bluetooth: Subscribe to notifications

third_party/WebKit/Source/modules/bluetooth/BluetoothSupplement.cpp

 // Copyright 2015 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 "config.h"
 #include "modules/bluetooth/BluetoothSupplement.h"
 
 #include "core/frame/LocalDOMWindow.h"
 #include "public/platform/Platform.h"
 
@@ skipping 10 matching lines @@
 }
 
 void BluetoothSupplement::provideTo(LocalFrame& frame, WebBluetooth* bluetooth)
 {
     OwnPtrWillBeRawPtr<BluetoothSupplement> bluetoothSupplement = adoptPtrWillBeNoop(new BluetoothSupplement(bluetooth));
     WillBeHeapSupplement<LocalFrame>::provideTo(frame, supplementName(), bluetoothSupplement.release());
 };
 
 WebBluetooth* BluetoothSupplement::from(ScriptState* scriptState)
 {
-    LocalDOMWindow* window = scriptState->domWindow();
+    return BluetoothSupplement::from(scriptState->domWindow());
+}
+
+WebBluetooth* BluetoothSupplement::from(ExecutionContext* executionContext)
+{
+    return BluetoothSupplement::from(executionContext->executingWindow());
+}
+
+WebBluetooth* BluetoothSupplement::from(LocalDOMWindow* window)
+{
     if (window && window->frame()) {
         BluetoothSupplement* supplement = static_cast<BluetoothSupplement*>(WillBeHeapSupplement<LocalFrame>::from(window->frame(), supplementName()));
         if (supplement && supplement->m_bluetooth)
             return supplement->m_bluetooth;
     }
     return Platform::current()->bluetooth();

[haraken, 2015/10/08 00:51:52]

We sometimes return a bluetooth of the frame and sometimes a bluetooth of the
platform, which seems undeterministic. Why is it ok?

In other words, what happens if:

- we return nullptr if the frame is already gone; or
- we unconditionally return Platform::current()->bluetooth()

?

[ortuno, 2015/10/08 16:07:45]

If the frame is gone then window->frame() would be false and we would return
Platform::current()->bluetooth()
There is actually no implementation of bluetooth() so that would return nullptr.
We moved web_bluetooth from child/ to renderer/ here:

https://codereview.chromium.org/1228283003

I think the intention was the following:

If there is a frame available then get the supplement from there. If frame is
not available then we can assume we are in a worker and get the supplement from
platform. Since we don't support workers yet we return nullptr in this case. Do
you think this can be done in a better way?

 }
 
 DEFINE_TRACE(BluetoothSupplement)
 {
     WillBeHeapSupplement<LocalFrame>::trace(visitor);
 }
 
 }; // blink