c++ - std::string as map key and efficiency of map.find() -

i have std::map<std::string,foo> , need find() elements when instead of std::string, have const char*. however, map.find("key_value") won't work, need std::string. so, need map.find(std::string{"key_value"}), create temporary std::string (with allocation , de-allocation) each time, making potentially inefficient. reasoning correct? if yes, how can improve situation?

i think using wrapper around const char* key map has own comparison function , can cheaply wrapped around const char*, no need allocations. idea? (note mapped values not copyable, movable).

you can avoid temporaries using 0-terminated strings , custom comparator map.
still, using unordered_map yield better indexing performance.

an alternative (much better), using wrapper, think.
there std::reference_wrapper task.

still, first rule of optimisation: not it.
second rule: not it.
third rule (only experts): after measurement , careful consideration, might anyway.

the downside is: must manually manage lifetimes of strings.

if going down road anyway, , not free string until free them all, consider using own custom allocator.