复制 std: : Vector: 更喜欢赋值还是 std: : copy？

我有两个向量:

std::vector<int> v1, v2;


// Filling v1
...

现在我需要复制 v1到 v2。有任何理由更喜欢

v2 = v1;

到

std::copy (v1.begin(), v1.end(), v2.begin());

(反之亦然) ？

107986 次浏览
小开

It's shorter.

std::copy is mainly meant for copying sections of containers. If you need to copy an entire container, you might as well use the copy constructor.

小开

The invocation of std::copy may try to access items beyond the end of the destination vector.

Use assignment.

It's not your job to micro-optimize: that's the library writer's responsibility, and ultimately the compiler's responsibility.

You can make your code arbitrarily fast if it doesn't have to be correct.

In the case of the copy, however, it's rather doubtful whether it even is faster, and it's certainly not correct for the general case.

小开

Assignment, by far. More generally, any time the size of the vector might change, or change the entire contents of the vector, you should prefer member functions. The only time std::copy would be appropriate is if you are only replacing a small range totally within the vector.

小开
最佳答案

Generally I would strongly prefer v2 = v1:

  1. It is shorter and makes the intent more clear
  2. std::copy won't work if v2 doesn't have the same length as v1 (it won't resize it, so it will retain some of the old elements best case (v2.size() > v1.size() and overwrite some random data used elsewhere in the program worst case
  3. If v1 is about to expire (and you use C++11) you can easily modify it to move the contents
  4. Performancewise assignment is unlikely to be slower then std::copy, since the implementers would probably use std::copy internally, if it gave a performance benefit.

In conclusion, std::copy is less expressive, might do the wrong thing and isn't even faster. So there isn't really any reason to use it here.

小开

If v2 isn't big enough you'll get a buffer overrun if you use copy as you have.

You can use a back insert iterator which will call push_back on v2. However this could lead to multiple reallocations depending upon how big v1 is.

copy(v1.begin(), v1.end(), back_inserter(v2));

You're better off letting vector manage things correctly. The assignment operator does this, as does vector::assign:

v2.assign(v1.begin(), v1.end());

I have an inkling that the assignment operator is implemented in terms of vector::assign.

小开

Assignement is clearer and internally uses std::copy (or unitizalized_copy _M_allocate_and_copy depending size and capacity) or so performances are the same.


提交答案