Suppress some no-op GraphicsContext calls

Suppress some no-op GraphicsContext calls

The CL avoids commiting deferred canvas saves for a few easy to catch
no-op GC calls.

This reduces the number of SkPicture draw ops by ~3%, and shows minor
(<1%) rasterization time gains.

R=schenney@chromium.org,senorblanco@chromium.org

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=165353

[Stephen Chennney, 2014-01-17 21:53:20]

LGTM with one nit.

https://codereview.chromium.org/136583009/diff/1/Source/platform/graphics/Gra...
File Source/platform/graphics/GraphicsContext.cpp (right):

https://codereview.chromium.org/136583009/diff/1/Source/platform/graphics/Gra...
Source/platform/graphics/GraphicsContext.cpp:1606: if (size.width() == 1 &&
size.height() == 1)
Not sure if the compiler automatically casts the 1 to 1.0f, but I feel more
comfortable with explicitly saying 1.0f.

[Stephen White, 2014-01-17 21:55:56]

On 2014/01/17 21:53:20, Stephen Chenney wrote:
> LGTM with one nit.
> 
>
https://codereview.chromium.org/136583009/diff/1/Source/platform/graphics/Gra...
> File Source/platform/graphics/GraphicsContext.cpp (right):
> 
>
https://codereview.chromium.org/136583009/diff/1/Source/platform/graphics/Gra...
> Source/platform/graphics/GraphicsContext.cpp:1606: if (size.width() == 1 &&
> size.height() == 1)
> Not sure if the compiler automatically casts the 1 to 1.0f, but I feel more
> comfortable with explicitly saying 1.0f.

Blink inherited this from the WebKit coding style:

"Unless required in order to force floating point math, do not append .0, .f and
.0f to floating point literals."

I actually agree with Stephen, though. It would be great if we changed the style
guide to remove this rule instead.

[f(malita), 2014-01-17 22:01:50]

On 2014/01/17 21:55:56, Stephen White wrote:
> On 2014/01/17 21:53:20, Stephen Chenney wrote:
> > LGTM with one nit.
> > 
> >
>
https://codereview.chromium.org/136583009/diff/1/Source/platform/graphics/Gra...
> > File Source/platform/graphics/GraphicsContext.cpp (right):
> > 
> >
>
https://codereview.chromium.org/136583009/diff/1/Source/platform/graphics/Gra...
> > Source/platform/graphics/GraphicsContext.cpp:1606: if (size.width() == 1 &&
> > size.height() == 1)
> > Not sure if the compiler automatically casts the 1 to 1.0f, but I feel more
> > comfortable with explicitly saying 1.0f.
> 
> Blink inherited this from the WebKit coding style:
> 
> "Unless required in order to force floating point math, do not append .0, .f
and
> .0f to floating point literals."
> 
> I actually agree with Stephen, though. It would be great if we changed the
style
> guide to remove this rule instead.

But, but, but... I worked so hard to adjust myself to the WK style.... :)

(SGTM, let's see if we can sneak it past the presubmit bot)

[Stephen Chennney, 2014-01-17 22:01:55]

On 2014/01/17 21:55:56, Stephen White wrote:
> On 2014/01/17 21:53:20, Stephen Chenney wrote:
> > LGTM with one nit.
> > 
> >
>
https://codereview.chromium.org/136583009/diff/1/Source/platform/graphics/Gra...
> > File Source/platform/graphics/GraphicsContext.cpp (right):
> > 
> >
>
https://codereview.chromium.org/136583009/diff/1/Source/platform/graphics/Gra...
> > Source/platform/graphics/GraphicsContext.cpp:1606: if (size.width() == 1 &&
> > size.height() == 1)
> > Not sure if the compiler automatically casts the 1 to 1.0f, but I feel more
> > comfortable with explicitly saying 1.0f.
> 
> Blink inherited this from the WebKit coding style:
> 
> "Unless required in order to force floating point math, do not append .0, .f
and
> .0f to floating point literals."
> 
> I actually agree with Stephen, though. It would be great if we changed the
style
> guide to remove this rule instead.

Actually, I should have suggested 1.f, which is my preference. But 1.0f is fine
too.

I never understood the reason for the WebKit rule apart from their general
"minimize the number of characters in a file except when naming methods".

[commit-bot: I haz the power, 2014-01-17 22:03:33]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/fmalita@chromium.org/136583009/2

[f(malita), 2014-01-17 22:47:41]

On 2014/01/17 22:03:33, I haz the power (commit-bot) wrote:
> CQ is trying da patch. Follow status at
> https://chromium-status.appspot.com/cq/fmalita%40chromium.org/136583009/2

Interesting, the drawDisplayList() optimization seems to cause a failure on
linux: clip-path-with-different-unittypes.svg. That's a clipping test that uses
sub-unit (object bounding box) values.

We can certainly skip the display list bounds check for now as we're not really
recording lots of pictures, but the failure may be indicative of a real display
list sizing issue (the bounding box should not be empty). I'll make a note and
look into it.

[commit-bot: I haz the power, 2014-01-17 22:49:54]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/fmalita@chromium.org/136583009/200001

[commit-bot: I haz the power, 2014-01-18 01:03:12]

Change committed as 165353