[MPIWG Fortran] Agenda for meeting today (in ~10 mins)

Malcolm Cohen malcolm at nag-j.co.jp
Sat Dec 6 19:02:45 CST 2014

Hi folks,

>3. Deprecate MPI_SIZEOF
>Bill: implementation of storage_size() is pretty trivial
>JeffH: there's also c_sizeof() (Bill says this is F03)

Actually C_SIZEOF  like STORAGE_SIZE was added in F2008; it does not appear 
in F2003.

>Bill: c_sizeof() uses C rules -- e.g., you get size of total array,
>etc. (returns answer in bytes, not bits).

Yes, but it only works for "interoperable" objects... so not on Fortran 
derived types that do not have the BIND attribute.  Some compilers might do 
C_SIZEOF on anything, but it is certainly not in the standard.  So 
STORAGE_SIZE(thingo)/STORAGE_SIZE(byte) would seem to be the recommended 
formulation, multiplied by SIZE(thingo) if it is an array and you want the 
total size of the array.

...........................Malcolm Cohen, NAG Oxford/Tokyo.

More information about the mpiwg-fortran mailing list