Reevaluating the coordinate dependency between time_slice/ggd and grids_ggd/grid in Equilibrium IDS

[prasad-sawantdesai]

In the equilibrium IDS, there is an assumption that elements in time_slice(:)/ggd(:) must have corresponding coordinate definitions in grids_ggd(:)/grid(:) at the matching time. However, in some of the use cases, during simulation, post-processing, or partial data generation, it is not always feasible or necessary to maintain this strict alignment. What would be the thought process to implement this strategy effectively in DD. Following is the simple example which explains this issue..

#include "ALClasses.h"

using namespace IdsNs;

int main(int argc, char *argv[])
{
  char uri[]="imas:mdsplus?path=./testdb";
  IdsNs::IDS ids;
  double t = 1.0;
  ids.open(uri, FORCE_CREATE_PULSE);

  ids._equilibrium.ids_properties.homogeneous_time = 0;
  
  ids._equilibrium.time_slice.resize(1);

  for (int i=0; i<5; i++) {
    std::cout << "slice " << i << "\n";
    
    ids._equilibrium.time_slice(0).time = t*i;
    ids._equilibrium.time_slice(0).convergence.iterations_n = 1;
    ids._equilibrium.time_slice(0).ggd.resize(1);
    ids._equilibrium.time_slice(0).ggd(0).psi.resize(1);
    ids._equilibrium.time_slice(0).ggd(0).psi(0).grid_index = 0;
    
    std::cout << "Pointer address: " << static_cast<const void*>(ids._equilibrium.grids_ggd.data()) << std::endl;
    if (i%2==0) {
      ids._equilibrium.grids_ggd.resize(1);
      ids._equilibrium.grids_ggd(0).time = t*i;
      ids._equilibrium.grids_ggd(0).grid.resize(1);
      ids._equilibrium.grids_ggd(0).grid(0).identifier.name = "unspecified";
      ids._equilibrium.grids_ggd(0).grid(0).identifier.index = 0;
      ids._equilibrium.grids_ggd(0).grid(0).identifier.description = "unspecified";
    }
    else {
      ids._equilibrium.grids_ggd.resize(0);
      // ids._equilibrium.grids_ggd(0).time = t*i;
      // ids._equilibrium.grids_ggd(0).grid.resize(0);
      // 
      // This causes exception 'time_slice(itime)/ggd(i1)' must have its coordinate in dimension 0 (any of grids_ggd(itime)/grid) filled.")
      }
    }
    ids._equilibrium.putSlice();
  }

  ids.close();
  return 0;
}

[imbeauf]

As discussed yesterday, your code snippet is correct, it's a bug in the validation routine.