[RFC] Document intrinsic modules?

[Walter Spector]

Tobias Burnus <burnus@net-b.de> wrote:
> I was wondering if one should document the intrinsic modules, gfortran
> ships with. Currently, we have: ISO_C_BINDING, ISO_FORTRAN_ENV, OMP_LIB
> and OMP_LIB_KINDS.

At least mention they are available.  The first two in a list of F2003
items supported, the latter are required by OpenMP.

> Expected are ISO_IEEE_* and ISO_VARYING_STRING;

Same here.  As F2003 features are available, folks should be made aware
of them.

> maybe also some modules
> in the spirit of the POSIX FORTRAN 77 language interfaces or of NAG's
> unix* modules.

If gfortran eventually does support the POSIX interfaces, and I hope that
it will, note that there was no standardized name for a F90 module.
The POSIX 1003.9 committee only addressed Fortran-77 compatibility - a
controversial aspect of their work.  But given that there were no working
F90 compilers at the time the POSIX Standard was being worked on, it is
somewhat understandable - if not especially forward-thinking.

Nonetheless, IRIX (and Cray) have had a F90 module called "PXF_DEFINITIONS"
available for many years.  And now that the Intel compiler supports PXF,
they implemented a F90 module called "IFPOSIX".  These modules only contain
INTERFACE blocks.  But are extremely helpful when developing code.

W.