Fix conics for gpu GrAAHairLinePathRender

Fix conics for gpu GrAAHairLinePathRender

This fixes one of the bugs discovered when rrects/ovals were changed to conics.
Specifcally the case where hairlines were being drawing randomly all of the place.
There is still a bug in a lot of our rrect/oval draws/clipping so the legacy flags
are still needed.

BUG=skia:3258
Committed: https://skia.googlesource.com/skia/+/14afb43b01e178b50674d16a5a62d2b7dbd600db

[egdaniel, 2014-12-22 16:30:43]

egdaniel@google.com changed reviewers:
+ bsalomon@google.com, reed@google.com

[reed1, 2014-12-22 16:32:13]

reed@google.com changed reviewers:
- reed@google.com

[reed1, 2014-12-22 16:32:14]

deferring to brian

[bsalomon, 2014-12-22 16:40:46]

https://codereview.chromium.org/822733003/diff/1/src/gpu/GrAAHairLinePathRend...
File src/gpu/GrAAHairLinePathRenderer.cpp (right):

https://codereview.chromium.org/822733003/diff/1/src/gpu/GrAAHairLinePathRend...
src/gpu/GrAAHairLinePathRenderer.cpp:40: // If a,b,c are the original control
points then the poly a0,a1,b0,c0,c1
really? a1 connects to b0? otherwise, lgtm

[egdaniel, 2014-12-22 16:52:17]

https://codereview.chromium.org/822733003/diff/1/src/gpu/GrAAHairLinePathRend...
File src/gpu/GrAAHairLinePathRenderer.cpp (right):

https://codereview.chromium.org/822733003/diff/1/src/gpu/GrAAHairLinePathRend...
src/gpu/GrAAHairLinePathRenderer.cpp:40: // If a,b,c are the original control
points then the poly a0,a1,b0,c0,c1
On 2014/12/22 16:40:46, bsalomon wrote:
> really? a1 connects to b0? otherwise, lgtm

yeah we draw the two "skinny" triangles on the side and then the "large" center
one (a1, b0, c1). Basically a triangle fan with b0 as the base.

[bsalomon, 2014-12-22 17:04:46]

https://codereview.chromium.org/822733003/diff/1/src/gpu/GrAAHairLinePathRend...
File src/gpu/GrAAHairLinePathRenderer.cpp (right):

https://codereview.chromium.org/822733003/diff/1/src/gpu/GrAAHairLinePathRend...
src/gpu/GrAAHairLinePathRenderer.cpp:40: // If a,b,c are the original control
points then the poly a0,a1,b0,c0,c1
On 2014/12/22 16:52:17, egdaniel wrote:
> On 2014/12/22 16:40:46, bsalomon wrote:
> > really? a1 connects to b0? otherwise, lgtm
> 
> yeah we draw the two "skinny" triangles on the side and then the "large"
center
> one (a1, b0, c1). Basically a triangle fan with b0 as the base.

ok, i think the comment is describing the bounding polygon before it is turned
into a tri-whatever.

[egdaniel, 2014-12-22 18:47:28]

The CQ bit was checked by egdaniel@google.com

[commit-bot: I haz the power, 2014-12-22 18:47:56]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/822733003/40001

[commit-bot: I haz the power, 2014-12-22 18:57:10]

Committed patchset #3 (id:40001) as
https://skia.googlesource.com/skia/+/14afb43b01e178b50674d16a5a62d2b7dbd600db