Age | Commit message (Collapse) | Author |
|
|
|
Original code only supports an animation gif whose frames are
all same size and all same color format. This modification supports
various type of animation gif specified in GIF89a.
<http://www.w3.org/Graphics/GIF/spec-gif89a.txt>
Change-Id: I517182d2d227c9330aaff1c4cebc6c02b31ff8a6
|
|
SkFlipPixelRef.cpp: SkPixelRef::Registrar::Registrar reg(...), names the constructor, not the type.
SkImageRef_GlobalPool.cpp: SkPixelRef::Registrar::Registrar reg(...), names the constructor, not the type.
Change-Id: I778005bc874b811cfa5be2fa1a0ad1ebe9b2450b
|
|
Change-Id: Ic9add408166bf9a0350db904e15912b723e55a15
|
|
Change-Id: I0871ad4aa742706d6bc0cef62925de1c2a20c99d
|
|
Change-Id: Ic7024b390883841714ac6fb5f6a11f87a9f707a6
|
|
|
|
The bug occurs when linear text is meassured. E.g. without this
correction, the following test case will fail
String s = new String("abcdef");
Paint p = new Paint();
p.setLinearText(true);
float width = p.measureText(s);
int numberOfCharacters = p.breakText(s, true, width, null);
assertTrue(numberOfCharacters == s.length());
The reason for this bug is that canonical text size is used to messure
linear text while the maxWidth is not scaled correspondingly.
Change-Id: Iede2d769e43e0914f3e877ee1f46eeeb6b5405ac
Signed-off-by: Peter Eliasson <peter.eliasson.x@sonyericsson.com>
|
|
GetUnitsPerEM can be used as convenience method to
find out the number of font units per EM square
without having to fetch the head table.
Change-Id: I6f09b86a13e020ff74156571782f4bc822b7c1f3
|
|
Change-Id: I064902aa9745342ccfe868331ccc000767e92fc0
|
|
|
|
|
|
|
|
Merge commit 'f393f015a3bd0709645b8f464ee41c82fc7624e2' into froyo-plus-aosp
* commit 'f393f015a3bd0709645b8f464ee41c82fc7624e2':
For issue #2651381, allow library prelinking.
|
|
Change-Id: Id386c3cf549c5856de1c0c0a91665b20e59672e0
|
|
Merge commit '49bb59032f09fd9724d4c17c6097337d59324904' into froyo-plus-aosp
* commit '49bb59032f09fd9724d4c17c6097337d59324904':
Update skia/emoji license notices for better accuracy
|
|
Change-Id: I8d2072f805d76c51824b65447def412d21ca99ca
|
|
Change-Id: I1e93e5d5d0f1c859819f3a84566647fd736e5fb5
|
|
Bug: 2638848
Change-Id: I4e028825bfb45b2a4c484899f3ec4cba3d96deb1
|
|
Without this this patch, an application could trigger a platform crash
by calling:
Typeface typeface = Typeface.createFromFile("");
boolean isBold = typeface.isBold();
The crash occurs since the find_name_and_style function does not
return an error code.
Change-Id: Id1d2a43b47d956c5a52f01e7d4b8cbb12631d59a
|
|
The drawPosTextOnPath method allows for text positioned on
a straight line to be drawn along a path.
This method can be used by layout engines to implement the
Canvas::drawTextOnPath method for complex scripts.
Change-Id: I2f9c22b8c97eeacb61cd6de7429ba875a1e7ade9
|
|
A missing color map for a GIF image could previously lead to a crash.
This commit adds a NULL check to prevent that from happening.
Change-Id: I5ecdfda979925a6de1eac6aae2bcae6a25b47f18
|
|
fix buffer overwrite in call to sk_memset16
Change-Id: I0400555d49e7f0d2237bf130ea01c9b8339a49a3
http://b/issue?id=2595706
|
|
Change-Id: I3b71be7e534d2313108f755df2b2fb474d240dad
|
|
Change-Id: I899361e402ed5865f680548c75bbb12f8de89504
|
|
Change-Id: Ib92980c1aab25e4dc037ea9119643dc04f3618f3
|
|
Change-Id: I050ea03bc67fd80660f8b6140c63a714a1d06e3f
|
|
This change makes the fast case for rects give the same result as drawPath (as it should)
fixes http://code.google.com/p/android/issues/detail?id=4086
Change-Id: Ia002be6700ce0894d7ecb303625d83c95b66ae03
|
|
Change-Id: I0c99b42cca2a9245b6d983f1d91c4d675f66fb14
|
|
Change-Id: I9ffdb50bde0c53070611c7dc626c384eca0906ea
|
|
This function will be used in WebKit (ucnv.cpp), which currently
embeds the string "docomo-emoji", which is not flexible.
Bug: 2327318
Change-Id: Iccbef2882ad931bde9b3ddefe34e81646e5b0686
|
|
fixes http://b/issue?id=2422924
Change-Id: Ia3abd981f7f5954204517880e69d44125dcaadb5
|
|
Change-Id: Ie6e18f698c2efc66110146af41afdb7cac5d8bd5
|
|
fixes http://b/issue?id=2136225
Change-Id: Ifa01a04406e1b53ffbd12a29a81445eeb60f0e60
|
|
consistent results between neon and residual loops"
|
|
disable neon dither proc, until it gives consistent results between neon and residual loops
Change-Id: I3db2230fb5c4aa8b9593abee6a8b9e25eaeef278
|
|
colorshader"
|
|
off-by-1 in alpha for colorshader
Change-Id: I6a6c75ac67725fad7cb17261236e1320cb5b1436
|
|
included implicitly via inter-library dependency. The library is also used
by the main executable.
Change-Id: Ib5562dbc481af6d95823c97e63355037e0049f7e
|
|
correctly treat src-alpha == 0 as a no-op."
|
|
src-alpha == 0 as a no-op.
update unittests to measure this
Change-Id: If4d61ac5f7ff3d7fc27cbc3f242dbdf7ff4e76be
http://b/issue?id=2333376
|
|
Change-Id: I1e49c30d483d678d393f99d0ce3af9a1ab0a9f53
|
|
detect that our compute t value in the chopper was imprecise, and clamp the segment
so we don't accidentally return a segment completely below the clip
Change-Id: I425ccc945ae7020834b066d2653c73ccd67a0989
http://b/issue?id=2034527
|
|
http://b/issue?id=2401896
Before the fix, the image referenced in the bug would take 30+ seconds to load on a large device,
and would crash a small device. On the small device, there was no RAM cap set, since the property
we queried for does not exist there. On the large device, the properties value was too small, so
I hard-code a larger one that seems to work better.
|
|
|
|
bitmap config
|
|
|
|
|
|
children
|
|
|