTreeNode::initializePage): Give objectReplacementCharacter zero width in the same way we do it for the various other characters.
11:33 Changeset [24739] by andersca

Reviewed by Adam.

Add FIXME comment.

  • Interfaces/DOMHTML.idl:
11:08 RunWebKitTestsDesign edited by aroben@apple.com
Added a description of how multiple Skipped files should be treated (diff)
10:27 Changeset [24738] by adachan

2007-07-27 Ada Chan <adachan@apple.com>

Reviewed by Adam.

Added methods to track user edited text fields now that
<rdar://problem/5359921> has been fixed.

  • DOMHTMLClasses.cpp: (DOMHTMLInputElement::QueryInterface): (DOMHTMLInputElement::isUserEdited): (DOMHTMLTextAreaElement::QueryInterface): (DOMHTMLTextAreaElement::isUserEdited):
  • DOMHTMLClasses.h:
  • Interfaces/DOMPrivate.idl:
07:31 Changeset [24737] by spadma

vbradley, reviewed by Sachin

DESC: AYNN-74AF84 : Username field not filled automatically

  • BrowserCore/Formcontrols/inc/FControlSelectListBox.h:
  • BrowserCore/Formcontrols/src/FControlInputSkin.cpp:
  • BrowserCore/Formcontrols/src/FControlInputWidget.cpp: (CControlInputWidget::OfferKeyEventL):
  • BrowserCore/Formcontrols/src/FControlSelectListBox.cpp: (CControlSelectListBox::UpdateContentL): (CControlSelectListBox::DrawListBox):
  • BrowserCore/Formcontrols/src/FControlSelectWidget.cpp: (CControlSelectWidget::ConstructL): (CControlSelectWidget::FilterContentL): (CControlSelectWidget::OfferKeyEventL):
06:59 Changeset [24736] by zbujtas

yadavall, reviewed by <Zalan>

DESC: Fixes for widget menu API

06:58 Changeset [24735] by lars

Enable XSLT support for the gtk port.

06:58 Changeset [24734] by lars

Premature end after finishing parsing is a fatal error.

06:58 Changeset [24733] by lars

add xlst related tests to Skipped list.

06:58 Changeset [24732] by lars

Simplify conversions that go from a QStringRef to a WebCore::String to not use a temporary QString.

06:57 Changeset [24731] by lars

Fix entity handling in the main document content (attributes still broken)

06:57 Changeset [24730] by lars

Correctly convert a null QString into a null WebCore::String. Add a String(const QStringRef&) constructor so that we can avoid converting QStringRef via QString to WebCore::String.

06:57 Changeset [24729] by lars

Parse version and encoding of the xml document.

06:56 Changeset [24728] by lars

Correctly or semi-correctly parse the public and system id.

Patch from Lars.

06:56 Changeset [24727] by lars

Fix uri parsing for attributes.

06:56 Changeset [24726] by lars

Fixing qualified name parsing - from Lars.

06:56 Changeset [24725] by lars

Parse dtd's

06:55 Changeset [24724] by lars

Implement error handling.

Add our copyrights

06:55 Changeset [24723] by lars

Making the stylesheets work (plus everything else basically)

06:55 Changeset [24722] by lars

Remove the libxml dependency when not compiling with XSLT.

Just ifdefing, displaying xml doesn't work.

05:26 Changeset [24721] by zecke


2007-07-27 Holger Hans Peter Freyther <zecke@selfish.org>

Reviewed by Mark.

Don't create an app bundle on OSX to keep WebKitTools/Scripts/run-launcher working.

  • QtLauncher/QtLauncher.pro:

2007-07-27 Holger Hans Peter Freyther <zecke@selfish.org>

Reviewed by Mark.

Correct the path of GdkLauncher and make checkFrameworks work on OSX
when building the Qt or Gtk+ port.

  • GdkLauncher/GdkLauncher.pro: Don't create an app bundle on OSX
  • Scripts/run-launcher:
  • Scripts/webkitdirs.pm: Don't add WebKit if we build the Qt or Gtk+ port.
05:26 Changeset [24720] by hausmann

Work around bug in the moc when used as preprocessor for generating the bindings. -DFOO=1 is needed instead of -DFOO to make the <hash>if FOO conditional match.

04:59 Changeset [24719] by bdash

2007-07-27 Patrick Hanna <phanna@email.unc.edu>

Reviewed by Darin Adler.

Change PluginBase::refresh to be static and update PluginsFunc::callAsFunction accordingly. Now we no longer
allocate an instance of a JSObject subclass on the stack which could lead to a crash during garbage collection.

  • bindings/js/kjs_navigator.cpp: (KJS::PluginsFunc::callAsFunction):
01:34 Changeset [24718] by hausmann

Always define JSChar to be unsigned short for the Qt builds, to ensure compatibility with UChar.

01:34 Changeset [24717] by hausmann

Fix snprintf hack for the Qt/Windows build with mingw-gcc. MingW has snprintf, so we don't need this helper function.

01:34 Changeset [24716] by hausmann

Make the DocTypeStrings.cpp generation work with MingW. Cleaned it up to be a regular extra compiler at the same time.

01:34 Changeset [24715] by hausmann

Fix compilation with Qt on Windows with MingW: Implemented currentThreadStackBase() for this platform.

01:33 Changeset [24714] by hausmann

Fix compilation with Qt on Windows with MingW: The MingW headers do not provide a prototype for a reentrant version of localtime. But since we don't use multiple threads for the Qt build we can use the plain localtime() function.

01:33 Changeset [24713] by hausmann

MingW build fixes: Do the "OBJECTS_TRAILING_SLASH" trick not only for win32-msvc but for all win32 mkspecs. Added LIB/INCLUDE hack from Qt to make us pick up 3rdparty libraries such as libxml from these environment variables.

01:33 Changeset [24712] by hausmann

Fix build-webkit for the Qt build on Windows with msvc/nmake by trying to detect the Qt mkspec and using "nmake" instead of "make" as build command.

01:33 Changeset [24711] by hausmann

Use $(MOVE) instead of mv to eliminated the shell dependency and replaced the long shell line to call bison and modify the css grammar file with a few lines of portable perl code.

01:33 Changeset [24710] by hausmann

Implemented currentTime() in the interpreter by using QDateTime, so that we don't need timeGetTime() on Windows and therefore also don't need to link against Winmm.dll.

01:32 Changeset [24709] by hausmann

Exclude DumpRenderTree from the Qt build on Windows for now.

01:32 Changeset [24708] by hausmann

For detecting the SVG support for the Qt build don't do the nm hack. Just always claim SVG support is enabled because that's what it is. The nm hack doesn't work on Windows anyway.

01:32 Changeset [24707] by hausmann

On make install on Windows copy the dll into the bin dir.

01:32 Changeset [24706] by hausmann

Link in libxml and libxslt on Windows.

01:32 Changeset [24705] by hausmann

Temporarily disable the ICO support for the Qt build on Windows.

01:32 Changeset [24704] by hausmann

Fix evaluation with ActiveState Perl 5.6. Somehow it tried to treat nameEntries as array.

01:31 Changeset [24703] by hausmann

Fix compilation with MSVC in the Qt build (missing header files).

01:31 Changeset [24702] by hausmann

Disable the favicon database on Windows for now, to eliminate the sqlite dependency.

01:31 Changeset [24701] by hausmann

Added custom preprocessor support to make_names.pl and use the moc for it in the Qt build.

01:31 Changeset [24700] by hausmann

Disable the batch mode for nmake to work around the clash of Node.c and Node.cpp both mapping to Node.obj

01:30 Changeset [24699] by hausmann

Fix compilation under Windows (removed sys/time.h dependency, use Qt instead).

01:30 Changeset [24698] by hausmann

Fix compilation on Windows.

01:30 Changeset [24697] by hausmann

Fix generation of ColorData.c on Windows.

01:30 Changeset [24696] by hausmann

Fix compilation under Qt/Windows.

01:30 Changeset [24695] by hausmann

Changed extendedAttributes to be a variable instead of a hash directly. Required to work with ActiveState perl.

01:29 Changeset [24694] by hausmann

Replace the use of snprintf with QByteArray to compile under msvc 2005 express.

01:29 Changeset [24693] by hausmann

Don't use pthread.h unless thread support is enabled.

01:29 Changeset [24692] by hausmann

Removed TCSystemMalloc from the Qt build, it's not necessary it seems.

01:29 Changeset [24691] by hausmann

Fix header guard generation for CSSGrammar.h under Windows

01:29 Changeset [24690] by hausmann

Fix build with moc as preprocessor. Don't rely the preprocessor preserving whitespace at the beginning of lines.

01:28 Changeset [24689] by hausmann

Added os-win32 to the include search path for the Qt windows build in order to provide the fake stdint.h header file.

01:28 Changeset [24688] by hausmann

Make make-css-file-arrays.pl work on Windows by not calling /usr/bin/gcc as preprocessor for the Qt build.

01:28 Changeset [24687] by hausmann

More gperf commandline fixes for Windows.

01:28 Changeset [24686] by hausmann

More quoting fixes for perl under Windows.

01:28 Changeset [24685] by hausmann

Fix command quoting under Windows

01:27 Changeset [24684] by hausmann

Fix cssmakeprops and cssmakevalue scripts to work on Windows without cygwin perl. Avoid inplace substitutions and direct invocation of /bin/sh. Also call gperf with arguments that GnuWin32's gperf understands.

01:27 Changeset [24683] by hausmann

Beautiful hacks to make more of the extra compiler magic work under Windows outside of cygwin.

01:27 Changeset [24682] by hausmann

Correctly propagate the custom preprocessor option through the idl parser and the codegenerator.

01:27 Changeset [24681] by hausmann

When constructing the absolute path when searching for IDL files in the list of search paths don't prepend a slash if $SOURCEROOT as environment variable is not set. In the Qt build we pass absolute search paths already and we don't set SOURCEROOT. Under Unix that gives a double slash for the directory which works fine, under Windows we get /c:\webkit\foo which of course does not work.

01:27 Changeset [24680] by hausmann

Relax the regular expression that tries to find constant declarations in the IDL source to not require leading whitespace. This makes it work with preprocessors that happen to eliminate leading whitespace.

01:26 Changeset [24679] by hausmann

Try to use the moc as preprocessor for the IDL bindings generator

01:26 Changeset [24678] by hausmann

Make it possible to specify the preprocessor the bindings generator uses instead of hardcoding /usr/bin/gcc (which is of course no option on Windows). Also try to use a more portable and Windows friendly way of opening the reading pipe to the preprocessor.

01:26 Changeset [24677] by hausmann

Don't use stdout redirection for make_names.pl as it doesn't work under Windows perl (outside cygwin).

01:26 Changeset [24676] by hausmann

Fix DocTypeStrings.cpp generation to work on Windows with cmd.exe

00:43 Changeset [24675] by hausmann

Work around aliasing bugs in WebCore by compiling with -fno-strict-aliasing for gcc builds for now.

00:28 Changeset [24674] by hausmann

Use -Wreturn-type only when compiling with g++

00:28 Changeset [24673] by hausmann

Don't use pkg-config on Windows for the Qt build

00:22 Changeset [24672] by aroben

Fix Bug 14773: REGRESSION (r24630): ASSERT_NOT_REACHED in DOMHTMLInputElement::name on Windows

Reviewed by Steve.

  • Interfaces/DOMHTML.idl: Change IDOMHTMLInputElement back to inheriting from IDOMElement. Safari 3 Beta uses this interface, so we can't change its vtable.
00:14 Changeset [24671] by hausmann

Fix compilation with MSVC.

00:11 Changeset [24670] by hausmann

Fix compilation with MSVC.


22:27 Changeset [24669] by ddkilzer

Added radar.

21:54 Changeset [24668] by ap

Reviewed by Oliver.

Add a logging channel for text input

  • Misc/WebKitLogging.h:
  • Misc/WebKitLogging.m: (WebKitInitializeLoggingChannelsIfNecessary):
  • WebView/WebHTMLView.mm: (-[WebHTMLView validAttributesForMarkedText]): (-[WebHTMLView textStorage]): (-[WebHTMLView characterIndexForPoint:]): (-[WebHTMLView firstRectForCharacterRange:]): (-[WebHTMLView selectedRange]): (-[WebHTMLView markedRange]): (-[WebHTMLView attributedSubstringFromRange:]): (-[WebHTMLView hasMarkedText]): (-[WebHTMLView unmarkText]): (-[WebHTMLView setMarkedText:selectedRange:]): (-[WebHTMLView doCommandBySelector:]): (-[WebHTMLView insertText:]):
21:18 Changeset [24667] by darin
  • http/tests/security/cross-frame-access-put-expected.txt: Checked in corrected results that match the checked-in test.
18:58 Changeset [24666] by darin
  • fix Tiger build
  • Misc/WebTypesInternal.h: For use inside the library, use NSInteger and NSUInteger, just like on Leopard. WebNSInteger and WebNSUInteger are still present, but they are used in public and private headers only.
  • Carbon/CarbonWindowAdapter.m:
  • Carbon/CarbonWindowFrame.m:
  • DefaultDelegates/WebScriptDebugServer.m:
  • History/WebBackForwardList.mm:
  • Misc/WebDownload.m:
  • Misc/WebSearchableTextView.m:
  • Plugins/WebBaseNetscapePluginView.mm:
  • WebCoreSupport/WebFrameLoaderClient.mm:
  • WebView/WebHTMLView.mm:
  • WebView/WebPDFView.mm:
  • WebView/WebView.mm:
  • WebView/WebViewInternal.h: Update all implementation files and internal headers to use NSInteger and NSUInteger rather than WebNSInteger and WebNSUInteger.
18:12 Changeset [24665] by sullivan

Reviewed by Kevin Decker

  • fixed <rdar://problem/5362989> Searching a pdf in some modes shows hilights for all instances of the word, even those not on the current page

It turns out that there were three different issues here, all contributing to incorrect display
of multiple matches for PDF pages in certain display modes. (1) in non-continuous display modes,
we weren't updating the match rects when the displayed page is changed with page up/down (e.g.);
(2) the mechanism to update the match rects on scrolling was busted except for the first scroll
away from 0,0; (3) the PDFKit API returns selection bounds for non-displayed pages in non-continuous
modes just as if they were the displayed pages. This patch fixes all three issues.

  • WebView/WebPDFView.h: made ivar name even longer

  • WebView/WebPDFView.mm: (-[WebPDFView setPDFDocument:]): updated for ivar name change (-[WebPDFView viewDidMoveToWindow]): observe page-change notifications as well as the others; this is necessary because in the non-continuous modes the view can be completely updated without any scrolling involved (problem 1) (-[WebPDFView viewWillMoveToWindow:]): stop observing page-change notifications (-[WebPDFView rectsForTextMatches]): skip any pages that aren't visible; this avoids treating matches on non-displayed non-continous modes as if they were on the displayed page (problem 3) (-[WebPDFView _PDFDocumentViewMightHaveScrolled:]): after checking whether scroll position has changed since we last checked it, remember the new one (d'oh!) (problem 2)

(-[WebPDFView _scaleOrDisplayModeOrPageChanged:]):
renamed to include page changes

(-[WebPDFView _visiblePDFPages]):
new method, returns the set of pages that are at least partly visible

17:51 Changeset [24664] by tristan

Reviewed by Anders Carlsson.

<rdar://problem/5350019> REGRESSION: Pages on milliyet.com are added to Back list when visited, causing Back button to just reload

  • http/tests/navigation/lockedhistory-iframe-expected.txt: Added.
  • http/tests/navigation/lockedhistory-iframe.html: Added.
17:46 Changeset [24663] by tristan

Reviewed by Anders Carlsson.

<rdar://problem/5350019> REGRESSION: Pages on milliyet.com are added to Back list when visited, causing Back button to just reload

  • loader/FrameLoader.cpp: (WebCore::FrameLoader::continueFragmentScrollAfterNavigationPolicy): If the policy action is FrameLoadTypeRedirectLockHistory, don't add the frame load url to the history.
17:26 Changeset [24662] by hyatt
17:14 Changeset [24661] by andersca

Reviewed by Geoff.

Add a FIXME about caching pages with plug-ins that have outstanding NPObjects.

  • loader/FrameLoader.cpp: (WebCore::FrameLoader::canCachePage):
17:12 Changeset [24660] by kmccullo

Reviewed by Sam.

  • Updated results as they are affecte by my previous checkin.
  • http/tests/security/cross-frame-access-put-expected.txt:
  • http/tests/security/resources/cross-frame-iframe-for-put-test.html:
17:02 Changeset [24659] by hyatt

FIx for <rdar://problem/5289721> CrashTracer: [REGRESSION] 2658 crashes in Safari at com.apple.WebCore: WebCore::RenderInline::height const + 16

Adding a null check for now and an assert. We don't know how/why this
bug happens, but hopefully the assert will help people to reproduce.

Reviewed by kevin

  • rendering/RenderInline.cpp: (WebCore::RenderInline::height):
16:52 Changeset [24658] by andersca

Reviewed by Oliver.

In Mail, a crash occurs at WTF::fastMalloc() when attempting to create a new HTML message from http://nytimes.com/

If the cache is disabled and there is already a resource with the given URL, return it instead of creating a new one.

  • loader/DocLoader.cpp: (WebCore::DocLoader::requestResource):
16:15 Changeset [24657] by adele

Reviewed by Geoff.

Build fix.

  • platform/network/cf/ResourceRequest.h:
15:47 Changeset [24656] by kmccullo

Reviewed by.

  • Fixed bot breakage.
  • fast/dom/Window/window-resize.html:
15:40 Changeset [24655] by andersca

Reviewed by John Sullivan.

Some frame load delegate methods not called when loading a page from the BF cache

These calls were moved to FrameLoader::begin, which is never called for pages loaded
from the BF cache. This happened in revision 24490.

  • loader/FrameLoader.cpp: (WebCore::FrameLoader::transitionToCommitted): Call delegate methods if the load is from the BF cache.
14:59 Changeset [24654] by kmccullo


Reviewed by Darin, Geoff and Sam.

  • <rdar://problem/5150461> Resizing the window larger than the screen causes no resize.
  • Tests to ensure we resize/move the window out of bounds correctly.
  • fast/dom/Window/window-resize-expected.txt: Added.
  • fast/dom/Window/window-resize.html: Added.
  • mac/leopard/Skipped:


Reviewed by Darin, Geoff and Sam.

  • <rdar://problem/5150461> Resizing the window larger than the screen causes no resize.
  • We now better match the behavior of other browsers.
  • bindings/js/kjs_window.cpp: (KJS::constrainToVisible): (KJS::WindowFunc::callAsFunction):


Reviewed by Darin, Geoff, Sam.

  • <rdar://problem/5150461> Resizing the window larger than the screen causes no resize.
  • Added delegate methods to intercept and fake the frame location so methods like resizeTo and moveTo can change the window location without actually making the window appear on-screen.
  • DumpRenderTree/UIDelegate.h:
  • DumpRenderTree/UIDelegate.m: (-[UIDelegate webView:setFrame:]): (-[UIDelegate webViewFrame:]):
14:43 Changeset [24653] by kmccullo

Reviewed by Brady.

  • Added leopard-specific results for <rdar://problem/5228256> REGRESSION(Leopard): layout test failures: security/block-test
  • mac/leopard/security: Added.
  • mac/leopard/security/block-test-expected.txt: Added.
14:35 Changeset [24652] by tristan

Reviewed by Adele Peterson.

<rdar://problem/5359921> WebCore should maintain a dirty flag for confirmation on unsaved changes

  • bindings/objc/DOMHTML.mm: (-[DOMHTMLInputElement _isUserEdited]): (-[DOMHTMLTextAreaElement _isUserEdited]): Renamed _isEdited to _isUserEdited since the functionality is now different

  • bindings/objc/DOMPrivate.h: Renamed _isEdited to _isUserEdited.

  • rendering/RenderTextControl.cpp: (WebCore::RenderTextControl::RenderTextControl): Defaulted m_userEdited to false

Set m_userEdited to false, since the change wasn't triggered first by an edit

Set m_userEdited to true, since it has been touched by the user

  • rendering/RenderTextControl.h: (WebCore::RenderTextControl::isUserEdited): (WebCore::RenderTextControl::setUserEdited): Added two new methods to set/return the userEdited boolean value.
14:21 Changeset [24651] by ggaren

Reviewed by Beth Dakin.

Fixed NULL-dereference crash seen on buildbot.

No bug filed because this is a part of <rdar://problem/5241167>.

  • history/PageCache.h: (WebCore::PageCache::get): NULL-check HistoryItems passed to get() and remove(), to match the old semantics of HashMap. For clarity, ASSERT that add() is not called with NULL arguments.
14:18 Changeset [24650] by kmccullo

Reviewed by Adam.

  • Lars wanted this check for Qt but it breaks platform specific layout tests.
  • Scripts/run-webkit-tests:
12:51 Changeset [24649] by andersca

Reviewed by Oliver Hunt.

<rdar://problem/5361106> Crash in user style sheet when loading webpage

Null check m_frame.

  • dom/Document.cpp: (WebCore::Document::stylesheetLoaded):
12:48 Changeset [24648] by spadma

condon, reviewed by ligman

DESC: Fix codescanner warnings

  • BrowserCore/Formcontrols/src/FControlInputSkin.cpp: (CFormInputSkin::ActivateL):
  • BrowserCore/Renderers/src/ImageRenderer.cpp: (CImageRenderer::~CImageRenderer):
  • BrowserCore/Renderers/src/TextRenderer.cpp: (CTextRenderer::~CTextRenderer):
  • BrowserView/src/FormDataLoginStore.cpp: (CFormDataLoginStore::ClearData):
  • BrowserView/src/LoadListeners.cpp: (CWmlResourceLoadListener::SetUrlL):
  • BrowserView/src/PageView.cpp: (CPageView::~CPageView):
  • ChangeLog:
  • Plugin/src/PluginSkin.cpp: (CPluginSkin::~CPluginSkin):
  • Plugin/src/PluginWin.cpp: (CPluginWin::CreateStreamL):
  • ResourceLoader/CacheSrc/HttpCacheEntry.cpp: (CHttpCacheEntry::~CHttpCacheEntry):
  • ResourceLoader/inc/FileTransaction.h:
  • ResourceLoader/src/BufferTransaction.cpp: (CBufferTransaction::HandleResolveComplete):
  • ResourceLoader/src/FileTransaction.cpp: (CFileTransaction::DoSubmitL): (CFileTransaction::ParseFileNameL): (CFileTransaction::ContentTypeLC):
  • ResourceLoader/src/HttpTransaction.cpp: (CHttpTransaction::MHFRunL):
  • ResourceLoader/src/UrlResponseHeaderInfo.cpp: (CUrlResponseHeaderInfo::SetResponseHeadersL):
10:36 Changeset [24647] by ggaren

Used `svn merge -r24646:24645' to revert project changes I accidentally
checked in.

  • WebCore.xcodeproj/project.pbxproj:
10:26 Changeset [24646] by ggaren

Reviewed by Darin Adler.

Fixed <rdar://problem/5241167> REGRESSION: PLT 1% slower due to r21793
(global page cache cap)

I just measured a .1%-.2% regression in PLT average, but a .4% win in
PLT fastest run, so I think this regression is licked.

Surprisingly, the combination of wtf data structures I originally used
for PageCache turned out to be too slow, so I reworked PageCache to use
a hand-crafted linked list for LRU tracking (instead of ListHashSet),
to store its CachedPages in the HistoryItems to which they corresponded
(instead of using a HashMap), and to ref/deref HistoryItems manually
(instead of using RefPtr, which would ref once for every reference in
the linked list).

  • history/HistoryItem.h: Added data members for PageCache data. Made PageCache a friend so it could access them. Ideally (from an encapsulation perspective), we would store this data an independent structure, but that's just too slow.
  • history/PageCache.cpp: Added a hand-crafted linked list for LRU tracking. (WebCore::PageCache::PageCache): (WebCore::PageCache::add): Adopt the HistoryItem's ref when it's added. (WebCore::PageCache::remove): deref() the Historyitem when it's removed. (WebCore::PageCache::prune): (WebCore::PageCache::addToLRUList): (WebCore::PageCache::removeFromLRUList):
  • history/PageCache.h: (WebCore::PageCache::get): Inlined just in case.
10:19 Changeset [24645] by andersca

Reviewed by John Sullivan.

Stop the test from dumping load callbacks as they differ on Tiger and Leopard.
This is done by removing the call to dumpResourceLoadCallbacks and moving the test away
from the "loading" directory.

  • http/tests/loading/cached-scripts-expected.txt: Removed.
  • http/tests/loading/cached-scripts.html: Removed.
  • http/tests/loading/resources/cached-scripts-subframe.html: Removed.
  • http/tests/misc/cached-scripts-expected.txt: Copied from http/tests/loading/cached-scripts-expected.txt.
  • http/tests/misc/cached-scripts.html: Copied from http/tests/loading/cached-scripts.html.
  • http/tests/misc/resources/cached-scripts-subframe.html: Copied from http/tests/loading/resources/cached-scripts-subframe.html.
09:21 Changeset [24644] by treat

Launch the context menu in the right place.

08:00 Changeset [24643] by treat

Patch from Qing Zhao that adds a signal for history notification.

07:47 Changeset [24642] by hausmann

Use QMAKE_LIBDIR instead of QMAKE_LFLAGS to specify the library search path for QtWebKit, as the former is portable and correctly transformed to /LIBPATH: with msvc for example.

06:26 BuildingQtOnWindows edited by hausmann@webkit.org
06:24 BuildingQtOnWindows edited by hausmann@webkit.org
06:06 BuildingQtOnWindows edited by hausmann@webkit.org
05:59 BuildingQtOnWindows edited by hausmann@webkit.org
05:55 BuildingQtOnWindows created by hausmann@webkit.org
05:51 QtWebKit edited by hausmann@webkit.org
02:41 Applications using WebKit edited by jcverdie@sand-labs.org
02:36 Changeset [24641] by hausmann

Ensure QtLauncher and DumpRenderTree are linked against the uninstalled QtWebKit library by making sure $$OUTPUT_DIR/lib appears before $QTDIR/lib in the library search path and rpath.

00:39 Changeset [24640] by rwlbuis

Reviewed by Nikolas.

Removing an attribute has no effect on SVG

Update the transform matrix also when transform attribute is empty.

00:11 Changeset [24639] by mjs
  • grammar fix pointed out by Mitz.
  • projects/goals.html:
Note: See TracTimeline for information about the timeline view.