an 36 ompi/include/mpif-externals.h ! defined with an integer value in mpif-common.h. an 47 ompi/include/mpif-sentinels.h ! (hence, the fact that this is an array of only 1 element does not an 51 ompi/mpi/fortran/use-mpi-f08/bindings/mpi-f-interfaces-bind.h ! expects an array. an 56 ompi/mpi/fortran/use-mpi-f08/bindings/mpi-f-interfaces-bind.h ! to be exactly equivalent to a single INTEGER, an array of mpi_f08 an 57 ompi/mpi/fortran/use-mpi-f08/bindings/mpi-f-interfaces-bind.h ! handles is exactly equivalent to an array of INTEGERS. So passing an 58 ompi/mpi/fortran/use-mpi-f08/bindings/mpi-f-interfaces-bind.h ! an address to the first MPI_VAL is exactly the same as passing an an 62 ompi/mpi/fortran/use-mpi-f08/bindings/mpi-f-interfaces-bind.h ! ompi/mpi/fortran/mpif-h) gets an (MPI_Fint*), and it's all good. an 76 ompi/mpi/fortran/use-mpi-f08/bindings/mpi-f-interfaces-bind.h ! The problem occurs because in the mpi_f08 module, an MPI_Status is an 85 ompi/mpi/fortran/use-mpi-f08/bindings/mpi-f-interfaces-bind.h ! an "interface" block for the PMPI_* subroutine that we want to call. an 87 ompi/mpi/fortran/use-mpi-f08/bindings/mpi-f-interfaces-bind.h ! is TYPE(MPI_Status) (rather than an INTEGER, an 115 ompi/mpi/fortran/use-mpi-f08/bindings/mpi-f-interfaces-bind.h ! The main difference here is that in Fortran, it is an error -- not a an 195 ompi/mpi/fortran/use-mpi-f08/bindings/mpi-f-interfaces-bind.h ! Note that we have an F08-specific C implementation function for an 174 opal/mca/btl/usnic/btl_usnic_map.c opal_process_name_t *an = &(a->proc_opal->proc_name); an 177 opal/mca/btl/usnic/btl_usnic_map.c if (an > bn) { an 179 opal/mca/btl/usnic/btl_usnic_map.c } else if (an < bn) {