Don't use GrDrawTarget's AA state bit to convey whether to do custom-aa

[bsalomon, 2011-10-26 21:32:09]

I was doing some experimenting with using MSAA for the offscreen AA. Relying on
GrDrawTarget's state bit to convey that anti-aliasing should be applied to code
outside GrDrawTarget was causing problems.

With this change the state bit only tells GrDrawTarget what it should do when it
is asked to draw (as do all the other state bits).

I also took the opportunity to add SK_OVERRIDE to GrPathRenderer subclasses.

[Stephen White, 2011-10-26 22:24:47]

LGTM

http://codereview.appspot.com/5314061/diff/2001/src/gpu/GrContext.cpp
File src/gpu/GrContext.cpp (right):

http://codereview.appspot.com/5314061/diff/2001/src/gpu/GrContext.cpp#newcode...
src/gpu/GrContext.cpp:1169: // will be axis-aligned, the rect won't land on
integer coords.
Nit:  missing "and".

http://codereview.appspot.com/5314061/diff/2001/src/gpu/GrDrawTarget.h
File src/gpu/GrDrawTarget.h (right):

http://codereview.appspot.com/5314061/diff/2001/src/gpu/GrDrawTarget.h#newcod...
src/gpu/GrDrawTarget.h:126: kHWAntialias_StateBit   = 0x02,
Not sure I understand this change.  Is this simply a rename to better reflect
what the bit does, or did its meaning change as well?

http://codereview.appspot.com/5314061/diff/2001/src/gpu/GrPathRenderer.h
File src/gpu/GrPathRenderer.h (right):

http://codereview.appspot.com/5314061/diff/2001/src/gpu/GrPathRenderer.h#newc...
src/gpu/GrPathRenderer.h:67: virtual bool canDrawPath(const GrDrawTarget::Caps&
targetCaps,
Random comment:  I like this change, since it narrows the API a bit.

http://codereview.appspot.com/5314061/diff/2001/src/gpu/GrPathRenderer.h#newc...
src/gpu/GrPathRenderer.h:83: * @param target  target that the path be rendered
to
Nit:  what happened to "will"?  It be gone.  :)

[Stephen White, 2011-10-26 22:27:03]

On 2011/10/26 22:24:47, Stephen White wrote:
>
http://codereview.appspot.com/5314061/diff/2001/src/gpu/GrDrawTarget.h#newcod...
> src/gpu/GrDrawTarget.h:126: kHWAntialias_StateBit   = 0x02,
> Not sure I understand this change.  Is this simply a rename to better reflect
> what the bit does, or did its meaning change as well?

Never mind; I re-read the patch description and now it's clear.

[bsalomon, 2011-10-27 12:35:08]

Thanks for the review! I fixed the nits and landed as r2539.