Unfork VectorPlatformCanvas.

Unfork VectorPlatformCanvas. 

Unfork VectorPlatformCanvas by making NativeMetafile know how to create an appropriate VectorPlatformDevice. This will also be useful when we have multiple NativeMetafile implemenations (each requiring a different VectorPlatformDevices). 

BUG=NONE
TEST=NONE

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=78662
Reverted: http://src.chromium.org/viewvc/chrome?view=rev&revision=78663

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=78812
Reverted: http://src.chromium.org/viewvc/chrome?view=rev&revision=78815

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=78859
Reverted: http://src.chromium.org/viewvc/chrome?view=rev&revision=78860

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=78935

[Lei Zhang, 2011-03-17 19:42:46]

http://codereview.chromium.org/6665046/diff/5020/printing/native_metafile.h
File printing/native_metafile.h (right):

http://codereview.chromium.org/6665046/diff/5020/printing/native_metafile.h#n...
printing/native_metafile.h:58: // VectorPlatformDevice implemenation bound to
the context created by
nit: typo.

http://codereview.chromium.org/6665046/diff/5020/printing/native_metafile.h#n...
printing/native_metafile.h:62: const float& scale_factor) = 0;
Why add scale_factor if it's going to go unused?

http://codereview.chromium.org/6665046/diff/5020/skia/ext/vector_platform_dev...
File skia/ext/vector_platform_device_linux.h (right):

http://codereview.chromium.org/6665046/diff/5020/skia/ext/vector_platform_dev...
skia/ext/vector_platform_device_linux.h:18: static skia::PlatformDevice*
CreateDevice(cairo_t* context, int width,
nit: do you need skia:: ? Same for the other vector_platform_device_* files.

[vandebo (ex-Chrome), 2011-03-17 19:53:43]

Mike, can you take a look, Lei requested a second set of eyes on this change.

http://codereview.chromium.org/6665046/diff/5020/printing/native_metafile.h
File printing/native_metafile.h (right):

http://codereview.chromium.org/6665046/diff/5020/printing/native_metafile.h#n...
printing/native_metafile.h:58: // VectorPlatformDevice implemenation bound to
the context created by
On 2011/03/17 19:42:46, Lei Zhang wrote:
> nit: typo.

Done.

http://codereview.chromium.org/6665046/diff/5020/printing/native_metafile.h#n...
printing/native_metafile.h:62: const float& scale_factor) = 0;
On 2011/03/17 19:42:46, Lei Zhang wrote:
> Why add scale_factor if it's going to go unused?

Two fold, one to make a uniform interface across platforms (Mac uses it)
(dpapad@ is changing StartPage() to be uniform across platforms and it will have
this same signature).  And because it could be used in some of the
implementations even though it isn't now.  Where we'd want to use it is to
squish one size document into another size, which we'll eventually need to do.

http://codereview.chromium.org/6665046/diff/5020/skia/ext/vector_platform_dev...
File skia/ext/vector_platform_device_linux.h (right):

http://codereview.chromium.org/6665046/diff/5020/skia/ext/vector_platform_dev...
skia/ext/vector_platform_device_linux.h:18: static skia::PlatformDevice*
CreateDevice(cairo_t* context, int width,
On 2011/03/17 19:42:46, Lei Zhang wrote:
> nit: do you need skia:: ? Same for the other vector_platform_device_* files.

Done.

[reed1, 2011-03-17 20:06:53]

Don't understand the big picture. Will try a chat or call.

http://codereview.chromium.org/6665046/diff/7001/skia/ext/vector_canvas.cc
File skia/ext/vector_canvas.cc (right):

http://codereview.chromium.org/6665046/diff/7001/skia/ext/vector_canvas.cc#ne...
skia/ext/vector_canvas.cc:13: setDevice(device);
standard pattern is

this->setDevice(device)->unref();

no biggie.

[vandebo (ex-Chrome), 2011-03-17 20:15:28]

On 2011/03/17 20:06:53, reed1 wrote:
> Don't understand the big picture. Will try a chat or call.

As we were talking about earlier, we will need to keep, for instance, both the
emf metafile class in addition to a new skia (pdf) metafile class.  Currently
construction (in print_web_view_helper) requires a PlatformSurface (an HDC on
windows), but the Skia implementation doesn't care about that.  This refactor
makes it irrelevant by having the metafile create the appropriate
VectorPlatformDevice.

Of course if we can get rid of the platform canvas/device code, this all goes
away.

[Lei Zhang, 2011-03-18 02:27:16]

LGTM