De-virtualize SkCanvas matrix ops.

De-virtualize SkCanvas matrix ops.

This moves the matrix management logic into non-virtual SkCanvas
methods, and turns the virtuals into protected notifiers.

R=reed@google.com,robertphillips@google.com
BUG=skia:2297
Committed: http://code.google.com/p/skia/source/detail?r=13799

[f(malita), 2014-03-12 22:23:47]

Preparing for lazy save, phase 2.

Why "did" and not "will"? Overrides in SkBBoxHierarchyRecord are accessing
post-op state (getTotalMatrix()), and refactoring in the context of a willXYX()
notifier would require duplicating the transform op. None of the other SkCanvas
subclasses care about ordering, so using post/"did" notifiers avoids BBH
recording overhead.

[reed1, 2014-03-13 12:03:08]

<bikeshed>I think I like pre- and post- better than will- and did- </bikeshed>

lgtm

[bsalomon, 2014-03-13 15:02:03]

On 2014/03/13 12:03:08, reed1 wrote:
> <bikeshed>I think I like pre- and post- better than will- and did- </bikeshed>
> 
> lgtm

lgtm. I like will and did. It seems less ambiguous and more clearly a
"notification" to the subclass. We already use "preTranslate", "postTranslate",
to mean something else (on matrix).

[reed1, 2014-03-13 15:04:41]

On 2014/03/13 15:02:03, bsalomon wrote:
> On 2014/03/13 12:03:08, reed1 wrote:
> > <bikeshed>I think I like pre- and post- better than will- and did-
</bikeshed>
> > 
> > lgtm
> 
> lgtm. I like will and did. It seems less ambiguous and more clearly a
> "notification" to the subclass. We already use "preTranslate",
"postTranslate",
> to mean something else (on matrix).

In this case, I propose the virtuals would be:

onPreConcat

indicating both pre-ness and protected-not-public-since-it-says-on-ness

[bsalomon, 2014-03-13 15:24:14]

On 2014/03/13 15:04:41, reed1 wrote:
> On 2014/03/13 15:02:03, bsalomon wrote:
> > On 2014/03/13 12:03:08, reed1 wrote:
> > > <bikeshed>I think I like pre- and post- better than will- and did-
> </bikeshed>
> > > 
> > > lgtm
> > 
> > lgtm. I like will and did. It seems less ambiguous and more clearly a
> > "notification" to the subclass. We already use "preTranslate",
> "postTranslate",
> > to mean something else (on matrix).
> 
> In this case, I propose the virtuals would be:
> 
> onPreConcat
> 
> indicating both pre-ness and protected-not-public-since-it-says-on-ness

To me that sounds like "subclass go ahead and implement pre-concating somthing."
Not, "subclass I'm about to concat something."

[reed1, 2014-03-13 15:52:20]

On 2014/03/13 15:24:14, bsalomon wrote:
> On 2014/03/13 15:04:41, reed1 wrote:
> > On 2014/03/13 15:02:03, bsalomon wrote:
> > > On 2014/03/13 12:03:08, reed1 wrote:
> > > > <bikeshed>I think I like pre- and post- better than will- and did-
> > </bikeshed>
> > > > 
> > > > lgtm
> > > 
> > > lgtm. I like will and did. It seems less ambiguous and more clearly a
> > > "notification" to the subclass. We already use "preTranslate",
> > "postTranslate",
> > > to mean something else (on matrix).
> > 
> > In this case, I propose the virtuals would be:
> > 
> > onPreConcat
> > 
> > indicating both pre-ness and protected-not-public-since-it-says-on-ness
> 
> To me that sounds like "subclass go ahead and implement pre-concating
somthing."
> Not, "subclass I'm about to concat something."

ok

[f(malita), 2014-03-13 19:58:19]

The CQ bit was checked by fmalita@chromium.org

[commit-bot: I haz the power, 2014-03-13 19:58:31]

CQ is trying da patch. Follow status at
https://skia-tree-status.appspot.com/cq/fmalita@chromium.org/195793012/1

[commit-bot: I haz the power, 2014-03-13 20:04:05]

Change committed as 13799