Minimal changes to not crash GPU with color fonts.

Minimal changes to not crash GPU with color fonts.

Committed: https://code.google.com/p/skia/source/detail?r=12400

[bungeman-skia, 2013-11-26 19:05:03]

Some minimal starting changes toward sane color font support.

[jvanverth1, 2013-11-26 19:55:34]

On 2013/11/26 19:05:03, bungeman1 wrote:
> Some minimal starting changes toward sane color font support.

This appears to produce reasonable results on the GPU ("reasonable" meaning we
need a new shader for this case, but otherwise it's quite colorful).

[reed1, 2013-11-26 19:58:02]

https://codereview.chromium.org/88993003/diff/1/src/gpu/SkGrFontScaler.cpp
File src/gpu/SkGrFontScaler.cpp (right):

https://codereview.chromium.org/88993003/diff/1/src/gpu/SkGrFontScaler.cpp#ne...
src/gpu/SkGrFontScaler.cpp:88: case SkMask::kARGB32_Format:
This may have us not crash, but its not logically correct, yes? A888 is 3 alpha
channels, not 1 (I think).

[bungeman-skia, 2013-11-26 20:43:30]

Committed patchset #2 manually as r12400.

[bungeman-skia, 2013-11-26 20:43:42]

https://codereview.chromium.org/88993003/diff/1/src/gpu/SkGrFontScaler.cpp
File src/gpu/SkGrFontScaler.cpp (right):

https://codereview.chromium.org/88993003/diff/1/src/gpu/SkGrFontScaler.cpp#ne...
src/gpu/SkGrFontScaler.cpp:88: case SkMask::kARGB32_Format:
On 2013/11/26 19:58:03, reed1 wrote:
> This may have us not crash, but its not logically correct, yes? A888 is 3
alpha
> channels, not 1 (I think).

Done. Added comment that this is temporary. These two are the same size, so it
works out. The gpu will draw the colors wrong, but this is already less wrong
than it is without this change.

[reed1, 2013-11-26 21:00:18]

https://code.google.com/p/skia/issues/detail?id=1869

lgtm