add peekPixels to canvas

add peekPixels to SkCanvas and SkSurface

This is motivated by a desire to remove call-sites that perform the folowing:

    SkBaseDevice* dev = canvas->getDevice();
    const SkBitmap& bitmap = dev->accessBitmap();
    ... read the bitmap

1. We'd like to deprecate getDevice() entirely
2. BaseDevice itself shouldn't know about bitmaps (only BitmapDevice should)
3. SkImage already has peekPixels()

[hal.canary, 2014-02-12 15:11:01]

https://codereview.chromium.org/159723006/diff/20001/include/core/SkCanvas.h
File include/core/SkCanvas.h (right):

https://codereview.chromium.org/159723006/diff/20001/include/core/SkCanvas.h#...
include/core/SkCanvas.h:130: *  RAM) return the const-address of those pixels,
and if not null, return
And if the canvas has a concept of "pixels" at all.

[reed1, 2014-02-12 15:15:43]

On 2014/02/12 15:11:01, Hal Canary wrote:
> https://codereview.chromium.org/159723006/diff/20001/include/core/SkCanvas.h
> File include/core/SkCanvas.h (right):
> 
>
https://codereview.chromium.org/159723006/diff/20001/include/core/SkCanvas.h#...
> include/core/SkCanvas.h:130: *  RAM) return the const-address of those pixels,
> and if not null, return
> And if the canvas has a concept of "pixels" at all.

"If the canvas has direct access to its pixels" -- is this sufficient?

[bsalomon, 2014-02-12 15:56:36]

lgtm

https://codereview.chromium.org/159723006/diff/20001/gm/gm.h
File gm/gm.h (right):

https://codereview.chromium.org/159723006/diff/20001/gm/gm.h#newcode12
gm/gm.h:12: //#include "SkBitmapDevice.h"
just remove it?

https://codereview.chromium.org/159723006/diff/20001/gm/imagefiltersclipped.cpp
File gm/imagefiltersclipped.cpp (right):

https://codereview.chromium.org/159723006/diff/20001/gm/imagefiltersclipped.c...
gm/imagefiltersclipped.cpp:62: SkCanvas canvas(fGradientCircle);
is the spacing really weird or is this just a codereview artifact? anyway, it'd
be nice to clean it up while we're modifying this. I seem similar odd whitespace
in files below with similar code...

[reed1, 2014-02-12 17:24:09]

https://codereview.chromium.org/159723006/diff/20001/gm/gm.h
File gm/gm.h (right):

https://codereview.chromium.org/159723006/diff/20001/gm/gm.h#newcode12
gm/gm.h:12: //#include "SkBitmapDevice.h"
On 2014/02/12 15:56:37, bsalomon wrote:
> just remove it?

Done.

https://codereview.chromium.org/159723006/diff/20001/gm/imagefiltersclipped.cpp
File gm/imagefiltersclipped.cpp (right):

https://codereview.chromium.org/159723006/diff/20001/gm/imagefiltersclipped.c...
gm/imagefiltersclipped.cpp:62: SkCanvas canvas(fGradientCircle);
On 2014/02/12 15:56:37, bsalomon wrote:
> is the spacing really weird or is this just a codereview artifact? anyway,
it'd
> be nice to clean it up while we're modifying this. I seem similar odd
whitespace
> in files below with similar code...

artifact I think

https://codereview.chromium.org/159723006/diff/20001/include/core/SkCanvas.h
File include/core/SkCanvas.h (right):

https://codereview.chromium.org/159723006/diff/20001/include/core/SkCanvas.h#...
include/core/SkCanvas.h:130: *  RAM) return the const-address of those pixels,
and if not null, return
On 2014/02/12 15:11:02, Hal Canary wrote:
> And if the canvas has a concept of "pixels" at all.

Done.

[reed1, 2014-02-12 17:24:16]

The CQ bit was checked by reed@google.com

[commit-bot: I haz the power, 2014-02-12 17:24:22]

CQ is trying da patch. Follow status at
https://skia-tree-status.appspot.com/cq/reed@google.com/159723006/140001

[commit-bot: I haz the power, 2014-02-12 17:29:38]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-02-12 17:29:38]

Retried try job too often on Build-Ubuntu12-GCC-x86_64-Release-Trybot for
step(s) BuildSkiaLib
http://108.170.219.164:10117/buildstatus?builder=Build-Ubuntu12-GCC-x86_64-Re...

[scroggo, 2014-02-12 18:40:57]

https://codereview.chromium.org/159723006/diff/140001/include/core/SkCanvas.h
File include/core/SkCanvas.h (right):

https://codereview.chromium.org/159723006/diff/140001/include/core/SkCanvas.h...
include/core/SkCanvas.h:81: *  (cpu or gpu), then the info's ColorType will be
kUnknown_SkColorType.
Something to think about - kNo_Config meant invalid in all cases, whereas
kUnknown does not.

This may be fine as is (at least for now), but maybe we'll find it makes sense
to have one type that is invalid, and one that is, perhaps neutral? i.e. a
picture doesn't really have an unknown color type - it has no inherent color
type.

https://codereview.chromium.org/159723006/diff/140001/include/core/SkCanvas.h...
include/core/SkCanvas.h:143: *  canvas canvas (or its parent surface if any)
will invalidate the
"canvas canvas" -> "canvas"

https://codereview.chromium.org/159723006/diff/140001/include/core/SkCanvas.h...
include/core/SkCanvas.h:1242: *  succed (though be slower, since it will return
a copy of the pixels).
succeed*

https://codereview.chromium.org/159723006/diff/140001/src/core/SkCanvas.cpp
File src/core/SkCanvas.cpp (right):

https://codereview.chromium.org/159723006/diff/140001/src/core/SkCanvas.cpp#n...
src/core/SkCanvas.cpp:1032: const SkBitmap& bm =
canvas->getDevice()->accessBitmap(false);
Is this just here for comparison?

https://codereview.chromium.org/159723006/diff/140001/tests/SurfaceTest.cpp
File tests/SurfaceTest.cpp (right):

https://codereview.chromium.org/159723006/diff/140001/tests/SurfaceTest.cpp#n...
tests/SurfaceTest.cpp:178: continue;   // gpu may not be enabled
This could hide errors if one of the others fails to get created.

[reed2, 2014-02-13 14:43:34]

https://codereview.chromium.org/159723006/diff/140001/include/core/SkCanvas.h
File include/core/SkCanvas.h (right):

https://codereview.chromium.org/159723006/diff/140001/include/core/SkCanvas.h...
include/core/SkCanvas.h:143: *  canvas canvas (or its parent surface if any)
will invalidate the
On 2014/02/12 18:40:58, scroggo wrote:
> "canvas canvas" -> "canvas"

Done.

https://codereview.chromium.org/159723006/diff/140001/include/core/SkCanvas.h...
include/core/SkCanvas.h:1242: *  succed (though be slower, since it will return
a copy of the pixels).
On 2014/02/12 18:40:58, scroggo wrote:
> succeed*

Done.

https://codereview.chromium.org/159723006/diff/140001/src/core/SkCanvas.cpp
File src/core/SkCanvas.cpp (right):

https://codereview.chromium.org/159723006/diff/140001/src/core/SkCanvas.cpp#n...
src/core/SkCanvas.cpp:1032: const SkBitmap& bm =
canvas->getDevice()->accessBitmap(false);
On 2014/02/12 18:40:58, scroggo wrote:
> Is this just here for comparison?

Just for local testing, will remove.

https://codereview.chromium.org/159723006/diff/140001/tests/SurfaceTest.cpp
File tests/SurfaceTest.cpp (right):

https://codereview.chromium.org/159723006/diff/140001/tests/SurfaceTest.cpp#n...
tests/SurfaceTest.cpp:178: continue;   // gpu may not be enabled
On 2014/02/12 18:40:58, scroggo wrote:
> This could hide errors if one of the others fails to get created.

Done.