I would like to switch our code from boost::shared_ptr to std::shared_ptr. If allowed, I would extend this to include all other boost:: classes that now have standard equivalents, such as boost::unordered_map.

boost works, so why bother? I see several advantages:

• It makes our code more standard, so less likely to confuse new developers and outsiders.
• Code that uses a mix of unique_ptr and shared_ptr would be less confusing.
• It would also give us a more accurate picture of what parts of boost we truly rely on, in hopes of eventually ending our use of boost entirely.

Acceptance of this RFC does not imply that it is a high priority. It simply gives us a go-ahead to write a ticket and implement the ticket when we have time. I have not looked deeply at the code to estimate how much effort is required, but I don't expect it to be very much. We have to change the definition of PTR and CONST_PTR and then clean up some additional code.