add optional cull-rect to patheffects, so they can do less work if their results

add optional cull-rect to patheffects, so they can do less work if their results
lie outside of the current clip-bounds (the cull rect).
Committed: https://code.google.com/p/skia/source/detail?r=7378

[reed1, 2013-01-23 15:40:59]

haven't used the cullrect yet, just plumbing it in

[robertphillips, 2013-01-23 15:57:21]

LGTM

[bsalomon, 2013-01-23 22:21:15]

On 2013/01/23 22:11:08, reed1 wrote:

lgtm

[reed1, 2013-01-24 19:01:25]

refactored to perform the cull before we check for a (special) line.

before

running bench [640 480]         giantdashline_diag_2  8888: cmsecs =  12.75
running bench [640 480]         giantdashline_vert_2  8888: cmsecs =  14.24
running bench [640 480]         giantdashline_diag_0  8888: cmsecs =  12.31
running bench [640 480]         giantdashline_vert_0  8888: cmsecs =  12.25
running bench [640 480]         giantdashline_hori_0  8888: cmsecs =  12.16

after

running bench [640 480]         giantdashline_diag_2  8888: cmsecs =  10.55
running bench [640 480]         giantdashline_vert_2  8888: cmsecs =  11.72
running bench [640 480]         giantdashline_hori_2  8888: cmsecs =   0.02
running bench [640 480]         giantdashline_diag_0  8888: cmsecs =  10.03
running bench [640 480]         giantdashline_vert_0  8888: cmsecs =   9.98
running bench [640 480]         giantdashline_hori_0  8888: cmsecs =   0.08

[reed1, 2013-01-24 19:54:03]

On 2013/01/24 19:01:25, reed1 wrote:
> refactored to perform the cull before we check for a (special) line.
> 
> before
> 
> running bench [640 480]         giantdashline_diag_2  8888: cmsecs =  12.75
> running bench [640 480]         giantdashline_vert_2  8888: cmsecs =  14.24
> running bench [640 480]         giantdashline_diag_0  8888: cmsecs =  12.31
> running bench [640 480]         giantdashline_vert_0  8888: cmsecs =  12.25
> running bench [640 480]         giantdashline_hori_0  8888: cmsecs =  12.16
> 
> after
> 
> running bench [640 480]         giantdashline_diag_2  8888: cmsecs =  10.55
> running bench [640 480]         giantdashline_vert_2  8888: cmsecs =  11.72
> running bench [640 480]         giantdashline_hori_2  8888: cmsecs =   0.02
> running bench [640 480]         giantdashline_diag_0  8888: cmsecs =  10.03
> running bench [640 480]         giantdashline_vert_0  8888: cmsecs =   9.98
> running bench [640 480]         giantdashline_hori_0  8888: cmsecs =   0.08

oops, out-of-date version of the test. Here are clean old and new

running bench [640 480]         giantdashline_diag_2  8888: cmsecs =  13.21
running bench [640 480]         giantdashline_vert_2  8888: cmsecs =  14.61
running bench [640 480]         giantdashline_diag_0  8888: cmsecs =  12.82
running bench [640 480]         giantdashline_vert_0  8888: cmsecs =  12.92
running bench [640 480]         giantdashline_hori_0  8888: cmsecs =  12.70

running bench [640 480]         giantdashline_diag_2  8888: cmsecs =  13.22
running bench [640 480]         giantdashline_vert_2  8888: cmsecs =  14.57
running bench [640 480]         giantdashline_hori_2  8888: cmsecs =   0.03
running bench [640 480]         giantdashline_diag_0  8888: cmsecs =  12.99
running bench [640 480]         giantdashline_vert_0  8888: cmsecs =  12.97
running bench [640 480]         giantdashline_hori_0  8888: cmsecs =   0.09

[robertphillips, 2013-01-24 20:03:54]

LGTM + nits/suggestions

https://codereview.appspot.com/7206044/diff/3002/src/effects/SkDashPathEffect...
File src/effects/SkDashPathEffect.cpp (right):

https://codereview.appspot.com/7206044/diff/3002/src/effects/SkDashPathEffect...
src/effects/SkDashPathEffect.cpp:90: 
outset_for_stroke?

https://codereview.appspot.com/7206044/diff/3002/src/effects/SkDashPathEffect...
src/effects/SkDashPathEffect.cpp:137: 
// remove excess line outside of the cull rect in a manner that will preserve
the phase behavior?

[reed1, 2013-01-24 20:41:11]

https://codereview.appspot.com/7206044/diff/3002/src/effects/SkDashPathEffect...
File src/effects/SkDashPathEffect.cpp (right):

https://codereview.appspot.com/7206044/diff/3002/src/effects/SkDashPathEffect...
src/effects/SkDashPathEffect.cpp:90: 
On 2013/01/24 20:03:54, robertphillips wrote:
> outset_for_stroke?

Done.

https://codereview.appspot.com/7206044/diff/3002/src/effects/SkDashPathEffect...
src/effects/SkDashPathEffect.cpp:137: 
On 2013/01/24 20:03:54, robertphillips wrote:
> // remove excess line outside of the cull rect in a manner that will preserve
> the phase behavior?

Done.