13 years ago2009-12-07 Benjamin Poulain <benjamin.poulain@nokia.com>
eric@webkit.org [Mon, 7 Dec 2009 15:31:42 +0000 (15:31 +0000)]
2009-12-07  Benjamin Poulain  <benjamin.poulain@nokia.com>

        Reviewed by Kenneth Rohde Christiansen.

        [Qt] Improve the autotests of QtWebkit

        Remove the calls to qWait() of the autotest of QWebView

        * tests/qwebview/tst_qwebview.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51767 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-07 Mikhail Naganov <mnaganov@chromium.org>
eric@webkit.org [Mon, 7 Dec 2009 15:02:17 +0000 (15:02 +0000)]
2009-12-07  Mikhail Naganov  <mnaganov@chromium.org>

        Reviewed by Pavel Feldman.

        [Chromium] Add declarations for DevTools message dispatching on IO thread
        (see http://codereview.chromium.org/460018)


        * public/WebDevToolsAgent.h:
        * public/WebDevToolsAgentClient.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51766 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-07 Maciej Stachowiak <mjs@apple.com>
mjs@apple.com [Mon, 7 Dec 2009 14:52:37 +0000 (14:52 +0000)]
2009-12-07  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Oliver Hunt.

        op_loop_if_less JIT codegen is broken for 64-bit

        * jit/JITOpcodes.cpp:
        (JSC::JIT::emit_op_loop_if_false): Fix codegen in this version - test was backwards.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51765 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago Reviewed by Xan Lopez.
kov@webkit.org [Mon, 7 Dec 2009 14:35:30 +0000 (14:35 +0000)]
    Reviewed by Xan Lopez.

        Build the new API test.

        [GTK] REGRESSION: webkit thinks it can render PDFs

        * GNUmakefile.am:


        Reviewed by Xan Lopez.

        Turn the MediaPlayer supported types blacklist into a whitelist

        [GTK] REGRESSION: webkit thinks it can render PDFs

        Covered by API test.

        * platform/graphics/gtk/MediaPlayerPrivateGStreamer.cpp:


        Reviewed by Xan Lopez.

        New test to make sure we do not regress this feature anymore. All
        test files were created by me.

        [GTK] REGRESSION: webkit thinks it can render PDFs

        * tests/resources/test.html: Added.
        * tests/resources/test.ogg: Added.
        * tests/resources/test.pdf: Added.
        * tests/resources/test.txt: Added.
        * tests/testmimehandling.c: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51764 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-07 Benjamin Poulain <benjamin.poulain@nokia.com>
eric@webkit.org [Mon, 7 Dec 2009 14:34:50 +0000 (14:34 +0000)]
2009-12-07  Benjamin Poulain  <benjamin.poulain@nokia.com>

        Reviewed by Kenneth Rohde Christiansen.

        [Qt] Improve the autotests of QtWebkit

        Refactor tst_qwebframe to remove qWait() and use
        the function waitForSignal() from util.h

        * tests/qwebframe/tst_qwebframe.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51763 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-07 Joanmarie Diggs <joanmarie.diggs@gmail.com>
eric@webkit.org [Mon, 7 Dec 2009 14:27:44 +0000 (14:27 +0000)]
2009-12-07  Joanmarie Diggs  <joanmarie.diggs@gmail.com>

        Reviewed by Xan Lopez.

        [Gtk] Expose the title attribute to assistive technologies

        * platform/gtk/accessibility/title-and-alt.html: Added.
        * platform/gtk/accessibility/title-and-alt-expected.txt: Added.
2009-12-07  Joanmarie Diggs  <joanmarie.diggs@gmail.com>

        Reviewed by Xan Lopez.

        [Gtk] Expose the title attribute to assistive technologies

        Expose 'alt' attribute from images as accessible name.
        Expose the 'title' core HTML attribute as accessible description.
        This is a modified version of the original fix submitted by Mario Sanchez Prada,
        adjusted so that it doesn't impact other platforms.

        * accessibility/gtk/AccessibilityObjectWrapperAtk.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51762 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-07 Benjamin Poulain <benjamin.poulain@nokia.com>
eric@webkit.org [Mon, 7 Dec 2009 14:13:45 +0000 (14:13 +0000)]
2009-12-07  Benjamin Poulain  <benjamin.poulain@nokia.com>

        Reviewed by Kenneth Rohde Christiansen.

        [Qt] Improve the autotests of QtWebkit

        Refactor the test of QGraphicsWebView:
        -make waitForSignal() available to all the tests.
        -remove QTest::qWait()

        * tests/qgraphicswebview/tst_qgraphicswebview.cpp:
        * tests/util.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51761 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoObject.create fails if properties on the descriptor are getters
oliver@apple.com [Mon, 7 Dec 2009 13:50:57 +0000 (13:50 +0000)]
Object.create fails if properties on the descriptor are getters

Reviewed by Maciej Stachowiak.

Correctly initialise the PropertySlots with the descriptor object.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51760 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago[Qt] Plugins: Force windowless mode when there is no native window handle
girish@forwardbias.in [Mon, 7 Dec 2009 13:44:35 +0000 (13:44 +0000)]
[Qt] Plugins: Force windowless mode when there is no native window handle

Reviewed by Simon Hausmann.

Inject wmode=opaque while instantiating the plugin for the case when the
webpage is not backed by a native window handle.


git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51759 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago[Qt] Fix support for generic font families
vestbo@webkit.org [Mon, 7 Dec 2009 13:29:20 +0000 (13:29 +0000)]
[Qt] Fix support for generic font families

Reviewed by Simon Hausmann.

By not propagating the resolved font family in FontCacheQt and
only concidering the font description we ended up passing the
generic font families to Qt directly. Since most systems don't
have a mapping for these font families we ended up using the
default font in most cases.

* platform/graphics/qt/FontCacheQt.cpp:
* platform/graphics/qt/FontPlatformData.h:
* platform/graphics/qt/FontPlatformDataQt.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51758 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoObject.getOwnPropertyDescriptor() allows cross-frame access
oliver@apple.com [Mon, 7 Dec 2009 13:06:18 +0000 (13:06 +0000)]
Object.getOwnPropertyDescriptor() allows cross-frame access

Reviewed by Maciej Stachowiak.

Make all implementations of getOwnPropertyDescriptor that have
cross domain restrictions simply fail immediately on cross domain
access, rather than trying to mimic the behaviour of normal
property access.

Test: http/tests/security/cross-frame-access-getOwnPropertyDescriptor.html

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51757 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoMoves JSC-specific functions from jni_utility and moves them to new jsc/jni_utility_p...
steveblock@google.com [Mon, 7 Dec 2009 12:36:56 +0000 (12:36 +0000)]
Moves JSC-specific functions from jni_utility and moves them to new jsc/jni_utility_private files.

Reviewed by Adam Barth.

Build fix only, no new tests.

* WebCore.xcodeproj/project.pbxproj: Modified. Add jni_utility_private.[cpp|h].
* bridge/jni/jni_instance.cpp: Modified. Include jni_utility_private.h for JSC.
* bridge/jni/jni_jsobject.mm: Modified. Include jni_utility_private.h for JSC.
* bridge/jni/jni_objc.mm: Modified. Include jni_utility_private.h for JSC.
* bridge/jni/jni_runtime.cpp: Modified. Include jni_utility_private.h for JSC.
* bridge/jni/jni_utility.cpp: Modified. Removed convertValueToJValue and convertArrayInstanceToJavaArray.
* bridge/jni/jni_utility.h: Modified. Removed convertValueToJValue and dispatchJNICall.
* bridge/jni/jsc: Added.
* bridge/jni/jsc/jni_utility_private.cpp: Added.
* bridge/jni/jsc/jni_utility_private.h: Added. Header for convertValueToJValue and dispatchJNICall.
* platform/android/TemporaryLinkStubs.cpp: Modified. Include jni_utility_private.h for JSC.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51756 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-07 Dirk Schulze <krit@webkit.org>
krit@webkit.org [Mon, 7 Dec 2009 12:04:00 +0000 (12:04 +0000)]
2009-12-07  Dirk Schulze  <krit@webkit.org>

        Not reviewed, adding myself to the reviewers list.

        * Scripts/modules/committers.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51755 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-06 Eric Seidel <eric@webkit.org>
eric@webkit.org [Mon, 7 Dec 2009 07:53:48 +0000 (07:53 +0000)]
2009-12-06  Eric Seidel  <eric@webkit.org>

        Reviewed by Adam Barth.

        Add an API for uploading results files to StatusBot

        Add Content-Type: plain/text which was forgotten
        from the previous commit when I landed with land-patches
        instead of land-diff.

        * QueueStatusServer/queue_status.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51754 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-06 Eric Seidel <eric@webkit.org>
eric@webkit.org [Mon, 7 Dec 2009 07:45:07 +0000 (07:45 +0000)]
2009-12-06  Eric Seidel  <eric@webkit.org>

        Reviewed by Adam Barth.

        Add an API for uploading results files to StatusBot

        * Scripts/modules/statusbot.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51753 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-06 Eric Seidel <eric@webkit.org>
eric@webkit.org [Mon, 7 Dec 2009 07:42:31 +0000 (07:42 +0000)]
2009-12-06  Eric Seidel  <eric@webkit.org>

        Reviewed by Adam Barth.

        Status Server needs a way to handle uploaded results

        * QueueStatusServer/queue_status.py: Add a ShowResults (results/*) command
        * QueueStatusServer/update_status.html: Add file upload.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51752 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-06 Shinichiro Hamaji <hamaji@chromium.org>
hamaji@chromium.org [Mon, 7 Dec 2009 07:06:06 +0000 (07:06 +0000)]
2009-12-06  Shinichiro Hamaji  <hamaji@chromium.org>

        Unreviewed. Revert r51748.

        Bugzilla should show images in git patches

        * PrettyPatch/PrettyPatch.rb:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51751 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-06 Adam Barth <abarth@webkit.org>
abarth@webkit.org [Mon, 7 Dec 2009 06:42:55 +0000 (06:42 +0000)]
2009-12-06  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        [bzt] BuildAttachment should not check the builders

        This is code that got copied here by accident when the class was created.

        * Scripts/modules/commands/download.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51750 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-06 Adam Barth <abarth@webkit.org>
abarth@webkit.org [Mon, 7 Dec 2009 06:42:31 +0000 (06:42 +0000)]
2009-12-06  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        [bzt] Make LandingSequence.update aware of ports

        This is required to make build work on the Chromium port because
        Chromium has a custom update-webkit.

        * Scripts/modules/landingsequence.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51749 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-06 Shinichiro Hamaji <shinichiro.hamaji@gmail.com>
hamaji@chromium.org [Mon, 7 Dec 2009 06:42:05 +0000 (06:42 +0000)]
2009-12-06  Shinichiro Hamaji  <shinichiro.hamaji@gmail.com>

        Reviewed by David Kilzer.

        Bugzilla should show images in git patches

        Show images in git patches using git-apply.

        * PrettyPatch/PrettyPatch.rb:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51748 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-06 Adam Barth <abarth@webkit.org>
abarth@webkit.org [Mon, 7 Dec 2009 05:20:13 +0000 (05:20 +0000)]
2009-12-06  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Add early warning system commands to bugzilla-tool.

        * Scripts/bugzilla-tool:
        * Scripts/modules/landingsequence.py:
        * Scripts/modules/webkitport.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51747 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-06 Adam Barth <abarth@webkit.org>
abarth@webkit.org [Mon, 7 Dec 2009 05:11:12 +0000 (05:11 +0000)]
2009-12-06  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        [bzt] Implement Qt EarlyWarningSystem and Chromium EarlyWarningSystem

        * Scripts/modules/commands/early_warning_system.py: Added.
        * Scripts/modules/commands/queues.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51746 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoREGRESSION (r51728): update-webkit fails when the current directory is
mitz@apple.com [Mon, 7 Dec 2009 03:27:54 +0000 (03:27 +0000)]
REGRESSION (r51728): update-webkit fails when the current directory is
not inside a Subversion working copy

Reviewed by Adele Peterson.

* Scripts/update-webkit: Invoke isSVN() in the correct working

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51745 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-06 Kent Tamura <tkent@chromium.org>
eric@webkit.org [Mon, 7 Dec 2009 02:45:54 +0000 (02:45 +0000)]
2009-12-06  Kent Tamura  <tkent@chromium.org>

        Reviewed by Darin Adler.

        Improve portability and readability of table-percent-height* tests.

        The purpose of these tests is to confirm that an element with
        height:N% inside a table cell has non-zero height. We don't need
        to check concrete pixel sizes of form controls.

        This updates the tests so that they check heights are non-zero and
        75% heights are same as corresponding 100% heights or actually 75%.

        * fast/replaced/table-percent-height-expected.txt:
        * fast/replaced/table-percent-height-text-controls-expected.txt:
        * fast/replaced/table-percent-height-text-controls.html:
        * fast/replaced/table-percent-height.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51744 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-06 Fumitoshi Ukai <ukai@chromium.org>
ukai@chromium.org [Mon, 7 Dec 2009 02:19:56 +0000 (02:19 +0000)]
2009-12-06  Fumitoshi Ukai  <ukai@chromium.org>

        Reviewed by Alexey Proskuryakov.

        Add tests for cross-origin WebSocket usage

        * websocket/tests/cross-origin-expected.txt: Added.
        * websocket/tests/cross-origin.html: Added.
        * websocket/tests/fixed-origin_wsh.py: Added.
        * websocket/tests/handshake-fail-by-cross-origin-expected.txt: Added.
        * websocket/tests/handshake-fail-by-cross-origin.html: Added.
        * websocket/tests/origin-test_wsh.py: Added.
        * websocket/tests/script-tests/cross-origin.js: Added.
        * websocket/tests/script-tests/frame-length-longer-than-buffer.js:
        * websocket/tests/script-tests/frame-length-skip.js:
        * websocket/tests/script-tests/handshake-error.js:
        * websocket/tests/script-tests/handshake-fail-by-cross-origin.js: Added.
        * websocket/tests/script-tests/simple.js:
        * websocket/tests/simple-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51743 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-06 Adam Barth <abarth@webkit.org>
abarth@webkit.org [Mon, 7 Dec 2009 01:09:32 +0000 (01:09 +0000)]
2009-12-06  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        [bzt] Rename AbstractTryQueue to AbstractReviewQueue

        * Scripts/modules/commands/queues.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51742 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-06 Maxime Simon <simon.maxime@gmail.com>
eric@webkit.org [Sun, 6 Dec 2009 17:40:06 +0000 (17:40 +0000)]
2009-12-06  Maxime Simon  <simon.maxime@gmail.com>

        Reviewed by Adam Barth.

        Build fix. Add in FrameLoaderClientHaiku the declaration of three functions introduced in r51644.

        * WebCoreSupport/FrameLoaderClientHaiku.cpp:
        * WebCoreSupport/FrameLoaderClientHaiku.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51741 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agotexImage2D pixel junk for transparency
oliver@apple.com [Sun, 6 Dec 2009 11:37:14 +0000 (11:37 +0000)]
texImage2D pixel junk for transparency

Reviewed by Maciej Stachowiak

Use kCGBlendModeCopy when drawing an image to the intermediate context
used to create a GL texture.

No test as we don't currently have any mechanism to retrieve pixel data
from the webgl context.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51740 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoNot reviewed, build fix.
mjs@apple.com [Sun, 6 Dec 2009 10:30:32 +0000 (10:30 +0000)]
Not reviewed, build fix.

Actually tested 64-bit *and* 32-bit build this time.

* jit/JITOpcodes.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51739 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoNot reviewed, build fix.
mjs@apple.com [Sun, 6 Dec 2009 10:09:55 +0000 (10:09 +0000)]
Not reviewed, build fix.

Really really fix 64-bit build for prior patch (actually tested this time).

* jit/JITOpcodes.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51738 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoNot reviewed, build fix.
mjs@apple.com [Sun, 6 Dec 2009 09:57:05 +0000 (09:57 +0000)]
Not reviewed, build fix.

Really fix 64-bit build for prior patch.

* jit/JITArithmetic.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51737 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoNot reviewed, build fix.
mjs@apple.com [Sun, 6 Dec 2009 09:51:12 +0000 (09:51 +0000)]
Not reviewed, build fix.

Fix 64-bit build for prior patch.

* jit/JITOpcodes.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51736 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-05 Maciej Stachowiak <mjs@apple.com>
mjs@apple.com [Sun, 6 Dec 2009 09:42:03 +0000 (09:42 +0000)]
2009-12-05  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Oliver Hunt.

        conway benchmark spends half it's time in op_less (jump fusion fails)

        <1% speedup on SunSpider and V8
        2x speedup on "conway" benchmark

        Two optimizations:
        1) Improve codegen for logical operators &&, || and ! in a condition context

        When generating code for combinations of &&, || and !, in a
        condition context (i.e. in an if statement or loop condition), we
        used to produce a value, and then separately jump based on its
        truthiness. Now we pass the false and true targets in, and let the
        logical operators generate jumps directly. This helps in four

        a) Individual clauses of a short-circuit logical operator can now
        jump directly to the then or else clause of an if statement (or to
        the top or exit of a loop) instead of jumping to a jump.

        b) It used to be that jump fusion with the condition of the first
        clause of a logical operator was inhibited, because the register
        was ref'd to be used later, in the actual condition jump; this no
        longer happens since a jump straight to the final target is
        generated directly.

        c) It used to be that jump fusion with the condition of the second
        clause of a logical operator was inhibited, because there was a
        jump target right after the second clause and before the actual
        condition jump. But now it's no longer necessary for the first
        clause to jump there so jump fusion is not blocked.

        d) We avoid generating excess mov statements in some cases.

        As a concrete example this source:

        if (!((x < q && y < q) || (t < q && z < q))) {
            // ...

        Used to generate this bytecode:

        [  34] less              r1, r-15, r-19
        [  38] jfalse            r1, 7(->45)
        [  41] less              r1, r-16, r-19
        [  45] jtrue             r1, 14(->59)
        [  48] less              r1, r-17, r-19
        [  52] jfalse            r1, 7(->59)
        [  55] less              r1, r-18, r-19
        [  59] jtrue             r1, 17(->76)

        And now generates this bytecode (also taking advantage of the second optimization below):

        [  34] jnless            r-15, r-19, 8(->42)
        [  38] jless             r-16, r-19, 26(->64)
        [  42] jnless            r-17, r-19, 8(->50)
        [  46] jless             r-18, r-19, 18(->64)

        Note the jump fusion and the fact that there's less jump
        indirection - three of the four jumps go straight to the target
        clause instead of indirecting through another jump.

        2) Implement jless opcode to take advantage of the above, since we'll now often generate
        a less followed by a jtrue where fusion is not forbidden.

        * parser/Nodes.h:
        (JSC::ExpressionNode::hasConditionContextCodegen): Helper function to determine
        whether a node supports special conditional codegen. Return false as this is the default.
        (JSC::ExpressionNode::emitBytecodeInConditionContext): Assert not reached - only really
        defined for nodes that do have conditional codegen.
        (JSC::UnaryOpNode::expr): Add const version.
        (JSC::LogicalNotNode::hasConditionContextCodegen): Returne true only if subexpression
        supports it.
        (JSC::LogicalOpNode::hasConditionContextCodegen): Return true.
        * parser/Nodes.cpp:
        (JSC::LogicalNotNode::emitBytecodeInConditionContext): Implemented - just swap
        the true and false targets for the child node.
        (JSC::LogicalOpNode::emitBytecodeInConditionContext): Implemented - handle jumps
        directly, improving codegen quality. Also handles further nested conditional codegen.
        (JSC::ConditionalNode::emitBytecode): Use condition context codegen when available.
        (JSC::IfNode::emitBytecode): ditto
        (JSC::IfElseNode::emitBytecode): ditto
        (JSC::DoWhileNode::emitBytecode): ditto
        (JSC::WhileNode::emitBytecode): ditto
        (JSC::ForNode::emitBytecode): ditto

        * bytecode/Opcode.h:
        - Added loop_if_false opcode - needed now that falsey jumps can be backwards.
        - Added jless opcode to take advantage of new fusion opportunities.
        * bytecode/CodeBlock.cpp:
        (JSC::CodeBlock::dump): Handle above.
        * bytecompiler/BytecodeGenerator.cpp:
        (JSC::BytecodeGenerator::emitJumpIfTrue): Add peephole for less + jtrue ==> jless.
        (JSC::BytecodeGenerator::emitJumpIfFalse): Add handling of backwrds falsey jumps.
        * bytecompiler/BytecodeGenerator.h:
        (JSC::BytecodeGenerator::emitNodeInConditionContext): Wrapper to handle tracking of
        overly deep expressions etc.
        * interpreter/Interpreter.cpp:
        (JSC::Interpreter::privateExecute): Implement the two new opcodes (loop_if_false, jless).
        * jit/JIT.cpp:
        (JSC::JIT::privateCompileMainPass): Implement JIT support for the two new opcodes.
        (JSC::JIT::privateCompileSlowCases): ditto
        * jit/JIT.h:
        * jit/JITArithmetic.cpp:
        (JSC::JIT::emitSlow_op_jless): ditto
        (JSC::JIT::emitBinaryDoubleOp): ditto
        * jit/JITOpcodes.cpp:
        (JSC::JIT::emitSlow_op_loop_if_less): ditto
        (JSC::JIT::emit_op_loop_if_false): ditto
        (JSC::JIT::emitSlow_op_loop_if_false): ditto
        * jit/JITStubs.cpp:
        * jit/JITStubs.h:
2009-12-05  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Oliver Hunt.

        conway benchmark spends half it's time in op_less (jump fusion fails)

        * fast/js/codegen-loops-logical-nodes-expected.txt:
        * fast/js/script-tests/codegen-loops-logical-nodes.js: Update to test some newly
        sensitive cases of codegen that were not already covered.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51735 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoWebCore
kov@webkit.org [Sun, 6 Dec 2009 01:50:28 +0000 (01:50 +0000)]

2009-12-03  Philippe Normand  <pnormand@igalia.com>

        Reviewed by Gustavo Noronha.

        Advertize audio/ogg correctly and refactored mime-type cache

        [GTK] Recognize oga as audio/ogg

        Test: media/media-can-play-ogg.html

        * platform/graphics/gtk/MediaPlayerPrivateGStreamer.cpp:


2009-12-05  Philippe Normand  <pnormand@igalia.com>

        Reviewed by Gustavo Noronha.

        New test for video.canPlayType and ogg containers.

        [GTK] Recognize oga as audio/ogg

        * media/media-can-play-ogg-expected.txt: Added.
        * media/media-can-play-ogg.html: Added.
        * platform/mac-leopard/Skipped:
        * platform/mac-snowleopard/Skipped:
        * platform/mac-tiger/Skipped:
        * platform/mac/Skipped:
        * platform/qt-mac/Skipped:
        * platform/qt-win/Skipped:
        * platform/qt/Skipped:
        * platform/win/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51734 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-05 Chris Jerdonek <chris.jerdonek@gmail.com>
eric@webkit.org [Sun, 6 Dec 2009 00:30:28 +0000 (00:30 +0000)]
2009-12-05  Chris Jerdonek  <chris.jerdonek@gmail.com>

        Reviewed by Darin Adler.

        Refactored web site documentation so that advice to add
        WebKitTools/Scripts to your shell path does not appear
        on multiple pages.


        * building/build.html:
        * building/checkout.html:
        * building/run.html:
        * coding/coding-style.html:
        * coding/contributing.html:
        * coding/scripts.html: Added.
        * quality/testing.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51733 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago Reviewed by Eric Seidel.
ap@apple.com [Sat, 5 Dec 2009 23:59:24 +0000 (23:59 +0000)]
    Reviewed by Eric Seidel.

        REGRESSION: websocket/tests/url-parsing.html is crashing on Leopard Bot

        * platform/network/cf/SocketStreamHandleCFNet.cpp:
        (WebCore::SocketStreamHandle::readStreamCallback): Call platformClose() to unschedule both
        streams, guaranteeing that there will be no callbacks after SocketStreamHandle is destroyed.
        (WebCore::SocketStreamHandle::writeStreamCallback): Edited a comment a bit.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51732 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoRubber-stamped by Mark Rowe.
mitz@apple.com [Sat, 5 Dec 2009 23:44:29 +0000 (23:44 +0000)]
Rubber-stamped by Mark Rowe.

Disabled a test because it puts the WebView into compositing mode,
which remains in effect for all subsequent repaint tests, causing them
to fail because DumpRenderTree always forces a full repaint when
generating the image of a WebView that is in compositing mode.

* fast/repaint/block-selection-gap-in-composited-layer.html: Removed.
* fast/repaint/block-selection-gap-in-composited-layer.html.disabled: Copied from LayoutTests/fast/repaint/block-selection-gap-in-composited-layer.html.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51731 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-05 Adam Barth <abarth@webkit.org>
abarth@webkit.org [Sat, 5 Dec 2009 21:55:12 +0000 (21:55 +0000)]
2009-12-05  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        [bzt] Implement ChromiumPort

        * Scripts/modules/webkitport.py:
        * Scripts/modules/webkitport_unittest.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51730 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-05 Adam Barth <abarth@webkit.org>
abarth@webkit.org [Sat, 5 Dec 2009 21:54:39 +0000 (21:54 +0000)]
2009-12-05  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Move update-webkit into BuildSteps

        We need to move update-webkit out of SCM.py because SCM isn't supposed to know
        that WebKit exists.  The proper place for the knowledge of the existence of
        update-webkit is in WebKitPort because some ports have specialized update
        scripts (analogous to build-webkit).

        * Scripts/modules/buildsteps.py:
        * Scripts/modules/commands/download.py:
        * Scripts/modules/landingsequence.py:
        * Scripts/modules/scm.py:
        * Scripts/modules/webkitport.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51729 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-05 Adam Barth <abarth@webkit.org>
abarth@webkit.org [Sat, 5 Dec 2009 21:54:05 +0000 (21:54 +0000)]
2009-12-05  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        update-webkit should call git svn rebase and resolve-ChangeLogs -f

        Teach update-webkit about Git.  I didn't add the call to
        resolve-ChangeLogs -f because sometimes that script goes bananas.  We
        can iterate from here, however.

        * Scripts/update-webkit:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51728 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-05 Adam Langley <agl@google.com>
abarth@webkit.org [Sat, 5 Dec 2009 17:50:53 +0000 (17:50 +0000)]
2009-12-05  Adam Langley  <agl@google.com>

        Reviewed by Adam Barth.

        Check that a CSS format() argument is of a valid type.


        * fast/css/url-format-non-string-expected.txt: Added
        * fast/css/url-format-non-string.html: Added.
2009-12-05  Adam Langley  <agl@google.com>

        Reviewed by Adam Barth.

        Check that a CSS format() argument is of a valid type.


        Test: fast/css/url-format-non-string.html

        * css/CSSParser.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51727 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoReviewed by Xan Lopez.
kov@webkit.org [Sat, 5 Dec 2009 16:56:05 +0000 (16:56 +0000)]
Reviewed by Xan Lopez.

Use a better URI for the unreachable test. localhost:80 is very
likely to be open.

* tests/testwebdatasource.c:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51726 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-05 Vincent Untz <vuntz@gnome.org>
kov@webkit.org [Sat, 5 Dec 2009 15:35:51 +0000 (15:35 +0000)]
2009-12-05  Vincent Untz <vuntz@gnome.org>

        Reviewed by Gustavo Noronha.

        Fixes race for builds with introspection enabled, and parallel

        * GNUmakefile.am:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51725 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Kent Hansen <kent.hansen@nokia.com>
eric@webkit.org [Sat, 5 Dec 2009 01:21:45 +0000 (01:21 +0000)]
2009-12-04  Kent Hansen  <kent.hansen@nokia.com>

        Reviewed by Darin Adler.

        JavaScript delete operator should return false for string properties

        * runtime/StringObject.cpp:
2009-12-04  Kent Hansen  <kent.hansen@nokia.com>

        Reviewed by Darin Adler.

        Add test for JavaScript string property deletion

        * fast/js/script-tests/string-property-deletion.js: Added.
        * fast/js/string-property-deletion-expected.txt: Added.
        * fast/js/string-property-deletion.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51724 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Marc-Antoine Ruel <maruel@chromium.org>
eric@webkit.org [Sat, 5 Dec 2009 00:46:28 +0000 (00:46 +0000)]
2009-12-04  Marc-Antoine Ruel  <maruel@chromium.org>

        Reviewed by Eric Seidel.

        Add reference to the GTK and QT build wiki pages


        * building/tools.html:
        * css/main.css:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51723 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Adam Langley <agl@google.com>
eric@webkit.org [Sat, 5 Dec 2009 00:39:27 +0000 (00:39 +0000)]
2009-12-04  Adam Langley  <agl@google.com>

        Reviewed by Eric Seidel.

        Chromium: make rounded borders heavier.

        Skia draws its rounded corners differently from the other ports.
        Whereas they have anti-aliased clipping regions, Skia only has 1-bit
        regions. (Which is technically more correct, but somewhat unhelpful
        for us.) Instead, with Skia we use a layer and collect all the
        clipping paths in effect. When the state is popped, we paint
        transparency outside the clipping paths and merge the layer down.

        This appears to cause rounded borders to look a little thin, which is
        addressed in this patch.

        This is well covered by existing tests but will require new baselines
        in the Chromium tree.


        * platform/graphics/skia/GraphicsContextSkia.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51722 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Yuzo Fujishima <yuzo@google.com>
eric@webkit.org [Sat, 5 Dec 2009 00:31:31 +0000 (00:31 +0000)]
2009-12-04  Yuzo Fujishima  <yuzo@google.com>

        Reviewed by Alexey Proskuryakov.

        Add test for web socket send.


        * websocket/tests/script-tests/send.js: Added.
        * websocket/tests/send-expected.txt: Added.
        * websocket/tests/send.html: Added.
        * websocket/tests/send_wsh.py: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51721 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 James Hawkins <jhawkins@chromium.org>
eric@webkit.org [Sat, 5 Dec 2009 00:24:28 +0000 (00:24 +0000)]
2009-12-04  James Hawkins  <jhawkins@chromium.org>

        Reviewed by Darin Fisher.

        Implement WebRegularExpression, a WebKit API for accessing

        * WebKit.gyp:
        * public/WebRegularExpression.h: Added.
        * public/WebTextCaseSensitivity.h: Added.
        * src/AssertMatchingEnums.cpp:
        * src/WebRegularExpression.cpp: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51720 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago[Chromium] Simplify zoom-related APIs and add a zoom level getter,
pkasting@chromium.org [Fri, 4 Dec 2009 23:56:21 +0000 (23:56 +0000)]
[Chromium] Simplify zoom-related APIs and add a zoom level getter,
part two: Remove deprecated APIs.

Reviewed by Darin Fisher.

* public/WebView.h:
* src/WebViewImpl.cpp:
* src/WebViewImpl.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51719 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Simon Fraser <simon.fraser@apple.com>
simon.fraser@apple.com [Fri, 4 Dec 2009 23:18:53 +0000 (23:18 +0000)]
2009-12-04  Simon Fraser  <simon.fraser@apple.com>

        Reviewed by Dan Bernstein.

        <rdar://problem/7441571> Hardware-backed layers are skipped when printing

        When printing, set the PaintBehaviorFlattenCompositingLayers flag so that compositing layers
        paint. This also causes 3D transforms to be flattened to 2D.

        Test: printing/compositing-layer-printing.html

        * page/FrameView.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51718 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoFix up some issues in my SVGListTraits change from yesterday.
oliver@apple.com [Fri, 4 Dec 2009 23:13:27 +0000 (23:13 +0000)]
Fix up some issues in my SVGListTraits change from yesterday.

Reviewed by Deth Bakin

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51717 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoSkip test that is failing because of new behaviour enforced by
kov@webkit.org [Fri, 4 Dec 2009 22:10:24 +0000 (22:10 +0000)]
Skip test that is failing because of new behaviour enforced by
libsoup. Bug reported.

* platform/gtk/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51716 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoAdd some missing methods for showing and hiding the
kenneth@webkit.org [Fri, 4 Dec 2009 22:00:50 +0000 (22:00 +0000)]
Add some missing methods for showing and hiding the
Web Inspector.

Reviewed by Adam Treat.

* DumpRenderTree/qt/DumpRenderTree.cpp:
* DumpRenderTree/qt/DumpRenderTree.h:
* DumpRenderTree/qt/LayoutTestControllerQt.cpp:
* DumpRenderTree/qt/LayoutTestControllerQt.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51715 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoSkip tests that require support for frame loader callbacks
kov@webkit.org [Fri, 4 Dec 2009 21:54:22 +0000 (21:54 +0000)]
Skip tests that require support for frame loader callbacks
reporting, and a new test.

* platform/gtk/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51714 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoUnreviewed build fix.
yael.aharon@nokia.com [Fri, 4 Dec 2009 21:35:33 +0000 (21:35 +0000)]
Unreviewed build fix.

[Qt] build fix after r51634 removed unused QBoxLayout include which included limits.h.

* DumpRenderTree/qt/DumpRenderTree.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51713 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Dirk Schulze <krit@webkit.org>
krit@webkit.org [Fri, 4 Dec 2009 21:19:11 +0000 (21:19 +0000)]
2009-12-04  Dirk Schulze  <krit@webkit.org>

        Reviewed by Gustavo Noronha Silva.

        Move the render result to svg/custom as a cross platform render result.

        * platform/mac/svg/custom/radial-gradient-with-outstanding-focalPoint-expected.txt: Removed.
        * svg/custom/radial-gradient-with-outstanding-focalPoint-expected.txt: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51712 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoNot reviewed: chromium build fix.
ajwong@chromium.org [Fri, 4 Dec 2009 21:13:34 +0000 (21:13 +0000)]
Not reviewed: chromium build fix.

Fix Chromium build break caused by moving of isDefaultPortForProtocol() and
portAllowed() into KURL.h/KURL.cpp.  Cloning code.

* platform/KURLGoogle.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51711 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Pavel Feldman <pfeldman@chromium.org>
eric@webkit.org [Fri, 4 Dec 2009 21:03:38 +0000 (21:03 +0000)]
2009-12-04  Pavel Feldman  <pfeldman@chromium.org>

        Reviewed by Timothy Hatcher.

        Web Inspector: Remove suppress highlight when restoring selected node.


        * inspector/front-end/ElementsPanel.js:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51710 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoWebKitTools: Implement the setAllowUniversalAccessFromFileURLs method
kenneth@webkit.org [Fri, 4 Dec 2009 21:01:48 +0000 (21:01 +0000)]
WebKitTools: Implement the setAllowUniversalAccessFromFileURLs method
for the Qt LayoutTestController.

Reviewed by Adam Treat.

* DumpRenderTree/qt/DumpRenderTree.cpp:
* DumpRenderTree/qt/LayoutTestControllerQt.cpp:
* DumpRenderTree/qt/LayoutTestControllerQt.h:

LayoutTests: Remove test that works after adding setAllowUniversalAccessFromFileURLs
to Qt's version of the LayoutTestController.

Reviewed by Adam Treat.

* platform/qt/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51709 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Dirk Schulze <krit@webkit.org>
krit@webkit.org [Fri, 4 Dec 2009 20:19:50 +0000 (20:19 +0000)]
2009-12-04  Dirk Schulze  <krit@webkit.org>

        Reviewed by Nikolas Zimmermann.

        Gradient SVG animation demonstrates tearing at animation extremes

        The focalPoint of a radial gradient is temporarly substracted by
        the centralPoint, if the focalPoint is not in the radius of the
        gradient. This is needed to calculate a new postion of the focalPoint
        according to the specification. But the new focalPoint needs to be
        moved by the centralPoint after this calculation, which is not the case
        at the moment. This patch fixes this issue. It was also introduced a
        deviation of maximal 0.2% for Cairo to get around the fixed point numbers
        in Cairo.

        W3C-SVG-1.1/pservers-grad-13-b needed an update. The missing adjustment
        of the new focalPoint caused wrong results for the last three tests.

        Test: svg/custom/radial-gradient-with-outstanding-focalPoint.svg

        * svg/SVGRadialGradientElement.cpp:

        Gradient SVG animation demonstrates tearing at animation extremes

        In SVG a focalPoint must be inside the radius of a radial gradient.
        It this isn't the case, we have to move the focalPoint into the radius.
        This checks the correct behavior of WebKit on false values for fx, fy.

        * platform/mac/svg/W3C-SVG-1.1/pservers-grad-13-b-expected.checksum:
        * platform/mac/svg/W3C-SVG-1.1/pservers-grad-13-b-expected.png:
        * platform/mac/svg/custom/radial-gradient-with-outstanding-focalPoint-expected.checksum: Added.
        * platform/mac/svg/custom/radial-gradient-with-outstanding-focalPoint-expected.png: Added.
        * platform/mac/svg/custom/radial-gradient-with-outstanding-focalPoint-expected.txt: Added.
        * svg/custom/radial-gradient-with-outstanding-focalPoint.svg: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51708 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Anton Muhin <antonm@chromium.org>
eric@webkit.org [Fri, 4 Dec 2009 20:16:02 +0000 (20:16 +0000)]
2009-12-04  Anton Muhin  <antonm@chromium.org>

        Reviewed by Dmitry Titov.

        Do not use WebCore::String::String(const UChar*, int length) to convert
        short v8 strings.

        Plus added string traits.

        * bindings/v8/V8Binding.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51707 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoREGRESSION: AX: buttons now extremely repetitive
cfleizach@apple.com [Fri, 4 Dec 2009 20:04:32 +0000 (20:04 +0000)]
REGRESSION: AX: buttons now extremely repetitive

Reviewed by Darin Adler.


Test: platform/mac/accessibility/button-shouldnot-have-axvalue.html

* accessibility/AccessibilityObject.h:
* accessibility/mac/AccessibilityObjectWrapper.mm:
(-[AccessibilityObjectWrapper accessibilityAttributeNames]):


* DumpRenderTree/AccessibilityUIElement.cpp:
* DumpRenderTree/AccessibilityUIElement.h:
* DumpRenderTree/gtk/AccessibilityUIElementGtk.cpp:
* DumpRenderTree/mac/AccessibilityUIElementMac.mm:
* DumpRenderTree/win/AccessibilityUIElementWin.cpp:


* platform/mac-leopard/accessibility/table-with-aria-role-expected.txt:
* platform/mac/accessibility/button-shouldnot-have-axvalue-expected.txt: Added.
* platform/mac/accessibility/button-shouldnot-have-axvalue.html: Added.
* platform/mac/accessibility/table-with-aria-role-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51706 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Zoltan Herczeg <zherczeg@inf.u-szeged.hu>
eric@webkit.org [Fri, 4 Dec 2009 20:01:20 +0000 (20:01 +0000)]
2009-12-04  Zoltan Herczeg  <zherczeg@inf.u-szeged.hu>

        Reviewed by Darin Adler.

        Check whether the right side of a string append is non-empty

        Especially beneficial for CachedScript::script() method
        which usually appends an empty string returned by flush()
        to its m_script member. m_script is often hundreds of kbytes
        which is duplated without reason.

        * platform/text/String.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51705 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoMail.app crashes when indenting table pasted from Numbers.app.
enrica@apple.com [Fri, 4 Dec 2009 19:49:37 +0000 (19:49 +0000)]
Mail.app crashes when indenting table pasted from Numbers.app.

Reviewed by John Sullivan.


Test: LayoutTests/editing/execCommand/indent-partial-table.html

* editing/IndentOutdentCommand.cpp:
(WebCore::IndentOutdentCommand::indentRegion): Added test to verify that the end of the paragraph
after the selection has not been moved.


* editing/execCommand/indent-partial-table-expected.txt: Added.
* editing/execCommand/indent-partial-table.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51704 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago Reviewed by Darin Adler.
ap@apple.com [Fri, 4 Dec 2009 19:16:15 +0000 (19:16 +0000)]
    Reviewed by Darin Adler.

        WebSocket should block the same ports that are blocked for resource loading

        Test: websocket/tests/url-parsing.html

        * page/SecurityOrigin.cpp:
        * page/SecurityOrigin.h:
        Move isDefaultPortForProtocol() to KURL, because that's a better place for it (SecurityOrigin
        is not even in WebCore/platform directory).

        * html/HTMLAnchorElement.cpp:
        Updated for the new location of isDefaultPortForProtocol().

        * platform/KURL.cpp:
        (WebCore::KURL::protocolIs): In an assertion, compare to "javascript" case-insensitively,
        since the function deosn't require lower case input.
        (WebCore::isDefaultPortForProtocol): Moved from SecurityOrigin.
        (WebCore::portAllowed): Moved from ResourceHandle.
        * platform/KURL.h:

        * platform/network/ResourceHandle.cpp: (WebCore::ResourceHandle::create): Updated for the
        new location of portAllowed().

        * websockets/WebSocket.cpp: (WebCore::WebSocket::connect): Per the spec, raise a SECURITY_ERR
        if trying to connect to a blocked port.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51703 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Jocelyn Turcotte <jocelyn.turcotte@nokia.com>
eric@webkit.org [Fri, 4 Dec 2009 18:12:46 +0000 (18:12 +0000)]
2009-12-04  Jocelyn Turcotte  <jocelyn.turcotte@nokia.com>

        Reviewed by Kenneth Rohde Christiansen.

        [Qt] Corrects QtLauncher style

        * QtLauncher/main.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51702 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Jocelyn Turcotte <jocelyn.turcotte@nokia.com>
eric@webkit.org [Fri, 4 Dec 2009 18:04:54 +0000 (18:04 +0000)]
2009-12-04  Jocelyn Turcotte  <jocelyn.turcotte@nokia.com>

        Reviewed by Kenneth Rohde Christiansen.

        [Qt] QtLauncher: add a menu to show or hide the web inspector.

        * QtLauncher/main.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51701 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Benjamin Poulain <benjamin.poulain@nokia.com>
eric@webkit.org [Fri, 4 Dec 2009 17:57:47 +0000 (17:57 +0000)]
2009-12-04  Benjamin Poulain  <benjamin.poulain@nokia.com>

        Reviewed by Kenneth Rohde Christiansen.

        [Qt] WebKit crashes when loading certain SVG images

        Update the test results for the new FontCache implementation.

        * platform/qt/css1/box_properties/float_elements_in_series-expected.txt:
        * platform/qt/css1/font_properties/font_variant-expected.txt:
        * platform/qt/css1/pseudo/firstline-expected.txt:
        * platform/qt/css1/pseudo/multiple_pseudo_elements-expected.txt:
        * platform/qt/css2.1/t1505-c524-font-var-00-b-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51700 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Benjamin Poulain <benjamin.poulain@nokia.com>
eric@webkit.org [Fri, 4 Dec 2009 17:50:42 +0000 (17:50 +0000)]
2009-12-04  Benjamin Poulain  <benjamin.poulain@nokia.com>

        Reviewed by Kenneth Rohde Christiansen.

        [Qt] WebKit crashes when loading certain SVG images

        Remove FontFallbackListQt and rely on the common FontFallbackList
        to handle the fonts. FontCache and FontPlatformData have been
        updated to work with the common FontFallbackList.

        In the previous implementation, FontPlatformDataCacheKey
        was a clone of FontPlatformData with the hashing
        capabilities added in order to use it as a key in the cache's
        hashmap. FontPlatformData has been modified to handle the hashing
        function directly so the data are not copied twice in memory.

        FontFallbackList::fontDataAt() from FontFallbackListQt was a copy of
        code from FontCache::getFontData() and FontFallbackList::fontDataAt().
        The behavior is similar except currFamily->family().length() was
        not tested and the fallback fonts selector were not used.

        Existing tests cover the change.

        * WebCore.pro:
        * platform/graphics/qt/FontCacheQt.cpp:
        * platform/graphics/qt/FontCustomPlatformData.cpp:
        * platform/graphics/qt/FontFallbackListQt.cpp:
        Removed. We now use the implementation from FontFallbackList.cpp
        * platform/graphics/qt/FontPlatformData.h:
        Add hashing capabilities to be able to use the data with the FontCache.
        This was previously done in FontCacheQt.cpp
        * platform/graphics/qt/FontPlatformDataQt.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51699 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Kenneth Rohde Christiansen <kenneth@webkit.org>
eric@webkit.org [Fri, 4 Dec 2009 17:43:29 +0000 (17:43 +0000)]
2009-12-04  Kenneth Rohde Christiansen  <kenneth@webkit.org>

        Reviewed by Antti Koivisto.

        Split out the renderPrivate in two methods, one for working on
        relative coordinates (relative to the viewport) and one for
        working on absolute coordinates. The latter is more effecient
        for implementing tiling, as you don't need translate the coords,
        and because it avoid clipping to the viewport.

        No behaviour changes, so no new tests.

        * Api/qwebframe.cpp:
        * Api/qwebframe_p.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51698 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoFix wrong assert that was only working through sheer luck.
treat@webkit.org [Fri, 4 Dec 2009 17:14:22 +0000 (17:14 +0000)]
Fix wrong assert that was only working through sheer luck.

Patch by Adam Treat <atreat@rim.com> on 2009-12-04
Reviewed by Dan Bernstein.

* platform/text/BidiContext.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51697 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago[Qt] Allow removing 'qrc' as a local security origin scheme
vestbo@webkit.org [Fri, 4 Dec 2009 15:36:52 +0000 (15:36 +0000)]
[Qt] Allow removing 'qrc' as a local security origin scheme

Reviewed by Simon Hausmann.


* page/SecurityOrigin.cpp:


* tests/qwebpage/tst_qwebpage.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51696 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago[Qt] Clean up argument parsing in the QtLauncher
vestbo@webkit.org [Fri, 4 Dec 2009 15:36:34 +0000 (15:36 +0000)]
[Qt] Clean up argument parsing in the QtLauncher

Reviewed by Simon Hausmann.

* QtLauncher/main.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51695 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Jocelyn Turcotte <jocelyn.turcotte@nokia.com>
eric@webkit.org [Fri, 4 Dec 2009 12:56:49 +0000 (12:56 +0000)]
2009-12-04  Jocelyn Turcotte  <jocelyn.turcotte@nokia.com>

        Reviewed by Kenneth Rohde Christiansen.

        [Qt] Prevent the inspector from closing its wrapping widget.
        This is not necessary anymore since we now hide the embedded
        close button.

        * WebCoreSupport/InspectorClientQt.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51694 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoAllow custom memory allocation control for WebCore's CollectionCache
zoltan@webkit.org [Fri, 4 Dec 2009 12:07:19 +0000 (12:07 +0000)]
Allow custom memory allocation control for WebCore's CollectionCache

Reviewed by Eric Seidel.

Inherits the following class from FastAllocBase because it is
instantiated by 'new':

struct name            - instantiated at: WebCore/'location'

struct CollectionCache - dom/Document.cpp:4029

* html/CollectionCache.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51693 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Steve Block <steveblock@google.com>
eric@webkit.org [Fri, 4 Dec 2009 11:30:41 +0000 (11:30 +0000)]
2009-12-04  Steve Block  <steveblock@google.com>

        Reviewed by Darin Adler.

        Fixes a crashing bug in Geolocation when a watch is cleared from some callbacks.

        * fast/dom/Geolocation/resources/permission-denied-already-clear-watch.js: Added. Tests clearing the watch from the error callback after a permission denied message when permissions have already been denied.
        * fast/dom/Geolocation/permission-denied-already-clear-watch.html: Added. HTML wrapper for above test.
        * fast/dom/Geolocation/permission-denied-already-clear-watch-expected.txt: Added. Expected result for above test.

        * fast/dom/Geolocation/resources/timeout-clear-watch.js: Added. Tests clearing the watch from the error callback due to a timeout.
        * fast/dom/Geolocation/timeout-clear-watch.html: Added. HTML wrapper for above test.
        * fast/dom/Geolocation/timeout-clear-watch-expected.txt: Added. Expected result for above test.

        * platform/gtk/Skipped: Modified. Skips new tests.
2009-12-04  Steve Block  <steveblock@google.com>

        Reviewed by Darin Adler.

        Fixes a crashing bug in Geolocation when a watch is cleared from some callbacks.

        In Geolocation::GeoNotifier::timerFired, a JS callback may be invoked if a
        fatal error was registered on this GeoNotifier or if this request has timed
        out. If the request is a watch, and is cleared by a call to clearWatch in the
        callback, the GeoNotifier object will be destroyed. We must therefore cache the
        m_geolocation member to allow us to call Geolocation::fatalErrorOccurred or

        Tests: fast/dom/Geolocation/permission-denied-already-clear-watch.html

        * page/Geolocation.cpp: Modified.
        (WebCore::Geolocation::GeoNotifier::timerFired): Modified. Cache the m_geolocation member before invoking the JS callback.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51692 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago[Qt] Re-enable audio/video tests after updated buildbot's Qt
ossy@webkit.org [Fri, 4 Dec 2009 10:37:26 +0000 (10:37 +0000)]
[Qt] Re-enable audio/video tests after updated buildbot's Qt
to Phonon supported one, and put 2 bad tests into skiplist.

Reviewed by Simon Hausmann.

* platform/qt/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51691 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoAllow custom memory allocation control for 2 classes of the platform directory in...
zoltan@webkit.org [Fri, 4 Dec 2009 10:33:55 +0000 (10:33 +0000)]
Allow custom memory allocation control for 2 classes of the platform directory in WebCore

Reviewed by Eric Seidel.

Inherits the following class from Noncopyable because it is instantiated
by 'new' and no need to be copyable:

class/struct name         - instantiated at: WebCore/'location'

class SharedTimer         - (its child class) workers/WorkerRunLoop.cpp:91

Inherits the following class from FastAllocBase because it is
instantiated by 'new':

class TransformOperations - rendering/style/RenderStyle.h:1175

Noncopyable.h's include added to SharedTimer.h.

* platform/SharedTimer.h:
* platform/graphics/transforms/TransformOperations.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51690 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Xan Lopez <xlopez@igalia.com>
xan@webkit.org [Fri, 4 Dec 2009 10:18:42 +0000 (10:18 +0000)]
2009-12-04  Xan Lopez  <xlopez@igalia.com>

        Reviewed by Gustavo Noronha.

        [GTK]Enable DNS prefetching

        Bump libsoup required version to 2.29.3 for DNS prefetching.

        * configure.ac:


2009-12-04  Xan Lopez  <xlopez@igalia.com>

        Reviewed by Gustavo Noronha.

        [GTK]Enable DNS prefetching

        Enable DNS prefetching.

        Based on a patch by José Millán.

        * platform/network/soup/DNSSoup.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51689 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoRoll out r51687 because it breaks the Leopard builds.
zoltan@webkit.org [Fri, 4 Dec 2009 08:56:25 +0000 (08:56 +0000)]
Roll out r51687 because it breaks the Leopard builds.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51688 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Zoltan Horvath <zoltan@webkit.org>
zoltan@webkit.org [Fri, 4 Dec 2009 08:38:39 +0000 (08:38 +0000)]
2009-12-04  Zoltan Horvath  <zoltan@webkit.org>

        Reviewed by Eric Seidel.

        Allow custom memory allocation control for 2 classes of the platform directory in WebCore

        Inherits the following class from Noncopyable because it is instantiated
        by 'new' and no need to be copyable:

        class/struct name         - instantiated at: WebCore/'location'

        class SharedTimer         - (its child class) workers/WorkerRunLoop.cpp:91

        Inherits the following class from FastAllocBase because it is
        instantiated by 'new':

        class TransformOperations - rendering/style/RenderStyle.h:1175

        * platform/SharedTimer.h:
        * platform/graphics/transforms/TransformOperations.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51687 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Yuzo Fujishima <yuzo@google.com>
eric@webkit.org [Fri, 4 Dec 2009 08:25:22 +0000 (08:25 +0000)]
2009-12-04  Yuzo Fujishima  <yuzo@google.com>

        Reviewed by Darin Fisher.

        Add test to see if unicode messages can be sent over web socket.


        * websocket/tests/script-tests/unicode.js: Added.
        * websocket/tests/unicode-expected.txt: Added.
        * websocket/tests/unicode.html: Added.
        * websocket/tests/unicode_wsh.py: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51686 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-04 Zoltan Horvath <zoltan@webkit.org>
zoltan@webkit.org [Fri, 4 Dec 2009 08:15:01 +0000 (08:15 +0000)]
2009-12-04  Zoltan Horvath  <zoltan@webkit.org>

        Reviewed by Eric Seidel.

        Allow custom memory allocation control for the svg directory in WebCore

        Inherits the following classes from Noncopyable because these are instantiated
        by 'new' and no need to be copyable:

        class/struct name           - instantiated at: WebCore/'location'

        struct ResourceSet          - svg/graphics/SVGResource.cpp:148
        class SVGFontData           - css/CSSFontFaceSource.cpp:156
        class SVGDocumentExtensions - dom/Document.cpp:3962
        class SVGViewSpec           - svg/SVGSVGElement.cpp:191

        * svg/SVGDocumentExtensions.h:
        * svg/SVGFontData.h:
        * svg/SVGViewSpec.h:
        * svg/graphics/SVGResource.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51685 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoWAI-ARIA: aria-activedescendant doesn't work as intended
cfleizach@apple.com [Fri, 4 Dec 2009 07:50:48 +0000 (07:50 +0000)]
WAI-ARIA: aria-activedescendant doesn't work as intended

Reviewed by Beth Dakin.


Fixes a number of issues regarding the "tree" role and aria-activedescendant.
1. The indexes were being reported incorrectly by treeitems.
2. aria-activedescendant changes were not being sent to the containing item.
3. The tree's selected rows need to consult aria-activedescendant.
4. Since a tree changes what it returns as its children (it returns its rows)
   the mac-specific array indexing methods need to correctly handle the tree case.

Tests: platform/mac/accessibility/aria-tree-activedescendant.html

* accessibility/AXObjectCache.h:
* accessibility/AccessibilityRenderObject.cpp:
* accessibility/mac/AXObjectCacheMac.mm:
* accessibility/mac/AccessibilityObjectWrapper.mm:
(-[AccessibilityObjectWrapper accessibilityAttributeValue:]):
(-[AccessibilityObjectWrapper accessibilityIndexOfChild:]):
(-[AccessibilityObjectWrapper accessibilityArrayAttributeValues:index:maxCount:]):
* accessibility/win/AXObjectCacheWin.cpp:


* platform/mac/accessibility/aria-tree-activedescendant-expected.txt: Added.
* platform/mac/accessibility/aria-tree-activedescendant.html: Added.
* platform/mac/accessibility/aria-tree-index-of-items-expected.txt: Added.
* platform/mac/accessibility/aria-tree-index-of-items.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51684 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-03 Zoltan Horvath <zoltan@webkit.org>
zoltan@webkit.org [Fri, 4 Dec 2009 07:48:59 +0000 (07:48 +0000)]
2009-12-03  Zoltan Horvath  <zoltan@webkit.org>

        Reviewed by Eric Seidel.

        Allow custom memory allocation control for the xml directory in WebCore

        Inherits the following class from Noncopyable because it is instantiated
        by 'new' and no need to be copyable:

        class/struct name               - instantiated at: WebCore/'location'

        struct XMLHttpRequestStaticData - xml/XMLHttpRequest.cpp:134

        Inherits the following classes from FastAllocBase because these are instantiated by 'new':

        class NodeTest                  - xml/XPathGrammar.y:258
        class NodeSet                   - xml/XPathValue.cpp:52
        struct EvaluationContext        - xml/XPathExpressionNode.cpp:40

        * xml/XMLHttpRequest.cpp:
        * xml/XPathExpressionNode.h:
        * xml/XPathNodeSet.h:
        * xml/XPathStep.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51683 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-03 Shinichiro Hamaji <hamaji@chromium.org>
hamaji@chromium.org [Fri, 4 Dec 2009 07:42:19 +0000 (07:42 +0000)]
2009-12-03  Shinichiro Hamaji  <hamaji@chromium.org>

        Reviewed by David Levin.

        check-webkit-style should check for camelCase variable names

        * Scripts/modules/cpp_style.py:
        * Scripts/modules/cpp_style_unittest.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51682 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoNew History changes do not compile for Chromium/V8
atwilson@chromium.org [Fri, 4 Dec 2009 06:40:56 +0000 (06:40 +0000)]
New History changes do not compile for Chromium/V8

Reviewed by Adam Barth.

Existing tests suffice (just trying to get code to compile).

* WebCore.gypi:
Added V8HistoryCustom.cpp and other missing files.
* bindings/v8/custom/V8CustomBinding.h:
Added custom handlers for History.pushState()/replaceState().
* bindings/v8/custom/V8HistoryCustom.cpp: Added.
Added custom handlers for History.pushState()/replaceState().
* history/BackForwardListChromium.cpp:
Stubbed out this routine for now - will implement in the future.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51681 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-03 Adam Barth <abarth@webkit.org>
eric@webkit.org [Fri, 4 Dec 2009 06:19:58 +0000 (06:19 +0000)]
2009-12-03  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        OwnPtr<XBLBindingManager> Document::m_bindingManager;

        The document actually owns the bindingManager.  That's what the code
        should say.

        * dom/Document.cpp:
        * dom/Document.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51680 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-03 Adam Barth <abarth@webkit.org>
abarth@webkit.org [Fri, 4 Dec 2009 05:55:46 +0000 (05:55 +0000)]
2009-12-03  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        OwnPtr<Tokenizer> Document::m_tokenizer;

        The document actually owns the tokenizer.  That's what the code should say.

        * dom/Document.cpp:
        * dom/Document.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51679 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-03 Adam Barth <abarth@webkit.org>
abarth@webkit.org [Fri, 4 Dec 2009 05:54:58 +0000 (05:54 +0000)]
2009-12-03  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        OwnPtr<RenderArena> Document::m_renderArena;

        The document owns the renderArena.  That's what the code should say.

        * dom/Document.cpp:
        (WebCore::Document::Document): Also, removed a redundant initialization
        of the tokenizer that I missed in my previous patch.
        * dom/Document.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51678 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoRevert r51633
atwilson@chromium.org [Fri, 4 Dec 2009 05:54:31 +0000 (05:54 +0000)]
Revert r51633

This reverts commit 4c9cc0b9a12d38654b9c687ffc1671c7307b397a.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51677 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years ago2009-12-03 Adam Barth <abarth@webkit.org>
abarth@webkit.org [Fri, 4 Dec 2009 05:28:33 +0000 (05:28 +0000)]
2009-12-03  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        The code should say that Document owns DocLoader

        It's the truth.

        * dom/Document.cpp:
        * dom/Document.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51676 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoTry and fix the Windows build.
barraclough@apple.com [Fri, 4 Dec 2009 02:53:05 +0000 (02:53 +0000)]
Try and fix the Windows build.

* JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def:  Export a symbol that should be exported.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51675 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoTry and fix the Mac build.
mrowe@apple.com [Fri, 4 Dec 2009 02:35:16 +0000 (02:35 +0000)]
Try and fix the Mac build.

* JavaScriptCore.exp:  Export a symbol that should be exported.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51674 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoFixing layout test bustage on platforms.
cfleizach@apple.com [Fri, 4 Dec 2009 02:28:47 +0000 (02:28 +0000)]
Fixing layout test bustage on platforms.

AX: VO just says "term" on many web sites

* platform/mac-leopard/accessibility/lists-expected.txt:
* platform/mac/accessibility/radio-button-group-members-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51673 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoREGRESSION(4.0.3-48777): Crash in JSC::ExecState::propertyNames() (Debug-only?)
oliver@apple.com [Fri, 4 Dec 2009 02:17:46 +0000 (02:17 +0000)]
REGRESSION(4.0.3-48777): Crash in JSC::ExecState::propertyNames() (Debug-only?)

Reviewed by Gavin Barraclough.

Work around odd GCC-ism and correct the scopechain for use by
calls made while a cachedcall is active on the callstack.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51672 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agohttps://bugs.webkit.org/show_bug.cgi?id=32136
barraclough@apple.com [Fri, 4 Dec 2009 02:15:18 +0000 (02:15 +0000)]
Add a rope representation to JSString.  Presently JSString always holds its data in UString form.
Instead, allow the result of a string concatenation to be represented in a tree form - with a
variable sized, reference-counted rope node retaining a set of UString::Reps (or other rope nopes).

Reviewed by Oliver "Brraaaaiiiinnnnnzzzzzzzz" Hunt.

Strings must still currently be resolved down to a flat UString representation before being used,
but by holding the string in a rope representation during construction we can avoid copying data
until we know the final size of the string.

~2% progression on SunSpider (~25% on date-format-xparb, ~20% on string-validate-input).

* JavaScriptCore.exp:

    - Update exports.

* interpreter/Interpreter.cpp:

    - Make use of new JSString::length() method to avoid prematurely resolving ropes.

* jit/JITOpcodes.cpp:

    - Switch the string length trampoline to read the length directly from JSString::m_length,
      rather than from the JSString's UString::Rep's 'len' property.

* jit/JITStubs.cpp:

    - Modify op_add such that addition of two strings, where either or both strings are already
      in rope representation, produces a rope as a result.

* runtime/JSString.cpp:
* runtime/JSString.h:
(JSC::JSString::Rope::operator new):

    - Add rope form.

* runtime/Operations.h:

    - Update string concatenation, and addition of ropes, to produce ropes.

* runtime/StringObject.cpp:

    - Make use of new JSString::length() method to avoid prematurely resolving ropes.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51671 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoAX: VO just says "term" on many web sites
cfleizach@apple.com [Fri, 4 Dec 2009 02:06:58 +0000 (02:06 +0000)]
AX: VO just says "term" on many web sites

Reviewed by Eric Seidel.


Test: platform/mac/accessibility/definition-list-term.html

* accessibility/AccessibilityRenderObject.cpp:
* accessibility/mac/AccessibilityObjectWrapper.mm:
(-[AccessibilityObjectWrapper subrole]):
(-[AccessibilityObjectWrapper roleDescription]):


* DumpRenderTree/AccessibilityUIElement.cpp:
* DumpRenderTree/AccessibilityUIElement.h:
* DumpRenderTree/gtk/AccessibilityUIElementGtk.cpp:
* DumpRenderTree/mac/AccessibilityUIElementMac.mm:
* DumpRenderTree/win/AccessibilityUIElementWin.cpp:


* platform/mac/accessibility/aria-grouping-roles-expected.txt:
* platform/mac/accessibility/aria-grouping-roles.html:
* platform/mac/accessibility/definition-list-term-expected.txt: Added.
* platform/mac/accessibility/definition-list-term.html: Added.
* platform/mac/accessibility/lists-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51670 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoImplement WAI-ARIA scrollbar role and related property aria-orientation
cfleizach@apple.com [Fri, 4 Dec 2009 01:00:07 +0000 (01:00 +0000)]
Implement WAI-ARIA scrollbar role and related property aria-orientation

Reviewed by Beth Dakin.


Test: accessibility/aria-scrollbar-role.html

* accessibility/AccessibilityObject.cpp:
* accessibility/AccessibilityObject.h:
* accessibility/AccessibilityRenderObject.cpp:
* accessibility/AccessibilityRenderObject.h:
* accessibility/mac/AccessibilityObjectWrapper.mm:
(-[AccessibilityObjectWrapper additionalAccessibilityAttributeNames]):
(-[AccessibilityObjectWrapper accessibilityAttributeValue:]):
* html/HTMLAttributeNames.in:


* DumpRenderTree/AccessibilityUIElement.cpp:
* DumpRenderTree/AccessibilityUIElement.h:
* DumpRenderTree/gtk/AccessibilityUIElementGtk.cpp:
* DumpRenderTree/mac/AccessibilityUIElementMac.mm:
* DumpRenderTree/win/AccessibilityUIElementWin.cpp:


* accessibility/aria-scrollbar-role-expected.txt: Added.
* accessibility/aria-scrollbar-role.html: Added.
* platform/gtk/Skipped:
* platform/win/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51669 268f45cc-cd09-0410-ab3c-d52691b4dbfc

13 years agoFixed <rdar://problem/7401617> Selection painting issue in hardware-
mitz@apple.com [Fri, 4 Dec 2009 00:16:50 +0000 (00:16 +0000)]
Fixed <rdar://problem/7401617> Selection painting issue in hardware-
accelerated layers
which is another part of https://bugs.webkit.org/show_bug.cgi?id=23628
Fix selection painting to do container-relative repaints

Reviewed by Darin Adler.

Test: fast/repaint/block-selection-gap-in-composited-layer.html

* rendering/RenderBlock.cpp:
(WebCore::RenderBlock::selectionGapRectsForRepaint): Compute and paint
gap rects in the coordinate space of the repaint container.
* rendering/RenderLayerCompositor.cpp:
(WebCore::RenderLayerCompositor::recursiveRepaintLayerRect): Added a
* rendering/RenderView.cpp:
(WebCore::RenderView::setSelection): Map block selection gap rects from
the repaint container’s coordinate space to the view’s coordinate space
before adding them to the cached selection bounds.
(WebCore::RenderView::clearSelection): Changed to use
repaintRectangleInViewAndCompositedLayers() so that the selection rect
is invalidated in composited layers as well.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@51668 268f45cc-cd09-0410-ab3c-d52691b4dbfc