use forward matrix to determine if we can ignore scale part of a matrix

use forward matrix to determine if we can ignore scale part of a matrix
Committed: https://code.google.com/p/skia/source/detail?r=8700

[reed1, 2013-04-15 19:21:51]

unittest on the way...

[robertphillips, 2013-04-15 19:43:02]

This is certainly clearer.

https://codereview.chromium.org/14018020/diff/1/src/core/SkBitmapProcState.cpp
File src/core/SkBitmapProcState.cpp (right):

https://codereview.chromium.org/14018020/diff/1/src/core/SkBitmapProcState.cp...
src/core/SkBitmapProcState.cpp:39: 
assert this now?

https://codereview.chromium.org/14018020/diff/1/src/core/SkBitmapProcState.cp...
src/core/SkBitmapProcState.cpp:67: 
Assert this now?

https://codereview.chromium.org/14018020/diff/1/src/core/SkBitmapProcState.cp...
src/core/SkBitmapProcState.cpp:93: static bool matrix_only_scale_translate(const
SkMatrix& m) {
This is done as "if (mask <= (SkMatrix::kTranslate_Mask |
SkMatrix::kScale_Mask))" elsewhere.

[reed1, 2013-04-15 20:00:26]

https://codereview.chromium.org/14018020/diff/1/src/core/SkBitmapProcState.cpp
File src/core/SkBitmapProcState.cpp (right):

https://codereview.chromium.org/14018020/diff/1/src/core/SkBitmapProcState.cp...
src/core/SkBitmapProcState.cpp:39: 
On 2013/04/15 19:43:02, robertphillips wrote:
> assert this now?

Done.

https://codereview.chromium.org/14018020/diff/1/src/core/SkBitmapProcState.cp...
src/core/SkBitmapProcState.cpp:93: static bool matrix_only_scale_translate(const
SkMatrix& m) {
On 2013/04/15 19:43:02, robertphillips wrote:
> This is done as "if (mask <= (SkMatrix::kTranslate_Mask |
> SkMatrix::kScale_Mask))" elsewhere.

Done.

[reed1, 2013-04-15 21:00:02]

ptal

[robertphillips, 2013-04-15 21:42:22]

lgtm. unit test?

[reed1, 2013-04-15 21:43:59]

yea... a unit test seems harder than I thought. I can certainly create a GM that
exercises this particular case, but testing when the fast case is correctly
chosen is harder.. suggestions?

[robertphillips, 2013-04-15 21:48:01]

I think a GM would also serve nicely. If this test ever becomes a predicate on
matrix(?) (e.g., is1To1/isFastPath) we could perhaps test that in the future.

[reed1, 2013-04-16 17:14:48]

Committed patchset #3 manually as r8700.

[robertphillips, 2013-04-17 21:32:26]

Reverted in r8731.