simplify logic in SkPerliNoiseShader for GPU when constant color

simplify logic in SkPerliNoiseShader for GPU when constant color

BUG=skia:3488
Committed: https://skia.googlesource.com/skia/+/ea8b55db889b19786008b7c613d238e7202485e5

[bsalomon, 2015-03-04 04:57:22]

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

[bsalomon, 2015-03-04 04:57:23]

Realized how simple this is after coding this:

https://codereview.chromium.org/978713002/

:-/

Maybe there is still a use for that.

https://codereview.chromium.org/974183002/diff/20001/src/effects/SkPerlinNois...
File src/effects/SkPerlinNoiseShader.cpp (right):

https://codereview.chromium.org/974183002/diff/20001/src/effects/SkPerlinNois...
src/effects/SkPerlinNoiseShader.cpp:969: uint32_t rgb = alpha >> 1;
Can I get away with this?

[bsalomon, 2015-03-04 14:55:50]

removed GM change.

[egdaniel, 2015-03-04 14:58:00]

lgtm. IMO the shift (aka mul by 128) is just as valid as the mul by 127.

[sugoi1, 2015-03-04 15:53:53]

sugoi@chromium.org changed reviewers:
+ sugoi@chromium.org

[sugoi1, 2015-03-04 15:53:53]

LGTM with the bit shift.

https://codereview.chromium.org/974183002/diff/20001/src/effects/SkPerlinNois...
File src/effects/SkPerlinNoiseShader.cpp (right):

https://codereview.chromium.org/974183002/diff/20001/src/effects/SkPerlinNois...
src/effects/SkPerlinNoiseShader.cpp:969: uint32_t rgb = alpha >> 1;
On 2015/03/04 04:57:23, bsalomon wrote:
> Can I get away with this?

I don't see why not, this should be fine.

[sugoi1, 2015-03-04 16:02:16]

On 2015/03/04 15:53:53, sugoi1 wrote:
> LGTM with the bit shift.
> 
>
https://codereview.chromium.org/974183002/diff/20001/src/effects/SkPerlinNois...
> File src/effects/SkPerlinNoiseShader.cpp (right):
> 
>
https://codereview.chromium.org/974183002/diff/20001/src/effects/SkPerlinNois...
> src/effects/SkPerlinNoiseShader.cpp:969: uint32_t rgb = alpha >> 1;
> On 2015/03/04 04:57:23, bsalomon wrote:
> > Can I get away with this?
> 
> I don't see why not, this should be fine.

Asking for a noise with 0 octaves has to be an extremely rare case anyway, I
doubt many people would do this, so even it this would cause a 1 bit change, I
doubt anyone would go crazy over it.

[reed1, 2015-03-04 17:15:03]

this CL still has SkColorFilter::CreateModeFilter. Is this fixing a different
aspect of the bug?

[sugoi1, 2015-03-04 17:43:59]

On 2015/03/04 17:15:03, reed1 wrote:
> this CL still has SkColorFilter::CreateModeFilter. Is this fixing a different
aspect of the bug?

Where is it? There was only one in this file AFAIK, and it's being removed,
isn't it?

[reed1, 2015-03-04 18:20:15]

heheheh, oops, I completely misread the diff. doh!

[bsalomon, 2015-03-04 18:57:10]

The CQ bit was checked by bsalomon@google.com

[bsalomon, 2015-03-04 18:57:11]

The patchset sent to the CQ was uploaded after l-g-t-m from sugoi@chromium.org,
egdaniel@google.com
Link to the patchset: https://codereview.chromium.org/974183002/#ps60001 (title:
"use >>")

[commit-bot: I haz the power, 2015-03-04 18:58:06]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/974183002/60001

[commit-bot: I haz the power, 2015-03-04 19:03:55]

Committed patchset #4 (id:60001) as
https://skia.googlesource.com/skia/+/ea8b55db889b19786008b7c613d238e7202485e5