Context Navigation

← Previous Period
Next Period →

Timeline

Jun 22, 2007:

4:06 PM Changeset in webkit [23747] by adele

2 edits in trunk/WebKit

Reviewed by Geoff and Darin.

Fix for: <rdar://problem/5239236> Other slash characters should not be permitted as part of a domain name

Misc/WebNSURLExtras.m: (isSlashOrPeriodLookalike): (allCharactersInIDNScriptWhiteList):

3:19 PM Changeset in webkit [23746] by andersca

2 edits in trunk/WebCore

Reviewed by Darin.

plugins/win/PluginStreamWin.cpp: (WebCore::PluginStreamWin::didFinishLoading): No need to fetch the size from resourceData.

3:16 PM Changeset in webkit [23745] by andersca

3 edits in trunk/WebCore

Reviewed by Darin.

<rdar://problem/5288617>
Crash playing a very large QuickTime movie inside Safari.

Get rid of m_completeDeliveryData and use the loader's resource data instead. This reduces the memory usage for
plug-ins which loads large files and use a transfer mode of NP_ASFILE or NP_ASFILEONLY.

plugins/win/PluginStreamWin.cpp: (WebCore::PluginStreamWin::PluginStreamWin): (WebCore::PluginStreamWin::~PluginStreamWin): (WebCore::PluginStreamWin::didReceiveData): (WebCore::PluginStreamWin::didFinishLoading):
plugins/win/PluginStreamWin.h:

3:03 PM Changeset in webkit [23744] by bdakin

6 edits
4 adds in trunk

LayoutTests:

Reviewed by Oliver and Darin.

Tests for <rdar://problem/5266999> Loading page from MangleMe fuzz
tester crashes in RenderTableRow::addChild (giant ROWSPAN)

The results of this test have changed because we now clamp rowspan
at 8190 to match Firefox.

fast/table/giantRowspan-expected.checksum:
fast/table/giantRowspan-expected.png:
fast/table/giantRowspan-expected.txt:

A new test that crashes our old heuristic for too-large rowspan,
but succeeds with the new one.

fast/table/giantRowspan2-expected.checksum: Added.
fast/table/giantRowspan2-expected.png: Added.
fast/table/giantRowspan2-expected.txt: Added.
fast/table/giantRowspan2.html: Added.

WebCore:

Reviewed by Oliver and Darin.

Fix for <rdar://problem/5266999> Loading page from MangleMe fuzz
tester crashes in RenderTableRow::addChild (giant ROWSPAN)

We already have code that tries to prevent allocating too-large
rowspans. The rowspan in this test, however, was small enough that
we thought it was okay in that code, even though clearly it was
still too large. I did a bunch of testing in Firefox, and it seems
that they set a hard limit on the maximum rowspan right in the DOM
-- 8k. So that is what this patch implements.

html/HTMLTableCellElement.cpp: (WebCore::HTMLTableCellElement::parseMappedAttribute):

2:57 PM Changeset in webkit [23743] by justing

4 edits
4 adds in trunk

LayoutTests:

Reviewed by Darin.

<rdar://problem/5272440> Adding To Do via Note received in e-mail adds newlines when pressing return

Demonstrates bug:

editing/deleting/5272440-expected.checksum: Added.
editing/deleting/5272440-expected.png: Added.
editing/deleting/5272440-expected.txt: Added.
editing/deleting/5272440.html: Added.

upstream() now works for the positions visually equivalent to [span, 0]
where span is an inline root editable element:

editing/deleting/delete-mixed-editable-content-001-expected.txt:

WebCore:

Reviewed by Darin.

<rdar://problem/5272440> Adding To Do via Note received in e-mail adds newlines on delete

Deletion inserts a placeholder after it removes full paragraphs if
one doesn't already exist. The code that looked for a pre-existing
placeholder didn't work because of a bug in enclosingNodeOfType.

editing/htmlediting.cpp: (WebCore::enclosingNodeOfType): We start looking at node->parentNode(), but don't go above an editing root.

12:47 PM Changeset in webkit [23742] by andersca

1 edit
1 add in trunk/LayoutTests

Add result.

fast/loader/user-style-sheet-resource-load-callbacks-expected.txt: Added.

12:04 PM Changeset in webkit [23741] by andersca

20 edits
2 adds in trunk

LayoutTests:

Reviewed by Geoff.

<rdar://problem/5244734>
WebKit shouldn't send delegate resource load methods for the user stylesheet (doing so causes assertions in Safari)

fast/loader/resources/user-style-sheet.css: Added.
fast/loader/user-style-sheet-resource-load-callbacks.html: Added.

WebCore:

Reviewed by Maciej and Geoff.

<rdar://problem/5244734>
WebKit shouldn't send delegate resource load methods for the user stylesheet (doing so causes assertions in Safari)

Add a sendResourceLoadCallbacks argument to the ResourceLoader and only send resource load callbacks if it's true. Make it so that
loading the user style sheet creates a resource loader with sendResourceLoadCallbacks set to false (done by adding a sendResourceLoadCallbacks
argument to a bunch of classes).

loader/Cache.cpp: (WebCore::createResource): (WebCore::Cache::requestResource):
loader/Cache.h:
loader/CachedCSSStyleSheet.cpp: (WebCore::CachedCSSStyleSheet::CachedCSSStyleSheet):
loader/CachedCSSStyleSheet.h:
loader/CachedResource.cpp: (WebCore::CachedResource::CachedResource):
loader/CachedResource.h: (WebCore::CachedResource::sendResourceLoadCallbacks):
loader/DocLoader.cpp: (WebCore::DocLoader::requestCSSStyleSheet): (WebCore::DocLoader::requestResource): (WebCore::DocLoader::checkCacheObjectStatus):
loader/DocLoader.h:
loader/MainResourceLoader.cpp: (WebCore::MainResourceLoader::MainResourceLoader):
loader/Request.cpp: (WebCore::Request::Request):
loader/Request.h: (WebCore::Request::sendResourceLoadCallbacks):
loader/ResourceLoader.cpp: (WebCore::ResourceLoader::ResourceLoader): (WebCore::ResourceLoader::willSendRequest): (WebCore::ResourceLoader::didReceiveResponse): (WebCore::ResourceLoader::didReceiveData): (WebCore::ResourceLoader::didFinishLoadingOnePart): (WebCore::ResourceLoader::didFail): (WebCore::ResourceLoader::didCancel):
loader/ResourceLoader.h: (WebCore::ResourceLoader::sendResourceLoadCallbacks):
loader/SubresourceLoader.cpp: (WebCore::SubresourceLoader::SubresourceLoader): (WebCore::SubresourceLoader::create):
loader/SubresourceLoader.h:
loader/loader.cpp: (WebCore::Loader::load): (WebCore::Loader::servePendingRequests):
loader/loader.h:
loader/mac/NetscapePlugInStreamLoaderMac.mm: (WebCore::NetscapePlugInStreamLoader::NetscapePlugInStreamLoader):

11:27 AM Changeset in webkit [23740] by thatcher

6 edits in trunk

LayoutTests:

Reviewed by Anders.

<rdar://problem/5228168> Leopard regression test failures: fast/applescript

New results to match the DumpRenderTree changes.

fast/AppleScript/001-expected.txt:
fast/AppleScript/array-expected.txt:
fast/AppleScript/date-expected.txt:

WebKitTools:

Reviewed by Anders.

<rdar://problem/5228168> Leopard regression test failures: fast/applescript

There were some changes in Leopard that made our test output inconsistent with Tiger.

NSAppleEventDescriptor no longer returns a stringValue for typeType descriptors
on Leopard, so output our own readable string for typeType descriptors.

NSArray's description also changed on Leopard to output more whitespace, so
make our own string representation for typeAEList descriptors. This requires
a special case typeUnicodeText too, so the output is a quoted string.

DumpRenderTree/AppleScriptController.m: (convertAEDescToObject):

10:28 AM Changeset in webkit [23739] by yongjzha

2 edits in S60/branches/3.1m/WebCore

2007-06-21 w3liu <wei.liu@nokia.com>

Reviewed by yongjun.zhang@nokia.com.
DESC: SCHY-72JNXK: 3.1 Voda: The alt text and broken images are not displayed correctly for IMG element.
http://bugs.webkit.org/show_bug.cgi?id=14243

WARNING: NO TEST CASES ADDED OR CHANGED

khtml/rendering/render_image.cpp: (RenderImage::setPixmap): (RenderImage::paint):

9:56 AM Changeset in webkit [23738] by thatcher

2 edits in trunk/WebCore

Reviewed by Tim Hatcher.

Bug 14305: Web Inspector marks local files with server-side compression tip
http://bugs.webkit.org/show_bug.cgi?id=14305

page/inspector/Resource.js: (WebInspector.Resource._shouldCompress): Prevent server-side compression tip from being given to local files

12:00 AM Changeset in webkit [23737] by bdash

3 edits
2 adds in trunk

2007-06-21 Mark Rowe <mrowe@apple.com>

Reviewed by Darin.

Fix for <rdar://problem/5266928> REGRESSION: crash/assert failure beneath RenderFrameSet::layOutAxis when loading

platform/StringImpl.cpp: (WebCore::StringImpl::toLengthArray): Simplify whitespace before checking for an empty string. This avoids returning a length of zero when the string is collapsed to empty by whitespace simplification.

2007-06-21 Mark Rowe <mrowe@apple.com>

Reviewed by Darin.

Test for <rdar://problem/5266928> REGRESSION: crash/assert failure beneath RenderFrameSet::layOutAxis when loading

fast/frames/frame-set-whitespace-attributes-expected.txt: Added.
fast/frames/frame-set-whitespace-attributes.html: Added.

Jun 21, 2007:

11:05 PM Changeset in webkit [23736] by hyatt

2 edits in trunk/WebCore

Fix large font size crash on Windows.

10:01 PM Changeset in webkit [23735] by andrew

2 edits in trunk/WebCore

Reviewed by Timothy.

Stop shift+scroll wheel performing a horizontal scroll in Web Inspector

page/inspector/inspector.css: Set body margin to 0

9:11 PM Changeset in webkit [23734] by thatcher

1 delete in trunk/WebKit/win/WebInspector

Remove some empty directories.

8:29 PM Changeset in webkit [23733] by staikos

9 edits in trunk

Patch from Adam Treat to implement default icons / resources for Qt. Also does
a small cleanup to QWebPage::icon. This is a great example of how QWebSettings
is not ideal.

6:40 PM Changeset in webkit [23732] by oliver

3 edits in trunk/WebCore

Reviewed by Darin.

Use RetainPtr for holding currentEvent. This removes explicit HardRetain/Release calls
and makes reference handling simpler in advance of IME refactoring.

WebCore.xcodeproj/project.pbxproj:
page/mac/EventHandlerMac.mm: (WebCore::currentEvent): (WebCore::EventHandler::currentNSEvent): (WebCore::EventHandler::wheelEvent): (WebCore::EventHandler::keyEvent): (WebCore::lastEventIsMouseUp): (WebCore::EventHandler::passMouseDownEventToWidget): (WebCore::EventHandler::eventLoopHandleMouseDragged): (WebCore::EventHandler::eventLoopHandleMouseUp): (WebCore::EventHandler::passSubframeEventToSubframe): (WebCore::EventHandler::passWheelEventToWidget): (WebCore::EventHandler::mouseDown): (WebCore::EventHandler::mouseDragged): (WebCore::EventHandler::mouseUp): (WebCore::EventHandler::mouseMoved):

5:41 PM Changeset in webkit [23731] by weinig

1 edit
1 delete in trunk/WebKit

Reviewed by Antti.

Remove empty directories

WebInspector/webInspector: directory removed.
WebInspector/webInspector/Images: directory removed

5:26 PM Changeset in webkit [23730] by staikos

7 edits in trunk/WebCore

Add rendered and stylable scrollbars for Qt

5:11 PM Changeset in webkit [23729] by sfalken

1 copy in tags/Safari-522.13.1b

New tag.

4:05 PM Changeset in webkit [23728] by ggaren

7 edits
2 adds in trunk

LayoutTests:

Reviewed by Dave Hyatt.

Fixed <rdar://problem/4977124> | http://bugs.webkit.org/show_bug.cgi?id=12646
REGRESSION: RapidWeaver's "Hello" HTML page doesn't display any DHTML
effects when clicking on its links (12646)

Added layout test for this bug:

fast/invalid/nestedh3s-rapidweaver-expected.txt: Added.
fast/invalid/nestedh3s-rapidweaver.html: Added.

Reversed the expected results for this test because I rolled out r7703:

fast/invalid/nestedh3s-expected.checksum:
fast/invalid/nestedh3s-expected.png:
fast/invalid/nestedh3s-expected.txt:
fast/invalid/nestedh3s.html:

WebCore:

Reviewed by Dave Hyatt.

Fixed <rdar://problem/4977124> | http://bugs.webkit.org/show_bug.cgi?id=12646
REGRESSION: RapidWeaver's "Hello" HTML page doesn't display any DHTML
effects when clicking on its links (12646)

I'm rolling our r7703 for these reasons:

It broke RapidWeaver.

WebKit is now orders of magnitude faster when handling deep nesting (5 seconds vs 3 minutes on my MBP in this particular case).

Firefox, IE, and HTML5 all allow nested heading tags.

Hyatt said so.

html/HTMLParser.cpp: (WebCore::HTMLParser::parseToken):

3:34 PM Changeset in webkit [23727] by weinig

2 edits in trunk/WebCore

Reviewed by Tim Hatcher.

Remove assert for now as it is causing the back button to not
work in debug builds.

page/InspectorController.cpp: (WebCore::InspectorController::didCommitLoad):

2:45 PM Changeset in webkit [23726] by justing

2 edits in trunk/WebKit

Reviewed by Tim.

<rdar://problem/5237524>
REGRESSION: Keyboard commands don't work in a message window until you click inside the message

When you open a message in its own window, Mail creates an empty
WebView, makes that WebView firstResponder and then sets off a
load inside that WebView.

When we're asked to create the empty WebView, we put an empty
WebHTMLView inside it (in r21367 we began creating a document
for empty frames). When Mail makes the WebView first responder
we make that empty WebHTMLView firstResponder. Then when the
load finishes we create a new WebHTMLView and set it as the
document view.

Inside _setDocumentView, if the old document view or one of its
descendants was the first responder, we'd makeFirstResponder:nil
so that the window wouldn't be left with a firstResponder that
was no longer inside of it. This change fixes the bug by instead
transferring firstResponder status to the new document view.

We could also fix this by not allowing the WebHTMLView to become
firstResponder when it's in the provisional state mentioned above.

WebView/WebFrameView.mm: (-[WebFrameView _setDocumentView:]):

2:35 PM Changeset in webkit [23725] by thatcher

1 copy in tags/Safari-5522.9

New tag.

2:26 PM Changeset in webkit [23724] by thatcher

4 edits in trunk

Versioning

2:14 PM Changeset in webkit [23723] by andersca

9 edits in trunk

LayoutTests:

Reviewed by Geoff.

Update test results.

http/tests/loading/bad-scheme-subframe-expected.txt:
http/tests/loading/bad-server-subframe-expected.txt:
http/tests/loading/basic-expected.txt:
http/tests/loading/empty-subframe-expected.txt:
http/tests/loading/simple-subframe-expected.txt:
webarchive/loading/test-loading-archive-expected.txt:

WebKitTools:

Reviewed by Geoff.

Don't dump didFirstLayout callbacks, they happen intermittently.

DumpRenderTree/FrameLoadDelegate.m:

1:30 PM Changeset in webkit [23722] by lars

2 edits in trunk/WebCore

Remove some debug output.

1:21 PM Changeset in webkit [23721] by andersca

2 edits in trunk/LayoutTests

Reviewed by Geoff.

Update test result.

fast/dom/xmlhttprequest-invalid-values-expected.txt:

12:28 PM Changeset in webkit [23720] by staikos

2 edits in trunk/WebCore

Fix the Qt Mac OS X build

12:19 PM Changeset in webkit [23719] by lars

26 edits in trunk/LayoutTestResults

update test results after my font refactoring.

12:01 PM Changeset in webkit [23718] by ap

6 edits
1 add in trunk

Reviewed by Anders.

http://bugs.webkit.org/show_bug.cgi?id=13905
REGRESSION: A Chinese dictionary widget has a problem with XMLHttpRequest response encoding

Tiger WebKit accidentally looked at HTML Http-Equiv META to determine the response encoding
(regardless of the actual response MIME type). This was corrected a while ago to match other
browsers and the draft XHR spec.

Now restoring this functionality for "text/html", with spec editor's blessing.

xml/xmlhttprequest.cpp: (WebCore::XMLHttpRequest::responseMIMEType): (WebCore::XMLHttpRequest::responseIsXML): (WebCore::XMLHttpRequest::didReceiveData):
xml/xmlhttprequest.h:

11:15 AM Changeset in webkit [23717] by sullivan

4 edits in trunk/WebKit

Reviewed by Adele

fixed <rdar://problem/5268673> REGRESSION: Context menu missing for PDF in frame when there's no selection

This was surprisingly interesting. It turns out that at least for Safari, the method [WebDefaultUIDelegate
webView:contextMenuItemsForElement:defaultMenuItems:] is only exercised from WebPDFView these days. It mimics
some of the code that was moved to WebCore as part of the Great Context Menu Refactoring of 2006, but is
independent of that code. And it was partly broken/incomplete, probably as a result of said refactoring.

DefaultDelegates/WebDefaultContextMenuDelegate.mm: (localizedMenuTitleFromAppKit()): deleted this function since I removed all callers (-[WebDefaultUIDelegate menuItemWithTag:target:representedObject:]): Handle WebMenuItemTagOpenFrameInNewWindow tag, since code later in this file was relying on it. Added ASSERT_NOT_REACHED to the default case, since returning nil from this method is bad. Also, use WebKit versions of menu title strings rather than AppKit versions. We added these strings to WebKit a while back so we don't need to sneakily find them in AppKit anymore. (-[WebDefaultUIDelegate webView:contextMenuItemsForElement:defaultMenuItems:]): make the WebView be the target of the Open Frame in New Window item; previously it was self, but that was just silly (probably a leftover from some code shuffling when context menus were reimplemented)

WebView/WebView.mm: (-[WebView _openFrameInNewWindowFromMenu:]): new method, used by Open Frame in New Window menu item.

WebView/WebViewInternal.h: Declare new method used by menu item. This isn't necessary for compilation, but could prevent the accidental deletion of the method implementation.

8:04 AM Changeset in webkit [23716] by spadma