Add implementation of CanvasSkia based on RenderText.

Add implementation of CanvasSkia based on RenderText.

This allows text drawing entirely through Skia, since RenderText ultimately uses
Skia to draw character glyphs.

This CL adds the new code, but does not yet enable it on any platforms. That
will be done in later CLs.

Note: Some functions, such as CanvasSkia::DrawStringWithHalo() and
PixelShouldGetHalo() are taken almost verbatim from the canvas_skia_win.cc
implementation.

BUG=105550
TEST=none

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=119086

[sky, 2012-01-23 23:01:17]

Don't you need to update some gyp files?

http://codereview.chromium.org/9074005/diff/19001/ui/gfx/canvas_skia_skia.cc
File ui/gfx/canvas_skia_skia.cc (right):

http://codereview.chromium.org/9074005/diff/19001/ui/gfx/canvas_skia_skia.cc#...
ui/gfx/canvas_skia_skia.cc:65: offset = text_size;
nit: I would be tempted to return here and in the other cases.

http://codereview.chromium.org/9074005/diff/19001/ui/gfx/canvas_skia_skia.cc#...
ui/gfx/canvas_skia_skia.cc:67: offset = available_height + text_size -
font.GetHeight() - 1;
Why the -1 here and +1 @ 71?

http://codereview.chromium.org/9074005/diff/19001/ui/gfx/canvas_skia_skia.cc#...
ui/gfx/canvas_skia_skia.cc:74: offset = half_height/2 - 2;
nit: 'half_height/2' -> 'half_height / 2'

[Alexei Svitkine (slow), 2012-01-24 16:22:24]

I've added the gyp file changes. Please take a look and let me know if these
make sense to you. Thanks!

http://codereview.chromium.org/9074005/diff/19001/ui/gfx/canvas_skia_skia.cc
File ui/gfx/canvas_skia_skia.cc (right):

http://codereview.chromium.org/9074005/diff/19001/ui/gfx/canvas_skia_skia.cc#...
ui/gfx/canvas_skia_skia.cc:65: offset = text_size;
On 2012/01/23 23:01:17, sky wrote:
> nit: I would be tempted to return here and in the other cases.

Done.

http://codereview.chromium.org/9074005/diff/19001/ui/gfx/canvas_skia_skia.cc#...
ui/gfx/canvas_skia_skia.cc:67: offset = available_height + text_size -
font.GetHeight() - 1;
On 2012/01/23 23:01:17, sky wrote:
> Why the -1 here and +1 @ 71?

These values were picked empirically by observing the differences between the
existing Windows implementation and this code.

The middle alignment case is particularly strange below. It needs the + 1 added
to the value before dividing by two, to get the same rounding behaviour as the
canvas_skia_win.cc code, but then ends up subtracting 2.

http://codereview.chromium.org/9074005/diff/19001/ui/gfx/canvas_skia_skia.cc#...
ui/gfx/canvas_skia_skia.cc:74: offset = half_height/2 - 2;
On 2012/01/23 23:01:17, sky wrote:
> nit: 'half_height/2' -> 'half_height / 2'

Done. I've also changed the variable name to better represent what it is (the
|half_height| name was from a version of this code that did the division by 2
earlier).

[sky, 2012-01-24 17:05:28]

LGTM

http://codereview.chromium.org/9074005/diff/19001/ui/gfx/canvas_skia_skia.cc
File ui/gfx/canvas_skia_skia.cc (right):

http://codereview.chromium.org/9074005/diff/19001/ui/gfx/canvas_skia_skia.cc#...
ui/gfx/canvas_skia_skia.cc:67: offset = available_height + text_size -
font.GetHeight() - 1;
On 2012/01/24 16:22:24, Alexei Svitkine wrote:
> On 2012/01/23 23:01:17, sky wrote:
> > Why the -1 here and +1 @ 71?
> 
> These values were picked empirically by observing the differences between the
> existing Windows implementation and this code.
> 
> The middle alignment case is particularly strange below. It needs the + 1
added
> to the value before dividing by two, to get the same rounding behaviour as the
> canvas_skia_win.cc code, but then ends up subtracting 2.

Please add a comment to this effect then.

[commit-bot: I haz the power, 2012-01-25 15:40:35]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/asvitkine@chromium.org/9074005/38001

[commit-bot: I haz the power, 2012-01-25 17:49:38]

Change committed as 119086