尝试push_back（）结构会导致2D向量中的信息不正确

Question

I have a 2D vector that I'm trying to populate with coordinates. I've defined a struct for the coordinates for my own sake, but for some reason the push_back() function is not pushing the correct y-coordinate onto the vector, instead only pushing the first one.

Here is the code in question: the rest of the code is not too important to this snippet.

struct coor2d {
  float x;
  float y;
};

// Inside a function 
int row = 5;
int col = 5;
float r_wide = 1.0/float(row);
float r_high = 1.0/float(col);

vector<vector<coor2d> > grid;
vector<coor2d> column;
for(int cr = 0; cr < row; cr++) {
  for(int cc = 0; cr < col; cc++) {
    coor2d temp;
    temp.x = (float(cc) * r_wide) + (r_wide/2.0);
    temp.y = ((float(cr) * r_high) + (r_high/2.0) * -1.0);
    // Here the temp.y value is correct
    column.push_back(temp);
    // Here the temp.y value is incorrect
  }
  grid.push_back(column);
}

The rest of the code depends upon this working properly. I assume I'm losing precision or have called something incorrectly here. I know I could make a constructor for coor2d, but I don't think that would solve this problem; however, I could be wrong.

An example of the problem manifesting:

Once through the first iteration of the for(cr < row) loop, the inner for(cc < col) loop outputs the proper x-coordinates, but after the column.push_back(temp) is complete, the y-coordinate acts as though cr is still 0.0f instead of 1.0f, outputting -0.1f into the vector instead of the proper -0.3f. This happens on any value of cr.

Can anyone shed some light on this problem? Thanks for any help!

Answer 1

As @Erik pointed out, you have a typo. This:

for(int cc = 0; cr < col; cc++) {

should be this:

for(int cc = 0; cc < col; cc++) {

also, I think you probably want to "reset" your column vector on each pass of the outer for-loop. I think the easy way is to just move it:

vector<vector<coor2d> > grid;
for(int cr = 0; cr < row; cr++) {
  vector<coor2d> column;  // move the column vector to here
  for(int cc = 0; cr < col; cc++) {

If you don't do this, the column vector just accumulates all the values you push onto it.

With those changes, I get what I consider to be "sane" output from this test program:

#include <iostream>
#include <vector>

struct coor2d {
  float x;
  float y;
};

int main(){
// Inside a function 
  int row = 5;
  int col = 5;
  float r_wide = 1.0/float(row);
  float r_high = 1.0/float(col);

  for(int cr = 0; cr < row; cr++) {
    std::vector<coor2d> column;
    for(int cc = 0; cc < col; cc++) {
      coor2d temp;
      temp.x = (float(cc) * r_wide) + (r_wide/2.0);
      temp.y = ((float(cr) * r_high) + (r_high/2.0) * -1.0);
    // Here the temp.y value is correct
      column.push_back(temp);
    // Here the temp.y value is incorrect
      std::cout << "temp.x: " << temp.x << " temp.y: " << temp.y << std::endl;
      std::cout << "vec.x:  " << column[cc].x << " vec.y:   " << column[cc].y << std::endl;
    }
  grid.push_back(column);
  }
}