simplify GrPathRenderer interface

[bsalomon, 2012-02-28 22:42:08]

This is the direction I'm thinking for the GrPathRenderer API. I haven't had a
chance to clean up the CL, modify the comments, or do any testing yet.

This will probably be preceded by a change to remove the offscreen AA code from
GrContext.cpp

[guanqun, 2012-02-29 07:04:31]

Fixing the missing "!", then I tested this CL, it seems OK. Aside from that
typo, it's LGTM.

Another question, are we determined to abandon OSAA code?

http://codereview.appspot.com/5706053/diff/1016/src/gpu/GrAAHairLinePathRende...
File src/gpu/GrAAHairLinePathRenderer.cpp (right):

http://codereview.appspot.com/5706053/diff/1016/src/gpu/GrAAHairLinePathRende...
src/gpu/GrAAHairLinePathRenderer.cpp:546: if (target->reserveVertexSpace(layout,
vertCnt, (void**)&verts)) {
Missing a "!" before target. :)

http://codereview.appspot.com/5706053/diff/1016/src/gpu/GrTesselatedPathRende...
File src/gpu/GrTesselatedPathRenderer.h (right):

http://codereview.appspot.com/5706053/diff/1016/src/gpu/GrTesselatedPathRende...
src/gpu/GrTesselatedPathRenderer.h:24: virtual bool drawPath(const SkPath& path,
I know this class is not used right now. But we should override the onDrawPath()
function instead, as it's a pure virtual member function from base class.

[bsalomon, 2012-02-29 13:57:22]

On 2012/02/29 07:04:31, guanqun wrote:
> Fixing the missing "!", then I tested this CL, it seems OK. Aside from that
> typo, it's LGTM.

Heh, yeah I stopped on this once it compiled :)

> 
> Another question, are we determined to abandon OSAA code?

I think so. It isn't enabled in Chrome anymore. If we were to re-enable it I'd
rather capture it in a class and make callable from within path renderers (or
other draw calls) rather than forcing the path renderers to be stateful.

> 
>
http://codereview.appspot.com/5706053/diff/1016/src/gpu/GrAAHairLinePathRende...
> File src/gpu/GrAAHairLinePathRenderer.cpp (right):
> 
>
http://codereview.appspot.com/5706053/diff/1016/src/gpu/GrAAHairLinePathRende...
> src/gpu/GrAAHairLinePathRenderer.cpp:546: if
(target->reserveVertexSpace(layout,
> vertCnt, (void**)&verts)) {
> Missing a "!" before target. :)
> 
>
http://codereview.appspot.com/5706053/diff/1016/src/gpu/GrTesselatedPathRende...
> File src/gpu/GrTesselatedPathRenderer.h (right):
> 
>
http://codereview.appspot.com/5706053/diff/1016/src/gpu/GrTesselatedPathRende...
> src/gpu/GrTesselatedPathRenderer.h:24: virtual bool drawPath(const SkPath&
path,
> I know this class is not used right now. But we should override the
onDrawPath()
> function instead, as it's a pure virtual member function from base class.

Will do when I update this later today.

[guanqun.lu_gmail.com, 2012-02-29 15:12:20]

On Wed, Feb 29, 2012 at 9:57 PM,  <bsalomon@google.com> wrote:
> I think so. It isn't enabled in Chrome anymore. If we were to re-enable
> it I'd rather capture it in a class and make callable from within path
> renderers (or other draw calls) rather than forcing the path renderers
> to be stateful.

Thanks for your input!

[bsalomon, 2012-03-02 21:50:47]

On 2012/02/29 15:12:20, guanqun.lu_gmail.com wrote:
> On Wed, Feb 29, 2012 at 9:57 PM,  <mailto:bsalomon@google.com> wrote:
> > I think so. It isn't enabled in Chrome anymore. If we were to re-enable
> > it I'd rather capture it in a class and make callable from within path
> > renderers (or other draw calls) rather than forcing the path renderers
> > to be stateful.
> 
> Thanks for your input!

Landed at r3312.