Why is the "ALL_CINT_FORTRAN_" macro commented out for the autogenerated code?

[MarcusTL12]

I see they are not commented out in libcint. Is it safe to just uncomment them in the gen-code.cl script? If not, what is the reason for the reduced compatibility between qcint vs libcint?

[MarcusTL12]

Update:
Naively uncommenting in the generation script does not work. This gives a bunch of compilation errors where functions are called with the wrong number of arguments.

It does however work to write short wrapper function manually in C that is doing essentially what the macro should be doing. Here I tested with some gradient 2-electron integrals that were not directly callable from fortran, and they work nicely with optimizer.

#include "cint.h"

int int2e_ip1_sph(double *buf, int *dims, int *shls, int *atm, int natm, int *bas,
              int nbas, double *env, CINTOpt *opt, double *cache);

void int2e_ip1_optimizer(CINTOpt **opt, int *atm, int natm, int *bas, int nbas,
                     double *env);

int int_wrapper_(double *buf, int *shls, int *atm, int *natm, int *bas,
                 int *nbas, double *env, void **opt) {
    return int2e_ip1_sph(buf, NULL, shls, atm, *natm, bas, *nbas, env,
                     *(CINTOpt **)(opt), NULL);
}

void opt_wrapper_(void **opt, int *atm, int *natm, int *bas, int *nbas,
                  double *env) {
    int2e_ip1_optimizer((CINTOpt **)opt, atm, *natm, bas, *nbas, env);
}

void del_opt_wrapper_(void **opt) { CINTdel_optimizer((CINTOpt **)opt); }

So there is seemingly no inherent problem to call from Fortran, just the macro not working. Would it take a lot of work to fix the macro?

[MarcusTL12]

Update 2:

So after comparing the gen-code.cl scripts from libcint and qcint, I noticed that in libcint the call of the macro looks like

"ALL_CINT_FORTRAN_(~a)~%"

while in qcint it looks like

"ALL_CINT_FORTRAN_(c~a)~%"

This prepending of a c to the name seems like an error as it then tries to generate functions with names "ccint2e_...". Removing this seemes to have fixed the problem.

[MarcusTL12]

Created a pull request (#15) with a possible fix.