Provide gtest printers for ui/gfx geometry types

ui/gfx/test/gfx_util.h

Index: ui/gfx/test/gfx_util.h
diff --git a/ui/gfx/test/gfx_util.h b/ui/gfx/test/gfx_util.h
index 999322ad4ee811dfd79dc57375a5fa5a0e5becef..5e4ec1f4b1436670b917d693e44420c127279896 100644
--- a/ui/gfx/test/gfx_util.h
+++ b/ui/gfx/test/gfx_util.h
@@ -11,7 +11,7 @@
 #include "testing/gtest/include/gtest/gtest.h"
 #include "third_party/skia/include/core/SkColor.h"
 #include "ui/gfx/box_f.h"
-
+#include "ui/gfx/geometry/rect.h"

[danakj, 2014/07/07 18:52:27]

do you mean rect_f.h?

[jamesr, 2014/07/07 19:37:25]

Whoops, yes. Will fix

 
 namespace gfx {
 
@@ -24,6 +24,14 @@ namespace gfx {
                                                const BoxF& lhs,
                                                const BoxF& rhs);
 
+#define EXPECT_RECTF_EQ(a, b) \
+  EXPECT_PRED_FORMAT2(::gfx::AssertRectFloatEqual, a, b)
+
+::testing::AssertionResult AssertRectFloatEqual(const char* lhs_expr,

[danakj, 2014/07/07 18:52:27]

Is this related to this CL somehow or here by mistake?

[jamesr, 2014/07/07 19:37:25]

It's related - code that was doing comparisons between two RectFs was using
ToString() prior to this patch to both get pretty failure values and to round
(via the floating point formatting) into a string.  The callsite for this
actually has very slightly different RectFs that stringify to the same thing but
are not exactly the same number. This makes the fuzziness a bit more explicit

+                                                const char* rhs_expr,
+                                                const RectF& lhs,
+                                                const RectF& rhs);
+
 #define EXPECT_SKCOLOR_EQ(a, b) \
   EXPECT_PRED_FORMAT2(::gfx::AssertSkColorsEqual, a, b)