Add FakeDisplayDelegate for off device usage.

services/ui/display/platform_screen_ozone.h

Index: services/ui/display/platform_screen_ozone.h
diff --git a/services/ui/display/platform_screen_ozone.h b/services/ui/display/platform_screen_ozone.h
index 521ead33aa394f005efe4fc06a817f032b788894..58aefe7bcceeaf48f8d85a0e8af6f6b90d953899 100644
--- a/services/ui/display/platform_screen_ozone.h
+++ b/services/ui/display/platform_screen_ozone.h
@@ -19,6 +19,7 @@
 #include "services/ui/public/interfaces/display/display_controller.mojom.h"
 #include "ui/display/chromeos/display_configurator.h"
 #include "ui/display/display.h"
+#include "ui/display/types/fake_display_controller.h"
 
 namespace display {
 
@@ -102,6 +103,14 @@ class PlatformScreenOzone
   ui::DisplayConfigurator display_configurator_;
   PlatformScreenDelegate* delegate_ = nullptr;
 
+  // If not null it provides a way to modify the display state when running off

[msw, 2016/09/15 19:30:08]

optional nit: "when running without a device (ie. linux desktop)" or similar?
(ok as-is if "off device" is already widely accepted terminology for that)

[kylechar, 2016/09/15 21:35:40]

It's the terminology Rob and I have been using, not sure if it's accepted. I've
added the clarification part though!

+  // device.
+  FakeDisplayController* fake_display_controller_ = nullptr;
+
+  // Tracks if we've made some type of change and want to wait for the display
+  // configuration to update before making further changes.
+  bool wait_for_display_update_ = false;

[msw, 2016/09/15 19:30:08]

optional nit: specialize this flag and comment for display mode changes? (this
flag shouldn't be re-used for other, potentially concurrent changes, right?)

[kylechar, 2016/09/15 21:35:40]

I'm not totally sure there is a case where we want concurrent changes?
ui::DisplayConfigurator posts a task with a 500ms delay before it handles the
display configuration changing. I'd imagine any user input that should trigger a
configuration change is going to want to wait until the last configuration
change is done.

It's possible that power state or content protection would happen concurrently
though? I'll change it to wait_for_display_config_update_ to be a bit clearer.

+
   // TODO(kylechar): These values can/should be replaced by DisplayLayout.
   int64_t primary_display_id_ = display::Display::kInvalidDisplayID;
   std::vector<DisplayInfo> cached_displays_;