add unittest for SkTreatAsSprite

add unittest for SkTreatAsSprite
Committed: https://code.google.com/p/skia/source/detail?r=6974

[robertphillips, 2013-01-02 18:33:07]

LGTM + nits/suggestions

https://codereview.appspot.com/7042044/diff/1/tests/DrawBitmapRectTest.cpp
File tests/DrawBitmapRectTest.cpp (right):

https://codereview.appspot.com/7042044/diff/1/tests/DrawBitmapRectTest.cpp#ne...
tests/DrawBitmapRectTest.cpp:20: // quick success check
SkMatrix::kTranslate_Mask == mat.getType ?

https://codereview.appspot.com/7042044/diff/1/tests/DrawBitmapRectTest.cpp#ne...
tests/DrawBitmapRectTest.cpp:41: // up by that amount.
kShift?

https://codereview.appspot.com/7042044/diff/1/tests/DrawBitmapRectTest.cpp#ne...
tests/DrawBitmapRectTest.cpp:47: 
kScale?

https://codereview.appspot.com/7042044/diff/1/tests/DrawBitmapRectTest.cpp#ne...
tests/DrawBitmapRectTest.cpp:74: 
I'm sure some compiler will be unhappy with just the "unsigned"

https://codereview.appspot.com/7042044/diff/1/tests/DrawBitmapRectTest.cpp#ne...
tests/DrawBitmapRectTest.cpp:74: 
SkRandom* - since it isn't const?

https://codereview.appspot.com/7042044/diff/1/tests/DrawBitmapRectTest.cpp#ne...
tests/DrawBitmapRectTest.cpp:91: 
SkRandom*?

https://codereview.appspot.com/7042044/diff/1/tests/DrawBitmapRectTest.cpp#ne...
tests/DrawBitmapRectTest.cpp:123: 
kTooMuchSubpixel?

[reed1, 2013-01-02 18:55:29]

uploaded version where bits is a parameter.

https://codereview.appspot.com/7042044/diff/1/tests/DrawBitmapRectTest.cpp
File tests/DrawBitmapRectTest.cpp (right):

https://codereview.appspot.com/7042044/diff/1/tests/DrawBitmapRectTest.cpp#ne...
tests/DrawBitmapRectTest.cpp:20: // quick success check
On 2013/01/02 18:33:07, robertphillips wrote:
> SkMatrix::kTranslate_Mask == mat.getType ?

or identity...

https://codereview.appspot.com/7042044/diff/1/tests/DrawBitmapRectTest.cpp#ne...
tests/DrawBitmapRectTest.cpp:41: // up by that amount.
On 2013/01/02 18:33:07, robertphillips wrote:
> kShift?

made this a parameter

https://codereview.appspot.com/7042044/diff/1/tests/DrawBitmapRectTest.cpp#ne...
tests/DrawBitmapRectTest.cpp:47: 
On 2013/01/02 18:33:07, robertphillips wrote:
> kScale?

no longer a compile-time value. see above.

https://codereview.appspot.com/7042044/diff/1/tests/DrawBitmapRectTest.cpp#ne...
tests/DrawBitmapRectTest.cpp:74: 
On 2013/01/02 18:33:07, robertphillips wrote:
> I'm sure some compiler will be unhappy with just the "unsigned"

we have 100s of examples of this use.

https://codereview.appspot.com/7042044/diff/1/tests/DrawBitmapRectTest.cpp#ne...
tests/DrawBitmapRectTest.cpp:74: 
On 2013/01/02 18:33:07, robertphillips wrote:
> SkRandom* - since it isn't const?

True, though we have about equal numbers of both patterns. I justify it to
myself by thinking of rand's seed as a mutable field.... gulp.

https://codereview.appspot.com/7042044/diff/1/tests/DrawBitmapRectTest.cpp#ne...
tests/DrawBitmapRectTest.cpp:123: 
On 2013/01/02 18:33:07, robertphillips wrote:
> kTooMuchSubpixel?

Hmmm, not sure that makes the call sites any clearer. I will think more on this.
Perhaps a comment (or calculation) that justifies the .1 and .02 would make this
nit clearer.

[bsalomon, 2013-01-02 18:59:32]

LGTM. (I kinda like using the k prefix for all kompile-time konstants).