On Monday 17 March 2008 14:40, Jeff Squyres wrote:

Craig mentioned last week that if we have these opaque Fortran types � (with their contents unspecified), that we could be guaranteed that � their first member would be at the first byte of the instance. If the type has the BIND(C) attribute, which it should, then it has exactly the same layout as a C struct would with the same members in the same order. So whatever C99 guarantees or does not guarantee applies (check the C standard :-). If it is not a BIND(C) datatype then there are no guarantees of any sort.

In C, would you make the C Interface: void MPI_Send(..., struct MPI_Comm comm,...) or do you want void MPI_Send(..., struct MPI_Comm *comm,...) ? Either one of these could be translated into an *equivalent* Fortran interface, where by equivalence I mean full binary compatibility. Aleks