[MITgcm-devel] 2 little things

[Ed Hill]

On Fri, 2005-02-04 at 09:35 -0500, Jean-Michel Campin wrote:
> Hi Ed,
> 
> 2 things:
> 
> 1) if I want to do a mpi test, 
> testreport -mpi -of my_optfile 
> is not enough to turn on the mpi switch, I also need to have 
> -DALLOW_USE_MPI -DALWAYS_USE_MPI in "my_optfile"
> 
> So, I was right when I was saying that this was the reason why the 
> pgf77 + mpi test was failling on itrda (from last Nov until end of 
> January)

Hi J-M,

Thanks!  I just fixed the "-mpi" argument passing.


> 2) I try to put check-bounds option with ifc on itrda and
> all the exp. that use mnc fail with this message:
> 
> > forrtl: severe (408): fort: (4): Variable VNAME has substring ending point 3 
> which is greater than the variable length of 2
> >
> > Image              PC        Routine            Line        Source
> > mitgcmuv           085E8488  Unknown               Unknown  Unknown
> >
> > Stack trace terminated abnormally.
> 
> It could be a compiler problem (since ifc + check-bounds work on cg01)
> but it might also mean that there is a small problem in mnc code.
> 
> I will check-in the optfile but, for now, with this "check-bounds" 
> option commented out.  If you want to investigate were this error 
> message comes from ...

I'm calling "shenanigans" on the compiler here.  I think its a compiler
error--that their "check-bounds" option is failing as it tries to handle
string constants being passed as arguments.

The variable "vname" gets used a lot in MNC.  They (the compiler
writers) need to provide me with an actual stack trace (so I can see
which subroutines are supposedly guilty) if they want me to take their
"check-bounds" error messages seriously.  ;-)

Ed

-- 
Edward H. Hill III, PhD
office:  MIT Dept. of EAPS;  Rm 54-1424;  77 Massachusetts Ave.
             Cambridge, MA 02139-4307
emails:  eh3 at mit.edu                ed at eh3.com
URLs:    http://web.mit.edu/eh3/    http://eh3.com/
phone:   617-253-0098
fax:     617-253-4464