Timeline
Aug 26, 2003:
- 11:41 PM Changeset in webkit [4900] by
-
- 5 edits in trunk
WebCore:
Reviewed by Maciej.
- follow-on to fix for 3310943 -- REGRESSION (Panther): textareas in forms sometimes draw blank (bugreporter)
- kwq/KWQListBox.mm: (QListBox::QListBox): Remove call to releaseGState. This was here to work around a problem we thought was an AppKit bug, but it turns out it was caused by NSView hackery in WebKit. We now do the WebKit part in a way that does not create the problem.
WebKit:
Reviewed by Maciej.
- fixed 3310943 -- REGRESSION (Panther): textareas in forms sometimes draw blank (bugreporter)
There are two fixes here. The more elegant and slightly faster one that requires a new AppKit,
and the less elegant one that works without AppKit support. By including both we don't have to
worry about timing of submission of WebKit vs. AppKit but we get the good, elegant fix. Later,
we can delete the less elegant fix.
- WebView.subproj/WebClipView.m: (+[WebClipView initialize]): Set up a boolean global so we only hav to do the "does AppKit support _focusRingVisibleRect" check one time. (-[WebClipView resetAdditionalClip]): Only do the renewGState thing if we don't have the _focusRingVisibleRect method, but if we do the renewGState thing, do it to self and all descendants using _web_renewGStateDeep. (-[WebClipView setAdditionalClip:]): Ditto. (-[WebClipView visibleRect]): Only limit this based on the additional clip if we don't have the _focusRingVisibleRect method. (-[WebClipView _focusRingVisibleRect]): Override the new method. Harmless if it's an old AppKit that doesn't have the method yet. (-[NSView _web_renewGStateDeep]): Implemented this helper method. We can get rid of it once we are entirely on the new AppKit.
- 10:47 PM Changeset in webkit [4899] by
-
- 3 edits in trunk/WebCore
Reviewed by Maciej.
- fixed 3031562 -- most forms don't work in certain configurations because of a bug in KWQVariant
- kwq/KWQVariant.mm: (QVariant::QVariant): Set the b element of the union, not the d element, in the constructor for bool.
- 6:23 PM Changeset in webkit [4898] by
-
- 2 adds in trunk/LayoutTests/fast/block/float
Add a test for floats/clearing/margin-collapsing under the new system
- 6:15 PM Changeset in webkit [4897] by
-
- 6 deletes in trunk/LayoutTests/fast/block/margin-collapse
Remove margin collapsing tests that are now buggy/invalid
- 6:09 PM Changeset in webkit [4896] by
-
- 4 edits in trunk/WebCore
Fix for 3395950, better support for self-collapsing blocks. Make them obey
clear properly and also change the definition of self-collapsing to match
the CSS2.1 specification.
Reviewed by gramps
- khtml/rendering/render_block.cpp:
- khtml/rendering/render_block.h:
- 5:47 PM Changeset in webkit [4895] by
-
- 2 adds in trunk/LayoutTests/fast/invalid
Add a layout test for <script/>
- 5:46 PM Changeset in webkit [4894] by
-
- 3 edits in trunk/WebCore
Fix for 3335175, www.calstone.com renders as blank. The fix is to deliberately
introduce an error into our HTML parsing code that causes us to honor self-closing
XML-style <script/> tags. This matches Mozilla and Opera behavior (both of them
have this bug), but it does not match WinIE.
Reviewed by mjs
- khtml/html/htmltokenizer.cpp:
- 4:57 PM Changeset in webkit [4893] by
-
- 2 adds in trunk/LayoutTests/fast/block/basic
Add a display:none test.
- 4:44 PM Changeset in webkit [4892] by
-
- 2 adds in trunk/LayoutTests/fast/invalid
Commit an h2/form test case.
- 4:43 PM Changeset in webkit [4891] by
-
- 3 edits in trunk/WebCore
Fix for 3286229, popup menu misplaced. <form> needs to be allowed as a child of
<h1>-<h6> tags. All other browsers allow this.
Reviewed by darin
- khtml/html/dtd.cpp:
- 4:13 PM Changeset in webkit [4890] by
-
- 3 edits in trunk/WebCore
Reviewed by Maciej.
- re-fixed 3366542 -- filename with non-ASCII name left out of Content-Disposition for <input type=file>
- khtml/html/html_formimpl.cpp: (HTMLFormElementImpl::formData): Added missing # sign.
- 4:09 PM Changeset in webkit [4889] by
-
- 2 edits in trunk/WebKit
Reviewed by Maciej.
- fixed 3321247 -- window size box disappears from Help window (caused by WebKit NSView hackery)
- WebView.subproj/WebHTMLViewPrivate.m: (-[WebHTMLView _recursiveDisplayRectIfNeededIgnoringOpacity:isVisibleRect:rectIsVisibleRectForView:topView:]), (-[WebHTMLView _recursiveDisplayAllDirtyWithLockFocus:visRect:]): Don't propagate dirty rects at all. This was causing problems because now the AppKit uses dirty regions, not dirty rects. In AppKit-722 and newer, _setDrawsDescendants: takes care of this for us so we don't have to do anything at all.
- 3:19 PM Changeset in webkit [4888] by
-
- 3 edits in trunk/WebCore
Fix for 3394107, make sure the display mutation code doesn't mutate objects with
display: none set.
Reviewed by darin
- khtml/css/cssstyleselector.cpp:
- 11:57 AM Changeset in webkit [4887] by
-
- 2 edits in trunk/WebKit
Reviewed by John.
- fixed 3392650 -- REGRESSION?: assertion fails trying Apple-hosted page load test while not on Apple network
- WebView.subproj/WebFramePrivate.m: (-[WebFrame _isLoadComplete]): Stop loading the provisional data source before blowing it away, in case there are some callbacks that haven't occurred yet. It's a waste of time to try to handle those additional callbacks, and can lead to failed assertions since the data source won't be hooked up to any frame any more.
Aug 25, 2003:
- 4:26 PM Changeset in webkit [4886] by
-
- 2 edits in trunk/WebKit
Fix for 3391609. Our rounding hack wasn't correctly reflected in the selection point
code. Piggy-backed on this fix are fixes for selection of letter-spacing, word-spacing
and justified text.
Reviewed by Hyatt.
- WebCoreSupport.subproj/WebTextRenderer.m: (initializeCharacterWidthIterator): (widthAndGlyphForSurrogate): (widthForNextCharacter): (-[WebTextRenderer _CG_floatWidthForRun:style:widths:fonts:glyphs:startGlyph:endGlyph:numGlyphs:]):
- 4:08 PM Changeset in webkit [4885] by
-
- 2 edits in trunk/WebKit
- fixed 3391264 -- REGRESSION (Panther): Back/Forward buttons not updating immediately after page load
The buttons are updated in response to window update notifications.
Jaguar was sending so many extra bogus notifications that it
masked the fact that we weren't ensuring that these notifications
were sent at all in the case of non-event-based interesting
changes that might affect menu items/toolbar items/etc.
Reviewed by Richard
- WebView.subproj/WebViewPrivate.m: (-[WebView _didStartProvisionalLoadForFrame:]): call -[NSApp setWindowsNeedUpdate:YES] so window update notices will be sent (-[WebView _didCommitLoadForFrame:]): ditto (-[WebView _didFinishLoadForFrame:]): ditto (-[WebView _didFailLoadWithError:forFrame:]): ditto (-[WebView _didFailProvisionalLoadWithError:forFrame:]): ditto
Aug 24, 2003:
- 6:57 PM Changeset in webkit [4884] by
-
- 3 edits in trunk/JavaScriptCore
Reviewed by John.
- fixed 3098350 -- opt. params to date methods are ignored (can't set end date in Exchange/Outlook web cal.)
- kjs/date_object.cpp: (DateProtoFuncImp::call): Added code to handle the optional parameters. Strangely, the table of functions already had the right number of parameters listed, but the code to look at the parameter values was missing.
- 6:44 PM Changeset in webkit [4883] by
-
- 2 edits in trunk/WebKit
Reviewed by John.
- fixed 3382179 -- REGRESSION: many images scroll down while loading (e.g., homepage.mac.com slide show)
- WebCoreSupport.subproj/WebImageRenderer.m: (-[WebImageRenderer drawClippedToValidInRect:fromRect:]): While loading, always draw one less line than the number NSImage says we have, since that last line is a partial line and draws unwanted white. This is not what the bug report complains about, but is a longstanding and very minor issue. Remove the code that adjust the Y origin. This is what the bug report is about. That adjustment is incorrect and was unknowingly compensating for some kind of NSImage bug that is now fixed. (-[WebImageRenderer tileInRect:fromPoint:]): Added an assertion that the WebImageRenderer is flipped. The code does assume that it's flipped.
Aug 22, 2003:
- 9:09 PM Changeset in webkit [4882] by
-
- 4 edits in trunk/WebKit
- fixed 3385837 -- REGRESSION: can't paste link from Safari into Keynote (paste at top level, not into text)
Reviewed by Darin
The problem was that we were using a single list of pasteboard
types for both "types we can read" and "types we can write", but
NSFilenamesPBoardType wasn't being written to, creating a bad
pasteboard. The fix is to split this list in two. We could do
this a little more elegantly if we weren't paranoid about
last-minute SPI changes breaking compatibility with Sherlock or
some other internal client.
- Misc.subproj/WebNSPasteboardExtras.h: add new _web_writableDragTypesForURL, commented various methods better.
- Misc.subproj/WebNSPasteboardExtras.m: (+[NSPasteboard _web_writableDragTypesForURL]): implement _web_writableDragTypesForURL; it's just like _web_dragTypesForURL but without the NSFilenamesPBoardType (-[NSPasteboard _web_writeURL:andTitle:withOwner:]): use _web_writableDragTypesForURL
- Misc.subproj/WebNSViewExtras.m: (-[NSView _web_dragPromisedImage:rect:URL:title:event:]): use _web_writableDragTypesForURL
- 3:02 PM Changeset in webkit [4881] by
-
- 3 edits in trunk/WebCore
Fix for 3388363, copying a word in a list item copied too much. The code that checked
for the bullet would add in too much text sometimes.
Reviewed by john
- kwq/KWQKHTMLPart.mm: (isTextFirstInListItem): (KWQKHTMLPart::attributedString):
- 12:28 PM Changeset in webkit [4880] by
-
- 2 edits in trunk/WebKit
Reviewed by Dave.
- fixed 3386051 -- REGRESSION: text "too dark" after the page is deactivated and reactivated (bugweb)
Cached clips were causing us to draw nothing for the top frames of framesets, resulting in darkened
text because we draw anti-aliased text twice without drawing a background.
- WebView.subproj/WebClipView.m: (-[WebClipView resetAdditionalClip]): Invalidate cached graphics state when changing the visible rect. (-[WebClipView setAdditionalClip:]): Ditto.
- 10:33 AM Changeset in webkit [4879] by
-
- 3 edits in trunk
95u version number
