OSDN Git Service
Steve Block [Wed, 5 Jan 2011 15:39:46 +0000 (15:39 +0000)]
Merge WebKit at r72805: Update WebKit revision
Note that this is a backwards merge from Chromium release 9.0.600.0
to 9.0.597.0, to align with the Chromium 9 stable release branch.
Change-Id: Idd1eee54590180c2efc16861453092295651eba3
Steve Block [Wed, 5 Jan 2011 14:52:42 +0000 (14:52 +0000)]
Merge WebKit at r72805: Cherry-pick WebKit change 73062
This change is the upstreaming of local changes in
WebCore/platform/network. It did not generate conflicts in the merge to
r73109 as both upstream and local versions matched at r73109. But since
it appears after r72805, it was inadvertantly removed as part of the
initial merge by Git to r72805.
See http://trac.webkit.org/changeset/73062
Note that this is a backwards merge from Chromium release 9.0.600.0
to 9.0.597.0, to align with the Chromium 9 stable release branch.
Change-Id: I6a266e1270cbc1f155065ae1c56c491ccf33e497
Steve Block [Wed, 5 Jan 2011 15:06:26 +0000 (15:06 +0000)]
Merge WebKit at r72805: Cherry-pick WebKit change 72948
This fixes CodeGeneratorV8.pm
It is a re-commit of 'Merge WebKit at r72274: Fix CodeGeneratorV8.pm',
which was inadvertantly removed as part of the initial merge by Git to
r72805.
See http://trac.webkit.org/changeset/72948
Note that this is a backwards merge from Chromium release 9.0.600.0
to 9.0.597.0, to align with the Chromium 9 stable release branch.
Change-Id: Iaa56a5a503aaaf4d996ebc44d66d557d12838cfa
Steve Block [Wed, 5 Jan 2011 12:20:13 +0000 (12:20 +0000)]
Merge WebKit at r72805: Update call sites to RenderFrame::docWidth()/docHeight()
This is effectively a revert of
"Merge Webkit at r73109: render frame docWidth/docHeight method changed."
See http://trac.webkit.org/changeset/72852 and
http://trac.webkit.org/changeset/73063
Note that this is a backwards merge from Chromium release 9.0.600.0
to 9.0.597.0, to align with the Chromium 9 stable release branch.
Change-Id: I81915808439847dce87beefdd527f77962e074b5
Steve Block [Wed, 5 Jan 2011 12:24:24 +0000 (12:24 +0000)]
Merge WebKit at r72805: Undo previous conflict resolution in RenderSVGResourceSolidColor.cpp
This is effectively a partial revert of
"Merge Webkit at r73109: Fix conflicts"
See http://trac.webkit.org/changeset/72926
Note that this is a backwards merge from Chromium release 9.0.600.0
to 9.0.597.0, to align with the Chromium 9 stable release branch.
Change-Id: I956ac20ac4985ed69921a8efad4a351ba0034e84
Steve Block [Wed, 5 Jan 2011 12:29:23 +0000 (12:29 +0000)]
Merge WebKit at r72805: Update GraphicsContext method signatures
Note that this is a partial revert of
"Merge Webkit at r73109: graphics method signature changed."
See http://trac.webkit.org/changeset/72926
Note that this is a backwards merge from Chromium release 9.0.600.0
to 9.0.597.0, to align with the Chromium 9 stable release branch.
Change-Id: I7e999ae384bd791c5953f5edcf494a008e6fceea
Steve Block [Wed, 5 Jan 2011 12:35:52 +0000 (12:35 +0000)]
Merge WebKit at r72805: Update EditorClient::getGuessesForWord()
This is a revert of
"Merge Webkit at r73109: getGuessForWord method signature change"
See http://trac.webkit.org/changeset/73088
Note that this is a backwards merge from Chromium release 9.0.600.0
to 9.0.597.0, to align with the Chromium 9 stable release branch.
Change-Id: If23862010b9e047f10991de5b428ca1b74b306b0
Steve Block [Wed, 5 Jan 2011 12:37:24 +0000 (12:37 +0000)]
Merge WebKit at r72805: Use Node::parent() rather than Node::parentNode()
This is a revert of
"Merge WebKit at r73109: use parentNode method."
See http://trac.webkit.org/changeset/72825
This reverts commit
23aa22bb0c8bf0d91c99d21c5dca4c64a4825a87.
Note that this is a backwards merge from Chromium release 9.0.600.0
to 9.0.597.0, to align with the Chromium 9 stable release branch.
Change-Id: I5ba4d0976c3b87b32326fdc8e2eb1cffdd7541cf
Steve Block [Wed, 5 Jan 2011 15:22:02 +0000 (15:22 +0000)]
Merge WebKit at r72805: Fix Makefiles
This is a partial revert of
"Merge Webkit at r73109: Fix Makefiles."
Note that this is a backwards merge from Chromium release 9.0.600.0
to 9.0.597.0, to align with the Chromium 9 stable release branch.
Change-Id: Iac42b5b4232b71fab322bf99e74e7fb6c7be10c5
Steve Block [Wed, 5 Jan 2011 12:15:11 +0000 (12:15 +0000)]
Merge WebKit at r72805: Initial merge by Git
Note that this is a backwards merge from Chromium release 9.0.600.0
to 9.0.597.0, to align with the Chromium 9 stable release branch.
Change-Id: I5d2bb4e8cee9d39ae8485abf48bdb55ecf8b3790
Cary Clark [Thu, 6 Jan 2011 13:07:46 +0000 (05:07 -0800)]
Merge "refine empty picture test to look for color rects"
Leon Scroggins [Wed, 5 Jan 2011 20:14:34 +0000 (15:14 -0500)]
Keep track of whether input fields have autocomplete.
Bug:
3300365
Requires a change in frameworks/base.
Change-Id: Ia466cae0c43b43817651e9d5903e18aae96a814f
Teng-Hui Zhu [Wed, 5 Jan 2011 17:19:00 +0000 (09:19 -0800)]
Merge "clip rect fix for iframe when GL on"
Cary Clark [Wed, 5 Jan 2011 15:58:18 +0000 (10:58 -0500)]
refine empty picture test to look for color rects
bug:
3296986
Change-Id: I171435f6285f671abc9f2e38ad00ebcabb783cb3
Derek Sollenberger [Wed, 5 Jan 2011 15:48:25 +0000 (07:48 -0800)]
Merge "Plugin API to report the onscreen visibility of the plugin."
Derek Sollenberger [Wed, 5 Jan 2011 14:40:04 +0000 (09:40 -0500)]
Plugin API to report the onscreen visibility of the plugin.
bug:
3324143
Change-Id: I4ad5837b4d79ee63bf53ce974a634d357130930e
Cary Clark [Wed, 5 Jan 2011 14:36:53 +0000 (09:36 -0500)]
use same matrix left when reading picture
In select text, the nav cache picture is read multiple times.
If the matrix is different, it's possible that the text
metrics will be off by one when the text position is near
to but not equal to an integer (e.g., 315.999969). Pinning
the matrix to zero makes the matrices the same in the bug case.
Use the simpler doIRect interface if the glyph rec isn't needed.
Check for contentRender() before dereferencing it.
bug:
3315620
Change-Id: I4bfaf9f9b7bf12f66b29f4609d744bceaf5b2964
Cary Clark [Wed, 29 Dec 2010 19:54:43 +0000 (14:54 -0500)]
start select all without tapping
For the keyboard interface to 'Select All',
establish the select text context by finding some
text on the page and returning its coordinates.
This allows tapping and keyboard interfaces to share
most of the select all implementation.
bug:
3270434
Change-Id: Ic080a8002115af9086a669dd366d0399b1082f3c
Steve Block [Wed, 5 Jan 2011 11:50:42 +0000 (03:50 -0800)]
am
c9f38c56: (-s ours) am
08fcb454: Cherry-pick WebKit change 60729 to avoid an AppCache crash - DO NOT MERGE
* commit '
c9f38c569b3b11991dc69847f941517d0164bff4':
Cherry-pick WebKit change 60729 to avoid an AppCache crash - DO NOT MERGE
Steve Block [Wed, 5 Jan 2011 11:48:41 +0000 (03:48 -0800)]
am
08fcb454: Cherry-pick WebKit change 60729 to avoid an AppCache crash - DO NOT MERGE
* commit '
08fcb454f4c35929b9ab4b984ce1a279a105c585':
Cherry-pick WebKit change 60729 to avoid an AppCache crash - DO NOT MERGE
Teng-Hui Zhu [Tue, 4 Jan 2011 23:13:55 +0000 (15:13 -0800)]
clip rect fix for iframe when GL on
First, for each layer, the clip rect should be intersecting with
all of its parents.
Then the rect should use pre-transformation info, the real
transformation will handle by the clip() function.
P.S: renaming variable "bounds" b/c there is a member function
using the same name.
bug:
3304761
Change-Id: Id5d602b55da904c63b37cb5436b5bf2f5526a94f
Derek Sollenberger [Tue, 4 Jan 2011 21:42:33 +0000 (13:42 -0800)]
Merge "Add full-screen support for OpenGL based plugins."
Patrick Scott [Tue, 4 Jan 2011 21:29:57 +0000 (13:29 -0800)]
Merge "Remove the div element check."
Derek Sollenberger [Tue, 4 Jan 2011 20:57:57 +0000 (15:57 -0500)]
Add full-screen support for OpenGL based plugins.
bug:
3321891
Change-Id: Ife1bbeed91b3285277bfe47c71f523d477891b98
Patrick Scott [Tue, 4 Jan 2011 20:58:53 +0000 (15:58 -0500)]
Remove the div element check.
This allows other elements to specify overflow.
Bug:
3310376
Change-Id: I039b07aebeefa9184b56950972222ffea1ed09c7
Steve Block [Tue, 4 Jan 2011 19:51:48 +0000 (11:51 -0800)]
Merge "Update JSDOMWindowBase to reflect recent upstreaming"
Patrick Scott [Tue, 4 Jan 2011 15:20:10 +0000 (10:20 -0500)]
Fix mlb shopping site not drawing correctly.
Return false in hasOverflowScroll if flattening is on.
Bug:
3309139
Change-Id: Ia4c981fab26154a38e66412d7e2938da32e7d718
Steve Block [Fri, 31 Dec 2010 13:19:41 +0000 (13:19 +0000)]
Update JSDOMWindowBase to reflect recent upstreaming
An Android-specific modification was upstreamed in
http://trac.webkit.org/changeset/74806
Bug:
3305994
Change-Id: Ie43475aa611259d2d2de39397e167e42b24434f3
Teng-Hui Zhu [Wed, 29 Dec 2010 17:53:03 +0000 (09:53 -0800)]
Merge "A WAR to get content showing up when GL enabled"
Teng-Hui Zhu [Tue, 28 Dec 2010 23:27:13 +0000 (15:27 -0800)]
A WAR to get content showing up when GL enabled
The emptiness check is not optimal for GL rendering yet.
That cause some content are not showing up when GL enabled.
A WAR is to just update the textures and use them without
emptiness check.
bug:
3296986
Change-Id: I5087dbc34b744ffbe24f4a4a6616f4d5715e1f11
Cary Clark [Tue, 28 Dec 2010 14:23:04 +0000 (09:23 -0500)]
join rects without gaps
While analyzing cursor rings, regions are built from text,
rects, and bitmaps to determine the areas inside and outside
the rings. These areas determine if the rings are fully visible,
and if they can be expanded into a larger bounds.
Empty input areas are defined only by rects drawn as their
background. In general, rects may be drawn as pieces that need
to be stitched together. Unlike glyphs, the stitching should
not allow gaps between rects.
By detecting gaps, and treating each unjoined rect as its own
area, a page with multiple successive input fields correctly
computes cursor rect bounds.
bug:
3116242
Change-Id: If702b692c99c96c0188b42a638f750da12af6df4
Cary Clark [Thu, 23 Dec 2010 20:04:20 +0000 (12:04 -0800)]
Merge "use text slop for ring test"
Cary Clark [Thu, 23 Dec 2010 19:47:34 +0000 (14:47 -0500)]
use text slop for ring test
The mTextSlop contains the cursor ring test area.
The mTestBounds contains the total test area.
Use the former when checking to see if text is
drawn outside of the cursor rings, but inside
the bounds.
bug:
3307120
Change-Id: Ice69ce3d70aabb435890a0fd8848ef3fe2de9796
Shimeng (Simon) Wang [Thu, 23 Dec 2010 18:58:40 +0000 (10:58 -0800)]
Merge WebKit at r73109: put correct ordering at Makefile.
Change-Id: I8953ea3eb6e78548c50953495cf688feaf9a8385
Steve Block [Thu, 23 Dec 2010 17:35:15 +0000 (17:35 +0000)]
Cherry-pick WebKit change 60729 to avoid an AppCache crash - DO NOT MERGE
This was observed when the AppCache database path was not set and hence
opening the database failed.
This change has already been pulled into master as part of the regular WebKit
merge process, so do not merge.
See http://trac.webkit.org/changeset/60729
Bug:
3305684
Change-Id: I5707839a9b0f886a24604f890ca4d575e311edbe
Shimeng (Simon) Wang [Thu, 9 Dec 2010 02:50:49 +0000 (18:50 -0800)]
Merge Webkit at r73109: render frame docWidth/docHeight method changed.
Change-Id: I492d321eea76f2d2c5278e44ae9c2fe37b5834a6
See:
http://trac.webkit.org/changeset/72852
http://trac.webkit.org/changeset/73063
Shimeng (Simon) Wang [Thu, 9 Dec 2010 01:48:05 +0000 (17:48 -0800)]
Merge Webkit at r73109: Fix Makefiles.
Change-Id: If8cc4cfc06399c67b08cf5157852de90876c3d55
Shimeng (Simon) Wang [Wed, 8 Dec 2010 19:32:25 +0000 (11:32 -0800)]
Merge Webkit at r73109: Fix conflicts
1. WebCore/Android.jscbindings.mk
http://trac.webkit.org/changeset/72590
http://trac.webkit.org/changeset/72344
2. WebCore/WebCore.vcproj/WebCore.vcproj
too many changes between 72274 and 73109
http://trac.webkit.org/log/trunk/WebCore/WebCore.vcproj/WebCore.vcproj
also due to Android change.
https://android-git.corp.google.com/g/#change,37374
3. WebCore/platform/android/PlatformBridge.h
http://trac.webkit.org/changeset/72631
4. WebCore/rendering/RenderSVGResourceSolidColor.cpp
http://trac.webkit.org/changeset/72926
our code is not upstreamed
5. WebCore/rendering/RenderView.h
http://trac.webkit.org/changeset/72852
http://trac.webkit.org/changeset/73063
Change-Id: Id7a1dbdfef6327b4fe08c9884122bf9decd039ca
Shimeng (Simon) Wang [Wed, 22 Dec 2010 18:07:38 +0000 (10:07 -0800)]
Merge WebKit at r73109: return true to support rich source info.
This is because:
http://trac.webkit.org/changeset/72360
which disables line info except in case of enabling JS debugger or inpspector.
This is a quick fix. Long term solution is to get this fixed upstream.
Change-Id: Ibc5dd89719628848ab4710fcc7c06453b60cc119
Shimeng (Simon) Wang [Thu, 16 Dec 2010 21:47:51 +0000 (13:47 -0800)]
Merge WebKit at r73109: Fix V8 JNI binding.
cherry pick the change for webkit bug:
https://bugs.webkit.org/show_bug.cgi?id=51156
And fix array bindings.
Change-Id: I64b89fadd10ac87fa998a79b2af4aa9db07a8318
Shimeng (Simon) Wang [Thu, 9 Dec 2010 22:48:17 +0000 (14:48 -0800)]
Merge WebKit at r73109: Update WebKit revision number.
Change-Id: I1748e24000110d5c667f729c25131d5e624de41f
Shimeng (Simon) Wang [Thu, 9 Dec 2010 19:43:06 +0000 (11:43 -0800)]
Merge Webkit at r73109: graphics method signature changed.
this is because:
http://trac.webkit.org/changeset/72926
http://trac.webkit.org/changeset/72528
Change-Id: I511eef7fece8e97ed4a199ca7b6f388124921b79
Shimeng (Simon) Wang [Thu, 9 Dec 2010 02:11:01 +0000 (18:11 -0800)]
Merge Webkit at r73109: add new method.
This is because:
http://trac.webkit.org/changeset/72631
This is to merge the following change into Android:
https://android-git.corp.google.com/g/#change,81198
Change-Id: Ifc7ca98388ca88828669371df501b616f8021e80
Shimeng (Simon) Wang [Thu, 9 Dec 2010 02:06:15 +0000 (18:06 -0800)]
Merge Webkit at r73109: getGuessForWord method signature change
This is because:
http://trac.webkit.org/changeset/73088
Change-Id: I4051bc71c638e74ede803b216d9035311cbaf198
Shimeng (Simon) Wang [Thu, 9 Dec 2010 02:03:15 +0000 (18:03 -0800)]
Merge WebKit at r73109: use parentNode method.
This is because of:
http://trac.webkit.org/changeset/72825/trunk/WebCore/dom/Node.h
which makes the parent method private.
Change-Id: I706c4cbc4e69bee990f9e138aeb7ad2ed2a3d42f
Shimeng (Simon) Wang [Wed, 8 Dec 2010 01:22:45 +0000 (17:22 -0800)]
Merge WebKit at r73109: Initial merge by git.
Change-Id: I61f1a66d9642e3d8405d3ac6ccab2a53421c75d8
Ben Murdoch [Wed, 22 Dec 2010 21:59:17 +0000 (13:59 -0800)]
am
1e5798a4: am
d436a8e4: Allow HTML5 video to seek.
* commit '
1e5798a41696c2d04b2a28eabf9e5914fb6856e2':
Allow HTML5 video to seek.
Ben Murdoch [Wed, 22 Dec 2010 21:56:43 +0000 (13:56 -0800)]
am
d436a8e4: Allow HTML5 video to seek.
* commit '
d436a8e4bf9a2c6eab6a09513ce13340c84ef10c':
Allow HTML5 video to seek.
Ben Murdoch [Wed, 22 Dec 2010 18:45:20 +0000 (18:45 +0000)]
Allow HTML5 video to seek.
Need to return the duration of the currently playing video to WebCore
to allow the video to seek correctly.
Bug:
3303699
Change-Id: I79f6a7e7217cf987d35b5b1a9313a918630be7a0
Cary Clark [Mon, 20 Dec 2010 17:33:36 +0000 (12:33 -0500)]
use copies of matrix, paint instead of references
The matrix and paint may be gone when later referenced
to determine the width of spaces. Replace the references
with copies.
When widening the initial selection to a word, the first
or last glyph could be skipped if the tap was one
glyph away. Detect this include the final glyph.
The bitmap used to detect a space between glyphs could
be too small for the test; make it as large as possible.
The test for finding the closest glyph when the test point
is inside the glyph's bounds should be triggered only
if the best glyph found so far is also a match.
If the initial hit point doesn't touch a word, abort.
Only measure the distance for test points if the test
text and best text have the same column state or if
the test point is outside of the current column.
The initial word selection needs to work differently from
extending the selection. Make the clip small for the
initial detect and widen it for the word expansion.
Build the selection region before expanding the word
since the column selection uses that data.
There is a companion bug fix in frameworks/base triggered
by this code, but neither is dependent on the other.
bug:
3293330
Change-Id: If9d1a3bd7ec5d1642b63dbab1845d2f3735ddc81
Kristian Monsen [Wed, 22 Dec 2010 12:09:14 +0000 (12:09 +0000)]
Removing the host resolver leak
The crash was fixed when some unsafe local changes were removed recently.
I think this was what removed the crash:
https://android-git.corp.google.com/g/#change,84795
Fix for bug
3243797
Change-Id: If23e55270f035831ef7781681b012dfdca0ac74d
Kristian Monsen [Wed, 22 Dec 2010 11:35:41 +0000 (03:35 -0800)]
Merge "Adding a method to delete a WebCache"
Kristian Monsen [Tue, 21 Dec 2010 16:31:19 +0000 (16:31 +0000)]
Adding a method to delete a WebCache
This is needed so the cache doesn't survive when closing
all private tabs.
Change-Id: I52ffaf89f8794466f98370c289b9a646de0fdb59
Bart Sears [Tue, 21 Dec 2010 21:52:48 +0000 (13:52 -0800)]
Turn iframe flattening back on
Until we can figure out why web gmail is having problems turn
iframe flattening back on.
Bug:
3294278
Change-Id: I7096be2a2178c6b56b5450defb405a265fc2b477
Jing Yu [Wed, 15 Dec 2010 22:28:35 +0000 (14:28 -0800)]
Override ALWAYS_INLINE attribute with inline.
ALWAYS_INLINE attribute was abused in libwebcore.so. By overriding
it with normal inline attribute, the size of libwebcore.so is
reduced by 3.71%, wihtout noticable performance regression on
webcore_test benchmark.
Change-Id: Icc1710726f28a93c3be660ba7c22c8c356748460
Patrick Scott [Tue, 21 Dec 2010 16:50:36 +0000 (11:50 -0500)]
Turn off flattening again now that scrolling is
fixed.
Change-Id: I2cdedf37cc7cff7b0cfe96dd65803902ca7e92c6
Patrick Scott [Tue, 21 Dec 2010 15:36:17 +0000 (10:36 -0500)]
Fix iframe scrolling.
Do the same check everywhere for enabling scrolling in an iframe. Make sure the
owner renderer is a RenderIFrame to avoid trying to composite an embedded
object. Also check if the owner layer has visible content since some iframes
are marked hidden.
Dirty the parent normal flow list when turning on overflow scrolling for divs.
Bug:
3280851
Bug:
3290276
Change-Id: I8d0a49324e06426c15abaf552874a32d6e2ee11a
Steve Block [Tue, 21 Dec 2010 14:50:21 +0000 (06:50 -0800)]
Merge "Set 660 permissions on Appcache database"
Kristian Monsen [Tue, 21 Dec 2010 14:45:30 +0000 (06:45 -0800)]
Merge "Revert "Revert "Merge "Add a method java can send proxy changes to""""
Steve Block [Tue, 21 Dec 2010 14:18:10 +0000 (14:18 +0000)]
Set 660 permissions on Appcache database
Bug:
3244281
Change-Id: I5c90f44c4e8c3cd7d1ae2652e5d9f864a69b5dfe
Steve Block [Tue, 21 Dec 2010 14:17:49 +0000 (06:17 -0800)]
Merge "Set 660 permissions on HTML5 Database API meta database"
Steve Block [Tue, 21 Dec 2010 11:55:38 +0000 (11:55 +0000)]
Set 660 permissions on HTML5 Database API meta database
Bug:
3244281
Change-Id: I7af7c47c92fc180435cff2416922916a1114c88f
Kristian Monsen [Tue, 21 Dec 2010 12:28:10 +0000 (12:28 +0000)]
Revert "Revert "Merge "Add a method java can send proxy changes to"""
This reverts commit
9c4386e62deef6236674a2b2073a1efd514d75ac.
Ben Murdoch [Tue, 21 Dec 2010 10:25:16 +0000 (02:25 -0800)]
Merge "Only download the same poster URL once."
Huahui Wu [Mon, 20 Dec 2010 23:02:48 +0000 (15:02 -0800)]
Merge "Revert "Merge "Add a method java can send proxy changes to"""
Huahui Wu [Mon, 20 Dec 2010 22:41:06 +0000 (14:41 -0800)]
Revert "Merge "Add a method java can send proxy changes to""
This reverts commit
c3a805c84165787f1693b3502ce139d6ea405307, reversing
changes made to
062390c915ca7282c88b435ad4fa7b6f157fb5ef.
Nicolas Roard [Sat, 18 Dec 2010 01:55:42 +0000 (17:55 -0800)]
Add clipping support
Change-Id: If26bb40c4f75caa0fc9146aecb9f06dd1bcecd44
Derek Sollenberger [Mon, 20 Dec 2010 19:55:52 +0000 (11:55 -0800)]
Merge "Prevent media layers from allocating textures."
Nicolas Roard [Sat, 18 Dec 2010 03:36:55 +0000 (19:36 -0800)]
Fix image drawing
Change-Id: I6dacf397f510948b4ecc3c180d2f61050dec00a2
Kristian Monsen [Mon, 20 Dec 2010 18:14:45 +0000 (10:14 -0800)]
Merge "Add a method java can send proxy changes to"
Derek Sollenberger [Mon, 20 Dec 2010 17:28:57 +0000 (12:28 -0500)]
Prevent media layers from allocating textures.
bug:
2907737
Change-Id: I3b20a91eda8b03572e111e3834d3abcaecaa17de
Steve Block [Mon, 20 Dec 2010 17:27:43 +0000 (09:27 -0800)]
Merge "Set 660 permissions on Geolocation cache DB"
Steve Block [Mon, 20 Dec 2010 13:25:03 +0000 (13:25 +0000)]
Set 660 permissions on Geolocation cache DB
Bug:
3256540
Change-Id: I2774f100c9e9c458ebdaf9b6d188c52e79fd2568
Kristian Monsen [Fri, 17 Dec 2010 12:05:47 +0000 (12:05 +0000)]
Add a method java can send proxy changes to
Used by, and uses, this in frameworks/base:
https://android-git.corp.google.com/g/#change,85836
Change-Id: I618f0f0cbeefc660152e60cced3940772f764e1c
Derek Sollenberger [Mon, 20 Dec 2010 13:26:56 +0000 (05:26 -0800)]
Merge "Add initial support for OpenGL plugins."
Derek Sollenberger [Fri, 17 Dec 2010 18:44:51 +0000 (13:44 -0500)]
Add initial support for OpenGL plugins.
bug:
2907737
Change-Id: I6f4f21146c834554db5ecb67fed8994f1cdc2cfe
Ben Murdoch [Mon, 20 Dec 2010 13:13:53 +0000 (13:13 +0000)]
Only download the same poster URL once.
WebKit r64884 introduced code that calls setPoster multiple times. On
Android this causes the Java side to attempt to download the poster many
times which for a long video results in OOM. So if the poster hasn't
changed, don't try to download it again.
Bug:
3277303
Change-Id: Ia401b78cf250af2c1fe0e600dea6d1da0971912a
Ben Murdoch [Mon, 20 Dec 2010 11:51:20 +0000 (11:51 +0000)]
Fix build when SVG is disabled.
Only build the SVG bindings if the environment varaible is set.
Bug:
3295269
Change-Id: I5ec7cb887afac0d394776487cf155fef6b566a6b
Ben Murdoch [Mon, 20 Dec 2010 11:47:48 +0000 (03:47 -0800)]
Merge "Fix crash in AutoFill"
Bart Sears [Mon, 20 Dec 2010 00:32:57 +0000 (16:32 -0800)]
Turn iframe flattening back on
Until we have iframe compositing re-enabled, go back to the
old iframe flattening.
Bug
3294278
Change-Id: I68d2d02d47b8f4ed1870f197703decfa3e94cf01
Patrick Scott [Fri, 17 Dec 2010 22:15:10 +0000 (14:15 -0800)]
Merge "Turn off iframe compositing until it is more stable."
Patrick Scott [Fri, 17 Dec 2010 22:12:51 +0000 (17:12 -0500)]
Turn off iframe compositing until it is more stable.
Change-Id: I2d54918b5910285bf7f1cfb408af8a6b038b65f9
Jeff Sharkey [Fri, 17 Dec 2010 19:59:48 +0000 (11:59 -0800)]
Merge "Fix two SIGSEGV bugs in request interception."
Ben Murdoch [Fri, 17 Dec 2010 17:42:21 +0000 (17:42 +0000)]
Fix crash in AutoFill
We recently changed WebKit to use one URLRequestContext per WebView.
However AutoFill did not get migrated to the new system and so
we have observed crashes due to a dangling pointer. We have one
AutoFill object per WebView and hence one Profile object per webview.
we access the URLRequestContext through the Profile so inject the
correct context in when the AutoFill object is created.
Needs a change in external/chromium, see
I004ab7ac9c6a72be1b62aa8e44322b97a371c136
Bug:
3291654
Change-Id: Iace23ae8c0240f0134894c12d2f7281161aa6150
Patrick Scott [Fri, 17 Dec 2010 15:16:35 +0000 (07:16 -0800)]
Merge "Use a lock around ioThread creation."
Patrick Scott [Fri, 17 Dec 2010 14:45:44 +0000 (09:45 -0500)]
Use a lock around ioThread creation.
This prevents use of the new thread before it actually starts running.
Change-Id: Idabb9cf81c62296534b696b82a08d3801aa16af8
Steve Block [Fri, 17 Dec 2010 14:37:39 +0000 (06:37 -0800)]
Merge "Don't strip HTTP response headers with empty values"
Steve Block [Fri, 17 Dec 2010 14:14:31 +0000 (14:14 +0000)]
Don't strip HTTP response headers with empty values
This change affects the Android HTTP stack only. We now match the XHR
spec - http://www.w3.org/TR/XMLHttpRequest/#the-getresponseheader-method
- and provide header strings even if they are empty.
Bug:
3294032
Change-Id: Ic42e5e97e0a86ae2d08680ad689a2ab249d6ecc7
Steve Block [Fri, 17 Dec 2010 14:05:29 +0000 (06:05 -0800)]
Merge "Fail when building with JSC and the Chrome HTTP stack"
Steve Block [Fri, 17 Dec 2010 13:20:29 +0000 (13:20 +0000)]
Fail when building with JSC and the Chrome HTTP stack
This seems better than silently switching to the Android HTTP stack,
which could be confusing.
Change-Id: I9396beeed1e5803fe5ab16ce0d232dc5d0683f0a
Jeff Sharkey [Fri, 17 Dec 2010 04:31:13 +0000 (20:31 -0800)]
Fix two SIGSEGV bugs in request interception.
When intercepting requests, properly ReleaseStringUTFChars for mimeType
and encoding after bringing over from Java. Also, when finishing an
intercepted request that failed, construct a valid WebResponse.
Change-Id: I2990df2108a8fdf45c103512a8b9a30f4d6992ce
Nicolas Roard [Thu, 16 Dec 2010 22:59:13 +0000 (14:59 -0800)]
Fix for layers repainting
Change-Id: I4072a5d560243d1d0053296b5128a9e411d5750a
Nicolas Roard [Wed, 8 Dec 2010 19:37:30 +0000 (11:37 -0800)]
Cleanup pass to make files compliant with webkit style
Change-Id: I9234be0c1559d5cc11af73ebd8cc14f98f6a4c2e
Nicolas Roard [Wed, 8 Dec 2010 19:37:30 +0000 (11:37 -0800)]
Layers implementation in GL.
- add textures management code for layers in TilesManager
- add the layers drawing GL code
- modify BaseLayerAndroid::drawGL() to paint the layers
- modify TexturesGenerator to process layers painting requests
Note: some further refactoring (when getting rid of SkLayer as a base
class) will let us remove some redudant function, e.g.
setAnchorPoint/setAnchorPointZ (bug:
3292285)
Change-Id: I3cc88575f331104d45c89e84ba27f3af755aa845
Patrick Scott [Thu, 16 Dec 2010 19:46:22 +0000 (11:46 -0800)]
Merge "Pass redirects through canHandleRequest."
Cary Clark [Thu, 16 Dec 2010 18:21:01 +0000 (13:21 -0500)]
improve text selection in scrollable divs
Detect which layer is hit by looking for scrollable layers.
In this case it isn't necessary to see what is drawn or where,
because the layer has already been prescreened by the parent clip.
Adjust the x/y hit by the layer's position. This corrects the
position when the layer scrolls.
Fix CachedLayer's debug output to be in sync with LayerAndroid.h
SelectText edge cases are corrected when there is not text to hit.
To find the closest character, first look at the edge of the
text line. If the hit is inside the line of text, measure from the
center of the character. This helps when textlines overlap.
bug:
3275625
bug:
3191699
Change-Id: I924cc0432f6dace32e0a7444fe69ef15dca2e77f
Patrick Scott [Thu, 16 Dec 2010 14:02:05 +0000 (09:02 -0500)]
Pass redirects through canHandleRequest.
Force the user gesture global to true in order for canHandleRequest to ask the
application.
Bug:
3278072
Change-Id: I1f821ba42c6c630465658b7044ee89c7255eb9a1
Teng-Hui Zhu [Wed, 15 Dec 2010 17:30:05 +0000 (09:30 -0800)]
Better thread safety protection on m_usable
m_usable is written in the thread of texture painting, it should
be better protected.
bug:
3215000
Change-Id: I5b1a5eaa4e469b6e5c840cd7299a1d955780729a
Nicolas Roard [Wed, 15 Dec 2010 21:31:34 +0000 (13:31 -0800)]
Implement a GLUtils::toSkMatrix() function
This also fix the rotation in software rendering.
Change-Id: Iffe1ed3666c9501861ec6793294bf5cde89026d7
Nicolas Roard [Wed, 15 Dec 2010 21:49:13 +0000 (13:49 -0800)]
Merge "Fix text selection"