[MITgcm-devel] Conversation with Ed about MNC Flags

[Jean-Michel Campin]

Hi Baylor,

Thanks for your comments. I agree that setting the flags for mnc / mdsio 
should be as easy as possible for a user point of view.
But I rather prefer an even simpler "default" behavior
than the one you suggested:

with useMNC=F or pkg/mnc not compiled or useMNC not in data.pkg
 then all output are written with mdsio
with useMNC=T in data.pkg and pkg/mnc successfully compiled, 
 then all output are written with mnc

And I don't seen the need for an useMDSIO flag since:
a) double writing of mdsio & mnc files is only provided 
 for debugging purpose.
b) not writing output that have been selected (with frequency
 and even sometime with file names) just because a user simply forgot 
 to comment out "useMDSIO=F" while MNC is not set properly (either
 not compiled or useMNC=T missing in data.pkg), does not seem 
 logical to me.
 And I insist on setting the flags so that this should not happen.

regarding pickup with mnc, I checked-in (~ a year ago) a small 
script (in contrib/jmc_script) to check that the restart is OK. 
I think it would be nice to have something similar for the mnc pickup.

And my last point is just to remind that mnc flags in pkg/ptracers
are not consistent with the rest of the code, and will have to 
be changed at some point, and the sooner the better:
> Date: Wed, 29 Dec 2004 10:25:44 -0500
> Subject: Re: [MITgcm-devel] .data and .meta grid files
  ...
> I think the priority should be to clean-up the flags that
> already exist (the things I talk to you last week about
> ptracers).
> The reason is that when a logic flag needs to be removed
> from a namelist, it's a pain for thoses who already use it
> (and it's more difficult to stop cleanly when a "retired"
> logical parameter is used in a data file, as opposed to
> an integer or real parameters).

Jean-Michel