1. remove references to (deprecated) SkGpuCanvas

1. remove references to (deprecated) SkGpuCanvas
2. remove references to setDevice (soon to be deprecated)
Committed: https://code.google.com/p/skia/source/detail?r=5751

[robertphillips, 2012-10-01 16:41:40]

LGTM + 2 cleanup possibilities

https://codereview.appspot.com/6597055/diff/1/gm/gmmain.cpp
File gm/gmmain.cpp (right):

https://codereview.appspot.com/6597055/diff/1/gm/gmmain.cpp#newcode325
gm/gmmain.cpp:325: }
Could we rename gc to canvas?

https://codereview.appspot.com/6597055/diff/1/gm/gmmain.cpp#newcode332
gm/gmmain.cpp:332: }
gcUnref -> canvasUnref?

[reed1, 2012-10-01 17:01:13]

ptal -- more files

[bsalomon, 2012-10-01 17:06:53]

On 2012/10/01 17:01:13, reed1 wrote:
> ptal -- more files

lgtm

[robertphillips, 2012-10-01 17:10:59]

LGTM + suggestions

https://codereview.appspot.com/6597055/diff/3001/gm/gmmain.cpp
File gm/gmmain.cpp (right):

https://codereview.appspot.com/6597055/diff/3001/gm/gmmain.cpp#newcode309
gm/gmmain.cpp:309: if (gRec.fBackend == kRaster_Backend) {
In that case we should use SkAutoTUnref here too.

https://codereview.appspot.com/6597055/diff/3001/gm/ninepatchstretch.cpp
File gm/ninepatchstretch.cpp (right):

https://codereview.appspot.com/6597055/diff/3001/gm/ninepatchstretch.cpp#newc...
gm/ninepatchstretch.cpp:17: SkDevice* dev;
For uniformity's sake we may want to just make dev an SkAutoTUnref to

https://codereview.appspot.com/6597055/diff/3001/tests/PremulAlphaRoundTripTe...
File tests/PremulAlphaRoundTripTest.cpp (right):

https://codereview.appspot.com/6597055/diff/3001/tests/PremulAlphaRoundTripTe...
tests/PremulAlphaRoundTripTest.cpp:50: GrContext* context) {
mv device inside loop?

https://codereview.appspot.com/6597055/diff/3001/tests/WritePixelsTest.cpp
File tests/WritePixelsTest.cpp (right):

https://codereview.appspot.com/6597055/diff/3001/tests/WritePixelsTest.cpp#ne...
tests/WritePixelsTest.cpp:303: 
Maybe change name to createDevice?

[reed1, 2012-10-01 17:48:38]

https://codereview.appspot.com/6597055/diff/3001/gm/gmmain.cpp
File gm/gmmain.cpp (right):

https://codereview.appspot.com/6597055/diff/3001/gm/gmmain.cpp#newcode309
gm/gmmain.cpp:309: if (gRec.fBackend == kRaster_Backend) {
On 2012/10/01 17:10:59, robertphillips wrote:
> In that case we should use SkAutoTUnref here too.

Done.

https://codereview.appspot.com/6597055/diff/3001/gm/ninepatchstretch.cpp
File gm/ninepatchstretch.cpp (right):

https://codereview.appspot.com/6597055/diff/3001/gm/ninepatchstretch.cpp#newc...
gm/ninepatchstretch.cpp:17: SkDevice* dev;
On 2012/10/01 17:10:59, robertphillips wrote:
> For uniformity's sake we may want to just make dev an SkAutoTUnref to

Done.

https://codereview.appspot.com/6597055/diff/3001/tests/PremulAlphaRoundTripTe...
File tests/PremulAlphaRoundTripTest.cpp (right):

https://codereview.appspot.com/6597055/diff/3001/tests/PremulAlphaRoundTripTe...
tests/PremulAlphaRoundTripTest.cpp:50: GrContext* context) {
On 2012/10/01 17:10:59, robertphillips wrote:
> mv device inside loop?

why?

[robertphillips, 2012-10-01 18:14:27]

https://codereview.appspot.com/6597055/diff/3001/tests/PremulAlphaRoundTripTe...
File tests/PremulAlphaRoundTripTest.cpp (right):

https://codereview.appspot.com/6597055/diff/3001/tests/PremulAlphaRoundTripTe...
tests/PremulAlphaRoundTripTest.cpp:50: GrContext* context) {
On 2012/10/01 17:48:39, reed1 wrote:
> On 2012/10/01 17:10:59, robertphillips wrote:
> > mv device inside loop?
> 
> why?

Uniformity again. You did it that way in ReadPixelsTest.cpp. Either way is
actually fine by me - I would just like it to be the same.