[Mitgcm-support] Re: Compilation

[mitgcm-support at dev.mitgcm.org]

Ibrahim Hoteit wrote:
> 
> Hi Patrick,
> 
> I got this error message during the compilation
> 
> > exfa_summary:
> > cat exch_recv_get_vec.F | ../tools/set64bitConst.sh |
> /usr/ccs/lib/cpp -P -DTARGET_SUN -DWORDLENGTH=4 -D_d=E -I. >
> -I../pkg/cal -I../pkg/debug -I../pkg/ecco -I../pkg/exf -I../pkg/flt
> -I../pkg/generic_advdiff -I../pkg/gmredi -I../pkg/kpp > -I../pkg/mdsio
> -I../pkg/mom_fluxform -I../pkg/mom_vecinv -I../pkg/monitor
> -I../pkg/obcs -I../pkg/rw -I../pkg/timeave > -I../eesupp/inc
> -I../model/inc > exch_recv_get_vec.f
> > "./CPP_EEMACROS.h", line 152: macro _d redefines previous macro at
> "command line"
> > "./EESUPPORT.h", line 179: Can't find include file mpif.h
> > "./EESUPPORT.h", line 179: Can't find include file mpif.h
> > make: *** [exch_recv_get_vec.f] Error 2

I assume you're compiling on SunOS, right?
I also guess that in your CPP_EEOPTIONS.h you have the lines
#define ALLOW_USE_MPI
#define  ALWAYS_USE_MPI
(maybe taking Virginie's option file?)

If so, change those lines to
#undef ALLOW_USE_MPI
#undef  ALWAYS_USE_MPI

If my above assumptions are not right, let me know.

> There is also a warning message
> 
> > f77: Warning: illegal option -xtypemap=real:64,double:64,integer:32
> ignored

That's exactly the problematic option I worked on on Mon.
(SunOS 5.7 does not know it, but SunOS 5.8 does)
Just leave it for now and see whether the code runs properly.
If not, try to replace  in bin/Makefile the option
"-xtypemap=real:64,double:64,integer:32" by "-r8 -i4" 
(SunOS 5.8 will create wrong results in some circumstances,
but you work under 5.7, I think).

Let me know the result.
Cheers
Patrick

> Ibrahim
>