Timeline
Jul 8, 2007:
- 11:29 PM Changeset in webkit [24105] by
-
- 5 edits1 add in trunk/WebCore
Reviewed by Sam.
- fixed <rdar://problem/5220706> REGRESSION (TOT): repro crash in -[WebView(WebViewInternal) _addObject:forIdentifier:] [14425] http://bugs.webkit.org/show_bug.cgi?id=14425
- bindings/js/kjs_window.cpp: (KJS::createWindow): No longer take an immediate argument - always do immediate loads on a newly created Window. Also, do a load of "" to make sure that the right info makes it to the app. (KJS::showModalDialog): Updated for above. (KJS::WindowFunc::callAsFunction): Updated for above.
- dom/Document.cpp: (WebCore::Document::shouldBeAllowedToLoadLocalResources): If our URL is about:blank, we're allowed if our opener is (since the opener must have written the contents).
- loader/FrameLoader.cpp: (WebCore::FrameLoader::changeLocation): Add a variant which takes a KURL, which it expects to be pre-completed. This is to avoid completing "" to the opener URL. (WebCore::FrameLoader::urlSelected): Allow loading empty URLs.
- loader/FrameLoader.h:
Test case is manual only, since it takes particular app behavior to reproduce:
- manual-tests/new-window-subresource-crash.html: Added.
- 3:43 PM Changeset in webkit [24104] by
-
- 2 edits in branches/wx-port-alpha/trunk/WebKit/wx/bindings/python
Expose WebView events to the Python bindings.
- 2:02 PM Changeset in webkit [24103] by
-
- 2 edits in trunk/WebCore
2007-07-08 Mitz Pettel <mitz@webkit.org>
Reviewed by Maciej.
- fix leaks introduced in r24099
- dom/Element.cpp: (WebCore::Element::~Element): Do not delete the cached computed style here. (WebCore::Element::detach): Delete the computed style here instead. (WebCore::Element::computedStyle): Removed a redundant ref() which was causing the leak. Changed to return 0 if the element is not attached, because in that case there is currently no way to tell when to delete the computed style.
- 12:36 PM Changeset in webkit [24102] by
-
- 2 edits in trunk/WebKitTools
2007-07-08 Mark Rowe <mrowe@apple.com>
Change name from WebKit/Qt to WebKit.
- CodeCoverage/regenerate-coverage-display:
- 11:55 AM Changeset in webkit [24101] by
-
- 2 edits in trunk/WebKit
- Misc/WebIconDatabase.mm: (-[WebIconDatabase init]): Removed the code to exclude the icon database from backups. We still want to do this, but in a way that only runs once ever, instead of once per launch, due to performance concerns.
- 11:47 AM Changeset in webkit [24100] by
-
- 3 edits in trunk/WebKitTools
Mark code coverage scripts as executable.
- 10:51 AM Changeset in webkit [24099] by
-
- 7 edits2 adds in trunk
LayoutTests:
Reviewed by Darin.
- test for http://bugs.webkit.org/show_bug.cgi?id=12384 getComputedStyle with td.style.display='none' reports for all properties except display
- fast/css/computed-style-without-renderer-expected.txt: Added.
- fast/css/computed-style-without-renderer.html: Added.
WebCore:
Reviewed by Darin.
- fix http://bugs.webkit.org/show_bug.cgi?id=12384 getComputedStyle with td.style.display='none' reports for all properties except display
Test: fast/css/computed-style-without-renderer.html
Implement rich computed style for elements that do not have a renderer.
- css/CSSComputedStyleDeclaration.cpp: (WebCore::getPositionOffsetValue): Changed to take a RenderStyle instead of a RenderObject. (WebCore::CSSComputedStyleDeclaration::getPropertyCSSValue): Changed to use the computed style returned by the node instead of going through the renderer. Changed properties that compute by measuring the renderer to fall back on returning the length from the RenderStyle if a renderer does not exist. (WebCore::CSSComputedStyleDeclaration::length): Changed to return the number of properties we can compute even if the element has no renderer. (WebCore::CSSComputedStyleDeclaration::copyInheritableProperties): Changed to use the computed style returned by the node instead of going through the renderer.
- dom/Element.cpp: (WebCore::ElementRareData::ElementRareData): Added a computed style member, used to cache computed style for elements that do not have a renderer. (WebCore::ElementRareData::resetComputedStyle): Added. (WebCore::Element::~Element): Delete the cached computed style. (WebCore::Element::recalcStyle): Added code to delete the cached computed style when necessary. (WebCore::Element::computedStyle): Added. Returns the style used in rendering the element, if available. Otherwise uses the style selector to compute style for the element.
- dom/Element.h:
- dom/Node.cpp: (WebCore::Node::computedStyle): Added. This method returns either the style used in rendering the node, or the style that would be used if the node had a renderer.
- dom/Node.h:
- 10:40 AM Changeset in webkit [24098] by
-
- 2 edits in trunk/WebCore
2007-07-08 Mitz Pettel <mitz@webkit.org>
Reviewed by Darin.
- fix http://bugs.webkit.org/show_bug.cgi?id=14544 Scroll wheel events are ignored when the cursor is over an iframe's border
- page/mac/EventHandlerMac.mm: (WebCore::EventHandler::passWheelEventToWidget): Changed to return false if the event coordinates are not inside the widget's NSView hierarchy.
- 8:37 AM Changeset in webkit [24097] by
-
- 10 edits4 adds in trunk
Reviewed by Mitz.
http://bugs.webkit.org/show_bug.cgi?id=10752
The inspector should not respond to Make Text Smaller/Bigger commands
Use the -webkit-text-size-adjust property to ignore text sizing
and do not take the zoom factor into account for line height when
-webkit-text-size-adjust is set to none.
- 4:49 AM Changeset in webkit [24096] by
-
- 2 edits in trunk/WebCore
Reviewed by Maciej.
- fix <rdar://problem/5318994> Integer overflow in WebCore's Base64.cpp:base64Encode
- platform/Base64.cpp: (WebCore::base64Encode): Check against a maximum size. (WebCore::base64Decode): Ditto.
- 4:47 AM Changeset in webkit [24095] by
-
- 8 edits in trunk
WebCore:
Reviewed by Maciej.
- fix <rdar://problem/5124665> WebCore secondary-thread assertion should use linked-on-or-after check instead of building on Tiger check
- platform/ThreadCheck.h: Renamed _WebCoreThreadViolationCheck to reportThreadViolation since it's in the WebCore namespace and also we shouldn't use leading underscores since those are reserved for the compiler and standard library. Added a new function named setDefaultThreadViolationBehavior and an enum for the parameter. Removed the threadViolationIsException parameter from WebCoreReportThreadViolation since we use a global variable for it now. Got rid of the unnecessary use of do while (0) in the WebCoreThreadViolationCheck macro -- since it's only a single function call it's already a statement and doesn't need to be turned into one with the do while (0) trick.
- platform/mac/ThreadCheck.mm: (WebCore::readThreadViolationBehaviorFromUserDefaults): Moved the code to read from NSUserDefaults into this helper function. Changed it to use stringForKey instead of objectForKey so we don't get a "method not found" if someone puts something other than a string in there. Updated to use the new globals. (WebCore::setDefaultThreadViolationBehavior): Added. (WebCore::reportThreadViolation): Simplified, now that most of the work is in the readThreadViolationBehaviorFromUserDefaults function. (WebCoreReportThreadViolation): Changed to use a global variable instead of parameter to determine how to report a violation. Added LogOnFirstThreadViolation, which will log only once for each function that violates.
- WebCore.exp: Updated for above changes.
WebKit:
Reviewed by Maciej.
- fix <rdar://problem/5124665> WebCore secondary-thread assertion should use linked-on-or-after check instead of building on Tiger check
- Misc/WebKitVersionChecks.h: Added WEBKIT_FIRST_VERSION_WITH_MAIN_THREAD_EXCEPTIONS.
- WebView/WebFrameView.mm: (-[WebFrameView initWithFrame:]): In first-time initialization block, check for binaries linked against older versions of WebKit, and set the default thread violation behavior to LogOnFirstThreadViolation.
- WebView/WebView.mm: (+[WebView initialize]): Improved comments.
- 2:54 AM Changeset in webkit [24094] by
-
- 3 edits4 adds in trunk
2007-07-08 Mitz Pettel <mitz@webkit.org>
Reviewed by Darin.
- fix http://bugs.webkit.org/show_bug.cgi?id=14221 Repro crash (ASSERTION FAILED: oldText in RenderBlock::updateFirstLetter() during relayout of :before content with first-letter style)
Test: fast/css-generated-content/before-with-first-letter.html
- rendering/RenderContainer.cpp: (WebCore::RenderContainer::updateBeforeAfterContentForContainer): If a child of the existing 'before' container is not text, do not assume that it is an image, since it can also be an inline wrapping a first-letter. In that case, do not update its style (so in particular, keep it a FIRST_LETTER type style) and leave it to updateFirstLetter() to take care of.
2007-07-08 Mitz Pettel <mitz@webkit.org>
Reviewed by Darin.
- test for http://bugs.webkit.org/show_bug.cgi?id=14221 Repro crash (ASSERTION FAILED: oldText in RenderBlock::updateFirstLetter() during relayout of :before content with first-letter style)
- fast/css-generated-content/before-with-first-letter-expected.checksum: Added.
- fast/css-generated-content/before-with-first-letter-expected.png: Added.
- fast/css-generated-content/before-with-first-letter-expected.txt: Added.
- fast/css-generated-content/before-with-first-letter.html: Added.
Jul 7, 2007:
- 6:40 PM Changeset in webkit [24093] by
-
- 4 edits in trunk
LayoutTests:
Reviewed by Oliver Hunt.
- update results after removing extraneous calls to isSafeScript for readonly attributes.
- http/tests/security/cross-frame-access-expected.txt:
WebCore:
Reviewed by Oliver Hunt.
Bailout early in Window::put() if attribute is readonly to avoid
extra calls to isSafeScript().
- bindings/js/kjs_window.cpp: (KJS::Window::put):
- 2:35 PM Changeset in webkit [24092] by
-
- 4 edits in trunk/LayoutTests
- updated results (I changed the test at the last minute, but didn't regenerate results)
- fast/canvas/fill-stroke-clip-reset-path-expected.checksum: Updated.
- fast/canvas/fill-stroke-clip-reset-path-expected.png: Updated.
- fast/canvas/fill-stroke-clip-reset-path-expected.txt: Updated.
- 1:42 PM Changeset in webkit [24091] by
-
- 4 edits4 adds in trunk
LayoutTests:
- test for <rdar://problem/5292364> REGRESSION: ZX Spectrum 3.0 widget - graphic anomalies in widget window frame
- fast/canvas/fill-stroke-clip-reset-path-expected.checksum: Added.
- fast/canvas/fill-stroke-clip-reset-path-expected.png: Added.
- fast/canvas/fill-stroke-clip-reset-path-expected.txt: Added.
- fast/canvas/fill-stroke-clip-reset-path.html: Added.
WebCore:
Reviewed by Oliver Hunt.
- fix <rdar://problem/5292364> REGRESSION: ZX Spectrum 3.0 widget - graphic anomalies in widget window frame
Updated Dashboard quirk so it affects fill, stroke, and clip, rather than just stroke.
Test: fast/canvas/fill-stroke-clip-reset-path.html
- html/CanvasRenderingContext2D.h: Added clearPathForDashboardBackwardCompatibilityMode.
- html/CanvasRenderingContext2D.cpp: (WebCore::CanvasRenderingContext2D::clearPathForDashboardBackwardCompatibilityMode): Added. (WebCore::CanvasRenderingContext2D::fill): Call the new function. (WebCore::CanvasRenderingContext2D::stroke): Replaced in-line code with a call to the new function. (WebCore::CanvasRenderingContext2D::clip): Call the new function.
- 1:14 PM Changeset in webkit [24090] by
-
- 4 edits4 adds in trunk
LayoutTests:
Reviewed by Darin.
Testcase for:
http://bugs.webkit.org/show_bug.cgi?id=8994
Canvas doesn't reset on resize
- fast/canvas/canvas-resize-reset-expected.checksum: Added.
- fast/canvas/canvas-resize-reset-expected.png: Added.
- fast/canvas/canvas-resize-reset-expected.txt: Added.
- fast/canvas/canvas-resize-reset.html: Added.
WebCore:
Reviewed and tweaked by Darin.
http://bugs.webkit.org/show_bug.cgi?id=8994
Canvas doesn't reset on resize
Also reset the rendering context when resetting the canvas.
- html/HTMLCanvasElement.cpp: (WebCore::HTMLCanvasElement::reset): (WebCore::HTMLCanvasElement::drawingContext):
- 3:11 AM Changeset in webkit [24089] by
-
- 2 edits in trunk/LayoutTests
Fixed a mismatch between test output and expected results (a typo that was only fixed in the former).
- fast/loader/early-load-cancel-expected.txt:
- 3:00 AM Changeset in webkit [24088] by
-
- 7 edits5 adds in trunk
Reviewed by Maciej.
http://bugs.webkit.org/show_bug.cgi?id=14525
Support exslt:node-set()
Test: fast/xsl/exslt-node-set.xml
- xml/XSLTExtensions.cpp: Added. (WebCore::exsltNodeSetFunction): A copy of exslt:node-set() implementation from libexslt 1.1.11 (shipped with Tiger). (WebCore::registerXSLTExtensions): This function registers any WebCore-provided extensions on a context. Currently, the only extension is exslt:node-set().
- xml/XSLTExtensions.h: Added.
- xml/XSLTProcessor.cpp: (WebCore::XSLTProcessor::transformToString): Register extensions when creating a context.
- WebCore.pro:
- WebCore.vcproj/WebCore.vcproj:
- WebCore.xcodeproj/project.pbxproj:
- WebCoreSources.bkl: Add XSLTExtensions.{h,cpp} to projects.