Help assigning memory to structure arrays

I need some help assigning memory to hold an array of structures with variable member arrays. Here is a part of the code

Code:

typedef struct { double *t_stp; int *grid_id; double *lv_x; double *lv_y; double *lv_z; }AA; int idx; static AA *load; if (idx == 0) { load = new AA[idx+1]; }else{ delete [] load; load = new AA[idx+1]; } loadcase[idx].grid_id = new int[nc]; loadcase[idx].lv_x = new double[nc]; loadcase[idx].lv_y = new double[nc]; loadcase[idx].lv_z = new double[nc];

idx and nc are passed in. Depending on the value of idx nc can hold different values. I am sure the way I am assigning memory to structure load is incorrect. But not sure how to increment it based on the value of idx.

Please help!

Have you considered the use of std::vector?

You do not need to use the typedef as you did since this is C++.

I couldn't get vectors to work. I was running into following messages which I didn't understand. If someone can provide the syntax, I am more than willing to switch to vectors

LINK : test.dll not found or not built by the last incremental link; performing full link
mysolver.def : warning LNK4102: export of deleting destructor 'public: virtual void * __ptr64 __cdecl std::bad_alloc::`scalar deleting destructor'(unsigned int)
__ptr64'; image may not run correctly
mysolver.def : warning LNK4102: export of deleting destructor 'public: virtual void * __ptr64 __cdecl std::length_error::`scalar deleting destructor'(unsigned i
nt) __ptr64'; image may not run correctly
mysolver.def : warning LNK4102: export of deleting destructor 'public: virtual void * __ptr64 __cdecl std::logic_error::`scalar deleting destructor'(unsigned in
t) __ptr64'; image may not run correctly
Creating library test.lib and object test.exp

OK. I converted my function to use vectors. Following is how I changed the function

Code:

struct AA { vector<int>grid_id; vector<double>lv_x; vector<double>lv_y; vector<double>lv_z; }; static AA *load; static int idx=1; vector<AA>load(5); for(ii=0;ii<n_loads;ii++) { for(jj=0;jj<ncases;jj++) { load[idx].grid_id.push_back(grids[ii][jj]); load[idx].lv_x.push_back(x_vec[ii][jj]); load[idx].lv_y.push_back(y_vec[ii][jj]); load[idx].lv_z.push_back(z_vec[ii][jj]); } }

The program crashes on the line "load[idx].grid_id.push_back(grids[ii][jj]);" Would appreciate if anyone can point me what I am doing wrong here.

static AA *load;
vector<AA>load(5);

Which load is the right load? Depends on their scope. If they're both in the same scope you'd probably get a compiler error. If the load defined as static AA * is local, then you dereference an uninitialized pointer when you execute

load[idx].grid_id.push_back(grids[ii][jj]);

And what is the point of static int idx = 1; do you know that arrays and vectors start at the zeroth place?

Thanks whiteflags. vector<AA>load(5) is the correct declaration. I fixed those errors and don't get the crash anymore. Now I want to increment the number of structures based on user input and not hard code it to 5. Any idea on what syntax for that would be.

push_back an AA instance on load until the "user input" indicates you should stop.

That's what I thought. But it results in the following error

Code:

load.push_back(AA); test.cxx(139) : error C2275: 'AA' : illegal use of this type as an expression test.cxx(18) : see declaration of 'AA' AA is declared as a structure struct AA { vector<int>grid_id; vector<double>lv_x; vector<double>lv_y; vector<double>lv_z; }; and load is declared as vector<AA>load;

My bad. I had to create an instance of AA. So I declared

AA test;

and then added the line

load.push_back(test);

Is this the correct way to code it. Now if I want to assign values to the members of each array, how do I do that.

load[0].grid_id.push_back(10) doesn't work. It

load[0] would not be where the new element is anyway. If you execute load.push_back(temp); before assigning everything temp needs to have, then the new element is referred to at load.back(); So create a reference to that and have fun doing what you could have done beforehand.

Hi Whiteflags,
Thanks for your continued help. One of the things I noticed while reading more about vectors is the resize method. I have modified to use resize to increase the structure arrays based on user input. This seems to be more straightforward and works as expected. Do you see any drawback of using this method to resize arrays? Following is my updated code segments

Code:

struct AA { vector<int>grid_id; vector<double>lv_x; vector<double>lv_y; vector<double>lv_z; }; static vector<AA>load; int idx; // increase the structure array based on the size of idx load.resize(idx); // initialize members of structure load for the idx structure element of load load[idx].grid_id.assign(ncases,0); load[idx].lv_x.assign(ncases,0.0); load[idx].lv_y.assign(ncases,0.0); load[idx].lv_z.assign(ncases,0.0); for(ii=0;ii<n_loads;ii++) { for(jj=0;jj<ncases;jj++) { load[idx].grid_id.push_back(grids[ii][jj]); load[idx].lv_x.push_back(x_vec[ii][jj]); load[idx].lv_y.push_back(y_vec[ii][jj]); load[idx].lv_z.push_back(z_vec[ii][jj]); } } // load the structure member arrays with values for(ii=0;ii<n_loads;ii++) { for(jj=0;jj<ncases;jj++) { load[idx].grid_id.at(jj) = grids[ii][jj]; load[idx].lv_x.at(jj) = x_vec[ii][jj]; load[idx].lv_y.at(jj) = y_vec[ii][jj]; load[idx].lv_z.at(jj) = z_vec[ii][jj]; } }