Please help! -- C-style 2D array containing std::vector

[a_sonhye]

I know it's weird but I would like to create a 2D c-style array. Each array element stores a vector of int. The result will produce 3D array of int.

I can compile my code but it seems like whenever I add some data into the vector and, later I want to display it, there is no data in the vector!!!

Please let me know what I did wrong. Thank you very much.

typedef std::vector<int> VEC_INT;

// Create array of vector
VEC_INT ** ppSubRegion;
VEC_INT * pSubRegion;
ppSubRegion = (VEC_INT **) malloc(SUBPATTERN * sizeof(VEC_INT *));
pSubRegion = (VEC_INT *) malloc(SUBPATTERN * SUBPATTERN * sizeof(VEC_INT));

for (int i = 0; i < SUBPATTERN; i++)
{
ppSubRegion[i] = &pSubRegion[i*SUBPATTERN];
}

// Add numbers into the vector according to some criteria. There is no error!
((VEC_INT) ppSubRegion[nRow][nCol]).push_back(iNumber);

// After display, it shows that the vector contains nothing.
(VEC_INT)(ppSubRegion[r][c]).size() <----- return 0!

Thank you very much

[JohnW@Wessex]

You are not creating any vectors at all with malloc.
Unless you are interfacing with some third party library that requires a parameter created with malloc, there is never any need to use it in C++ code.

[a_sonhye]

My vector is just stl vector.

typedef vector<int> VEC_INT;

[JohnW@Wessex]

Why not this?

typedef vector<int> VEC_INT;

VEC_INT subRegion[SUBPATTERN][SUBPATTERN];

subRegion[nRow][nCol].push_back(iNumber);

size_t size = subRegion[nRow][nCol].size();

or

vector<vector<VEC_INT>> subRegion(SUBPATTERN, vector<VEC_INT>(SUBPATTERN));

subRegion[nRow][nCol].push_back(1);

size_t size = subRegion[nRow][nCol].size();

or some variation thereof?

[a_sonhye]

Hi, JohnW. Thank you very much.

I'm new for STL.
And my previous code was programmed that way but at that time, I use the struct below so the code work fine.
But now, I replace this struct with std::vector so I think i will work the same.

struct TEXTURE_FEATURES
{
double textureFeature[NUM_TEXTURES];
};

Also, I need 3D array not 2D array.
That's why I declare 2D-array and each element of array storing vector.

[JohnW@Wessex]

For explain further...

vector<vector<VEC_INT>>

A vector of vector of vector<int> or 3D vector.


subRegion(SUBPATTERN, vector<VEC_INT>(SUBPATTERN));

Calls the row vector constructor, with SUBPATTERN elements in the row,
each with a value of vector<VEC_INT>(SUBPATTERN)


vector<VEC_INT>(SUBPATTERN)

Constructs a column, with SUBPATTERN elements in the column,
each with a value of VEC_INT (empty int vector).

[a_sonhye]

Thank you very much.

I got it after copying the code and compile it. Now it's more clear to me. But, can I ask one more question? Just for my curious.

From my code at the top most post, how can I assign a vector to my 2D pointer, ppSubRegion[nRow][nCol]???

Thank you very much.

[Lindley]

Well, first off, be absolutely sure that assigning a vector is really what you want to do. It can be an expensive operation. If you need to move a vector from one place to another, it's almost always cheaper to swap it with the one that's already there instead. Of course, if you need two copies of the vector then that won't work.

Depending on which you actually want:

VEC_INT v;

ppSubRegion[nRow][nCol] = v;
ppSubRegion[nRow][nCol].swap(v);

[a_sonhye]

Hello, Lindley.

Thank you. I tried it and it work. Thanks ^^;;