[MITgcm-devel] [MITgcm-cvs] MITgcm/pkg/mnc CVS Commit
Jean-Michel Campin
jmc at mit.edu
Tue Jan 10 12:50:13 EST 2017
Hi Gael,
I think there is potential problems with this addition of:
CALL ALL_PROC_DIE( myThid )
in S/R MNC_HANDLE_ERR:
a) Since MNC_HANDLE_ERR is called in many places in pkg/mnc, we are not sure
that all procs will hit the same error at the same time (e.g., a proc-specific
file might be missing for one proc but not for all procs). So it seems dangerous
to assume that this will never happen (at least from what I understand of pkg/mnc).
b) you pass the argument "myThid" to this S/R call but most of pkg/mnc runs single
thread. I recommend to check the comments/description within eesupp/src/all_proc_die.F
regarding the argument "myThArg".
So, in case we would ignore the first problem (a), this call would have to be changed to
CALL ALL_PROC_DIE( 0 )
Cheers,
Jean-Michel
On Tue, Jan 10, 2017 at 10:44:07AM -0500, Gael Forget wrote:
> Update of /u/gcmpack/MITgcm/pkg/mnc
> In directory baudelaire:/srv/scratch/gforget/MITgcm_201503/pkg/mnc
>
> Modified Files:
> mnc_utils.F
> Log Message:
>
> - call ALL_PROC_DIE before 'stop', refer to routine name in stop message
>
>
>
> _______________________________________________
> MITgcm-cvs mailing list
> MITgcm-cvs at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-cvs
More information about the MITgcm-devel
mailing list