Answering How to self-copy a vector? has got me a bit confused about iterator invalidation. Some literature says “if you use insert, push_back, etc. consider all iterators invalid”. Thats clear, it might cause the vector to grow which invalidates iterators. What about the special case where I know there is going to be enough room?
first try:
myvec.reserve(myvec.size()*3); //does this protect me from iterator invalidation?
vector<string>::iterator it = myvec.end();
myvec.insert(myvec.end(), myvec.begin(), it);
myvec.insert(myvec.end(), myvec.begin(), it);
After some excellent answers second try:
auto size = myvec.size();
myvec.reserve(size*3); //does this protect me from iterator invalidation?
myvec.insert(myvec.end(), myvec.begin(), myvec.begin()+size);
myvec.insert(myvec.end(), myvec.begin(), myvec.begin()+size);
After more excellent answers third try:
auto size = myvec.size();
myvec.reserve(size*3); //does this protect me from iterator invalidation?
back_insert_iterator< vector<string> > back_it (myvec);
copy (myvec.begin(),myvec.begin()+size,back_it);
copy (myvec.begin(),myvec.begin()+size,back_it);
This quote from Josuttis’ “C++ Standard Library Reference”:
Inserting or removing elements
invalidates references, pointers, and
iterators that refer to the following
element. If an insertion causes
reallocation, it invalidates all
references, iterators, and pointers.
suggests that my code is safe and defined behavior. Is there a passage in the standard which guaranties this?
The past-the-end iterator is always a bit special. I’d be careful. The standard says this (23.3.6.5):
The key here is “before the insertion point”. Since your original
itis not before the insertion point (since it is the insertion point), I wouldn’t bank on it remaining valid.