Left over cleanup from r4416 (fix for lingering AA clip mask bug)

This CL consolidates duplicate code in the GrClipMaskManager and GrSoftwarePathRender into a helper method in GrSWMaskHelper. Left over from: http://codereview.appspot.com/6351055/

[bsalomon, 2012-07-03 18:02:50]

http://codereview.appspot.com/6356058/diff/1/src/gpu/GrSWMaskHelper.h
File src/gpu/GrSWMaskHelper.h (right):

http://codereview.appspot.com/6356058/diff/1/src/gpu/GrSWMaskHelper.h#newcode83
src/gpu/GrSWMaskHelper.h:83: static void DrawTextureOnRect(GrAutoScratchTexture*
ast,
Let's move your comment from the impl file to the header.

Would it make sense to combine the two helpers into one? The AST param to the
new guy feels a bit weird since the function detaches it. Up till now we only
use AST* as an out param (and in general we want to move away from using AST).
We also seem to be relying on the caller to not have done anything between the
calls that could have led to the cache recycling the texture.

If the above idea is wrong-headed then let's make it really obvious what each
does and how they are to be used together. Maybe rename:
DrawPathMaskToTexture()
DrawMaskTextureToTarget()
and mention in the comment that the second is intended to consume the texture
generated by the first.

You don't reset the draw state. I suppose that is on purpose to let the caller
specify view matrix, blend, etc. But we are making some assumptions coming into
this function. One is that stage GrPaint::kTotalStages is available and that the
draw state's "first coverage stage" setting is <= GrPaint::kTotalStages. These
should go away as the shader refactor comes along (with separate appendable
lists of stages for color and coverage). However, we should document the
assumptions here. Or perhaps the stage to use should be a param? Or is the state
that is interesting small enough that it could be passed in to the function and
the function resets the draw state and therefore doesn't need to make these
assumptions?

[robertphillips, 2012-07-09 13:57:28]

I believe this addresses all the earlier issues.

http://codereview.appspot.com/6356058/diff/1/src/gpu/GrSWMaskHelper.h
File src/gpu/GrSWMaskHelper.h (right):

http://codereview.appspot.com/6356058/diff/1/src/gpu/GrSWMaskHelper.h#newcode83
src/gpu/GrSWMaskHelper.h:83: static void DrawTextureOnRect(GrAutoScratchTexture*
ast,
Comment moved.

The SW path offsets the path between the two functions so I would like to leave
them alone - rather than add a flag. I have modified the "ast" usage to be
closer to how this will all ultimately work - so it should be more palatable.

I have renamed the two functions to make their behavior clearer.

I have documented the assumptions. In the straight-ahead SW path rendering case
the draw state can be arbitrary.

http://codereview.appspot.com/6356058/diff/7/src/gpu/GrClipMaskManager.cpp
File src/gpu/GrClipMaskManager.cpp (right):

http://codereview.appspot.com/6356058/diff/7/src/gpu/GrClipMaskManager.cpp#ne...
src/gpu/GrClipMaskManager.cpp:373: 
In this case the ClipMaskManager has set up the draw state with the correct
offset and blend mode to effect the boolean operations.

[bsalomon, 2012-07-09 14:28:28]

LGTM

http://codereview.appspot.com/6356058/diff/7/src/gpu/GrSWMaskHelper.cpp
File src/gpu/GrSWMaskHelper.cpp (right):

http://codereview.appspot.com/6356058/diff/7/src/gpu/GrSWMaskHelper.cpp#newco...
src/gpu/GrSWMaskHelper.cpp:212: GrAssert(NULL ==
drawState->getTexture(kPathMaskStage));
I think this should be isStageEnabled() to be forward-compatible with what Tom
is doing.

[TomH, 2012-07-09 14:53:14]

http://codereview.appspot.com/6356058/diff/7/src/gpu/GrSWMaskHelper.cpp
File src/gpu/GrSWMaskHelper.cpp (right):

http://codereview.appspot.com/6356058/diff/7/src/gpu/GrSWMaskHelper.cpp#newco...
src/gpu/GrSWMaskHelper.cpp:212: GrAssert(NULL ==
drawState->getTexture(kPathMaskStage));
Probably so!
Well, !isStageEnabled()

[robertphillips, 2012-07-09 16:33:12]

committed as r4474

http://codereview.appspot.com/6356058/diff/7/src/gpu/GrSWMaskHelper.cpp
File src/gpu/GrSWMaskHelper.cpp (right):

http://codereview.appspot.com/6356058/diff/7/src/gpu/GrSWMaskHelper.cpp#newco...
src/gpu/GrSWMaskHelper.cpp:212: GrAssert(NULL ==
drawState->getTexture(kPathMaskStage));
On 2012/07/09 14:53:14, TomH wrote:
> Probably so!
> Well, !isStageEnabled()

Done.