Timeline
Jun 19, 2006:
- 10:17 PM Changeset in webkit [14920] by
-
- 8 edits2 adds in trunk
Reviewed by Darin.
- http://bugzilla.opendarwin.org/show_bug.cgi?id=5499 Page reload does not send any cache control headers
WebCore:
- bindings/js/kjs_window.cpp: (KJS::LocationFunc::callAsFunction): Call scheduleRefresh() for Location::Reload.
- page/Frame.h: Add scheduleRefresh(). Don't derive from TransferJob or implement its methods - that was used only for storing response HTTP headers, which was an overkill.
- page/FramePrivate.h: Replace TransferJob with a HashMap for storing response headers.
- page/Frame.cpp: (WebCore::Frame::didOpenURL): Don't needlessly change d->m_cachePolicy. Don't create a TransferJob. (WebCore::Frame::stopLoading): Directly access the metadata map, instead of going through a TransferJob. (WebCore::Frame::receivedFirstData): Ditto. (WebCore::Frame::addMetaData): Ditto. (WebCore::Frame::scheduleRefresh): A new function that schedules a refresh, similarly to what scheduleRedirection() does. (WebCore::Frame::changeLocation): Set request.reload attribute based on the current cache policy.
WebKit:
- WebView/WebFrame.m: (-[WebFrame _addExtraFieldsToRequest:mainResource:alwaysFromRequest:]): Set a proper Cache-Control header for reload requests. (-[WebFrame loadRequest:]): Reset loadType to WebFrameLoadTypeStandard (after a reload, it stayed at WebFrameLoadTypeReload, so _addExtraFieldsToRequest erroneously added a Cache-Control header to them).
LayoutTests:
- http/tests/misc/refresh-headers-expected.txt: Added.
- http/tests/misc/refresh-headers.php: Added.
- 8:20 PM Changeset in webkit [14919] by
-
- 2 edits in trunk/WebCore
WebCore:
Reviewed by Darin. Patch by Ben Goodger.
http://bugzilla.opendarwin.org/show_bug.cgi?id=9501
Windows build fails with link error CharsetTable not defined
- platform/make-charset-table.pl: (process_iana_charsets): Move test for equality to alias "None" until after normalization steps to ensure escape characters in the input data don't interfere with processing. Convert nearby tabs to 4 spaces in this function.
- 7:43 PM Changeset in webkit [14918] by
-
- 12 edits6 adds in trunk/WebCore
Reviewed by Darin.
- fixed http://bugzilla.opendarwin.org/show_bug.cgi?id=9488 "Animated GIFs do not respect transforms in SVG"
http://bugzilla.opendarwin.org/show_bug.cgi?id=6946
"SVG shows invalidation issues in WebKit"
http://www.treebuilder.de/default.asp?file=441875.xml
"Invalidation issues with "SVG 3d" demo"
http://code.google.com/webstats/2005-12/pages.html
"SVG text doesn't repaint correctly"
- kcanvas/KCanvasContainer.cpp: (WebCore::KCanvasContainer::computeAbsoluteRepaintRect): Override base class, and apply appropriate transforms, so damage rects in transformed SVG content get propagated up properly. (WebCore::KCanvasContainer::getAbsoluteRepaintRect): note a FIXME; this method seems wrong.
- kcanvas/KCanvasContainer.h: Prototype new method.
- css/svg.css: Don't apply overflow:hidden to foreignObject, since that makes it a RenderLayer so it paints twice.
- kcanvas/RenderForeignObject.cpp: (WebCore::RenderForeignObject::paint): Transform the damage rect before passing it down to HTML content, so everything paints that is supposed to. Also handle opacity here since we won't get layers. (WebCore::RenderForeignObject::computeAbsoluteRepaintRect): Override base class, and apply appropriate transforms, so damage rects in HTML embedded in SVG get propagated up properly. (WebCore::RenderForeignObject::requiresLayer): Never use a RenderLayer. (WebCore::RenderForeignObject::layout): Make sure to dirty our previous bounds when layout changes, as by transform.
- kcanvas/RenderForeignObject.h:
- kcanvas/RenderSVGImage.cpp: (WebCore::RenderSVGImage::paint): Transform the damage rect when painting. Also handle opacity here since we won't get layers. (WebCore::RenderForeignObject::computeAbsoluteRepaintRect): Override base class, and apply appropriate transforms, so damage rects in SVG images get propagated up properly. (WebCore::RenderSVGImage::translationForAttributes): New helper method, factored out of below. (WebCore::RenderSVGImage::translateForAttributes): Use above. (WebCore::RenderSVGImage::requiresLayer): Never use a RenderLayer. (WebCore::RenderSVGImage::layout): Make sure to properly dirty the old bounds, accounting for transforms. (WebCore::RenderSVGImage::relativeBBox): Correct bbox computation.
- kcanvas/RenderSVGImage.h:
- kcanvas/RenderSVGText.cpp: (WebCore::RenderSVGText::paint): Transform incoming damage rect. Handle opacity here since we won't get a layer. (WebCore::RenderSVGText::computeAbsoluteRepaintRect): Apply transforms. (WebCore::RenderSVGText::requiresLayer): Never use a RenderLayer. (WebCore::RenderSVGText::layout): Make sure to dirty the old bounds.
- platform/cg/GraphicsContextCG.cpp: (WebCore::GraphicsContext::roundToDevicePixels): Instead of transforming rect to device space and back to user space, do this for the origin and lower right corner of the rect. Otherwise the rect will get inflated if user space is rotated or skewed.
- kcanvas/RenderPath.cpp: (WebCore::RenderPath::layout): Make sure to dirty the old bounds. (WebCore::RenderPath::paint): Hhandle opacity here.
Bonus fix:
- kcanvas/RenderSVGImage.cpp: (WebCore::RenderSVGImage::nodeAtPoint): Fix hit testing. Wasn't applying the x/y attribute transform.
- 4:31 PM Changeset in webkit [14917] by
-
- 1 edit in trunk/WebKit/WebView/WebView.h
Build fix.
- 3:58 PM Changeset in webkit [14916] by
-
- 40 edits20 copies2 moves in trunk/WebCore
- 2:58 PM Changeset in webkit [14915] by
-
- 3 edits in trunk/WebKit
Reviewed by Darin.
- added mechanism to notify UIDelegate when part of the webview is redrawn. For now, it only works for HTML views.
- WebView/WebUIDelegatePrivate.h: Define a new UIDelegate method -webView:didDrawRect:
- WebView/WebHTMLView.m: (-[WebView drawSingleRect:]): Call through to UIDelegate if it implements that method. I tested that this does not impact PLT numbers in the case where the delegate implements the method but does nothing in it.
- 2:39 PM Changeset in webkit [14914] by
-
- 4 edits3 deletes in trunk/WebCore
Reviewed by Brady.
- removed a few empty/unused files Brady found by tracking down the nm error message
- editing/VisibleRange.cpp: Removed.
- editing/VisibleRange.h: Removed.
- kcanvas/KCanvasImage.cpp: Removed.
- WebCore.vcproj/WebCore/WebCore.vcproj: Removed VisibleRange.cpp/h.
- WebCore.xcodeproj/project.pbxproj: Removed VisibleRange.cpp/h and KCanvasImage.cpp.
- WebCoreSources.bkl: Removed VisibleRange.cpp.
- 12:23 PM Changeset in webkit [14913] by
-
- 2 edits in trunk/JavaScriptCore
2006-06-19 Anders Carlsson <acarlsson@apple.com>
Reviewed by John.
- kjs/interpreter.cpp: (KJS::TimeoutChecker::pauseTimeoutCheck): (KJS::TimeoutChecker::resumeTimeoutCheck): Fix argument order in setitimer calls.
- 9:42 AM Changeset in webkit [14912] by
-
- 2 edits in trunk/WebCore
Reviewed by Darin.
- http://bugzilla.opendarwin.org/show_bug.cgi?id=9492 Fix or remove window.debug.
It was only available in newly created windows, and became undefined after changing the URL
or reloading. Let's try removing it; hopefully, no-one will notice.
- bindings/js/kjs_proxy.cpp: (WebCore::KJSProxy::initScriptIfNeeded): Don't add window.debug.
- 9:15 AM Changeset in webkit [14911] by
-
- 9 edits in trunk
Reviewed by Darin.
- http://bugzilla.opendarwin.org/show_bug.cgi?id=8972 REGRESSION: invalid UTF-8 sequences are not displayed
Test: LayoutTests/fast/encoding/invalid-UTF-8.html
- dom/xml_tokenizer.cpp: (WebCore::getXHTMLEntity): Properly null-terminate the result. This didn't matter before, because the garbage at the end was guaranteed to be invalid UTF-8, and was omitted in appendOmittingUnwanted();
- platform/StreamingTextDecoder.cpp: (WebCore::StreamingTextDecoder::appendOmittingBOM): Only omit BOM characters. Renamed back from appendOmittingUnwanted();
(WebCore::StreamingTextDecoder::convertUsingICU):
- platform/StreamingTextDecoder.h: Update for the above function renaming.
- 3:51 AM Changeset in webkit [14910] by
-
- 2 edits in trunk/WebKit
WebKit:
Reviewed by Maciej.
- fix http://bugzilla.opendarwin.org/show_bug.cgi?id=9452 Assertion failure in -[WebFramePrivate setProvisionalDataSource:]
- WebView/WebFrame.m: (-[WebFrame _checkLoadCompleteForThisFrame]): Avoid re-entering the delegate's -[webView:didFailProvisionalLoadWithError:forFrame].
Jun 18, 2006:
- 5:00 PM Changeset in webkit [14909] by
-
- 3 edits in trunk/WebKitTools
WebKitTools:
Reviewed by darin.
http://bugzilla.opendarwin.org/show_bug.cgi?id=9485
Teach svn-apply and svn-unapply to use full path names
- Scripts/svn-apply: Changed to use full path names. Cleaned up code.
- Scripts/svn-unapply: Ditto.
- 4:54 PM Changeset in webkit [14908] by
-
- 2 edits in trunk/WebCore
WebCore:
Reviewed by darin.
http://bugzilla.opendarwin.org/show_bug.cgi?id=9498
Remove dead code left in JSHTMLDocument::getOwnPropertySlot() from r14298 (Bug 7838)
- bindings/js/kjs_html.cpp: (KJS::JSHTMLDocument::getOwnPropertySlot): Remove dead code.
- 3:20 PM Changeset in webkit [14907] by
-
- 4 edits in trunk
LayoutTests:
Reviewed by ggaren.
- fast/history/clicked-link-is-visited.html: Changed to use DOM mouse event.
WebCore:
Reviewed by ggaren.
- manual-tests/redraw-page-cache-visited-links.html: Changed to use DOM mouse event.
- 10:58 AM Changeset in webkit [14906] by
-
- 5 edits in trunk/WebKit
WebKit:
Reviewed by xenon.
- http://bugzilla.opendarwin.org/show_bug.cgi?id=9479 Disassociate the inspector from the frame when it detaches from its parent
- WebInspector/WebInspector.m: (-[NSWindow setWebFrame:]): Added code to (de)register with the WebFrame the inspector is (no longer) targeting. (-_webFrameDetached:): Added. Moved the code that was previously in -[inspectedWindowWillClose:] here. This is called by the WebFrame when it is detached from its parent.
- WebInspector/WebInspectorInternal.h:
- WebView/WebFrame.m: (-[WebFramePrivate dealloc]): (-[WebFrame _detachFromParent]): Added code to notify all registered inspectors that the WebFrame is detaching. (-[WebFrame _addInspector:]): Added. (-[WebFrame _removeInspector:]): Added.
- WebView/WebFrameInternal.h:
- 10:54 AM Changeset in webkit [14905] by
-
- 7 edits7 adds in trunk
LayoutTests:
Reviewed by ggaren.
- test and results for http://bugzilla.opendarwin.org/show_bug.cgi?id=9150 DumpRenderTree should be able to keep URL history during runs
- fast/history/clicked-link-is-visited-expected.checksum: Added.
- fast/history/clicked-link-is-visited-expected.png: Added.
- fast/history/clicked-link-is-visited-expected.txt: Added.
- fast/history/clicked-link-is-visited.html: Added.
- fast/history/resources/clicked-link-is-visited-2.html: Added.
WebCore:
Reviewed by ggaren.
- Update a manual test for http://bugzilla.opendarwin.org/show_bug.cgi?id=9150 DumpRenderTree should be able to keep URL history during runs
- manual-tests/redraw-page-cache-visited-links.html: Added note about window.history hack. Fixed coordinates for mouse events. Added call to keepWebHistory().
- manual-tests/resources/redraw-page-cache-visited-links-2.html: Added note about window.history hack.
WebKitTools:
Reviewed by ggaren.
- http://bugzilla.opendarwin.org/show_bug.cgi?id=9150 DumpRenderTree should be able to keep URL history during runs
Test: LayoutTests/fast/history/clicked-link-is-visited.html
- DumpRenderTree/DumpRenderTree.m: Add keepWebHistory function to layoutTestController (+[LayoutTestController isSelectorExcludedFromWebScript:]): Updated for keepWebHistory. (-[LayoutTestController keepWebHistory]): Added. We only set optional shared history if it is currently nil since keepWebHistory() might be called more than once incidentally for the same test. (dumpRenderTree): Set optional shared history in WebHistory to nil by default.
- DumpRenderTree/DumpRenderTree.xcodeproj/project.pbxproj: Xcode 2.3 clean up.
- 10:11 AM Changeset in webkit [14904] by
-
- 16 edits in trunk
2006-06-18 Anders Carlsson <acarlsson@apple.com>
Reviewed by Geoff.
- kjs/interpreter.cpp: (KJS::TimeoutChecker::pauseTimeoutCheck): Do nothing if the timeout check hasn't been started.
(KJS::TimeoutChecker::resumeTimeoutCheck):
Do nothing if the timeout check hasn't been started.
Use the right signal handler when unblocking.
(KJS::Interpreter::handleTimeout):
pause/resume the timeout check around the call to
shouldInterruptScript().
WebCore:
2006-06-17 Anders Carlsson <acarlsson@apple.com>
Reviewed by Geoff.
- bindings/js/kjs_binding.cpp: (KJS::ScriptInterpreter::ScriptInterpreter): Set the default script timeout.
(KJS::ScriptInterpreter::shouldInterruptScript):
New function which asks the frame if the script should be interrupted.
- bindings/js/kjs_binding.h:
- bindings/js/kjs_events.cpp: (KJS::JSAbstractEventListener::handleEvent):
- bindings/js/kjs_proxy.cpp: (WebCore::KJSProxy::evaluate): Add calls to startTimeoutCheck/stopTimeoutCheck
- bindings/js/kjs_window.cpp: (KJS::WindowFunc::callAsFunction): Add calls to pauseTimeoutCheck/unpauseTimeoutCheck
(KJS::ScheduledAction::execute):
Add calls to startTimeoutCheck/stopTimeoutCheck
- bridge/mac/FrameMac.h:
- bridge/mac/FrameMac.mm: (WebCore::FrameMac::shouldInterruptJavaScript): New function which asks the bridge if the script should be interrupted.
- bridge/mac/WebCoreFrameBridge.h:
- page/Frame.h: Add function declarations.
WebKit:
2006-06-18 Anders Carlsson <acarlsson@apple.com>
Reviewed by Geoff.
- WebCoreSupport/WebFrameBridge.m: (-[WebFrameBridge shouldInterruptJavaScript]): Ask the UI delegate if the script should be interrupted.
- WebView/WebUIDelegatePrivate.h: Declare webViewShouldInterruptJavaScript: delegate method
- 12:00 AM Changeset in webkit [14903] by
-
- 3 edits2 adds in trunk
Reviewed by ggaren.
- http://bugzilla.opendarwin.org/show_bug.cgi?id=8131 Some properties and methods of window and document objects cannot be converted to a string
Most of the properties were already fixed, this just adds adds a prototype to window.debug.
Test: fast/dom/everything-to-string.html
- bindings/js/kjs_proxy.cpp: (WebCore::TestFunctionImp::TestFunctionImp): (WebCore::KJSProxy::initScriptIfNeeded):