Smart pointers (namely unique_ptr and shared_ptr) were added in C++11 and have become an impactful part of modern C++, helping to alleviate many issues that accompany heap allocated memory. Smart pointers can essentially be used as a quick and easy RAII wrapper class, allowing the memory to be deleted once the object goes out of scope and the destructor is called.
I would like to talk about the differences between unique_ptr and shared_ptr. The benefit of having these two types of pointers is that it ensures the programmer explicitly shows the ownership of the memory. If you read unique_ptr then you know the memory will only ever have one pointer managing it, making it much simpler to manage. A shared pointer on the other hand can have many owners, something that can lead to issues so must be explicit and clear. It's important to note that pointers that do not having any kind of heap allocations to manage can and frankly should be raw pointers, again for the sake of being explicit.
My favourite part of smart pointers is the beautifully simple way that unique_ptr forces its uniqueness once created with just a single keyword... "delete" which it uses to explicitly delete the copy constructor and copy assignment, only allowing move semantics to be used which would keep it unique.
Smart pointers also progressed into having their own functions for allocating memory on the heap. These being make_unique and make_shared which should always be used over new and delete when using smart pointers. In C++14 and onwards, new and delete should be avoided, unless placement new and delete are required for more specific memory management.
Back