Add flipped gradient branch to two point conical gradient

Add flipped gradient branch to two point conical gradient

BUG=skia:

Committed: http://code.google.com/p/skia/source/detail?r=14235

Committed: http://code.google.com/p/skia/source/detail?r=14271

[egdaniel, 2014-04-16 17:10:14]

This is a pre-step to my main gradients change. This patch orders the gradient
so that the smallest radius for conical gradients is passed in as the start
circle. So if the input is the opposite we have to flip everything and keep
track if we did a flip so that we render everything correctly. This patch still
uses most of the old render paths on cpu/gpu plus a check to see if it was
flipped.

Mike K- Can you check that all the picture versioning stuff seems to be correct?

[mtklein, 2014-04-16 17:23:34]

https://codereview.chromium.org/227623004/diff/40001/src/effects/gradients/Sk...
File src/effects/gradients/SkTwoPointConicalGradient.cpp (right):

https://codereview.chromium.org/227623004/diff/40001/src/effects/gradients/Sk...
src/effects/gradients/SkTwoPointConicalGradient.cpp:357: if
(buffer.pictureVersion() > kNewTwoPointConicalVersion || 0 ==
buffer.pictureVersion()) {
I'd personally just write the literal 24 until you find need for it in two
places.

https://codereview.chromium.org/227623004/diff/40001/src/effects/gradients/Sk...
src/effects/gradients/SkTwoPointConicalGradient.cpp:360: // Sort gradient by
radius size for old pictures
Mark this with // V23_COMPATIBILITY_CODE so we can grep for it later?

[egdaniel, 2014-04-16 17:53:58]

small bug fix caused by change to SkTwoPointConicalGradient earlier today, plus
mikes suggestions

[mtklein, 2014-04-16 17:54:59]

On 2014/04/16 17:53:58, egdaniel wrote:
> small bug fix caused by change to SkTwoPointConicalGradient earlier today,
plus
> mikes suggestions

Picture version code lgtm

[bsalomon, 2014-04-16 19:12:55]

lgtm

https://codereview.chromium.org/227623004/diff/60001/src/effects/gradients/Sk...
File src/effects/gradients/SkGradientShader.cpp (right):

https://codereview.chromium.org/227623004/diff/60001/src/effects/gradients/Sk...
src/effects/gradients/SkGradientShader.cpp:822: SkColor colorsNew[colorCount];
woah.. this looks like a compiler extension. colorCount isn't a compile time
constant.

[bsalomon, 2014-04-16 19:14:08]

On 2014/04/16 19:12:55, bsalomon wrote:
> lgtm
> 
>
https://codereview.chromium.org/227623004/diff/60001/src/effects/gradients/Sk...
> File src/effects/gradients/SkGradientShader.cpp (right):
> 
>
https://codereview.chromium.org/227623004/diff/60001/src/effects/gradients/Sk...
> src/effects/gradients/SkGradientShader.cpp:822: SkColor colorsNew[colorCount];
> woah.. this looks like a compiler extension. colorCount isn't a compile time
> constant.

I didn't mean to write lgtm.. the colorCount thing needs to be fixed.

[mtklein, 2014-04-16 19:20:50]

https://codereview.chromium.org/227623004/diff/60001/src/effects/gradients/Sk...
File src/effects/gradients/SkGradientShader.cpp (right):

https://codereview.chromium.org/227623004/diff/60001/src/effects/gradients/Sk...
src/effects/gradients/SkGradientShader.cpp:822: SkColor colorsNew[colorCount];
On 2014/04/16 19:12:56, bsalomon wrote:
> woah.. this looks like a compiler extension. colorCount isn't a compile time
> constant.

Yeah, this will be more or less supported depending on the compiler.  It's a C99
feature, and may make it into C++17.  I actually like it, but we can't write it.
 If nothing else, it's safer than alloca...

[bsalomon, 2014-04-16 21:14:07]

lgtm

[egdaniel, 2014-04-17 13:12:13]

The CQ bit was checked by egdaniel@google.com

[commit-bot: I haz the power, 2014-04-17 13:12:30]

CQ is trying da patch. Follow status at
https://skia-tree-status.appspot.com/cq/egdaniel@google.com/227623004/80001

[commit-bot: I haz the power, 2014-04-17 13:28:09]

Change committed as 14235

[egdaniel, 2014-04-17 13:56:54]

A revert of this CL has been created in
https://codereview.chromium.org/239993006/ by egdaniel@google.com.

The reason for reverting is: Breaking all CompareGMs.

[egdaniel, 2014-04-18 14:16:11]

Main differences in new version is the for TwoPointConicalGradients we have its
asAGradient function return an unsorted version of the gradient (thus it
supports pdf and any other users of it).

Depends on https://codereview.chromium.org/241173005/ landing which removes the
asAGradient dependency on the gpu side.

Also is the ignore file set up correctly?

[bsalomon, 2014-04-18 14:31:44]

lgtm

https://codereview.chromium.org/227623004/diff/100001/src/effects/gradients/S...
File src/effects/gradients/SkGradientShaderPriv.h (right):

https://codereview.chromium.org/227623004/diff/100001/src/effects/gradients/S...
src/effects/gradients/SkGradientShaderPriv.h:171: // Flip the Takes in Color
info for src which contains count colors at the start
Flip the Takes?

[egdaniel, 2014-04-21 12:55:17]

The CQ bit was checked by egdaniel@google.com

[commit-bot: I haz the power, 2014-04-21 12:55:20]

CQ is trying da patch. Follow status at
https://skia-tree-status.appspot.com/cq/egdaniel@google.com/227623004/210001

[commit-bot: I haz the power, 2014-04-21 13:10:28]

Change committed as 14271