Hi,
I have a problem with my HPF/f95 compiler. The problem appears because
my HPF compiler translates HPF codes into old Fortran 77 with some
Fortran 90 extensions, and my Lahey lf95 compiler usually handles this
mix very well (thanks to backwards compatibility), until today. The
problem is that my HPF compiler passes arrays that are optional
arguments and not PRESENT as a simple number (rank 0 array) that tells
it that the argument was not present.
The code essence of the generated Fortran 90 code is:
MODULE test
CONTAINS
SUBROUTINE option (array)
INTEGER, DIMENSION (2) :: array
END SUBROUTINE option
END MODULE test
PROGRAM optional_args
EXTERNAL option
INTEGER :: value
value=1
CALL option (value)
END PROGRAM optional_args
LF95 complains that there is a shape mistmatch on the argument of option
and aborts compilation. If the subroutine option is not in a module and
does not have an explicit (OK, I got this term now?) interface, LF95
does not complain and the program runs OK (sort of like compiling it
with g77).
I need to fix this behaviour and make LF95 accept such mismatches. What
does the standard say? Anybody have a hint, a switch or something like
this.
Thanks a lot,
Aleksandar
--
_____________________________________________
Aleksandar Donev
Physics Department
Michigan State University
East Lansing, MI 48824-1116
E-mail: [log in to unmask]
Work phone: (517) 432-6770
_____________________________________________
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|