Have a common implementation of cc::OutputSurface

cc/test/fake_output_surface.cc

 // Copyright 2012 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "cc/test/fake_output_surface.h"
 
 namespace cc {
 
 FakeOutputSurface::FakeOutputSurface(
     scoped_ptr<WebKit::WebGraphicsContext3D> context3d, bool has_parent)
-    : num_sent_frames_(0) {
-  context3d_ = context3d.Pass();
+    : OutputSurface(context3d.Pass()),
+      num_sent_frames_(0) {
   capabilities_.has_parent_compositor = has_parent;
 }
 
 FakeOutputSurface::FakeOutputSurface(
     scoped_ptr<SoftwareOutputDevice> software_device, bool has_parent)
-    : num_sent_frames_(0) {
-  software_device_ = software_device.Pass();
+    : OutputSurface(software_device.Pass()),
+      num_sent_frames_(0) {
   capabilities_.has_parent_compositor = has_parent;
 }
 
 FakeOutputSurface::~FakeOutputSurface() {}
 
-bool FakeOutputSurface::BindToClient(OutputSurfaceClient* client) {

[danakj, 2013/01/24 02:03:06]

Can you keep this function and still call makeContextCurrent() here after
calling the super class. I'm just paranoid about changes in cc/ make tests into
no-ops by changing behaviour over time.

-  if (!context3d_)
-    return true;
-  DCHECK(client);
-  if (!context3d_->makeContextCurrent())
-    return false;
-  client_ = client;
-  return true;
-}
-
-const struct OutputSurface::Capabilities& FakeOutputSurface::Capabilities()
-    const {
-  return capabilities_;
-}
-
-WebKit::WebGraphicsContext3D* FakeOutputSurface::Context3D() const {
-  return context3d_.get();
-}
-
-SoftwareOutputDevice* FakeOutputSurface::SoftwareDevice() const {
-  return software_device_.get();
-}
-
 void FakeOutputSurface::SendFrameToParentCompositor(
     CompositorFrame* frame) {
   frame->AssignTo(&last_sent_frame_);
   ++num_sent_frames_;
 }
 
 }  // namespace cc