Mixing Smart and Raw Pointers

Yes, it is common and often necessary to mix smart pointers and raw pointers in the same program. The key is to establish clear conventions around ownership and responsibility.

When a function uses a raw pointer, it is typically signaling that it wants access to the resource, but is not taking ownership of it. For example:

1#include <memory>
2#include <iostream>
3
4void PrintName(std::string* Name) {
5  std::cout << *Name << '\n';
6}
7
8int main() {
9  auto Name{std::make_unique<std::string>(
10    "Gandalf")};
11  PrintName(Name.get());
12}

1Gandalf

Here, main is maintaining ownership of the resource, and PrintName is simply accessing it.

Conversely, when ownership needs to be transferred, smart pointers are used:

1#include <memory>
2#include <utility>
3#include <iostream>
4
5void StoreNameElsewhere(std::unique_ptr<
6  std::string> Name) {
7  // Store name in file, database etc
8  std::cout << "Storing " << *Name << "\n";
9}
10
11int main() {
12  auto Name{std::make_unique<std::string>(
13    "Gandalf")};
14  StoreNameElsewhere(std::move(Name));
15}

1Storing Gandalf

In summary:

• Use smart pointers for ownership
• Use raw pointers for non-owning access
• Be clear about whether a function is taking ownership or not
• Don't delete resources you don't own