⚠ Archived content — this site is no longer maintained.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

Timeline



Oct 20, 2002:

11:11 PM Changeset in webkit [2391] by cblu
  • 3 edits in trunk/WebKit

Fixed: 3025847 - Can't use any menu command keys after clicking on a plugin

Don't give plug-ins command-modified keys anymore because the ability wasn't worth the annoyance. If only plug-ins truthfully responded t events! Anyway, filed 3080103 so that I can look into a way to make this work for everybody.

  • Plugins.subproj/WebBaseNetscapePluginView.m:
11:03 PM Changeset in webkit [2390] by darin
  • 4 edits in trunk/WebCore

Redid my fix. It was just an ordering thing. The image had to already be
installed as the current image before calling ref(), because in the cached
case, you get the loading complete callback right then.

  • khtml/rendering/render_image.cpp: (RenderImage::updateFromElement): Call deref on the old image later. Basically, instead of moving the ref of the new image up, I moved the deref of the old image down.
10:39 PM Changeset in webkit [2389] by darin
  • 4 edits in trunk/WebCore

Roll out my render_image.cpp change. It was causing a big problem
getting images from the cache. I need to study further to figure out why.

  • khtml/rendering/render_image.cpp: (RenderImage::updateFromElement): Roll back the change.
10:03 PM Changeset in webkit [2388] by cblu
  • 4 edits in trunk/WebKit

Fixed: 3025868 - Plugin context menu disappears then reappears if you click off of it

Also fixed some rare dragging weirdness in WebHTMLView. If you attempted to drag a plug-in after dragging an image, the image would drag again.

  • Plugins.subproj/WebBaseNetscapePluginView.m: (-[WebBaseNetscapePluginView mouseDragged:]): subclass and do nothing to prevent calling mouseDragged in WebHTMLView (-[WebBaseNetscapePluginView performKeyEquivalent:]): no changes to this method, removed menuForEvent, wasn't needed after all
  • Plugins.subproj/WebNetscapePluginStream.m: (-[WebNetscapePluginStream stop]): minor clean-up
9:02 PM Changeset in webkit [2387] by hyatt
  • 4 edits in trunk/WebCore

Rewrite the clearing code for floats. Rewrote the
collapsing margin code to really work with floating and
clearing. This fixes the right hand side of wired.com.

  • khtml/rendering/render_flow.cpp: (RenderFlow::layoutBlockChildren): (RenderFlow::checkClear): (RenderFlow::addChild):
5:41 PM Changeset in webkit [2386] by darin
  • 4 edits in trunk/WebCore
  • fixed 3007961 -- Crash in DOM::HTMLImageElementImpl::parseAttribute
  • khtml/rendering/render_image.cpp: (RenderImage::updateFromElement): Be sure to call ref on the new image before calling deref on the old image. That's because deref on the old image causes a cache flush, which might cause the image to go away in certain cases.
5:08 PM Changeset in webkit [2385] by cblu
  • 4 edits in trunk/WebCore

Fixed possible build failure.

  • kwq/KWQDOMNode.cpp: (isImage): remove unused variable.
4:54 PM Changeset in webkit [2384] by darin
  • 3 edits in trunk/WebKit
  • fixed a minor problem with autorelease I just ran into
  • WebView.subproj/WebHTMLViewPrivate.m: (+[WebHTMLView load]): Create an autorelease pool for the benefit of the class initialize functions that will be called.
4:18 PM Changeset in webkit [2383] by cblu
  • 4 edits in trunk/WebCore

Partial fix for:
3023814 - contexual menus and dragging doesn't work for some images

The image map case is still broken because I need to find a way to get the RenderImage for an AREA node. Anyone know how?

  • kwq/KWQDOMNode.cpp: (isImage): Also return true if the node is an INPUT tag with TYPE set to "image"
3:10 PM Changeset in webkit [2382] by cblu
  • 13 edits in trunk/WebKit

Fixed 3050010: Should show contextual menu for WebImageView
Also made drags from WebImageView work.

  • Misc.subproj/WebNSViewExtras.h:
  • Misc.subproj/WebNSViewExtras.m: added _web_setPromisedImageDragImage which does the hackery to extend the promised file drag for images
  • WebView.subproj/WebControllerPrivate.h:
  • WebView.subproj/WebControllerPrivate.m: (-[WebController _menuForElement:]): added, gets menu items from delegate and creates menu
  • WebView.subproj/WebHTMLView.m: (-[WebHTMLView menuForEvent:]): simplified, gets element and calls -[WebController _menuForElement:] (-[WebHTMLView dragImage:at:offset:event:pasteboard:source:slideBack:]): call _web_setPromisedImageDragImage: when dragging image
  • WebView.subproj/WebImageRepresentation.h:
  • WebView.subproj/WebImageRepresentation.m: (-[WebImageRepresentation dealloc]): release the URL (-[WebImageRepresentation URL]): added so WebImageView can get the URL (-[WebImageRepresentation setDataSource:]): retain the URL
  • WebView.subproj/WebImageView.h: use new drag bools
  • WebView.subproj/WebImageView.m: (-[WebImageView initWithFrame:]): ditto (-[WebImageView setAcceptsDrags:]): ditto (-[WebImageView acceptsDrags]): ditto (-[WebImageView setAcceptsDrops:]): ditto (-[WebImageView acceptsDrops]): ditto (-[WebImageView controller]): new (-[WebImageView menuForEvent:]): new, creates element and calls -[WebController _menuForElement:] (-[WebImageView mouseDragged:]): drag promised file (-[WebImageView dragImage:at:offset:event:pasteboard:source:slideBack:]): call _web_setPromisedImageDragImage: (-[WebImageView namesOfPromisedFilesDroppedAtDestination:]): new
12:59 PM Changeset in webkit [2381] by darin
  • 7 edits in trunk/WebKit
  • fixed a problem where you would get garbage bits when resizing while loading a standalone image

Cleaned up some of the complexity of how we were handling resizing.

  • WebView.subproj/WebDynamicScrollBarsView.m: (-[WebDynamicScrollBarsView updateScrollers]): Do the update twice to properly handle the case where a disappearing scroller makes a second scroller appear or vice versa.
  • WebView.subproj/WebFramePrivate.m: (-[WebFrame _setState:]): Remove the code that was trying to set the setDrawsBackground flag on the scroll view. It was doing it wrong, causing bugs with drawing non-HTML views while loading. Also, it wasn't working as an optimization, since the window was still drawing the metallic background behind. We can revisit adding an optimization at the WebHTMLView level if it creates a measurable speed increase, but we must be careful to test loading with non-HTML views along with the HTML cases.
  • WebView.subproj/WebHTMLView.m: (-[WebHTMLView _setNeedsLayoutToYes:]): Added. Calls setNeedsLayout:YES. (-[WebHTMLView viewWillMoveToSuperview:]): Added. Registers so we get _setNeedsLayoutToYes: calls when the superview's size changes. This allows us to do a layout at all the right times. (-[WebHTMLView layout]): Call setNeedsDisplay:YES, because if we do a layout without a complete display the screen will look bad. Also do a reapplyStyles, because if there are pending style changes we want to apply them before doing a layout since they could affect the layout. (-[WebHTMLView drawRect:]): Remove the special case that says to always relayout during live resize, because it works without a special case now. For the other live resize case, turn it from a special case into a general one. Any time layout is done, make sure to draw the entire view, not just the passed-in rectangle, and call setNeedsDisplay:NO so we don't do it all over again.
  • WebView.subproj/WebImageView.m: (-[WebImageView layout]): Explicitly set frame size to 0,0 when there is no image.
  • WebView.subproj/WebView.m: Remove now-unnecessary viewWillStartLiveResize, setFrame:, viewDidEndLiveResize, isOpaque, and drawRect overrides. (-[WebView initWithFrame:]): Don't call setDrawsBackground:NO for the reasons above. Don't set scroll bar visibility because the dynamic scroll bars view handles that already.
9:09 AM Changeset in webkit [2380] by darin
  • 4 edits in trunk/WebCore
  • fixed 3076390 -- crash in DOM::DocumentImpl::setParsing loading thread detail page at groups.apple.com
  • khtml/khtml_part.cpp: (KHTMLPart::openURL): Don't do any redirect special case if the document is nil. Just do a normal load in that case.
8:40 AM Changeset in webkit [2379] by darin
  • 5 edits in trunk/WebCore
  • fixed 3079864 -- REGRESSION: assertion failure on any page with a form element
  • khtml/html/html_formimpl.cpp: (HTMLGenericFormElementImpl::attach): Roll back the parent call to use NodeBaseImpl::attach. Not sure why I changed that at the last minute before my last checkin, but it was *bad*.
  • fixed a crash I observed when clicking on an about:blank page
  • kwq/WebCoreBridge.mm: (-[WebCoreBridge elementAtPoint:]): Check for nil renderer.

Oct 19, 2002:

10:10 PM Changeset in webkit [2378] by darin
  • 3 edits in trunk/WebKit
  • fixed 3073693 -- flash drawn to screen even when window minimized
  • Plugins.subproj/WebBaseNetscapePluginView.m: (-[WebBaseNetscapePluginView start]): Set up additional notifications. Don't start the null event sender if the window is miniaturized. (-[WebBaseNetscapePluginView windowDidMiniaturize:]): Stop the null event sender. (-[WebBaseNetscapePluginView windowDidDeminiaturize:]): Start the null event sender.
4:41 PM Changeset in webkit [2377] by darin
  • 8 edits in trunk/WebCore
  • fixed 3075032 -- textarea contents sometimes aren't shown

The root of the problem is that we weren't delivering close calls to
the render objects in the case where we created them late due to our FOUC
fix. I taught the DOM implementation to remember that a close call is
due, and deliver it when the render object is created. We can probably
replace the closeEntireTree hack with this too. I'll talk to Dave about
that on Monday.

  • khtml/xml/dom_nodeimpl.h: Add m_rendererNeedsClose and closeRenderer.
  • khtml/xml/dom_nodeimpl.cpp: (NodeImpl::NodeImpl): Initialize m_rendererNeedsClose to false. (NodeImpl::closeRenderer): Call close on the renderer if it's created, otherwise set m_rendererNeedsClose. (NodeImpl::attach): If the renderer is created and m_rendererNeedsClose is true call close on the renderer and set it to false.
  • khtml/html/html_formimpl.cpp: (HTMLGenericFormElementImpl::attach): Rearrange this so that the base class attach is called before updateFromElement is called. This allows the base attach to do a close for us before we update.
  • khtml/html/htmlparser.cpp: (KHTMLParser::insertNode): Call closeRenderer on the DOM node rather than calling close on the render node. (KHTMLParser::popOneBlock): Ditto.
  • khtml/xml/xml_tokenizer.cpp: (XMLHandler::endElement): Ditto. (XMLTokenizer::finish): Ditto.
10:19 AM Changeset in webkit [2376] by darin
  • 3 edits in trunk/WebKit
  • WebView.subproj/WebFramePrivate.m: (-[WebFrame _transitionToCommitted]): Removed stray printf.
1:11 AM Changeset in webkit [2375] by hyatt
  • 5 edits in trunk/WebCore

A couple of whitespace fixes. What's really odd about these
fixes is that they *work* but only if you load the page by
hitting "Back". I suspect that I've now fixed the bugs on
KHTML's end, and we have some sort of whitepsace dropping
problem on our end.

  • khtml/html/htmlparser.cpp: (KHTMLParser::insertNode):
  • khtml/html/htmltokenizer.cpp: (HTMLTokenizer::write):
Note: See TracTimeline for information about the timeline view.