Timeline



Aug 3, 2002:

1:21 PM Changeset in webkit [1728] by darin
  • 4 edits in trunk/WebCore
  • fixed 2943519 - • is rendered as an asterisk rather than a bullet
  • khtml/html/htmltokenizer.cpp: (fixUpChar): Replaced the macro of this name by an inline function that follows the Unicode specification. The macro was both making Windows Latin 1 extensions character codes (like 149 for bullet) work, but also mapping to characters that were suitable for Qt. Since OS X has no problem with the real Unicode values, we don't need that hack.

Aug 2, 2002:

10:53 PM Changeset in webkit [1727] by darin
  • 46 edits
    6 adds in trunk

WebCore:

  • khtml/misc/decoder.cpp: (Decoder::setEncoding): Put some hacks for handling Hebrew text encoding inside ifndef APPLE_CHANGES. It's not clear that these or the equivalent are needed for WebCore (they are related to the as-yet unimplemented bidirectional text support) and they were making us crash.
  • fixed 2949235 -- checkboxes in wrong place until you scroll away and back

I fixed this once before, then removed that fix in favor of a simpler one.
But the simpler one didn't work, and I didn't test it properly. Now I have
a fix that works without the drawbacks of the original.

  • khtml/rendering/render_replaced.cpp: (RenderWidget::layout): Remove the call to move the widget aside. This does no good because it's only called if the widget is told it needs to layout, which only happens if it's resized.
  • khtml/rendering/render_root.cpp: (RenderRoot::layout): Call a KWQ function so we can do some work at layout time.
  • kwq/KWQKHTMLPartImpl.mm: (moveWidgetsAside): Added. Function used recursively to move all widgets aside. (KWQKHTMLPartImpl::layout): Added. Called at layout time; moves all the widgets aside. They are moved back when print functions are called.

WebCore support for feature where client redirects are treated as if the
page was continuing to load. A big part of this was changing redirection
so that it uses QTimer, allowing us to use more of the original KHTMLPart
code. I added a feature to QTimer that KWQKHTMLPartImpl uses to monitor
what KHTMLPart does with the timer.

  • kwq/qt/qobject.h: Changed "slots" definition so that all slots are public (to help KWQSlot get at them). Added a prefix for the slot and signal strings, so that we can detect the case where it's trying to connect a slot to another slot. Implement the convenience version of connect here in the header. Implement blockSignals. Remove setTarget. Make target mutable. Add new m_signalListHead and m_signalsBlocked.
  • kwq/KWQObject.mm: (QObject::findSignal): Added. Helper function that locates a signal by name. (QObject::connect): Connects a signal to a slot using the new KWQSignal and KWQSlot. (QObject::disconnect): Disconnects, similar to the above. (QObject::QObject): Initialize target, m_signalListHead, and m_signalsBlocked. (QObject::~QObject): Assert that m_signalListHead is 0.
  • khtml/khtml_part.cpp: (KHTMLPart::init): Move the code that connects the redirection timer out of APPLE_CHANGES. (KHTMLPart::~KHTMLPart): Remove the APPLE_CHANGES around the code that stops the redirection timer. (KHTMLPart::restoreURL): Remove the APPLE_CHANGES around the code that stops the redirection timer. (KHTMLPart::openURL): Move the APPLE_CHANGES so the code that stops the redirection timer will run. (KHTMLPart::closeURL): Remove the APPLE_CHANGES around the code that stops the redirection timer. (KHTMLPart::scheduleRedirection): Remove most of this function from the APPLE_CHANGES. We still probably need to revisit the handling of m_bComplete.
  • kwq/KWQSignal.h: Added.
  • kwq/KWQSignal.mm: Added.
  • kwq/KWQSlot.h: Added.
  • kwq/KWQSlot.mm: Added.
  • WebCore.pbproj/project.pbxproj: Added KWQSignal.h/mm, KWQSlot.h/mm.
  • khtml/misc/loader.h: Simplify APPLE_CHANGES ifdefs since all slots are now public.
  • force-clean-timestamp: Touch this since header files were changed.
  • khtml/ecma/kjs_window.cpp: (Window::get): Disable the "name" part of this. We can re-enable it if we need it, but every QObject had a name because of this one piece of code, and I don't think it was effectively setting the frame name. (Window::put): Disable the "name" part of this for the same reason. (History::getValueProperty): Disable the "length" part of this. It was calling some unimplemented stuff. We can re-enable this and implement what we need in KWQ then.
  • khtml/rendering/render_form.h: Remove RenderTextArea::performAction, because I changed this one case to use the new connect machinery. We can convert all the other form items to use the new machinery, and then eliminate performAction altogether.
  • khtml/rendering/render_form.cpp: (RenderSubmitButton::RenderSubmitButton): Remove unneeded setTarget. For the moment, connect takes care of this, and in the long run we won't need to set the target at all. (RenderLineEdit::RenderLineEdit): Ditto.
  • khtml/rendering/render_frames.h:
  • khtml/rendering/render_frames.cpp: Put partLoadingErrorNotify inside ifndef APPLE_CHANGES because it uses a bunch of Qt stuff that's not used anywhere else, and we never call it.
  • kwq/KWQKHTMLPartImpl.h: Remove scheduleRedirection, timerEvent, and m_redirectionTimer, add redirectionTimerStartedOrStopped and layout.
  • kwq/KWQKHTMLPartImpl.mm: (redirectionTimerMonitor): Added. Small stub that calls redirectionTimerStartedOrStopped. (KWQKHTMLPartImpl::KWQKHTMLPartImpl): Set up a monitor on the redirection timer. (KWQKHTMLPartImpl::~KWQKHTMLPartImpl): Remove now-unneeded killTimer call. (KWQKHTMLPartImpl::jumpToSelection): Simplify by using dynamic_cast instead of render_name(). (KWQKHTMLPartImpl::redirectionTimerStartedOrStopped): Added. Calls bridge to report client redirect events whenever the redirection timer is started or stopped.
  • kwq/qt/qtextedit.h: Define a KWQSignal for textChanged, and a function that calls it.
  • kwq/KWQTextEdit.mm: (QTextEdit::QTextEdit): Set up a KWQSignal for textChanged.
  • kwq/KWQTextArea.mm: (-[KWQTextArea textDidEndEditing:]): Instead of calling emitAction, call textChanged, which will emit the signal using the new signal machinery.
  • kwq/qt/qtimer.h:
  • kwq/KWQTimer.mm: (+[KWQTimerTarget targetWithQTimer:]): Added. Used to implement QTimer in terms of NSTimer. (-[KWQTimerTarget timerFired:]): Added. Used to implement QTimer in terms of NSTimer. (QTimer::QTimer): Added. (QTimer::isActive): Implemented. (QTimer::start): Implemented. (QTimer::stop): Implemented. (QTimer::setMonitor): Implemented. (QTimer::fire): Implemented.
  • kwq/WebCoreBridge.h: Changed name of gotoAnchor to scrollToAnchor. Added client redirect methods.
  • kwq/WebCoreBridge.mm: (-[WebCoreBridge scrollToAnchor:]): Changed name.

WebKit:

WebKit support for feature where client redirects are treated as if the
page was continuing to load.

  • WebCoreSupport.subproj/WebBridge.m: (-[WebBridge reportClientRedirectTo:delay:fireDate:]): Added. Passes the client redirect along to the location change handler. (-[WebBridge reportClientRedirectCancelled]): Passes client redirect cancel along to the location change handler. (-[WebBridge dataSourceChanged]): Remove hack related to dummy data source. (-[WebBridge dataSource]): Remove hack related to dummy data source.
  • WebView.subproj/WebLocationChangeHandler.h: Added client redirect methods. Added class for base implementation of protocol. Updated stale comment.
  • WebView.subproj/WebLocationChangeHandler.m: Added. Provides base implementation of the WebLocationChangeHandler protocol so that clients don't need to implement all the methods if they are only interested in some of them.
  • WebKit.pbproj/project.pbxproj: Added WebLocationChangeHandler.m.
  • WebView.subproj/WebFramePrivate.m: (-[WebFrame _goToItem:withFrameLoadType:]): Call scrollToAnchor under its new name.
  • Misc.subproj/WebIconLoader.h:
  • Misc.subproj/WebIconLoader.m:
  • WebView.subproj/WebControllerPolicyHandler.m:
  • WebView.subproj/WebControllerPolicyHandlerPrivate.h: Replaced MyCompanyName with the actual name of our company.

WebBrowser:

  • fixed 2948387 -- stop button flashes to "go" twice while loading citibank.com
  • LocationChangeHandler.m: (-[LocationChangeHandler locationChangeStartedForDataSource:]): Call _setInstantRedirectPending:NO when it's a load of the main frame. (-[LocationChangeHandler receivedPageTitle:forDataSource:]): Make this call through to the load monitor even for frames other than the main one. (-[LocationChangeHandler serverRedirectTo:forDataSource:]): Make this call through to the load monitor even for frames other than the main one. (-[LocationChangeHandler clientRedirectTo:delay:fireDate:forDataSource:]): Added. Call _setInstantRedirectPending:YES when it's a no-delay redirect of the main frame. (-[LocationChangeHandler clientRedirectCancelledForDataSource:]): Added. Call _setInstantRedirectPending:NO when it's the main frame, and also post a notification.
  • BrowserDocument.h: Added BrowserDocumentClientRedirectCancelledNotification and _instantRedirectPending instance variable.
  • BrowserDocument.m: (-[BrowserDocument isLoading]): Return YES if _instantRedirectPending. (-[BrowserDocument _setInstantRedirectPending:]): Set _instantRedirectPending.
  • BrowserDocumentPrivate.h: Declare _setInstantRedirectPending.
  • BrowserWindowController.h: Renamed frameLoadRedirected to mainFrameRedirected.
  • BrowserWindowController.m: (-[BrowserWindowController windowDidLoad]): Connect the new BrowserDocumentClientRedirectCancelledNotification to frameLoadPossiblyFinished. (-[BrowserWindowController mainFrameRedirected]): Renamed from frameLoadRedirected and removed the parameter since it is called only for the main frame. (-[BrowserWindowController frameLoadPossiblyFinished:]): Renamed from frameLoadFinished since a frame load is not necessarily finished when this is called. Also move the work back in here from respondToFrameLoadFinished to undo my previous attempt at this feature.
  • Test/PageLoadTestRunner.m: (-[PageLoadTestRunner init]): Connect the new BrowserDocumentClientRedirectCancelledNotification to pageLoadPossiblyDone. (-[PageLoadTestRunner pageLoadPossiblyDone:]): Renamed from pageLoadDone since the page load is not necessarily finished when this is called. Also move the work back in here from respondToPageLoadDone to undo my previous attempt at this feature.
  • BrowserNSControlExtras.h:
  • BrowserNSControlExtras.m:
  • ContextMenuHandler.h:
  • ContextMenuHandler.m:
  • TitleBarButton.h:
  • TitleBarButton.m: Replaced MyCompanyName with the actual name of our company.
5:45 PM Changeset in webkit [1726] by hyatt
  • 19 edits in trunk/WebCore

Lots of form control changes.

  • Inputs and textareas now respond to setFont and can have their font set by the Web page explicitly. These fields emulate the WinIE behavior of defaulting to the UI system font and only picking up new fonts if they are explicitly specified on the element itself.
  • Buttons and selects use the small versions now.
  • Stopped trying to align controls along baselines of their interior text. Konqueror is the only browser that does this, and it looks bad on many pages. With this patch controls align along their bottom borders (which extend slightly below the baseline of surrounding text). This matches Gecko, MacIE, and WinIE behavior.
  • Stopped using monospace for textareas and inputs.
  • Fixed a bug with inputs not calling setScrollable on their cells.
  • khtml/css/html4.css:
  • khtml/rendering/render_form.cpp: (RenderSubmitButton::calcMinMaxWidth): (RenderSubmitButton::baselinePosition): (RenderLineEdit::baselinePosition): (RenderSelect::baselinePosition):
  • khtml/rendering/render_form.h:
  • kwq/KWQButton.mm: (QButton::QButton): (QButton::sizeHint):
  • kwq/KWQComboBox.mm: (QComboBox::init):
  • kwq/KWQLineEdit.mm: (QLineEdit::cursorPosition): (QLineEdit::setFont):
  • kwq/KWQNSTextField.mm: (-[KWQNSTextField initWithFrame:widget:]):
  • kwq/KWQView.mm: (-[KWQNSButtonCell cellSizeForBounds:]): (+[KWQNSButton initialize]):
  • kwq/WebCoreTextRendererFactory.h:
  • kwq/WebCoreTextRendererFactory.m: (-[WebCoreTextRendererFactory fontWithFamily:traits:size:]):
  • kwq/qt/qbutton.h:
  • kwq/qt/qlineedit.h:
2:44 PM Changeset in webkit [1725] by sullivan
  • 3 edits in trunk/WebKit
  • fixed 3015705 -- Command-up-arrow and -down-arrow should go to beginning and end of web page
  • WebView.subproj/WebView.m: (-[WebView keyDown:]): Added checks for command key.

Aug 1, 2002:

6:51 PM Changeset in webkit [1724] by rjw
  • 11 edits in trunk/WebKit

Effort towards 3014555 (going back to frame), but not fixed.
Need to think more about appropriate solution and compromises.

  • History.subproj/WebBackForwardList.h:
  • History.subproj/WebBackForwardList.m: (-[WebBackForwardList backEntryAtIndex:]):
  • History.subproj/WebHistoryItem.h:
  • History.subproj/WebHistoryItem.m: (-[WebHistoryItem initWithURL:title:]): (-[WebHistoryItem initWithURL:title:image:]): (-[WebHistoryItem initWithURL:target:parent:title:image:]): (-[WebHistoryItem dealloc]): (-[WebHistoryItem parent]): (-[WebHistoryItem setTitle:]): (-[WebHistoryItem setTarget:]): (-[WebHistoryItem setParent:]): (-[WebHistoryItem setDisplayTitle:]):
  • WebCoreSupport.subproj/WebBridge.m:
  • WebView.subproj/WebController.m: (-[WebController _goToItem:withFrameLoadType:]):
  • WebView.subproj/WebFramePrivate.h:
  • WebView.subproj/WebFramePrivate.m: (-[WebFrame _transitionToCommitted]): (-[WebFrame _isLoadComplete]): (-[WebFrame _scrollToTop]):
4:44 PM Changeset in webkit [1723] by cblu
  • 3 edits in trunk/WebKit

WebKit:

Only send command key events if the plug-in is in the responder chain.

  • Plugins.subproj/WebPluginView.m: (-[WebPluginView isInResponderChain]): (-[WebPluginView performKeyEquivalent:]): (-[WebPluginView setWindow]): added some more logging

WebBrowser:

Fixed: 2897678 - send urls from LS to new windows

  • AppController.m: (-[AppController handleGURLEvent:withReplyEvent:]):
3:40 PM Changeset in webkit [1722] by darin
  • 4 edits in trunk/WebCore
  • fixed 2841277 -- support for charset=gb2312
  • kwq/character-sets.txt: Added "euc-cn" to the list of aliases for "gb2312", since that's the name used by CFStringEncoding for that character set.
2:48 PM Changeset in webkit [1721] by mjs
  • 7 edits in trunk/WebCore

Remove KWQKHTMLPartImpl's setBaseURL method and m_baseURL field,
since nothing should really need them any more.

  • kwq/KWQKHTMLPartImpl.h:
  • kwq/KWQKHTMLPartImpl.mm: (KWQKHTMLPartImpl::begin): Set base URL on document, if it exists. (KWQKHTMLPartImpl::requestFrame): Get base URL from document. (KWQKHTMLPartImpl::requestObject): Ditto. (KWQKHTMLPartImpl::frames): Ditto. (KWQKHTMLPartImpl::setBaseURL): Removed.
2:26 PM Changeset in webkit [1720] by cblu
  • 3 edits in trunk/WebKit

Only send command keyDown once by ignoring the keyDown that comes after performKeyEquivalent.

  • Plugins.subproj/WebPluginView.m: (-[WebPluginView keyDown:]):
2:09 PM Changeset in webkit [1719] by cblu
  • 3 edits in trunk/WebKit

Fixed: 2970530 - Command keys are not getting passed in to plugins

  • Plugins.subproj/WebPluginView.m: (-[WebPluginView performKeyEquivalent:]):
1:31 PM Changeset in webkit [1718] by rjw
  • 18 edits in trunk

WebKit:

More back/forward word. Support for restoring scroll position.
We still have a big remaining issue. We don't record the entire
frame tree in back/forward items. This means going back or forward
to a frame will looose the context of the containing frameset,
see 3014555.

  • History.subproj/WebBackForwardList.m: (-[WebBackForwardList currentEntry]):
  • History.subproj/WebHistoryItem.h:
  • History.subproj/WebHistoryItem.m: (-[WebHistoryItem anchor]): (-[WebHistoryItem setAnchor:]):
  • WebCoreSupport.subproj/WebBridge.m: (-[WebBridge reportError:]):
  • WebView.subproj/WebController.m: (-[WebController _goToItem:withFrameLoadType:]): (-[WebController goBack]): (-[WebController goForward]):
  • WebView.subproj/WebFrame.m: (-[WebFrame setProvisionalDataSource:]):
  • WebView.subproj/WebFramePrivate.h:
  • WebView.subproj/WebFramePrivate.m: (-[WebFrame _transitionToCommitted]): (-[WebFrame _isLoadComplete]): (-[WebFrame _goToItem:withFrameLoadType:]): (-[WebFrame _restoreScrollPosition]): (-[WebFrame _scrollToTop]):

WebBrowser:

Removed old obsolete restore scroll position code.

  • LocationChangeHandler.m: (-[LocationChangeHandler locationChangeStartedForDataSource:]):

WebCore:

Added method to allow WebKit to direct khtml to
goto to an anchor point. Need for back/forward.

  • kwq/KWQKHTMLPartImpl.mm: (KWQKHTMLPartImpl::urlSelected):
  • kwq/WebCoreBridge.h:
  • kwq/WebCoreBridge.mm: (-[WebCoreBridge gotoAnchor:]):

WebFoundation:

Convenience method to remove fragment (target anchor)
from URL.

  • Misc.subproj/WebNSURLExtras.h:
  • Misc.subproj/WebNSURLExtras.m: (-[NSURL _web_URLWithoutFragment]):
1:09 PM Changeset in webkit [1717] by kocienda
  • 4 edits in trunk/WebCore

Tweaked WebCore cache settings. The WebCore cache will now grow
to 1 MB (was 512k). The maximum cacheable item is 16k instead of 40k.

  • khtml/misc/loader.cpp
11:19 AM Changeset in webkit [1716] by kocienda
  • 1 edit in trunk/WebKit/Misc.subproj/WebFileDatabase.m

Added socket loader to get uncached HTTP content without using CFNetwork.

Zoom!

11:03 AM Changeset in webkit [1715] by cblu
  • 3 edits in trunk/WebKit

Fixed:
2938010 - PLUGINS: right-click support
2938011 - PLUGINS: extended key events support
2970523 - Arrow key events not correctly passed to applets

  • Plugins.subproj/WebPluginView.m: (-[WebPluginView modifiersForEvent:]): (-[WebPluginView getCarbonEvent:withEvent:]): (-[WebPluginView mouseDown:]): (-[WebPluginView keyUp:]): (-[WebPluginView keyDown:]): (-[WebPluginView menuForEvent:]):
10:47 AM Changeset in webkit [1714]
  • 3 copies in tags/Alexander-15

This commit was manufactured by cvs2svn to create tag 'Alexander-15'.

10:47 AM Changeset in webkit [1713] by sheridan
  • 8 edits in trunk

Alex-15 versions & changelog markers

8:18 AM Changeset in webkit [1712] by gramps
  • 3 edits in trunk/WebKit

Fixed deployment build compile problem.

2:59 AM Changeset in webkit [1711] by mjs
  • 8 edits in trunk/WebCore

Restored the original KHTMLPart::write() now that none of our
changes in the impl version are beneficial in any way.

  • khtml/khtml_part.cpp: (KHTMLPart::write): Restore original code in lieue of calling impl version.
  • kwq/KWQKHTMLPartImpl.h:
  • kwq/KWQKHTMLPartImpl.mm: (KWQKHTMLPartImpl::openURL): Remove initialization of m_decodingStarted, since it is now gone. (KWQKHTMLPartImpl::write): Removed.
2:36 AM Changeset in webkit [1710] by mjs
  • 7 edits in trunk/WebCore

Removed m_documentSource and remaining ill-conceived uses thereof.

  • kwq/KWQKHTMLPartImpl.h:
  • kwq/KWQKHTMLPartImpl.mm: (KWQKHTMLPartImpl::openURL): Don't initialize it. (KWQKHTMLPartImpl::write): Don't use it.
1:30 AM Changeset in webkit [1709] by mjs
  • 13 edits in trunk

WebCore:

Remove stuff formerly useful for "View Reconstructed Source"
feature but now useless.

  • kwq/KWQKHTMLPartImpl.h:
  • kwq/KWQKHTMLPartImpl.mm: (KWQKHTMLPartImpl::documentSource): Removed.
  • kwq/WebCoreBridge.h:
  • kwq/WebCoreBridge.mm: (-[WebCoreBridge documentTextFromDOM]): Removed.

WebKit:

Remove stuff formerly useful for "View Reconstructed Source"
feature but now useless.

  • WebView.subproj/WebDataSource.h:
  • WebView.subproj/WebDataSource.m: (-[WebDataSource documentTextFromDOM]): Removed.
12:52 AM Changeset in webkit [1708] by mjs
  • 16 edits in trunk/WebKit

Changed things around so that WebFrame, not WebDataSource, owns
WebBridge. This allows us to remove the dummy data source.

  • WebView.subproj/WebFrame.m: (-[WebFrame initWithName:webView:provisionalDataSource:controller:]): Remove all dummy data source-related crud, but call _changeBridge to create the initial bridge. (-[WebFrame setProvisionalDataSource:]): Remove dummy data source-related epicycles.
  • WebView.subproj/WebFramePrivate.h: Add bridge member to private class.
  • WebView.subproj/WebFramePrivate.m: (-[WebFramePrivate setDataSource:]): Don't remove old data source from frame, this is no longer needed and may even be harmful. (-[WebFrame _transitionToCommitted]): Remove much now-unneeded code. (-[WebFrame _isLoadComplete]): Get location change handler from controller. Don't call end on the bridge if the document is not html. (-[WebFrame _changeBridge]): Free the old bridge and make a new one. (-[WebFrame _bridge]): Get it directly, not from the data source.
  • WebView.subproj/WebDataSourcePrivate.h: Remove everything related to the `dummy' concept.
  • WebView.subproj/WebDataSourcePrivate.m: (-[WebDataSource _setFinalURL:]): Reorder code a bit. (-[WebDataSource _bridge]): Get bridge from frame. (-[WebDataSource _commitIfReady]): No more need to set the bridge's frame. (-[WebDataSource _makeRepresentation]): Set data source on the rep. (-[WebDataSource _startLoading:]): Get location change handler from controller. (-[WebDataSource _setTitle:]): Likewise. (-[WebDataSource receivedPageIcon:]): Likewise. (-[WebDataSource _loadIcon]): Likewise.
  • WebView.subproj/WebMainResourceClient.m: (-[WebMainResourceClient WebResourceHandle:dataDidBecomeAvailable:]): Get location change handler from the controller.
  • WebView.subproj/WebViewPrivate.m: (-[WebView _makeDocumentViewForDataSource:]): Change *and* commit the data source. This could probably be collapsed down to one method now.
  • WebView.subproj/WebDocument.h: Added setDataSource: method to the WebDocumentRepresenation protocol. For now it is only useful for the HTML representation, but could potentially be useful for other representation classes. It should also allow the data source to be removed as a parameter from other methods, but I have not done that yet. It may be even better still to pass the data source as a paremeter to the init method.
  • WebView.subproj/WebHTMLRepresentation.m: (-[WebHTMLRepresentation init]): Remove allocation of bridge. (-[WebHTMLRepresentation dealloc]): Remove release of bridge. (-[WebHTMLRepresentation setDataSource:]): Store unretained reference to the data source's frame's bridge.
  • WebView.subproj/WebHTMLViewPrivate.m: (-[WebHTMLView _bridge]): Fetch the bridge from the frame, not the data source.
  • WebView.subproj/WebImageRepresentation.m: (-[WebImageRepresentation setDataSource:]): Added empty implementation.
  • WebView.subproj/WebTextRepresentation.m: (-[WebTextRepresentation setDataSource:]): Added empty implementation.
  • Plugins.subproj/WebPluginStream.m: (-[WebPluginStream setDataSource:]): Added empty implementation.

Jul 31, 2002:

2:57 PM Changeset in webkit [1707] by darin
  • 13 edits in trunk/WebCore
  • kwq/qt/qptrvector.h:
  • kwq/qt/qvector.h:
  • kwq/KWQVectorImpl.h:
  • kwq/KWQVectorImpl.mm:
  • force-clean-timestamp: New implementation of QVector that does not use CFArray. Should fix some data() anomalies. Shares wierdnesses with the original Qt class.
  • WebCore-tests.exp: Remove old QVector entry points and add new ones.
  • kwq/KWQKHTMLPartImpl.mm: (KWQKHTMLPartImpl::write): Did a tiny optimization by initializing the QString rather than first assigning and then initializing. Also reworded some comments and rearranged code to make it clearer that m_decodingStarted is now used only to decide when to call determineParseMode.
  • khtml/css/cssparser.cpp: Removed the APPLE_CHANGES around a change that's fine for non-Apple use as well.

Jul 30, 2002:

7:38 PM Changeset in webkit [1706] by mjs
  • 13 edits in trunk

WebCore:

Fixes of various bugs that prevented Alexander from running the
Mozilla page load test.

  • fixed 3008682 - Alexander cannot run Mozilla-PLT
  • khtml/misc/decoder.cpp: (Decoder::decode): Don't drop the accumulated buffer if the document happens to be unicode!
  • kwq/KWQCString.mm: (operator==): Fix reversed logic when comparing an empty QCString to a const char *.
  • kwq/KWQKHTMLPartImpl.mm: (KWQKHTMLPartImpl::slotData): Set complete flag later so that initializing it to false doesn't clobber setting it. (KWQKHTMLPartImpl::begin): Initialize complete flag to false. (KWQKHTMLPartImpl::write): Don't buffer things up for the decoder - the decoder does that internally.
  • kwq/WebCoreBridge.h, kwq/WebCoreBridge.mm: Remove setURL: method.

WebKit:

Fixes of various bugs that prevented Alexander from running the
Mozilla page load test.

  • fixed 3008682 - Alexander cannot run Mozilla-PLT
  • WebCoreSupport.subproj/WebBridge.m: (-[WebBridge dataSourceChanged]): Don't call setURL: to set a URL that we got from a redirect; just pass it directly to openURL:. The inputURL is irrelevant.
  • WebView.subproj/WebDataSourcePrivate.m: (-[WebDataSource _setFinalURL:]): Don't call setURL: on the bridge; instead, assert that the data source is committed, because it would be a mistake to get a redirect after having received data.
7:14 PM Changeset in webkit [1705] by rjw
  • 4 edits in trunk/WebCore

Don't create unnecessary unicode string. Of the
95695 unnecessary unicode allocations we do loading
the base test, this usage accounted for 7185. Most
of the others come from DOMString to QConstString to
QString to unicode() conversion during parsing. All
that unnecessary work needs to be removed.

  • kwq/KWQColor.mm: (decodeColorFromHexColorString):
4:53 PM Changeset in webkit [1704] by hyatt
  • 4 edits in trunk/WebCore

Ok, this time i've got it. I hope. I think. Fingers crossed.
Putting the table recalc check back in one more time. :)

  • khtml/rendering/render_table.cpp: (RenderTable::addColumns): (RenderTable::addColInfo):
4:30 PM Changeset in webkit [1703] by hyatt
  • 4 edits in trunk/WebCore

Backing myself out again. No crashes, but now pages stall.
#*&*#@%*&!!!!

  • khtml/rendering/render_table.cpp: (RenderTable::recalcColInfo): (RenderTable::addColInfo):
4:07 PM Changeset in webkit [1702] by hyatt
  • 5 edits in trunk/WebCore

Putting my table patch back in. Also optimizing the code
to avoid deleting and recreating the same colinfo over
and over again. Now colinfos can recalc their own
information.

  • khtml/rendering/render_table.cpp: (RenderTable::addColumns): (RenderTable::recalcColInfo): (RenderTable::addColInfo): (RenderTable::calcMinMaxWidth):
  • khtml/rendering/render_table.h:
3:38 PM Changeset in webkit [1701] by rjw
  • 4 edits in trunk/WebCore

Inline critical function.

  • kwq/KWQString.mm: (compareToLatinCharacter):
2:03 PM Changeset in webkit [1700] by rjw
  • 3 edits in trunk/WebKit

i Remove annoying log, it's really annoying. Added FIX ME

comment instead.

  • WebView.subproj/WebFramePrivate.m: (-[WebFrame _transitionToCommitted]):
12:05 PM Changeset in webkit [1699] by rjw
  • 4 edits in trunk/WebCore

Wow! Avoid creation of unicode string when only a single character
is needed. Simple change for big gains, 7% - 10% improvement
on cached pages. Other similar gains may remain to be had.

  • khtml/css/cssparser.cpp:
12:04 PM Changeset in webkit [1698] by sullivan
  • 3 edits in trunk/WebKit

WebKit:

  • fixed 3012083 "Looked for URLs on the pasteboard, but found none" spam dragging nil-target bookmark
  • Misc.subproj/WebNSViewExtras.m: (-[NSView _web_bestURLForDraggingInfo:]): This method was getting data for types on the pasteboard without first checking whether the types were on the pasteboard, a no-no. Restructured this method and tweaked it here and there also.

(-[NSView _web_dragOperationForDraggingInfo:]): moved a fast
test in front of a slow test

WebBrowser:

  • fixed 3011887 -- dropping favorite on a favorites folder makes copy but should move
  • FavoriteButton.m: (-[FavoriteButton determineDragOperation:]): New helper routine to decide between move & copy based on drag source and option key state. (-[FavoriteButton draggingEntered:]), (-[FavoriteButton draggingUpdated:]): use determineDragOperation.
  • ToolbarButtonItem.m: (+[ToolbarButtonItem itemWithIdentifier:button:action:]): tweaked this while investigating some other bug.
11:04 AM Changeset in webkit [1697] by darin
  • 4 edits in trunk/WebCore
  • khtml/css/makevalues: Fail if we don't find gperf.
3:00 AM Changeset in webkit [1696] by hyatt
  • 4 edits in trunk/WebCore

Back myself out. I seem to have stumbled onto another bug.
Until I figure it out, turn this back off.

  • khtml/rendering/render_table.cpp: (RenderTable::addColumns): (RenderTable::addColInfo):
1:51 AM Changeset in webkit [1695] by hyatt
  • 4 edits in trunk/WebCore

Fix for 3000604, 2902173, 2921313. Ensure that tables recalc their
column widths like they are supposed to. This fixes livepage.apple.com,
applemuseum.com, and many other pages (that didn't have bugs filed).
Yay!

  • khtml/rendering/render_table.cpp: (RenderTable::recalcColInfo): (RenderTable::addColInfo):
Note: See TracTimeline for information about the timeline view.