⚠ Archived content — this site is no longer maintained.   Current WebKit documentation is at docs.webkit.org.

Timeline



Jun 2, 2003:

3:59 PM Changeset in webkit [4471] by rjw
  • 4 edits in trunk/WebKit

Fix for 3250352.

Reviewed by Chris.

  • WebView.subproj/WebDataSourcePrivate.m: (-[WebDataSource _startLoading:]): Check respondsToSelector: before calling.
  • WebView.subproj/WebResourceLoadDelegate.h:
  • WebView.subproj/WebUIDelegate.h: Headerdoc tweaks.
3:41 PM Changeset in webkit [4470] by hyatt
  • 3 edits in trunk/WebCore

Fix for 3276099. value() was wrong for radio buttons. Needed
to get it from the attribute since I was no longer caching
anything in m_value.

Reviewed by kocienda

  • khtml/html/html_formimpl.cpp: (HTMLInputElementImpl::value):
2:29 PM Changeset in webkit [4469] by kocienda
  • 3 edits in trunk/WebCore

Code written by darin
Reviewed by me

Fix for this bug:

<rdar://problem/3272162>: repro crash in khtml::CachedImage::checkNotify() on particular page

  • khtml/ecma/kjs_html.cpp: (Image::putValue): ref onLoadListener when setting a value to onLoad (Image::~Image): deref onLoadListener when destructing the image.
2:12 PM Changeset in webkit [4468] by hyatt
  • 4 edits in trunk/WebCore

Fix for 3263679, color: # doesn't parse. Fix is to add an
error-handling case for # by itself to parser.y.

Reviewed by gramps/ken

  • khtml/css/parser.cpp:
  • khtml/css/parser.y:
1:18 PM Changeset in webkit [4467] by sheridan
  • 3 edits in trunk

v82u

12:47 PM Changeset in webkit [4466]
  • 3 copies in tags/Safari-81

This commit was manufactured by cvs2svn to create tag 'Safari-81'.

12:47 PM Changeset in webkit [4465] by sheridan
  • 8 edits in trunk

v81 stamp

11:42 AM Changeset in webkit [4464] by darin
  • 3 edits in trunk/WebCore

Reviewed by Ken.

  • fixed 3256787 -- HOMEPAGE:\000 characters being added by Safarin in Form submission, causing an exception in property list parsing
  • kwq/KWQTextCodec.mm: (QTextCodec::fromUnicode): Encode \ as 5C instead of 815F when encoding Shift-JIS. This should make the Apple Store work again. The only thing we lose is the ability to send an actual Shift-JIS backslash (815F), but I think that is almost never needed. In other browsers you can't even type a backslash in Shift-JIS.
10:56 AM Changeset in webkit [4463] by cblu
  • 4 edits in trunk/WebKit

Fixed: <rdar://problem/3154910>: No video when viewing QT plug-in content at some pages but audio works

This fix works around QT plug-in bug 3275755, but I think the fix is logical and worth keeping even after 3275755 is fixed.

Eric Carlson:
The problem happens when you call NPP_SetWindow with a 0 width or height more than once. The first call to NPP_SetWindow always seems to have width and height set to 0, but the next call sometimes has it set to the correct values (those in the EMBED tag). This is when it draws successfully. It seems to me that the fix is to always pass the correct width and height to NPP_SetWindow. You always position the plug-in far offscreen (1000000, -52) and set the clip region to an empty rect (48576, 52, 48576, 52) so there isn't really any danger of the plug-in drawing anyway. Additionally, you pass the correct width and height in the call to NPP_New before the first call to NPP_SetWindow.

Reviewed by john, darin.

  • Plugins.subproj/WebBaseNetscapePluginView.h:
  • Plugins.subproj/WebBaseNetscapePluginView.m: (-[WebBaseNetscapePluginView saveAndSetPortStateForUpdate:]): use the NSView width and height if greater than 0, else use the tag specified width and height (-[WebBaseNetscapePluginView isNewWindowEqualToOldWindow]): new (-[WebBaseNetscapePluginView setWindow]): NPP_SetWindow may be expensive, only call it if it has changed
  • Plugins.subproj/WebNetscapePluginPackage.m: (-[WebNetscapePluginPackage launchRealPlayer]): tweak, no need to store error code since it is ignored
9:25 AM Changeset in webkit [4462] by kocienda
  • 3 edits in trunk/WebCore

top level:

Reviewed by John

  • configure.in: Removed unneeded configure-time checks for Foundation. Added check to see if there is a Foundation tree checked out. Added HAVE_FOUNDATION_SOURCES macro to config.h.

Tests:

Reviewed by John

Tweaked path in Makefile.am so that tests can find private
Foundation headers using new configure-time checks.

Modified tests to be "skippable" based on whether there is
a Foundation tree checked out and HAVE_FOUNDATION_SOURCES
is set in config.h.

Modified test harness so that it knows how to skip tests.

  • CookieManager/Makefile.am:
  • CookieManager/wkcookiemanager-test.m: (main):
  • WebFoundation-Misc/Makefile.am:
  • WebFoundation-Misc/ifnsarrayextensions-test.m: (main):
  • WebFoundation-Misc/ifnscalendardateextensions-test.m: (main):
  • WebFoundation-Misc/ifnsobjectextensions-test.m: (main):
  • WebFoundation-Misc/ifnsstringextensions-test.m: (main):
  • WebFoundation-Misc/ifnsurlextensions-test.m: (main):
  • WebFoundation-Misc/ifreadwritelock-test.m: (main):
  • WebFoundation-Misc/ifrecursivereadwritelock-test.m: (main):
  • WebFoundation-Misc/ifurlparsing-test.m: (main):
  • harness.c: (runtest):

WebCore:

Reviewed by NOBODY (OOPS!).

  • config.h:
3:07 AM Changeset in webkit [4461] by hyatt
  • 3 edits in trunk/WebCore

Fix for 3275136. Make sure that when the checked attribute
is set dynamically (something we were ignoring before my previous
patch) that we also properly update the radio button array in
the document.

Reviewed by gramps

  • khtml/html/html_formimpl.cpp: (HTMLInputElementImpl::parseAttribute): (HTMLInputElementImpl::attach): (HTMLInputElementImpl::reset):

May 30, 2003:

7:04 PM Changeset in webkit [4460] by mjs
  • 3 edits in trunk/WebCore

Reviewed by Chris.

  • fixed 3169067 - PEOPLESOFT: left frame missing; frame RenderRoot height and width are both zero
  • khtml/khtml_part.cpp: (KHTMLPart::gotoAnchor):
6:54 PM Changeset in webkit [4459] by mjs
  • 3 edits in trunk/WebCore

Reviewed by Ken.

  • khtml/xml/dom_docimpl.cpp: (DocumentImpl::getElementById): Return null when the ID is the empty string. Otherwise we'll return the first element in the document, usually the top-level HTML element, which is wrong and does not match other browsers.
5:37 PM Changeset in webkit [4458] by rjw
  • 11 edits in trunk

WebKit:

Fixed 3272516. Items are now expired from the
b/f cache if they are older than 30 minutes. This
number was pulled out of our #!$es. Also did
some cleanup of the b/f cache code.

Reviewed by Ken.

  • WebCoreSupport.subproj/WebBridge.m: (-[WebBridge saveDocumentToPageCache:]):
  • WebView.subproj/WebFramePrivate.h:
  • WebView.subproj/WebFramePrivate.m: (-[WebFrame _transitionToCommitted:]): (-[WebFrame _createPageCacheForItem:]): (-[WebFrame _setState:]): (-[WebFrame _loadItem:withLoadType:]): (-[WebFrame _continueLoadRequestAfterNavigationPolicy:formState:]):
  • WebView.subproj/WebPreferences.m: (+[WebPreferences initialize]): (-[WebPreferences _resourceTimedLayoutEnabled]): (-[WebPreferences _backForwardCacheExpirationInterval]):
  • WebView.subproj/WebPreferencesPrivate.h:
  • WebView.subproj/WebView.h:
  • WebView.subproj/WebPreferences.h: Cleaned up some headerdoc comments.

WebCore:

As part of some cleanup when fixing 3272516 changed use of
@"" for cache keys to NSString declarations.

Reviewed by Ken.

  • kwq/WebCoreBridge.mm: (-[WebCoreBridge openURL:reload:contentType:refresh:lastModified:pageCache:]):
4:21 PM Changeset in webkit [4457] by hyatt
  • 6 edits in trunk/WebCore

Fix for 3269129 and 3269220. setAttribute('value') didn't work
for inputs. Basically defaultValue and defaultChecked have never
worked. This patch fixes the problem by making sure that for both
value and checked you have a two-tier system where you look first
for the value as altered by the control (or by setting the DOM
property) and if that isn't set then you look at the default value.

This allows setAttribute (which changes the *default*) to work in
the cases where the control has not overridden the default
with a different value.

Reviewed by NOBODY (OOPS!).

  • khtml/html/html_formimpl.cpp: (HTMLInputElementImpl::HTMLInputElementImpl): (HTMLInputElementImpl::state): (HTMLInputElementImpl::parseAttribute): (HTMLInputElementImpl::attach): (HTMLInputElementImpl::reset): (HTMLInputElementImpl::setChecked): (HTMLInputElementImpl::value): (HTMLInputElementImpl::setValue): (HTMLInputElementImpl::isCheckedRadioButtonForDocument):
  • khtml/html/html_formimpl.h:
  • khtml/rendering/render_form.cpp: (RenderLineEdit::RenderLineEdit): (RenderLineEdit::calcMinMaxWidth): (RenderLineEdit::updateFromElement): (RenderLineEdit::slotTextChanged):
  • khtml/rendering/render_form.h:
4:08 PM Changeset in webkit [4456] by darin
  • 4 edits in trunk/WebCore

Reviewed by John.

  • added "sjis" as a synonym for "Shift_JIS"; this might fix 3235161, HOMEPAGE: Publishing page with Japanese characters w/ Safari generates an error, but I am currently unable to test the fix.
  • kwq/mac-encodings.txt: Added sjis to the Shift_JIS line.
  • kwq/KWQCharsetData.c: Regenerated.
10:41 AM Changeset in webkit [4455] by hyatt
  • 4 edits in trunk/WebCore

Fix for 3263156, stack overflow at www.liceo.edu.mx. This
site nests about 500 bold tags, only closes some of them,
and then opens 500 more, etc.

The fix for this problem is to stop honoring tags in the parser
after you see 20 identical tags. We do this only for fontstyle HTML
tags.

Reviewed by john/darin

  • khtml/html/htmlparser.cpp: (KHTMLParser::getElement): (KHTMLParser::allowNestedRedundantTag):
  • khtml/html/htmlparser.h:
10:23 AM Changeset in webkit [4454] by sullivan
  • 3 edits in trunk/WebCore

WebCore:

Reviewed by Chris

  • khtml/rendering/break_lines.cpp: Removed #ifdef/#else for which the two sides were identical that I noticed in passing.

WebBrowser:

  • fixed 3273444 -- Window title leaks when window is closed

Reviewed by Chris

  • BrowserWindow.m: (-[BrowserWindow dealloc]): release _title here, because there's a leftover retain from the last setTitle to enable our delay-title-change-for-performance hack.

May 29, 2003:

7:55 PM Changeset in webkit [4453] by mjs
  • 3 edits in trunk/WebCore

Reviewed by Richard.

  • fixed 3270373 - MALLORY: repro crash in +[KWQObjectTimerTarget stopDeferringTimers] while changing folder
  • kwq/KWQObject.mm: (QObject::killTimer): Remove the timer from the deferred list too - clients expect that a timer will never be delivered after the object was killed. (QObject::killTimers): Use killTimer on each timer for this object instead of trying to be tricky. (-[KWQObjectTimerTarget timerFired]): Don't add a timer to the deferred list more than once, because there's no benefit in ganging up multiple firings while deferred, and it makes other things simpler. (+[KWQObjectTimerTarget stopDeferringTimers]): Make sure to remove the timer before firing it, since it may also cancel itself, and we don't want to remove an extra timer in that case.
6:55 PM Changeset in webkit [4452] by rjw
  • 18 edits in trunk

WebKit:

* Public API change *
This API addition will be used (in some fashion) for Scott's WWDC
demo.

Implemented 'estimatedProgress' method on WebView. This should
eventually replace the broken algorithm on WebBrowser. Maybe
for panther.

Reviewed by Chris.

  • WebKit.exp:
  • WebView.subproj/WebBaseResourceHandleDelegate.m: (-[WebBaseResourceHandleDelegate connection:didReceiveResponse:]): (-[WebBaseResourceHandleDelegate connection:didReceiveData:]): (-[WebBaseResourceHandleDelegate connectionDidFinishLoading:]): (-[WebBaseResourceHandleDelegate connection:didFailWithError:]): (-[WebBaseResourceHandleDelegate cancelWithError:]):
  • WebView.subproj/WebDataSourcePrivate.m: (-[WebDataSource _startLoading:]):
  • WebView.subproj/WebFramePrivate.h:
  • WebView.subproj/WebFramePrivate.m: (-[WebFrame _isLoadComplete]): (-[WebFrame _numPendingOrLoadingRequests:]):
  • WebView.subproj/WebView.h:
  • WebView.subproj/WebView.m: (-[WebView estimatedProgress]):
  • WebView.subproj/WebViewPrivate.h:
  • WebView.subproj/WebViewPrivate.m: (-[WebViewPrivate dealloc]): (-[WebView _isPerformingProgrammaticFocus]): (-[WebView _progressStarted]): (-[WebView _progressCompleted]): (-[WebView _incrementProgressForConnection:data:]): (-[WebView _completeProgressForConnection:]):

Code cleanup. Moved variable initialization into block that
check for non-nil self.

  • WebCoreSupport.subproj/WebImageRenderer.m: (-[WebImageRenderer initWithMIMEType:]): (-[WebImageRenderer initWithData:MIMEType:]): (-[WebImageRenderer initWithContentsOfFile:]):

WebCore:

Support for the 'estimatedProgress' method on
WebView. Added bridge access to the number of pending or still
loading requests (from KHTML's perspective). This is used to
more accurately guess at the total remaining bytes to load.

Reviewed by Chris.

  • kwq/KWQLoader.h:
  • kwq/KWQLoader.mm: (KWQServeRequest): (KWQNumberOfPendingOrLoadingRequests):
  • kwq/WebCoreBridge.h:
  • kwq/WebCoreBridge.mm: (-[WebCoreBridge numPendingOrLoadingRequests]):
5:59 PM Changeset in webkit [4451] by rjw
  • 2 edits in trunk/WebKit

Fixed 3272226. The shared image factory was being released
when any renderer had 0 reps!

Reviewed by John.

  • WebCoreSupport.subproj/WebImageRendererFactory.m: (-[WebImageRendererFactory imageRendererWithMIMEType:]):
3:54 PM Changeset in webkit [4450] by cblu
  • 3 edits in trunk/WebKit

WebKit:

Fixed: 3151216 - Safari crashes on Drag&Drop if plugin dialog is open

Reviewed by rjw.

  • Misc.subproj/WebNSViewExtras.m: (-[NSView _web_dragOperationForDraggingInfo:]): return NSDragOperationNone is the app has a modal window so the current page can't be changed with a drag
  • WebView.subproj/WebView.m: (-[WebView draggingEntered:]): return _web_dragOperationForDraggingInfo

WebBrowser:

Fixed: 3151216 - Safari crashes on Drag&Drop if plugin dialog is open

Reviewed by rjw.

  • BrowserApplication.m: (-[BrowserApplication beginModalSessionForWindow:]): disable dragging app-wide (-[BrowserApplication endModalSession:]): renable dragging app-wide
3:44 PM Changeset in webkit [4449] by kocienda
  • 3 edits in trunk/WebCore

Reviewed by Maciej

Fix for this bug:

<rdar://problem/3271697>: Search fails at travel site

  • khtml/html/html_formimpl.cpp: (HTMLFormElementImpl::parseAttribute): Allow form method to be set to GET from javascript.
3:42 PM Changeset in webkit [4448] by cblu
  • 2 edits in trunk/WebKit

Fixed: 3273109 - leak from functionPointerForTVector in -[WebNetscapePluginPackage load]

  • Plugins.subproj/WebNetscapePluginPackage.m: (-[WebNetscapePluginPackage load]): free the main function after using it
3:37 PM Changeset in webkit [4447] by rjw
  • 2 edits in trunk/WebKit

Fixed 3273115. Always use pixel dimensions,
not absolute dimensions.

Reviewed by Ken.

  • WebCoreSupport.subproj/WebImageRenderer.m: (-[WebImageRenderer incrementalLoadWithBytes:length:complete:]):
1:00 PM Changeset in webkit [4446] by rjw
  • 2 edits in trunk/WebKit

Fix 3272292. Ensure that loadStatus is always
initialized to NSImageRepLoadStatusUnknownType.

Reviewed by Chris.

  • WebCoreSupport.subproj/WebImageRenderer.m: (-[WebImageRenderer initWithMIMEType:]): (-[WebImageRenderer initWithData:MIMEType:]): (-[WebImageRenderer initWithContentsOfFile:]):
12:09 PM Changeset in webkit [4445] by sullivan
  • 3 edits in trunk/WebCore
  • fixed MIME type leaks that were most of 3272566 (several leaks after visiting amazon.com and emptying cache)

Reviewed by Darin

  • kwq/KWQPixmap.mm: (QPixmap::QPixmap): (QPixmap::operator=): fixed two leaks of MIME string
8:59 AM Changeset in webkit [4444] by darin
  • 5 edits in trunk/WebCore

Reviewed by John.

  • fixed 3269564 -- dynamically-created forms look right, but don't post values when submitted
  • khtml/html/html_formimpl.h: Remove useless virtual setParent function. Since the inherited setParent is not virtual, it does no good to make the subclass override it with a virtual function.
  • khtml/html/html_formimpl.cpp: Remove useless virtual setParent. (HTMLGenericFormElementImpl::attach): Add code here, cribbed from setParent, to add a new form element to the form it's in. (HTMLButtonElementImpl::attach): A copy of the code is also needed here.
  • WebCore.pbproj/project.pbxproj: Let Project Builder be Project Builder.
8:57 AM Changeset in webkit [4443] by kocienda
  • 5 edits in trunk/WebKit

Reviewed by Richard

Fix for this bug:

Radar 3260323 (Some links at nike.com cause assertion failure
(connectionDidFinishLoading sent after cancel))

Added a flag which is set when a load is cancelled. This flag prevents
bad behvior when loads that finish cause the load itself to be
cancelled (which could happen with a javascript that changes the
window location). This is used to prevent both the body of
cancelWithError: and the body of connectionDidFinishLoading: running
for a single delegate. Cancelling wins.

  • WebView.subproj/WebBaseResourceHandleDelegate.h:
  • WebView.subproj/WebBaseResourceHandleDelegate.m: (-[WebBaseResourceHandleDelegate connectionDidFinishLoading:]): (-[WebBaseResourceHandleDelegate cancelWithError:]):
Note: See TracTimeline for information about the timeline view.