bluetooth: macOS: Adding log in BluetoothAdapterMac::DisconnectGatt()

device/bluetooth/bluetooth_adapter_mac.mm

Index: device/bluetooth/bluetooth_adapter_mac.mm
diff --git a/device/bluetooth/bluetooth_adapter_mac.mm b/device/bluetooth/bluetooth_adapter_mac.mm
index 667bbd6e06425cd14a23461624b662165cbff694..d7660f6028f3a1832e6325c12652d022dfbbc6b3 100644
--- a/device/bluetooth/bluetooth_adapter_mac.mm
+++ b/device/bluetooth/bluetooth_adapter_mac.mm
@@ -641,12 +641,16 @@ BluetoothAdapterMac::RetrieveGattConnectedDevicesWithService(
 
 void BluetoothAdapterMac::CreateGattConnection(
     BluetoothLowEnergyDeviceMac* device_mac) {
+  DCHECK(device_mac);

[scheib, 2017/03/30 22:48:41]

This DCHECK is redundant with the "*device_mac" isn't it? If device_mac is null
then a crash will occur and the stack should be produced blaming that line, it
should be easy to understand. -- If I'm wrong, by all means use a DCHECK to make
it clear.

[jlebel, 2017/03/30 23:16:08]

It will crash, but somewhere in operator<< related to
BluetoothLowEnergyDeviceMac. My idea was to make it very clear. There is also
one difference, null will be caught by the DCHECK, a deallocated device will
crash in the operator<<.

Do you think I should remove the DCHECK?

[jlebel, 2017/03/31 07:39:12]

Done.

+  VLOG(1) << *device_mac << ": Create gatt connection.";
   [low_energy_central_manager_ connectPeripheral:device_mac->peripheral_
                                          options:nil];
 }
 
 void BluetoothAdapterMac::DisconnectGatt(
     BluetoothLowEnergyDeviceMac* device_mac) {
+  DCHECK(device_mac);
+  VLOG(1) << *device_mac << ": Disconnect gatt.";
   [low_energy_central_manager_
       cancelPeripheralConnection:device_mac->peripheral_];
 }