https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79330
--- Comment #5 from anlauf at gcc dot gnu.org ---
(In reply to kargls from comment #4)
> (In reply to Jerry DeLisle from comment #3)
> > Still fails on trunk (16)
>
> In looking at the code, I'm wondering if it conforms to the Fortran standard.
>
> module caf_openmp_interface
> implicit none
> character(len=*), parameter :: PREFIX="_gfortran_"
>
> ! Clearly, prefix is declared outside of the interface.
>
> interface
>
> ! Interface is a new namespace and one normally needs to import
> ! symbols from outside the interface. But, here, prefix is not
> ! within the namespace of the subroutine (?) .
>
> module subroutine this_image() bind(C,name=PREFIX//"caf_this_image")
> implicit none
> end subroutine
> end interface
> end module
>
>
> What to other compilers do with the code?
NAG and Intel and Nvidia accept this version (using "module subroutine").
However, for an "ordinary" subroutine they give:
NAG Fortran Compiler Release 7.2(Shin-Urayasu) Build 7200
Error: pr79330.f90, line 8: Implicit type for PREFIX in THIS_IMAGE
detected at NONE@<end-of-statement>
Warning: pr79330.f90, line 10: Symbol PREFIX referenced but never set
and
pr79330.f90(10): error #6054: A CHARACTER data type is required in this
context. [PREFIX]
end subroutine this_image
^
pr79330.f90(10): error #8066: The NAME= specifier in BIND statement or in the
BIND attribute specification of a type declaration statement must be a
character initialization expression.
end subroutine this_image
^
pr79330.f90(7): error #6404: This name does not have a type, and must have an
explicit type. [PREFIX]
subroutine this_image() bind(C,name=PREFIX//"caf_this_image")
-----------------------------------------^
and
NVFORTRAN-S-0146-Expression must be character type (pr79330.f90: 7)
NVFORTRAN-F-1250-The NAME= specifier in BIND statement or in the BIND attribute
specification of a type declaration statement must be a character
initialization expression - prefix. (pr79330.f90: 7)
respectively.
