Replace edge types with GrEdgeEffect.

Replace edge types with GrEdgeEffect. 

This strips out last of the edge types and the fixed function edge attribute and replaces them with using GrEdgeEffect. Also fixes a minor bug when checking attribute counts -- it was using kAttribIndexCount instead of kVertexAttribCnt. 

Original Author: jvanverth@google.com
Committed: https://code.google.com/p/skia/source/detail?r=8392

[bsalomon, 2013-03-25 19:50:49]

This is mostly the same as Jim's change. I just check the caps in TestCreate(),
assert on caps in emitCode(), and updated some of the comments that were copied
from GrDrawState.

[bsalomon, 2013-03-25 19:54:14]

I have some cleanup I'd like to do around fDiscardIfOutsideEdge and hiding
GrGLShaderBuilder::fHeader but am saving for followup CLs.

[robertphillips, 2013-03-25 20:05:40]

lgtm

https://codereview.chromium.org/13069003/diff/1/src/gpu/effects/GrEdgeEffect.cpp
File src/gpu/effects/GrEdgeEffect.cpp (right):

https://codereview.chromium.org/13069003/diff/1/src/gpu/effects/GrEdgeEffect....
src/gpu/effects/GrEdgeEffect.cpp:12: 
I agree with Brian that this should really be broken up into 3 effects.

https://codereview.chromium.org/13069003/diff/1/src/gpu/effects/GrEdgeEffect.h
File src/gpu/effects/GrEdgeEffect.h (right):

https://codereview.chromium.org/13069003/diff/1/src/gpu/effects/GrEdgeEffect....
src/gpu/effects/GrEdgeEffect.h:22: enum EdgeType {
device ...

https://codereview.chromium.org/13069003/diff/1/src/gpu/effects/GrEdgeEffect....
src/gpu/effects/GrEdgeEffect.h:43: static GrEffectRef* Create(EdgeType type) {
is "(stroked/filled)" comment relevant?

[bsalomon, 2013-03-26 13:24:02]

Replace edge types with GrEdgeEffect. 

This strips out last of the edge types and the fixed function edge attribute and
replaces them with using GrEdgeEffect. Also fixes a minor bug when checking
attribute counts -- it was using kAttribIndexCount instead of kVertexAttribCnt. 

Original Author: jvanvert@google.com

[bsalomon, 2013-03-26 13:41:50]

Sorry for the several additional patch uploads after moving this to another
machine. Only patchsets 1 and 4 are of interest. As noted below, I had to change
the static init of the effects.

(Also, I've corrected Jim's email address in the SVN log locally from the way it
appears in the gcl-generated message)

https://codereview.chromium.org/13069003/diff/1/src/gpu/effects/GrEdgeEffect.h
File src/gpu/effects/GrEdgeEffect.h (right):

https://codereview.chromium.org/13069003/diff/1/src/gpu/effects/GrEdgeEffect....
src/gpu/effects/GrEdgeEffect.h:22: enum EdgeType {
On 2013/03/25 20:05:40, robertphillips wrote:
> device ...

Done.

https://codereview.chromium.org/13069003/diff/1/src/gpu/effects/GrEdgeEffect....
src/gpu/effects/GrEdgeEffect.h:43: static GrEffectRef* Create(EdgeType type) {
On 2013/03/25 20:05:40, robertphillips wrote:
> is "(stroked/filled)" comment relevant?

Done.

https://codereview.chromium.org/13069003/diff/13002/src/gpu/effects/GrEdgeEff...
File src/gpu/effects/GrEdgeEffect.h (right):

https://codereview.chromium.org/13069003/diff/13002/src/gpu/effects/GrEdgeEff...
src/gpu/effects/GrEdgeEffect.h:45: static GrEffectRef*
gEdgeEffectRef[kEdgeTypeCount] = {
XCode/clang did not like the SkAutoTUnref<GrEffectRef> constructor list. It
gives an error about SkNoncopyable having a private copy cons. It's as if it was
going to construct temp SkAutoTUnrefs and then copy cons them into the array.
That's definitely not what we want because the temp's destructor will unref.

I had checked on windows that the ref cnt was correct after the static
initialization.

This is uglier but works with Xcode.

[robertphillips, 2013-03-26 14:09:20]

LGTM + nits

https://codereview.chromium.org/13069003/diff/13002/src/gpu/effects/GrEdgeEff...
File src/gpu/effects/GrEdgeEffect.cpp (right):

https://codereview.chromium.org/13069003/diff/13002/src/gpu/effects/GrEdgeEff...
src/gpu/effects/GrEdgeEffect.cpp:107: } else {
convert the "someday" comment into a "TODO"?

https://codereview.chromium.org/13069003/diff/13002/src/gpu/effects/GrEdgeEff...
File src/gpu/effects/GrEdgeEffect.h (right):

https://codereview.chromium.org/13069003/diff/13002/src/gpu/effects/GrEdgeEff...
src/gpu/effects/GrEdgeEffect.h:67: 
We don't usually bother with inlines on these

[bsalomon, 2013-03-26 14:45:30]

https://codereview.chromium.org/13069003/diff/13002/src/gpu/effects/GrEdgeEff...
File src/gpu/effects/GrEdgeEffect.cpp (right):

https://codereview.chromium.org/13069003/diff/13002/src/gpu/effects/GrEdgeEff...
src/gpu/effects/GrEdgeEffect.cpp:107: } else {
On 2013/03/26 14:09:20, robertphillips wrote:
> convert the "someday" comment into a "TODO"?

Done.

https://codereview.chromium.org/13069003/diff/13002/src/gpu/effects/GrEdgeEff...
File src/gpu/effects/GrEdgeEffect.h (right):

https://codereview.chromium.org/13069003/diff/13002/src/gpu/effects/GrEdgeEff...
src/gpu/effects/GrEdgeEffect.h:67: 
On 2013/03/26 14:09:20, robertphillips wrote:
> We don't usually bother with inlines on these

Done.

[bsalomon, 2013-03-26 14:45:41]

Committed manually as r8392 (presubmit successful).