[System-Keyboard-Lock] Forward navigator functions to RenderFrameHost

third_party/WebKit/public/platform/modules/keyboard_lock/keyboard_lock.mojom

Index: third_party/WebKit/public/platform/modules/keyboard_lock/keyboard_lock.mojom
diff --git a/third_party/WebKit/public/platform/modules/keyboard_lock/keyboard_lock.mojom b/third_party/WebKit/public/platform/modules/keyboard_lock/keyboard_lock.mojom
new file mode 100644
index 0000000000000000000000000000000000000000..77b384f45af735b75035c4cc6c5bbc651f31ccc0
--- /dev/null
+++ b/third_party/WebKit/public/platform/modules/keyboard_lock/keyboard_lock.mojom
@@ -0,0 +1,21 @@
+// Copyright 2017 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.
+
+module blink.mojom;
+
+// The browser side service to process navigator.requestKeyLock() and
+// navigator.cancelKeyLock() web APIs. See http://crbug.com/680809.
+interface KeyboardLockService {
+  // Registers a set of string-formatted key codes
+  // (https://www.w3.org/TR/uievents/#interface-keyboardevent) to the platform
+  // dependent native API, so the web page can receive these key codes
+  // thereafter.
+  // The reason will only be provided if the request is rejected.

[dcheng, 2017/04/26 13:27:18]

Is it necessary to return a freeform string here? Maybe the return value should
actually be a status enum?

[Hzj_jie, 2017/04/26 22:05:56]

If we are only talking about the Chrome implementation, this function will never
false and the reason. So the design of the return values is to ensure we are
following the spec only.
I will add a comment here to ensure it has been updated once the spec defines
the error types.

[dcheng, 2017/04/27 02:53:10]

I think this shouldn't stop us from using an enum for now:
- We can define an enum for the return types (even if the Chrome implementation
currently can't fail, I imagine it can in the future). Right now, it could just
be SUCCESS and FAILED as the two entries.
- The mojo method itself should just return an enum status
- The mapping of status code -> string should happen in the browser. See
WebBluetooth for an example.

[Hzj_jie, 2017/04/28 00:10:31]

This will be done in a new change immediately.

+  RequestKeyLock(array<string> key_codes)
+    => (bool allowed_by_user_or_browser, string reason);
+
+  // Removes all reserved keys. This function is expected to never fail, the
+  // callback indicates the finish of the processing only.

[dcheng, 2017/04/26 13:27:18]

There's no callback -- is this comment correct?

[Hzj_jie, 2017/04/26 22:05:56]

Sorry, I forget to update the comment.

+  CancelKeyLock();
+};