Create a semi-stable API for capturing the state of an SkCanvas and reconstructing that state acros…

gm/canvasstate.cpp

Index: gm/canvasstate.cpp
diff --git a/gm/canvasstate.cpp b/gm/canvasstate.cpp
index 5dc90ef2ae6a6c711f459fc3cbb72b99de31f0c2..50a897d05d5dc7d10d8360af4d76638d9e00995e 100644
--- a/gm/canvasstate.cpp
+++ b/gm/canvasstate.cpp
@@ -11,6 +11,8 @@
 #include "SkPath.h"
 #include "SkRect.h"
 
+#include "SkCanvasUtils.h"

[djsollen, 2013/08/28 15:45:12]

need to move this inline with other includes.

+
 namespace skiagm {
 
 /*
@@ -155,9 +157,18 @@ private:
     // draw a rect within the layer's bounds and again outside the layer's bounds
     void drawTestPattern(SkCanvas* canvas, U8CPU layerAlpha, SkCanvas::SaveFlags flags) {
         canvas->saveLayerAlpha(&fRect, layerAlpha, flags);
-        canvas->drawRect(fRect, fBluePaint);
-        canvas->translate(0, fRect.height() + 10);
-        canvas->drawRect(fRect, fBluePaint);
+
+        SkCanvas* newCanvas = canvas;

[djsollen, 2013/08/28 15:45:12]

need to add comment as to why we do this here.

[mtklein, 2013/08/28 18:20:46]

Yeah, I'm confused. :)  The name newCanvas makes me thing we're allocating a new
one.

+        SkCanvasState* state = SkCanvasUtils::CaptureCanvasState(canvas);
+        if (state) {
+            newCanvas = SkCanvasUtils::CreateFromCanvasState(state);
+            SkCanvasUtils::ReleaseCanvasState(state);
+        }
+
+        newCanvas->drawRect(fRect, fBluePaint);

[scroggo, 2013/08/28 17:22:01]

I know this is just a test, so if it is testing what we want it to that's fine.

Shouldn't we report if CaptureCanvasState failed?

[djsollen, 2013/08/29 15:09:19]

This was really just jammed in here and I'm convinced that GM isn't the right
place for this as it wants us to draw with a few canvas types that we can't draw
with.  I'm going to create a test that has the behavior that I want.

+        newCanvas->translate(0, fRect.height() + 10);
+        newCanvas->drawRect(fRect, fBluePaint);
+
         canvas->restore();
     }