I am getting an exception with the following code snippet:
// matlab mex program ...
float *ptr=new float[n*m];
.
.
.
std::vector<float> v(n*m);
v.assign(ptr);
How do I correctly assign ptr to the vector v?
-
ptr=new *float[nm] -> ptr=new float[nm];user3394578– user33945782014-03-22 21:43:56 +00:00Commented Mar 22, 2014 at 21:43
Add a comment
|
1 Answer
You can use either the constructor or assign, as you tried. However, you need to pass the length in some way. In the standard library, it's a common idiom to do this by passing the begin and end of a sequence - but those can be bare pointers. Like this:
std::vector<float> v; // Note you do not need the size here
v.assign(ptr, ptr+n*m);
or just:
std::vector<float> v(ptr, ptr+n*m);
It also seems like you have an error in your allocation:
float *ptr=new float[n*m]; // Allocate floats, not pointers to floats
2 Comments
user3394578
How copy the pointer in pointer of std::vector<float> ?? avoiding too many data copy in the vector<float>.
ltjax
You cannot, as that would violate encapsulation. However, you can get the a pointer to the data in the vector and only use that instead of using new to allocate memory: float* ptr = v.data();