Change scoped_ptr to a type alias for std::unique_ptr on OS_WIN

printing/backend/printing_info_win.h

Index: printing/backend/printing_info_win.h
diff --git a/printing/backend/printing_info_win.h b/printing/backend/printing_info_win.h
index c91d837724935c352d36861faca9531ffce88346..896ed12a763be49d43e5cee6baa25431ed0a0c8d 100644
--- a/printing/backend/printing_info_win.h
+++ b/printing/backend/printing_info_win.h
@@ -26,7 +26,7 @@ class PrinterInfo {
  public:
   bool Init(HANDLE printer) {
     buffer_.reset(GetPrinterInfo(printer, level));
-    return buffer_;
+    return buffer_ != nullptr;

[Vitaly Buka (NO REVIEWS), 2016/03/04 18:25:17]

is any warning without this?
I thought Chromium prefer checking pointers without comparing with null.

[dcheng, 2016/03/04 18:28:59]

It depends on the reviewer =P

With C++11 and explicit operator bool, this can no longer implicitly convert,
since it's not a place where a "context conversion to bool" is allowed (an
example of somewhere where it is allowed is if (x) { ... }).

So we have several choices:
- return static_cast<bool>(buffer_);
- return !!buffer_;  (my personal preferred option)
- return buffer_ != nullptr;  (some reviewers expressed that this might be
clearer)

[Vitaly Buka (NO REVIEWS), 2016/03/04 18:34:39]

Thanks. I like !! too, but current one is lgtm too.

[joedow, 2016/03/04 20:46:19]

I like !! as well to boolify but I have seen feedback both ways.

   }
 
   const PrinterInfoType* get() const {
@@ -44,7 +44,7 @@ class DriverInfo {
  public:
   bool Init(HANDLE printer) {
     buffer_.reset(GetDriverInfo(printer, level));
-    return buffer_;
+    return buffer_ != nullptr;
   }
 
   const DriverInfoType* get() const {