Is the sole difference between boost::scoped_ptr<T>
and std::unique_ptr<T>
the fact that std::unique_ptr<T>
has move semantics whereas boost::scoped_ptr<T>
is just a get/reset smart pointer?
问题:
回答1:
No, but that is the most important difference.
The other major difference is that unique_ptr
can have a destructor object with it, similarly to how shared_ptr
can. Unlike shared_ptr
, the destructor type is part of the unique_ptr
's type (the way allocators are part of STL container types).
A const unique_ptr
can effectively do most of what a scoped_ptr
can do; indeed, unlike scoped_ptr
, a const unique_ptr
cannot be rebound with a reset
call.
Also, unique_ptr<T>
can work on a T
which is an incomplete type. The default deleter type requires that T
be complete when you do anything to the unique_ptr
that potentially invokes the deleter. You therefore have some freedom to play games about where that happens, depending on the situation.
回答2:
unique_ptr
owns an object exclusively.It is non-copyable but supports transfer-of-ownership. It was introduced as replacement for the now deprecated auto_ptr
.
scoped_ptr
is neither copyable nor movable. It is the preferred choice when you want to make sure pointers are deleted when going out of scope.