I'm learning about the standard library algorithms recently and have a question about the function fill_n(iter, n, val)
. This function requires the container has at least n
elements starting from iter
.
Here is the testing code:
// Version 1, Error
vector<int> vec;
vec.reserve(10); // Only allocate space for at least 10 elements
fill_n(vec.begin(), 10, 0);
// Version 2, OK
vector<int> vec;
vec.resize(10); // Value initialized 10 elements
fill_n(vec.begin(), 10, 0);
// Version 3, OK
vector<int> vec;
fill_n(back_inserter(vec), 10, 0); // Push back 10 elements via back_inserter
Why the version 1 code is error while version 2 & 3 are not?
reserve
only reserves space, but the size of the vector remains unchanged. The iterator returned by begin
can not be incremented past the end of the vector, and because it is the (unchanged) size that determines where the end of the vector is, you get an error.
Version 1 does NOT work because:
std::reserve
modifies the capacity of the vector and NOT it's size. std::fill_n
requires that the container have the correct size beforehand.
Version 2 works because:
std::resize
does modify the size of the vector and not just its capacity.
Version 3 works because:
std::back_inserter
will call push_back
on the vector which adds to the vector and modifies it's size accordingly.
reserve
does not initialize anything. It just reserve some space so no reallocation happens each time new item is pushed.
So the solution to tell fill_n
to push its result directly to the vector
at the end for example.
Change this:
// Version 1, Error
vector<int> vec;
vec.reserve(10); // Only allocate space for at least 10 elements
fill_n(vec.begin(), 10, 0);
To:
// Version 1, Corrected
vector<int> vec;
vec.reserve(10); // Only allocate space for at least 10 elements
fill_n(std::back_inserter(vec), 10, 0);