Timeline

Nov 1, 2002:

10:46 PM Changeset in webkit [2539] by darin

• 3 edits in trunk/WebKit

• Plugins.subproj/WebBaseNetscapePluginView.m: (-[WebBaseNetscapePluginView sendEvent:]): Fixed uninitialized variable problem.

6:16 PM Changeset in webkit [2538] by darin

• 5 edits in trunk/WebCore

• applied the same "subclass responsibility" pattern here as in some of the other WebCore classes

• kwq/WebCoreImageRendererFactory.h: Move the subclass responsibility into a protocol.
• kwq/WebCoreImageRendererFactory.m: Remove now-unneeded stubs.

6:12 PM Changeset in webkit [2537] by darin

• 9 edits in trunk/WebKit

• fixed 3090239 -- crash on window close

• WebView.subproj/WebFramePrivate.m: (-[WebFrame _detachFromParent]): Nil out the bridge and release it so we don't close the URL twice.

• fixed 3058598 -- animated gif animates repeatedly in Alex, but only once in other browsers

• WebCoreSupport.subproj/WebImageRenderer.h: Added fields used by bug workaround.
• WebCoreSupport.subproj/WebImageRenderer.m: (-[WebImageRenderer blockHasGIFExtensionSignature:length:]): Helper method for bug workaround. (-[WebImageRenderer checkDataForGIFExtensionSignature:]): Checks incoming data for the GIF extension signatures. (-[WebImageRenderer initWithData:]): Override to call checkDataForGIFExtensionSignature. (-[WebImageRenderer incrementalLoadWithBytes:length:complete:]): Call checkDataForGIFExtensionSignature. (-[WebImageRenderer repetitionCount]): Return 1 if we were going to return 0, but we didn't see a GIF extension signature. This workaround can be removed once we require a system new enough to have the bug fix for bug 3090341. (-[WebImageRenderer nextFrame:]): Stop on last frame, don't wrap around to first.

• WebCoreSupport.subproj/WebImageRendererFactory.h: Update for changes to WebCore.
• WebCoreSupport.subproj/WebImageRendererFactory.m: Tweaks.

• WebView.subproj/WebImageRepresentation.m: Update imports.
• WebView.subproj/WebImageView.m: Update imports.

5:16 PM Changeset in webkit [2536] by rjw

• 10 edits in trunk

i Implemented rendering of diacriticals.

• Misc.subproj/WebUnicode.h:
• Misc.subproj/WebUnicode.m: (_unicodeDigitValue): (_unicodeCategory): (_unicodeDirection): (_unicodeJoining): (_unicodeDecompositionTag): (_unicodeMirrored): (_unicodeCombiningClass): (_unicodeLower): (_unicodeUpper):
• WebCoreSupport.subproj/WebTextRenderer.m: (-[WebTextRenderer drawCharacters:stringLength:fromCharacterPosition:toCharacterPosition:atPoint:withPadding:withTextColor:backgroundColor:rightToLeft:]): (-[WebTextRenderer _floatWidthForCharacters:stringLength:fromCharacterPosition:numberOfCharacters:withPadding:applyRounding:attemptFontSubstitution:widths:fonts:glyphs:numGlyphs:]): (-[WebTextRenderer extendCharacterToGlyphMapToInclude:]):

Removed old implementation.

• kwq/KWQChar.mm: (QChar::direction):

Renamed JoiningCenter -> JoiningCausing.

• kwq/WebCoreUnicode.h:

5:10 PM Changeset in webkit [2535] by darin

• 7 edits in trunk

JavaScriptCore:

• kjs/object.cpp: Make the same change Maciej just did, but to the other constructor right next to the one he changed.

WebFoundation:

• English.lproj/StringsNotToBeLocalized.txt: Update for recent changes.

WebKit:

• English.lproj/StringsNotToBeLocalized.txt: Update for recent changes.

4:28 PM Changeset in webkit [2534] by darin

• 8 edits in trunk/WebCore

• fixed 3062598 -- meta refresh of 0.1s fails

KHTML was using an integer for the delay.

• khtml/khtml_part.h: Make scheduleRedirection take a double.
• khtml/khtmlpart_p.h: Make m_delayRedirect be a double.
• khtml/khtml_part.cpp: (KHTMLPart::slotData): Use a double when parsing meta refresh. (KHTMLPart::checkCompleted): Cast to int when passing to timer start. (KHTMLPart::scheduleRedirection): Cast to int when passing to timer start; also check for absurd values and do nothing. (KHTMLPart::slotParentCompleted): Cast to int when passing to timer start.

• khtml/xml/dom_docimpl.cpp: (DocumentImpl::processHttpEquiv): Use a double when parsing meta refresh.

• kwq/KWQKDebug.h: Add overloads for unsigned int and double to keep us compiling now that we use a double in one place.

4:13 PM Changeset in webkit [2533] by cblu

• 9 edits in trunk/WebKit

WebKit:

Check that the plug-in is loaded before calling it.

• Plugins.subproj/WebBaseNetscapePluginStream.h:
• Plugins.subproj/WebBaseNetscapePluginStream.m: (-[WebBaseNetscapePluginStream dealloc]): (-[WebBaseNetscapePluginStream setPluginPointer:]): (-[WebBaseNetscapePluginStream setResponse:]): (-[WebBaseNetscapePluginStream receivedData:]): (-[WebBaseNetscapePluginStream destroyStreamWithReason:]): (-[WebBaseNetscapePluginStream finishedLoadingWithData:]):
• Plugins.subproj/WebBasePluginPackage.h:
• Plugins.subproj/WebBasePluginPackage.m: (-[WebBasePluginPackage isLoaded]):
• Plugins.subproj/WebNetscapePluginDocumentView.m: (-[WebNetscapePluginDocumentView setDataSource:]):
• Plugins.subproj/WebNetscapePluginPackage.m: (-[WebNetscapePluginPackage isLoaded]):
• Plugins.subproj/WebNetscapePluginRepresentation.m: (-[WebNetscapePluginRepresentation receivedData:withDataSource:]):

WebBrowser:

Use the display name of the dataSource not the document when reporting plug-in errors.

• LoadProgressMonitor.m: (-[LoadProgressMonitor pluginFailedWithError:dataSource:]):

3:17 PM Changeset in webkit [2532] by cblu

• 16 edits
  2 adds in trunk/WebKit

WebFoundation:

• Misc.subproj/WebError.m:

WebKit:

• Moved plug-in error handling to WebResourceLoadDelegate
• Report plug-in load failure, java load failure, and plug-in not found errors.
• Added WebPluginError, subclass of WebError.
• Attempted fix: 3090675 - Standalone WMP (Window Media Player) content crashes Alexander

• English.lproj/Localizable.strings:
• Misc.subproj/WebKitErrors.h:
• Plugins.subproj/WebBaseNetscapePluginView.m: (-[WebBaseNetscapePluginView sendEvent:]): don't send event to null function
• Plugins.subproj/WebNetscapePluginDocumentView.m: (-[WebNetscapePluginDocumentView setDataSource:]): report plug-in load failure
• Plugins.subproj/WebNullPluginView.h:
• Plugins.subproj/WebNullPluginView.m: take a WebPluginError (-[WebNullPluginView dealloc]): (-[WebNullPluginView viewDidMoveToWindow]):
• Plugins.subproj/WebPluginError.h: Added.
• Plugins.subproj/WebPluginError.m: Added. (-[WebPluginErrorPrivate dealloc]): (-[WebPluginError dealloc]): (-[WebPluginError pluginPageURL]): (-[WebPluginError pluginName]): (-[WebPluginError MIMEType]):
• WebCoreSupport.subproj/WebBridge.m: (-[WebBridge viewForPluginWithURL:attributes:baseURL:MIMEType:]): report plug-in load failure (-[WebBridge viewForJavaAppletWithFrame:attributes:baseURL:]): report plug-in load failure
• WebKit.pbproj/project.pbxproj:
• WebView.subproj/WebController.m: (-[WebResourceLoadDelegate pluginFailedWithError:dataSource:]): added, does nothing
• WebView.subproj/WebControllerPolicyDelegate.h:
• WebView.subproj/WebResourceLoadDelegate.h:
• WebView.subproj/WebView.m: (+[WebView initialize]): added new error strings

WebBrowser:

• Moved plug-in error handling to WebResourceLoadDelegate

• BrowserWebController.m:
• English.lproj/Localizable.strings:
• LoadProgressMonitor.m: (-[LoadProgressMonitor pluginNotFoundSheetEnded:returnCode:contextInfo:]): (-[LoadProgressMonitor pluginFailedWithError:dataSource:]): display special sheets for new errors.

3:00 PM Changeset in webkit [2531] by darin

• 4 edits in trunk/WebCore

• fixed 3090249 -- random crash on page load
• fixed 3090286 -- assert _job in KWQResourceLoader cancel

• kwq/KWQResourceLoader.mm: (-[KWQResourceLoader cancel]): This function needs to handle the case where the load is already cancelled. I think this also caused Avie's crash.

2:09 PM Changeset in webkit [2530] by sullivan

• 5 edits in trunk/WebKit

WebKit:

• some weaning of WebBookmark API from WebBookmarkGroup

• Bookmarks.subproj/WebBookmark.m: (-[WebBookmark dealloc]): set group to nil instead of asserting that it's non-nil

• Bookmarks.subproj/WebBookmarkGroup.h: removed obsolete ivar _bookmarksByID
• Bookmarks.subproj/WebBookmarkGroup.m: (-[WebBookmarkGroup initWithFile:]): removed obsolete ivar _bookmarksByID (-[WebBookmarkGroup dealloc]): removed obsolete ivar _bookmarksByID (-[WebBookmarkGroup removeBookmark:]): removed this unnecessary method; callers need to use [[bookmark parent] removeChild:bookmark] instead

WebBrowser:

• fixed bug Darin found in my recent checkin that caused rearranging bookmarks in toolbar to fail an assertion in [WebBookmarkGroup removeBookmark:]
• fixed a leak I found in the process (bookmark rearranged in toolbar would leak)

• BookmarksController.m: (-[BookmarksController updateBookmarkSources]): use [[bookmark parent] removeChild:bookmark] instead of now-obsolete [[bookmark group] removeBookmark:bookmark]

• BookmarksViewController.m: (-[BookmarksViewController removeBookmarks:]), (-[BookmarksViewController redoRemoveBookmarks:]), (-[BookmarksViewController undoNewBookmark:]), (-[BookmarksViewController moveBookmarks:toNewFolder:]): use [[bookmark parent] removeChild:bookmark] instead of now-obsolete [[bookmark group] removeBookmark:bookmark]

• BrowserWebBookmarkExtras.m: (-[WebBookmark deleteWithUndo]): use [[bookmark parent] removeChild:bookmark] instead of now-obsolete [[bookmark group] removeBookmark:bookmark]

• FavoriteButton.m: (-[FavoriteButton draggedImage:endedAt:operation:]): don't use undoable delete in the drag-to-rearrange case; added FIXME explaining why

• FavoritesBar.m: (-[FavoritesBar deleteFavoritesWithURLsMatchingBookmarks:]), (-[FavoritesBar titleChangeSheetDidEnd:returnCode:contextInfo:]): use [[bookmark parent] removeChild:bookmark] instead of now-obsolete [[bookmark group] removeBookmark:bookmark] (-[FavoritesBar performDragOperation:]): autorelease the copy of the bookmark in the dragged FavoriteButton

12:45 PM Changeset in webkit [2529] by trey

• 7 edits in trunk/WebKit

Moved involved code for loading URLs from WebBridge up to WebFrame, with a little
consolidation along the way. I need to get all this stuff in one place in prep
for doing back-forward properly with frames. There should be no change in
functionality as a result of these changes.

• WebCoreSupport.subproj/WebBridge.m: (-[WebBridge loadURL:reload:]): (-[WebBridge postWithURL:data:contentType:]): (-[WebBridge createChildFrameNamed:withURL:renderPart:allowsScrolling:marginWidth:marginHeight:]): Much of the guts of these is moved to WebFramePrivate.
• WebView.subproj/WebDataSourcePrivate.h:
• WebView.subproj/WebDataSourcePrivate.m: (-[WebDataSource _setIsClientRedirect:]): (-[WebDataSource _isClientRedirect]): Keep state of whether we're processing a redirect in DataSource. I had previously to do this with a loadType, but that failed because cliRedir is orthogonal.
• WebView.subproj/WebFramePrivate.h: WebFrameLoadTypeClientRedirect is gone.
• WebView.subproj/WebFramePrivate.m: (-[WebFramePrivate init]): Initialize loadType explicitly. (-[WebFrame _transitionToCommitted]): Test [dataSource _isClientRedirect] instead of looking at loadType. (-[WebFrame _isLoadComplete]): Nuke WebFrameLoadTypeClientRedirect (-[WebFrame _goToItem:withFrameLoadType:]): Nuke WebFrameLoadTypeClientRedirect. (-[WebFrame _loadRequest:]): Helper method moved from WebBridge. (-[WebFrame _loadURL:loadType:clientRedirect:]): Core impl moved up from WebBridge. (-[WebFrame _postWithURL:data:contentType:]): Core impl moved up from WebBridge.

11:51 AM Changeset in webkit [2528] by darin

• 4 edits in trunk/WebKit

• implemented loop counts

• WebCoreSupport.subproj/WebImageRenderer.h: Added repetitionsComplete field.
• WebCoreSupport.subproj/WebImageRenderer.m: (-[WebImageRenderer repetitionCount]): Get NSImageLoopCount. (-[WebImageRenderer nextFrame:]): Check repetitionCount and stop once we hit it.

Oct 31, 2002:

5:15 PM Changeset in webkit [2527] by darin

• 9 edits in trunk/WebCore

• fixed huge leak when doing page load test

• khtml/ecma/kjs_binding.h:
• khtml/ecma/kjs_binding.cpp: Remove ScriptInterpreter::mark(). We don't want to mark the elements of m_domObjects, because we don't want to keep them around the entire lifetime of the interpreter. This is fine because we already have forgetDOMObject(), which takes care of when they are deallocated.

• khtml/ecma/kjs_dom.cpp:
• khtml/ecma/kjs_dom.h:
• khtml/ecma/kjs_html.cpp:
• khtml/ecma/kjs_html.h: Changed lots of parameters to pass by const reference for a bit of a code size and performance gain.

5:10 PM Changeset in webkit [2526] by darin

• 3 edits in trunk/WebKit

• WebView.subproj/WebFrame.m: (-[WebFrame setController:]): Add an assert, hoping to catch a reported bug earlier or disprove one theory.

2:59 PM Changeset in webkit [2525] by trey

• 10 edits in trunk/WebKit

Added API to support 3072505 - SnapBack for user-entered page address
Refixed 3041616 - Extra page added in back button history
Previous fix was stepped on by a later checkin. Re-established that fix, then
went on to make sure we have the final URL and title in the BF item in the
case of a client redirect. Also handles redirects that happen within frames.
Fixed 3078560 - div added to "back" list

• History.subproj/WebBackForwardList.h:
• History.subproj/WebBackForwardList.m: (-[WebBackForwardList containsEntry:]): Added new API for 3072505.
• WebCoreSupport.subproj/WebBridge.h:
• WebCoreSupport.subproj/WebBridge.m: (-[WebBridge reportClientRedirectTo:delay:fireDate:]): (-[WebBridge reportClientRedirectCancelled]): (-[WebBridge loadURL:reload:]): Added logging. Renamed flag. Establish the right loadtype and set BF items from the old dataSource onto the new dataSource in the case of a client redirect.
• WebView.subproj/WebDataSourcePrivate.h:
• WebView.subproj/WebDataSourcePrivate.m: (-[WebDataSource _setRequest:]): Add logging. (-[WebDataSource _addBackForwardItem:]): (-[WebDataSource _addBackForwardItems:]): (-[WebDataSource _backForwardItems]): New setters/getters to support setting up the new dataSource of a client redirect.
• WebView.subproj/WebFramePrivate.h: New loadType WebFrameLoadTypeClientRedirect.
• WebView.subproj/WebFramePrivate.m: (-[WebFrame _transitionToCommitted]): Update the URL in the BF list in the client redirect case. (-[WebFrame _isLoadComplete]): NOP case for WebFrameLoadTypeClientRedirect (-[WebFrame _goToItem:withFrameLoadType:]): ASSERT_NOT_REACHED case for WebFrameLoadTypeClientRedirect

2:15 PM Changeset in webkit [2524] by rjw

• 9 edits in trunk

Corrected extern definitions of lookup tables.

• Misc.subproj/WebUnicode.h:
• Misc.subproj/WebUnicode.m: (_unicodeDirection):
• WebCoreSupport.subproj/WebTextRenderer.m: (+[WebTextRenderer initialize]):

Turned on use of lookup tables.

• kwq/KWQChar.mm: (QChar::direction): (QChar::mirrored): (QChar::mirroredChar):

12:27 PM Changeset in webkit [2523] by mjs

• 4 edits in trunk/JavaScriptCore

• fixed 3082660 - REGRESSION: one ListImp leaks opening/closing nearly empty web page

• kjs/object.cpp: Set gc allowed on freshly created ListImp, since there is no List wrapper for it.

11:12 AM Changeset in webkit [2522] by darin

• 5 edits in trunk/JavaScriptCore

• kjs/grammar.y: Fix the APPLE_CHANGES thing here too.
• kjs/grammar.cpp: Regenerated this file.

10:43 AM Changeset in webkit [2521] by mjs

• 5 edits in trunk/WebCore

• fixed 3075803 - yahoo customize layout - delete, ordering of items doesn't work

• khtml/html/html_formimpl.h:
• khtml/html/html_formimpl.cpp: (HTMLSelectElementImpl::recalcStyle): New method to insure the select element updates if the contents of the option subelements are changed.

10:05 AM Changeset in webkit [2520]

• 3 copies in tags/Alexander-30

This commit was manufactured by cvs2svn to create tag 'Alexander-30'.

10:05 AM Changeset in webkit [2519] by sheridan

• 9 edits in trunk

Alex-30 version markers